CHANGES revision 142425
1 2 OpenSSL CHANGES 3 _______________ 4 5 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 6 7 *) Avoid a race condition when CRLs are checked in a multi threaded 8 environment. This would happen due to the reordering of the revoked 9 entries during signature checking and serial number lookup. Now the 10 encoding is cached and the serial number sort performed under a lock. 11 Add new STACK function sk_is_sorted(). 12 [Steve Henson] 13 14 *) Add Delta CRL to the extension code. 15 [Steve Henson] 16 17 *) Various fixes to s3_pkt.c so alerts are sent properly. 18 [David Holmes <d.holmes@f5.com>] 19 20 *) Reduce the chances of duplicate issuer name and serial numbers (in 21 violation of RFC3280) using the OpenSSL certificate creation utilities. 22 This is done by creating a random 64 bit value for the initial serial 23 number when a serial number file is created or when a self signed 24 certificate is created using 'openssl req -x509'. The initial serial 25 number file is created using 'openssl x509 -next_serial' in CA.pl 26 rather than being initialized to 1. 27 [Steve Henson] 28 29 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 30 31 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 32 by using the Codenomicon TLS Test Tool (CAN-2004-0079) 33 [Joe Orton, Steve Henson] 34 35 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 36 (CAN-2004-0112) 37 [Joe Orton, Steve Henson] 38 39 *) Make it possible to have multiple active certificates with the same 40 subject in the CA index file. This is done only if the keyword 41 'unique_subject' is set to 'no' in the main CA section (default 42 if 'CA_default') of the configuration file. The value is saved 43 with the database itself in a separate index attribute file, 44 named like the index file with '.attr' appended to the name. 45 [Richard Levitte] 46 47 *) X509 verify fixes. Disable broken certificate workarounds when 48 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 49 keyUsage extension present. Don't accept CRLs with unhandled critical 50 extensions: since verify currently doesn't process CRL extensions this 51 rejects a CRL with *any* critical extensions. Add new verify error codes 52 for these cases. 53 [Steve Henson] 54 55 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 56 A clarification of RFC2560 will require the use of OCTET STRINGs and 57 some implementations cannot handle the current raw format. Since OpenSSL 58 copies and compares OCSP nonces as opaque blobs without any attempt at 59 parsing them this should not create any compatibility issues. 60 [Steve Henson] 61 62 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 63 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 64 this HMAC (and other) operations are several times slower than OpenSSL 65 < 0.9.7. 66 [Steve Henson] 67 68 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 69 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 70 71 *) Use the correct content when signing type "other". 72 [Steve Henson] 73 74 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 75 76 *) Fix various bugs revealed by running the NISCC test suite: 77 78 Stop out of bounds reads in the ASN1 code when presented with 79 invalid tags (CAN-2003-0543 and CAN-2003-0544). 80 81 Free up ASN1_TYPE correctly if ANY type is invalid (CAN-2003-0545). 82 83 If verify callback ignores invalid public key errors don't try to check 84 certificate signature with the NULL public key. 85 86 [Steve Henson] 87 88 *) New -ignore_err option in ocsp application to stop the server 89 exiting on the first error in a request. 90 [Steve Henson] 91 92 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 93 if the server requested one: as stated in TLS 1.0 and SSL 3.0 94 specifications. 95 [Steve Henson] 96 97 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 98 extra data after the compression methods not only for TLS 1.0 99 but also for SSL 3.0 (as required by the specification). 100 [Bodo Moeller; problem pointed out by Matthias Loepfe] 101 102 *) Change X509_certificate_type() to mark the key as exported/exportable 103 when it's 512 *bits* long, not 512 bytes. 104 [Richard Levitte] 105 106 *) Change AES_cbc_encrypt() so it outputs exact multiple of 107 blocks during encryption. 108 [Richard Levitte] 109 110 *) Various fixes to base64 BIO and non blocking I/O. On write 111 flushes were not handled properly if the BIO retried. On read 112 data was not being buffered properly and had various logic bugs. 113 This also affects blocking I/O when the data being decoded is a 114 certain size. 115 [Steve Henson] 116 117 *) Various S/MIME bugfixes and compatibility changes: 118 output correct application/pkcs7 MIME type if 119 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 120 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 121 of files as .eml work). Correctly handle very long lines in MIME 122 parser. 123 [Steve Henson] 124 125 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 126 127 *) Countermeasure against the Klima-Pokorny-Rosa extension of 128 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 129 a protocol version number mismatch like a decryption error 130 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 131 [Bodo Moeller] 132 133 *) Turn on RSA blinding by default in the default implementation 134 to avoid a timing attack. Applications that don't want it can call 135 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 136 They would be ill-advised to do so in most cases. 137 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 138 139 *) Change RSA blinding code so that it works when the PRNG is not 140 seeded (in this case, the secret RSA exponent is abused as 141 an unpredictable seed -- if it is not unpredictable, there 142 is no point in blinding anyway). Make RSA blinding thread-safe 143 by remembering the creator's thread ID in rsa->blinding and 144 having all other threads use local one-time blinding factors 145 (this requires more computation than sharing rsa->blinding, but 146 avoids excessive locking; and if an RSA object is not shared 147 between threads, blinding will still be very fast). 148 [Bodo Moeller] 149 150 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 151 ENGINE as defaults for all supported algorithms irrespective of 152 the 'flags' parameter. 'flags' is now honoured, so applications 153 should make sure they are passing it correctly. 154 [Geoff Thorpe] 155 156 *) Target "mingw" now allows native Windows code to be generated in 157 the Cygwin environment as well as with the MinGW compiler. 158 [Ulf Moeller] 159 160 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 161 162 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 163 via timing by performing a MAC computation even if incorrrect 164 block cipher padding has been found. This is a countermeasure 165 against active attacks where the attacker has to distinguish 166 between bad padding and a MAC verification error. (CAN-2003-0078) 167 168 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 169 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 170 Martin Vuagnoux (EPFL, Ilion)] 171 172 *) Make the no-err option work as intended. The intention with no-err 173 is not to have the whole error stack handling routines removed from 174 libcrypto, it's only intended to remove all the function name and 175 reason texts, thereby removing some of the footprint that may not 176 be interesting if those errors aren't displayed anyway. 177 178 NOTE: it's still possible for any application or module to have it's 179 own set of error texts inserted. The routines are there, just not 180 used by default when no-err is given. 181 [Richard Levitte] 182 183 *) Add support for FreeBSD on IA64. 184 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 185 186 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 187 Kerberos function mit_des_cbc_cksum(). Before this change, 188 the value returned by DES_cbc_cksum() was like the one from 189 mit_des_cbc_cksum(), except the bytes were swapped. 190 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 191 192 *) Allow an application to disable the automatic SSL chain building. 193 Before this a rather primitive chain build was always performed in 194 ssl3_output_cert_chain(): an application had no way to send the 195 correct chain if the automatic operation produced an incorrect result. 196 197 Now the chain builder is disabled if either: 198 199 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 200 201 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 202 203 The reasoning behind this is that an application would not want the 204 auto chain building to take place if extra chain certificates are 205 present and it might also want a means of sending no additional 206 certificates (for example the chain has two certificates and the 207 root is omitted). 208 [Steve Henson] 209 210 *) Add the possibility to build without the ENGINE framework. 211 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 212 213 *) Under Win32 gmtime() can return NULL: check return value in 214 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 215 [Steve Henson] 216 217 *) DSA routines: under certain error conditions uninitialized BN objects 218 could be freed. Solution: make sure initialization is performed early 219 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 220 Nils Larsch <nla@trustcenter.de> via PR#459) 221 [Lutz Jaenicke] 222 223 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 224 checked on reconnect on the client side, therefore session resumption 225 could still fail with a "ssl session id is different" error. This 226 behaviour is masked when SSL_OP_ALL is used due to 227 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 228 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 229 followup to PR #377. 230 [Lutz Jaenicke] 231 232 *) IA-32 assembler support enhancements: unified ELF targets, support 233 for SCO/Caldera platforms, fix for Cygwin shared build. 234 [Andy Polyakov] 235 236 *) Add support for FreeBSD on sparc64. As a consequence, support for 237 FreeBSD on non-x86 processors is separate from x86 processors on 238 the config script, much like the NetBSD support. 239 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 240 241 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 242 243 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 244 OpenSSL 0.9.7.] 245 246 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 247 code (06) was taken as the first octet of the session ID and the last 248 octet was ignored consequently. As a result SSLv2 client side session 249 caching could not have worked due to the session ID mismatch between 250 client and server. 251 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 252 PR #377. 253 [Lutz Jaenicke] 254 255 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 256 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 257 removed entirely. 258 [Richard Levitte] 259 260 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 261 seems that in spite of existing for more than a year, many application 262 author have done nothing to provide the necessary callbacks, which 263 means that this particular engine will not work properly anywhere. 264 This is a very unfortunate situation which forces us, in the name 265 of usability, to give the hw_ncipher.c a static lock, which is part 266 of libcrypto. 267 NOTE: This is for the 0.9.7 series ONLY. This hack will never 268 appear in 0.9.8 or later. We EXPECT application authors to have 269 dealt properly with this when 0.9.8 is released (unless we actually 270 make such changes in the libcrypto locking code that changes will 271 have to be made anyway). 272 [Richard Levitte] 273 274 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 275 octets have been read, EOF or an error occurs. Without this change 276 some truncated ASN1 structures will not produce an error. 277 [Steve Henson] 278 279 *) Disable Heimdal support, since it hasn't been fully implemented. 280 Still give the possibility to force the use of Heimdal, but with 281 warnings and a request that patches get sent to openssl-dev. 282 [Richard Levitte] 283 284 *) Add the VC-CE target, introduce the WINCE sysname, and add 285 INSTALL.WCE and appropriate conditionals to make it build. 286 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 287 288 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 289 cygssl-x.y.z.dll, where x, y and z are the major, minor and 290 edit numbers of the version. 291 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 292 293 *) Introduce safe string copy and catenation functions 294 (BUF_strlcpy() and BUF_strlcat()). 295 [Ben Laurie (CHATS) and Richard Levitte] 296 297 *) Avoid using fixed-size buffers for one-line DNs. 298 [Ben Laurie (CHATS)] 299 300 *) Add BUF_MEM_grow_clean() to avoid information leakage when 301 resizing buffers containing secrets, and use where appropriate. 302 [Ben Laurie (CHATS)] 303 304 *) Avoid using fixed size buffers for configuration file location. 305 [Ben Laurie (CHATS)] 306 307 *) Avoid filename truncation for various CA files. 308 [Ben Laurie (CHATS)] 309 310 *) Use sizeof in preference to magic numbers. 311 [Ben Laurie (CHATS)] 312 313 *) Avoid filename truncation in cert requests. 314 [Ben Laurie (CHATS)] 315 316 *) Add assertions to check for (supposedly impossible) buffer 317 overflows. 318 [Ben Laurie (CHATS)] 319 320 *) Don't cache truncated DNS entries in the local cache (this could 321 potentially lead to a spoofing attack). 322 [Ben Laurie (CHATS)] 323 324 *) Fix various buffers to be large enough for hex/decimal 325 representations in a platform independent manner. 326 [Ben Laurie (CHATS)] 327 328 *) Add CRYPTO_realloc_clean() to avoid information leakage when 329 resizing buffers containing secrets, and use where appropriate. 330 [Ben Laurie (CHATS)] 331 332 *) Add BIO_indent() to avoid much slightly worrying code to do 333 indents. 334 [Ben Laurie (CHATS)] 335 336 *) Convert sprintf()/BIO_puts() to BIO_printf(). 337 [Ben Laurie (CHATS)] 338 339 *) buffer_gets() could terminate with the buffer only half 340 full. Fixed. 341 [Ben Laurie (CHATS)] 342 343 *) Add assertions to prevent user-supplied crypto functions from 344 overflowing internal buffers by having large block sizes, etc. 345 [Ben Laurie (CHATS)] 346 347 *) New OPENSSL_assert() macro (similar to assert(), but enabled 348 unconditionally). 349 [Ben Laurie (CHATS)] 350 351 *) Eliminate unused copy of key in RC4. 352 [Ben Laurie (CHATS)] 353 354 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 355 [Ben Laurie (CHATS)] 356 357 *) Fix off-by-one error in EGD path. 358 [Ben Laurie (CHATS)] 359 360 *) If RANDFILE path is too long, ignore instead of truncating. 361 [Ben Laurie (CHATS)] 362 363 *) Eliminate unused and incorrectly sized X.509 structure 364 CBCParameter. 365 [Ben Laurie (CHATS)] 366 367 *) Eliminate unused and dangerous function knumber(). 368 [Ben Laurie (CHATS)] 369 370 *) Eliminate unused and dangerous structure, KSSL_ERR. 371 [Ben Laurie (CHATS)] 372 373 *) Protect against overlong session ID context length in an encoded 374 session object. Since these are local, this does not appear to be 375 exploitable. 376 [Ben Laurie (CHATS)] 377 378 *) Change from security patch (see 0.9.6e below) that did not affect 379 the 0.9.6 release series: 380 381 Remote buffer overflow in SSL3 protocol - an attacker could 382 supply an oversized master key in Kerberos-enabled versions. 383 (CAN-2002-0657) 384 [Ben Laurie (CHATS)] 385 386 *) Change the SSL kerb5 codes to match RFC 2712. 387 [Richard Levitte] 388 389 *) Make -nameopt work fully for req and add -reqopt switch. 390 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 391 392 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 393 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 394 395 *) Make sure tests can be performed even if the corresponding algorithms 396 have been removed entirely. This was also the last step to make 397 OpenSSL compilable with DJGPP under all reasonable conditions. 398 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 399 400 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 401 to allow version independent disabling of normally unselected ciphers, 402 which may be activated as a side-effect of selecting a single cipher. 403 404 (E.g., cipher list string "RSA" enables ciphersuites that are left 405 out of "ALL" because they do not provide symmetric encryption. 406 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 407 [Lutz Jaenicke, Bodo Moeller] 408 409 *) Add appropriate support for separate platform-dependent build 410 directories. The recommended way to make a platform-dependent 411 build directory is the following (tested on Linux), maybe with 412 some local tweaks: 413 414 # Place yourself outside of the OpenSSL source tree. In 415 # this example, the environment variable OPENSSL_SOURCE 416 # is assumed to contain the absolute OpenSSL source directory. 417 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 418 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 419 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 420 mkdir -p `dirname $F` 421 ln -s $OPENSSL_SOURCE/$F $F 422 done 423 424 To be absolutely sure not to disturb the source tree, a "make clean" 425 is a good thing. If it isn't successfull, don't worry about it, 426 it probably means the source directory is very clean. 427 [Richard Levitte] 428 429 *) Make sure any ENGINE control commands make local copies of string 430 pointers passed to them whenever necessary. Otherwise it is possible 431 the caller may have overwritten (or deallocated) the original string 432 data when a later ENGINE operation tries to use the stored values. 433 [G�tz Babin-Ebell <babinebell@trustcenter.de>] 434 435 *) Improve diagnostics in file reading and command-line digests. 436 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 437 438 *) Add AES modes CFB and OFB to the object database. Correct an 439 error in AES-CFB decryption. 440 [Richard Levitte] 441 442 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 443 allows existing EVP_CIPHER_CTX structures to be reused after 444 calling EVP_*Final(). This behaviour is used by encryption 445 BIOs and some applications. This has the side effect that 446 applications must explicitly clean up cipher contexts with 447 EVP_CIPHER_CTX_cleanup() or they will leak memory. 448 [Steve Henson] 449 450 *) Check the values of dna and dnb in bn_mul_recursive before calling 451 bn_mul_comba (a non zero value means the a or b arrays do not contain 452 n2 elements) and fallback to bn_mul_normal if either is not zero. 453 [Steve Henson] 454 455 *) Fix escaping of non-ASCII characters when using the -subj option 456 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 457 [Lutz Jaenicke] 458 459 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 460 form for "surname", serialNumber has no short form. 461 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 462 therefore remove "mail" short name for "internet 7". 463 The OID for unique identifiers in X509 certificates is 464 x500UniqueIdentifier, not uniqueIdentifier. 465 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 466 [Lutz Jaenicke] 467 468 *) Add an "init" command to the ENGINE config module and auto initialize 469 ENGINEs. Without any "init" command the ENGINE will be initialized 470 after all ctrl commands have been executed on it. If init=1 the 471 ENGINE is initailized at that point (ctrls before that point are run 472 on the uninitialized ENGINE and after on the initialized one). If 473 init=0 then the ENGINE will not be iniatialized at all. 474 [Steve Henson] 475 476 *) Fix the 'app_verify_callback' interface so that the user-defined 477 argument is actually passed to the callback: In the 478 SSL_CTX_set_cert_verify_callback() prototype, the callback 479 declaration has been changed from 480 int (*cb)() 481 into 482 int (*cb)(X509_STORE_CTX *,void *); 483 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 484 i=s->ctx->app_verify_callback(&ctx) 485 has been changed into 486 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 487 488 To update applications using SSL_CTX_set_cert_verify_callback(), 489 a dummy argument can be added to their callback functions. 490 [D. K. Smetters <smetters@parc.xerox.com>] 491 492 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 493 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 494 495 *) Add and OPENSSL_LOAD_CONF define which will cause 496 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 497 This allows older applications to transparently support certain 498 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 499 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 500 load the config file and OPENSSL_add_all_algorithms_conf() which will 501 always load it have also been added. 502 [Steve Henson] 503 504 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 505 Adjust NIDs and EVP layer. 506 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 507 508 *) Config modules support in openssl utility. 509 510 Most commands now load modules from the config file, 511 though in a few (such as version) this isn't done 512 because it couldn't be used for anything. 513 514 In the case of ca and req the config file used is 515 the same as the utility itself: that is the -config 516 command line option can be used to specify an 517 alternative file. 518 [Steve Henson] 519 520 *) Move default behaviour from OPENSSL_config(). If appname is NULL 521 use "openssl_conf" if filename is NULL use default openssl config file. 522 [Steve Henson] 523 524 *) Add an argument to OPENSSL_config() to allow the use of an alternative 525 config section name. Add a new flag to tolerate a missing config file 526 and move code to CONF_modules_load_file(). 527 [Steve Henson] 528 529 *) Support for crypto accelerator cards from Accelerated Encryption 530 Processing, www.aep.ie. (Use engine 'aep') 531 The support was copied from 0.9.6c [engine] and adapted/corrected 532 to work with the new engine framework. 533 [AEP Inc. and Richard Levitte] 534 535 *) Support for SureWare crypto accelerator cards from Baltimore 536 Technologies. (Use engine 'sureware') 537 The support was copied from 0.9.6c [engine] and adapted 538 to work with the new engine framework. 539 [Richard Levitte] 540 541 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 542 make the newer ENGINE framework commands for the CHIL engine work. 543 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 544 545 *) Make it possible to produce shared libraries on ReliantUNIX. 546 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 547 548 *) Add the configuration target debug-linux-ppro. 549 Make 'openssl rsa' use the general key loading routines 550 implemented in apps.c, and make those routines able to 551 handle the key format FORMAT_NETSCAPE and the variant 552 FORMAT_IISSGC. 553 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 554 555 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 556 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 557 558 *) Add -keyform to rsautl, and document -engine. 559 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 560 561 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 562 BIO_R_NO_SUCH_FILE error code rather than the generic 563 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 564 [Ben Laurie] 565 566 *) Add new functions 567 ERR_peek_last_error 568 ERR_peek_last_error_line 569 ERR_peek_last_error_line_data. 570 These are similar to 571 ERR_peek_error 572 ERR_peek_error_line 573 ERR_peek_error_line_data, 574 but report on the latest error recorded rather than the first one 575 still in the error queue. 576 [Ben Laurie, Bodo Moeller] 577 578 *) default_algorithms option in ENGINE config module. This allows things 579 like: 580 default_algorithms = ALL 581 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 582 [Steve Henson] 583 584 *) Prelminary ENGINE config module. 585 [Steve Henson] 586 587 *) New experimental application configuration code. 588 [Steve Henson] 589 590 *) Change the AES code to follow the same name structure as all other 591 symmetric ciphers, and behave the same way. Move everything to 592 the directory crypto/aes, thereby obsoleting crypto/rijndael. 593 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 594 595 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 596 [Ben Laurie and Theo de Raadt] 597 598 *) Add option to output public keys in req command. 599 [Massimiliano Pala madwolf@openca.org] 600 601 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 602 (up to about 10% better than before for P-192 and P-224). 603 [Bodo Moeller] 604 605 *) New functions/macros 606 607 SSL_CTX_set_msg_callback(ctx, cb) 608 SSL_CTX_set_msg_callback_arg(ctx, arg) 609 SSL_set_msg_callback(ssl, cb) 610 SSL_set_msg_callback_arg(ssl, arg) 611 612 to request calling a callback function 613 614 void cb(int write_p, int version, int content_type, 615 const void *buf, size_t len, SSL *ssl, void *arg) 616 617 whenever a protocol message has been completely received 618 (write_p == 0) or sent (write_p == 1). Here 'version' is the 619 protocol version according to which the SSL library interprets 620 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 621 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 622 the content type as defined in the SSL 3.0/TLS 1.0 protocol 623 specification (change_cipher_spec(20), alert(21), handshake(22)). 624 'buf' and 'len' point to the actual message, 'ssl' to the 625 SSL object, and 'arg' is the application-defined value set by 626 SSL[_CTX]_set_msg_callback_arg(). 627 628 'openssl s_client' and 'openssl s_server' have new '-msg' options 629 to enable a callback that displays all protocol messages. 630 [Bodo Moeller] 631 632 *) Change the shared library support so shared libraries are built as 633 soon as the corresponding static library is finished, and thereby get 634 openssl and the test programs linked against the shared library. 635 This still only happens when the keyword "shard" has been given to 636 the configuration scripts. 637 638 NOTE: shared library support is still an experimental thing, and 639 backward binary compatibility is still not guaranteed. 640 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 641 642 *) Add support for Subject Information Access extension. 643 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 644 645 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 646 additional bytes when new memory had to be allocated, not just 647 when reusing an existing buffer. 648 [Bodo Moeller] 649 650 *) New command line and configuration option 'utf8' for the req command. 651 This allows field values to be specified as UTF8 strings. 652 [Steve Henson] 653 654 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 655 runs for the former and machine-readable output for the latter. 656 [Ben Laurie] 657 658 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 659 of the e-mail address in the DN (i.e., it will go into a certificate 660 extension only). The new configuration file option 'email_in_dn = no' 661 has the same effect. 662 [Massimiliano Pala madwolf@openca.org] 663 664 *) Change all functions with names starting with des_ to be starting 665 with DES_ instead. Add wrappers that are compatible with libdes, 666 but are named _ossl_old_des_*. Finally, add macros that map the 667 des_* symbols to the corresponding _ossl_old_des_* if libdes 668 compatibility is desired. If OpenSSL 0.9.6c compatibility is 669 desired, the des_* symbols will be mapped to DES_*, with one 670 exception. 671 672 Since we provide two compatibility mappings, the user needs to 673 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 674 compatibility is desired. The default (i.e., when that macro 675 isn't defined) is OpenSSL 0.9.6c compatibility. 676 677 There are also macros that enable and disable the support of old 678 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 679 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 680 are defined, the default will apply: to support the old des routines. 681 682 In either case, one must include openssl/des.h to get the correct 683 definitions. Do not try to just include openssl/des_old.h, that 684 won't work. 685 686 NOTE: This is a major break of an old API into a new one. Software 687 authors are encouraged to switch to the DES_ style functions. Some 688 time in the future, des_old.h and the libdes compatibility functions 689 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 690 default), and then completely removed. 691 [Richard Levitte] 692 693 *) Test for certificates which contain unsupported critical extensions. 694 If such a certificate is found during a verify operation it is 695 rejected by default: this behaviour can be overridden by either 696 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 697 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 698 X509_supported_extension() has also been added which returns 1 if a 699 particular extension is supported. 700 [Steve Henson] 701 702 *) Modify the behaviour of EVP cipher functions in similar way to digests 703 to retain compatibility with existing code. 704 [Steve Henson] 705 706 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 707 compatibility with existing code. In particular the 'ctx' parameter does 708 not have to be to be initialized before the call to EVP_DigestInit() and 709 it is tidied up after a call to EVP_DigestFinal(). New function 710 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 711 EVP_MD_CTX_copy() changed to not require the destination to be 712 initialized valid and new function EVP_MD_CTX_copy_ex() added which 713 requires the destination to be valid. 714 715 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 716 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 717 [Steve Henson] 718 719 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 720 so that complete 'Handshake' protocol structures are kept in memory 721 instead of overwriting 'msg_type' and 'length' with 'body' data. 722 [Bodo Moeller] 723 724 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 725 [Massimo Santin via Richard Levitte] 726 727 *) Major restructuring to the underlying ENGINE code. This includes 728 reduction of linker bloat, separation of pure "ENGINE" manipulation 729 (initialisation, etc) from functionality dealing with implementations 730 of specific crypto iterfaces. This change also introduces integrated 731 support for symmetric ciphers and digest implementations - so ENGINEs 732 can now accelerate these by providing EVP_CIPHER and EVP_MD 733 implementations of their own. This is detailed in crypto/engine/README 734 as it couldn't be adequately described here. However, there are a few 735 API changes worth noting - some RSA, DSA, DH, and RAND functions that 736 were changed in the original introduction of ENGINE code have now 737 reverted back - the hooking from this code to ENGINE is now a good 738 deal more passive and at run-time, operations deal directly with 739 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 740 dereferencing through an ENGINE pointer any more. Also, the ENGINE 741 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 742 they were not being used by the framework as there is no concept of a 743 BIGNUM_METHOD and they could not be generalised to the new 744 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 745 ENGINE_cpy() has been removed as it cannot be consistently defined in 746 the new code. 747 [Geoff Thorpe] 748 749 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 750 [Steve Henson] 751 752 *) Change mkdef.pl to sort symbols that get the same entry number, 753 and make sure the automatically generated functions ERR_load_* 754 become part of libeay.num as well. 755 [Richard Levitte] 756 757 *) New function SSL_renegotiate_pending(). This returns true once 758 renegotiation has been requested (either SSL_renegotiate() call 759 or HelloRequest/ClientHello receveived from the peer) and becomes 760 false once a handshake has been completed. 761 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 762 sends a HelloRequest, but does not ensure that a handshake takes 763 place. SSL_renegotiate_pending() is useful for checking if the 764 client has followed the request.) 765 [Bodo Moeller] 766 767 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 768 By default, clients may request session resumption even during 769 renegotiation (if session ID contexts permit); with this option, 770 session resumption is possible only in the first handshake. 771 772 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 773 more bits available for options that should not be part of 774 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 775 [Bodo Moeller] 776 777 *) Add some demos for certificate and certificate request creation. 778 [Steve Henson] 779 780 *) Make maximum certificate chain size accepted from the peer application 781 settable (SSL*_get/set_max_cert_list()), as proposed by 782 "Douglas E. Engert" <deengert@anl.gov>. 783 [Lutz Jaenicke] 784 785 *) Add support for shared libraries for Unixware-7 786 (Boyd Lynn Gerber <gerberb@zenez.com>). 787 [Lutz Jaenicke] 788 789 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 790 be done prior to destruction. Use this to unload error strings from 791 ENGINEs that load their own error strings. NB: This adds two new API 792 functions to "get" and "set" this destroy handler in an ENGINE. 793 [Geoff Thorpe] 794 795 *) Alter all existing ENGINE implementations (except "openssl" and 796 "openbsd") to dynamically instantiate their own error strings. This 797 makes them more flexible to be built both as statically-linked ENGINEs 798 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 799 Also, add stub code to each that makes building them as self-contained 800 shared-libraries easier (see README.ENGINE). 801 [Geoff Thorpe] 802 803 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 804 implementations into applications that are completely implemented in 805 self-contained shared-libraries. The "dynamic" ENGINE exposes control 806 commands that can be used to configure what shared-library to load and 807 to control aspects of the way it is handled. Also, made an update to 808 the README.ENGINE file that brings its information up-to-date and 809 provides some information and instructions on the "dynamic" ENGINE 810 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 811 [Geoff Thorpe] 812 813 *) Make it possible to unload ranges of ERR strings with a new 814 "ERR_unload_strings" function. 815 [Geoff Thorpe] 816 817 *) Add a copy() function to EVP_MD. 818 [Ben Laurie] 819 820 *) Make EVP_MD routines take a context pointer instead of just the 821 md_data void pointer. 822 [Ben Laurie] 823 824 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 825 that the digest can only process a single chunk of data 826 (typically because it is provided by a piece of 827 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 828 is only going to provide a single chunk of data, and hence the 829 framework needn't accumulate the data for oneshot drivers. 830 [Ben Laurie] 831 832 *) As with "ERR", make it possible to replace the underlying "ex_data" 833 functions. This change also alters the storage and management of global 834 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 835 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 836 index counters. The API functions that use this state have been changed 837 to take a "class_index" rather than pointers to the class's local STACK 838 and counter, and there is now an API function to dynamically create new 839 classes. This centralisation allows us to (a) plug a lot of the 840 thread-safety problems that existed, and (b) makes it possible to clean 841 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 842 such data would previously have always leaked in application code and 843 workarounds were in place to make the memory debugging turn a blind eye 844 to it. Application code that doesn't use this new function will still 845 leak as before, but their memory debugging output will announce it now 846 rather than letting it slide. 847 848 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 849 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 850 has a return value to indicate success or failure. 851 [Geoff Thorpe] 852 853 *) Make it possible to replace the underlying "ERR" functions such that the 854 global state (2 LHASH tables and 2 locks) is only used by the "default" 855 implementation. This change also adds two functions to "get" and "set" 856 the implementation prior to it being automatically set the first time 857 any other ERR function takes place. Ie. an application can call "get", 858 pass the return value to a module it has just loaded, and that module 859 can call its own "set" function using that value. This means the 860 module's "ERR" operations will use (and modify) the error state in the 861 application and not in its own statically linked copy of OpenSSL code. 862 [Geoff Thorpe] 863 864 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 865 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 866 the operation, and provides a more encapsulated way for external code 867 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 868 to use these functions rather than manually incrementing the counts. 869 870 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 871 [Geoff Thorpe] 872 873 *) Add EVP test program. 874 [Ben Laurie] 875 876 *) Add symmetric cipher support to ENGINE. Expect the API to change! 877 [Ben Laurie] 878 879 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 880 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 881 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 882 These allow a CRL to be built without having to access X509_CRL fields 883 directly. Modify 'ca' application to use new functions. 884 [Steve Henson] 885 886 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 887 bug workarounds. Rollback attack detection is a security feature. 888 The problem will only arise on OpenSSL servers when TLSv1 is not 889 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 890 Software authors not wanting to support TLSv1 will have special reasons 891 for their choice and can explicitly enable this option. 892 [Bodo Moeller, Lutz Jaenicke] 893 894 *) Rationalise EVP so it can be extended: don't include a union of 895 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 896 (similar to those existing for EVP_CIPHER_CTX). 897 Usage example: 898 899 EVP_MD_CTX md; 900 901 EVP_MD_CTX_init(&md); /* new function call */ 902 EVP_DigestInit(&md, EVP_sha1()); 903 EVP_DigestUpdate(&md, in, len); 904 EVP_DigestFinal(&md, out, NULL); 905 EVP_MD_CTX_cleanup(&md); /* new function call */ 906 907 [Ben Laurie] 908 909 *) Make DES key schedule conform to the usual scheme, as well as 910 correcting its structure. This means that calls to DES functions 911 now have to pass a pointer to a des_key_schedule instead of a 912 plain des_key_schedule (which was actually always a pointer 913 anyway): E.g., 914 915 des_key_schedule ks; 916 917 des_set_key_checked(..., &ks); 918 des_ncbc_encrypt(..., &ks, ...); 919 920 (Note that a later change renames 'des_...' into 'DES_...'.) 921 [Ben Laurie] 922 923 *) Initial reduction of linker bloat: the use of some functions, such as 924 PEM causes large amounts of unused functions to be linked in due to 925 poor organisation. For example pem_all.c contains every PEM function 926 which has a knock on effect of linking in large amounts of (unused) 927 ASN1 code. Grouping together similar functions and splitting unrelated 928 functions prevents this. 929 [Steve Henson] 930 931 *) Cleanup of EVP macros. 932 [Ben Laurie] 933 934 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 935 correct _ecb suffix. 936 [Ben Laurie] 937 938 *) Add initial OCSP responder support to ocsp application. The 939 revocation information is handled using the text based index 940 use by the ca application. The responder can either handle 941 requests generated internally, supplied in files (for example 942 via a CGI script) or using an internal minimal server. 943 [Steve Henson] 944 945 *) Add configuration choices to get zlib compression for TLS. 946 [Richard Levitte] 947 948 *) Changes to Kerberos SSL for RFC 2712 compliance: 949 1. Implemented real KerberosWrapper, instead of just using 950 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 951 2. Implemented optional authenticator field of KerberosWrapper. 952 953 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 954 and authenticator structs; see crypto/krb5/. 955 956 Generalized Kerberos calls to support multiple Kerberos libraries. 957 [Vern Staats <staatsvr@asc.hpc.mil>, 958 Jeffrey Altman <jaltman@columbia.edu> 959 via Richard Levitte] 960 961 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 962 already does with RSA. testdsa.h now has 'priv_key/pub_key' 963 values for each of the key sizes rather than having just 964 parameters (and 'speed' generating keys each time). 965 [Geoff Thorpe] 966 967 *) Speed up EVP routines. 968 Before: 969encrypt 970type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 971des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 972des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 973des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 974decrypt 975des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 976des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 977des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 978 After: 979encrypt 980des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 981decrypt 982des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 983 [Ben Laurie] 984 985 *) Added the OS2-EMX target. 986 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 987 988 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 989 to support NCONF routines in extension code. New function CONF_set_nconf() 990 to allow functions which take an NCONF to also handle the old LHASH 991 structure: this means that the old CONF compatible routines can be 992 retained (in particular wrt extensions) without having to duplicate the 993 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 994 [Steve Henson] 995 996 *) Enhance the general user interface with mechanisms for inner control 997 and with possibilities to have yes/no kind of prompts. 998 [Richard Levitte] 999 1000 *) Change all calls to low level digest routines in the library and 1001 applications to use EVP. Add missing calls to HMAC_cleanup() and 1002 don't assume HMAC_CTX can be copied using memcpy(). 1003 [Verdon Walker <VWalker@novell.com>, Steve Henson] 1004 1005 *) Add the possibility to control engines through control names but with 1006 arbitrary arguments instead of just a string. 1007 Change the key loaders to take a UI_METHOD instead of a callback 1008 function pointer. NOTE: this breaks binary compatibility with earlier 1009 versions of OpenSSL [engine]. 1010 Adapt the nCipher code for these new conditions and add a card insertion 1011 callback. 1012 [Richard Levitte] 1013 1014 *) Enhance the general user interface with mechanisms to better support 1015 dialog box interfaces, application-defined prompts, the possibility 1016 to use defaults (for example default passwords from somewhere else) 1017 and interrupts/cancellations. 1018 [Richard Levitte] 1019 1020 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 1021 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 1022 [Steve Henson] 1023 1024 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 1025 tidy up some unnecessarily weird code in 'sk_new()'). 1026 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 1027 1028 *) Change the key loading routines for ENGINEs to use the same kind 1029 callback (pem_password_cb) as all other routines that need this 1030 kind of callback. 1031 [Richard Levitte] 1032 1033 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 1034 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 1035 than this minimum value is recommended. 1036 [Lutz Jaenicke] 1037 1038 *) New random seeder for OpenVMS, using the system process statistics 1039 that are easily reachable. 1040 [Richard Levitte] 1041 1042 *) Windows apparently can't transparently handle global 1043 variables defined in DLLs. Initialisations such as: 1044 1045 const ASN1_ITEM *it = &ASN1_INTEGER_it; 1046 1047 wont compile. This is used by the any applications that need to 1048 declare their own ASN1 modules. This was fixed by adding the option 1049 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 1050 needed for static libraries under Win32. 1051 [Steve Henson] 1052 1053 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 1054 setting of purpose and trust fields. New X509_STORE trust and 1055 purpose functions and tidy up setting in other SSL functions. 1056 [Steve Henson] 1057 1058 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 1059 structure. These are inherited by X509_STORE_CTX when it is 1060 initialised. This allows various defaults to be set in the 1061 X509_STORE structure (such as flags for CRL checking and custom 1062 purpose or trust settings) for functions which only use X509_STORE_CTX 1063 internally such as S/MIME. 1064 1065 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 1066 trust settings if they are not set in X509_STORE. This allows X509_STORE 1067 purposes and trust (in S/MIME for example) to override any set by default. 1068 1069 Add command line options for CRL checking to smime, s_client and s_server 1070 applications. 1071 [Steve Henson] 1072 1073 *) Initial CRL based revocation checking. If the CRL checking flag(s) 1074 are set then the CRL is looked up in the X509_STORE structure and 1075 its validity and signature checked, then if the certificate is found 1076 in the CRL the verify fails with a revoked error. 1077 1078 Various new CRL related callbacks added to X509_STORE_CTX structure. 1079 1080 Command line options added to 'verify' application to support this. 1081 1082 This needs some additional work, such as being able to handle multiple 1083 CRLs with different times, extension based lookup (rather than just 1084 by subject name) and ultimately more complete V2 CRL extension 1085 handling. 1086 [Steve Henson] 1087 1088 *) Add a general user interface API (crypto/ui/). This is designed 1089 to replace things like des_read_password and friends (backward 1090 compatibility functions using this new API are provided). 1091 The purpose is to remove prompting functions from the DES code 1092 section as well as provide for prompting through dialog boxes in 1093 a window system and the like. 1094 [Richard Levitte] 1095 1096 *) Add "ex_data" support to ENGINE so implementations can add state at a 1097 per-structure level rather than having to store it globally. 1098 [Geoff] 1099 1100 *) Make it possible for ENGINE structures to be copied when retrieved by 1101 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 1102 This causes the "original" ENGINE structure to act like a template, 1103 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 1104 operational state can be localised to each ENGINE structure, despite the 1105 fact they all share the same "methods". New ENGINE structures returned in 1106 this case have no functional references and the return value is the single 1107 structural reference. This matches the single structural reference returned 1108 by ENGINE_by_id() normally, when it is incremented on the pre-existing 1109 ENGINE structure. 1110 [Geoff] 1111 1112 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 1113 needs to match any other type at all we need to manually clear the 1114 tag cache. 1115 [Steve Henson] 1116 1117 *) Changes to the "openssl engine" utility to include; 1118 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 1119 about an ENGINE's available control commands. 1120 - executing control commands from command line arguments using the 1121 '-pre' and '-post' switches. '-post' is only used if '-t' is 1122 specified and the ENGINE is successfully initialised. The syntax for 1123 the individual commands are colon-separated, for example; 1124 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 1125 [Geoff] 1126 1127 *) New dynamic control command support for ENGINEs. ENGINEs can now 1128 declare their own commands (numbers), names (strings), descriptions, 1129 and input types for run-time discovery by calling applications. A 1130 subset of these commands are implicitly classed as "executable" 1131 depending on their input type, and only these can be invoked through 1132 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 1133 can be based on user input, config files, etc). The distinction is 1134 that "executable" commands cannot return anything other than a boolean 1135 result and can only support numeric or string input, whereas some 1136 discoverable commands may only be for direct use through 1137 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 1138 pointers, or other custom uses. The "executable" commands are to 1139 support parameterisations of ENGINE behaviour that can be 1140 unambiguously defined by ENGINEs and used consistently across any 1141 OpenSSL-based application. Commands have been added to all the 1142 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 1143 control over shared-library paths without source code alterations. 1144 [Geoff] 1145 1146 *) Changed all ENGINE implementations to dynamically allocate their 1147 ENGINEs rather than declaring them statically. Apart from this being 1148 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 1149 this also allows the implementations to compile without using the 1150 internal engine_int.h header. 1151 [Geoff] 1152 1153 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 1154 'const' value. Any code that should be able to modify a RAND_METHOD 1155 should already have non-const pointers to it (ie. they should only 1156 modify their own ones). 1157 [Geoff] 1158 1159 *) Made a variety of little tweaks to the ENGINE code. 1160 - "atalla" and "ubsec" string definitions were moved from header files 1161 to C code. "nuron" string definitions were placed in variables 1162 rather than hard-coded - allowing parameterisation of these values 1163 later on via ctrl() commands. 1164 - Removed unused "#if 0"'d code. 1165 - Fixed engine list iteration code so it uses ENGINE_free() to release 1166 structural references. 1167 - Constified the RAND_METHOD element of ENGINE structures. 1168 - Constified various get/set functions as appropriate and added 1169 missing functions (including a catch-all ENGINE_cpy that duplicates 1170 all ENGINE values onto a new ENGINE except reference counts/state). 1171 - Removed NULL parameter checks in get/set functions. Setting a method 1172 or function to NULL is a way of cancelling out a previously set 1173 value. Passing a NULL ENGINE parameter is just plain stupid anyway 1174 and doesn't justify the extra error symbols and code. 1175 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 1176 flags from engine_int.h to engine.h. 1177 - Changed prototypes for ENGINE handler functions (init(), finish(), 1178 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 1179 [Geoff] 1180 1181 *) Implement binary inversion algorithm for BN_mod_inverse in addition 1182 to the algorithm using long division. The binary algorithm can be 1183 used only if the modulus is odd. On 32-bit systems, it is faster 1184 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 1185 roughly 5-15% for 256-bit moduli), so we use it only for moduli 1186 up to 450 bits. In 64-bit environments, the binary algorithm 1187 appears to be advantageous for much longer moduli; here we use it 1188 for moduli up to 2048 bits. 1189 [Bodo Moeller] 1190 1191 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 1192 could not support the combine flag in choice fields. 1193 [Steve Henson] 1194 1195 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 1196 extensions from a certificate request to the certificate. 1197 [Steve Henson] 1198 1199 *) Allow multiple 'certopt' and 'nameopt' options to be separated 1200 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 1201 file: this allows the display of the certificate about to be 1202 signed to be customised, to allow certain fields to be included 1203 or excluded and extension details. The old system didn't display 1204 multicharacter strings properly, omitted fields not in the policy 1205 and couldn't display additional details such as extensions. 1206 [Steve Henson] 1207 1208 *) Function EC_POINTs_mul for multiple scalar multiplication 1209 of an arbitrary number of elliptic curve points 1210 \sum scalars[i]*points[i], 1211 optionally including the generator defined for the EC_GROUP: 1212 scalar*generator + \sum scalars[i]*points[i]. 1213 1214 EC_POINT_mul is a simple wrapper function for the typical case 1215 that the point list has just one item (besides the optional 1216 generator). 1217 [Bodo Moeller] 1218 1219 *) First EC_METHODs for curves over GF(p): 1220 1221 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 1222 operations and provides various method functions that can also 1223 operate with faster implementations of modular arithmetic. 1224 1225 EC_GFp_mont_method() reuses most functions that are part of 1226 EC_GFp_simple_method, but uses Montgomery arithmetic. 1227 1228 [Bodo Moeller; point addition and point doubling 1229 implementation directly derived from source code provided by 1230 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 1231 1232 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 1233 crypto/ec/ec_lib.c): 1234 1235 Curves are EC_GROUP objects (with an optional group generator) 1236 based on EC_METHODs that are built into the library. 1237 1238 Points are EC_POINT objects based on EC_GROUP objects. 1239 1240 Most of the framework would be able to handle curves over arbitrary 1241 finite fields, but as there are no obvious types for fields other 1242 than GF(p), some functions are limited to that for now. 1243 [Bodo Moeller] 1244 1245 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 1246 that the file contains a complete HTTP response. 1247 [Richard Levitte] 1248 1249 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 1250 change the def and num file printf format specifier from "%-40sXXX" 1251 to "%-39s XXX". The latter will always guarantee a space after the 1252 field while the former will cause them to run together if the field 1253 is 40 of more characters long. 1254 [Steve Henson] 1255 1256 *) Constify the cipher and digest 'method' functions and structures 1257 and modify related functions to take constant EVP_MD and EVP_CIPHER 1258 pointers. 1259 [Steve Henson] 1260 1261 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 1262 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 1263 [Bodo Moeller] 1264 1265 *) Modify EVP_Digest*() routines so they now return values. Although the 1266 internal software routines can never fail additional hardware versions 1267 might. 1268 [Steve Henson] 1269 1270 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 1271 1272 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 1273 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 1274 1275 ASN1 error codes 1276 ERR_R_NESTED_ASN1_ERROR 1277 ... 1278 ERR_R_MISSING_ASN1_EOS 1279 were 4 .. 9, conflicting with 1280 ERR_LIB_RSA (= ERR_R_RSA_LIB) 1281 ... 1282 ERR_LIB_PEM (= ERR_R_PEM_LIB). 1283 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 1284 1285 Add new error code 'ERR_R_INTERNAL_ERROR'. 1286 [Bodo Moeller] 1287 1288 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 1289 suffices. 1290 [Bodo Moeller] 1291 1292 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 1293 sets the subject name for a new request or supersedes the 1294 subject name in a given request. Formats that can be parsed are 1295 'CN=Some Name, OU=myOU, C=IT' 1296 and 1297 'CN=Some Name/OU=myOU/C=IT'. 1298 1299 Add options '-batch' and '-verbose' to 'openssl req'. 1300 [Massimiliano Pala <madwolf@hackmasters.net>] 1301 1302 *) Introduce the possibility to access global variables through 1303 functions on platform were that's the best way to handle exporting 1304 global variables in shared libraries. To enable this functionality, 1305 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 1306 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 1307 is normally done by Configure or something similar). 1308 1309 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 1310 in the source file (foo.c) like this: 1311 1312 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 1313 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 1314 1315 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 1316 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 1317 1318 OPENSSL_DECLARE_GLOBAL(int,foo); 1319 #define foo OPENSSL_GLOBAL_REF(foo) 1320 OPENSSL_DECLARE_GLOBAL(double,bar); 1321 #define bar OPENSSL_GLOBAL_REF(bar) 1322 1323 The #defines are very important, and therefore so is including the 1324 header file everywhere where the defined globals are used. 1325 1326 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 1327 of ASN.1 items, but that structure is a bit different. 1328 1329 The largest change is in util/mkdef.pl which has been enhanced with 1330 better and easier to understand logic to choose which symbols should 1331 go into the Windows .def files as well as a number of fixes and code 1332 cleanup (among others, algorithm keywords are now sorted 1333 lexicographically to avoid constant rewrites). 1334 [Richard Levitte] 1335 1336 *) In BN_div() keep a copy of the sign of 'num' before writing the 1337 result to 'rm' because if rm==num the value will be overwritten 1338 and produce the wrong result if 'num' is negative: this caused 1339 problems with BN_mod() and BN_nnmod(). 1340 [Steve Henson] 1341 1342 *) Function OCSP_request_verify(). This checks the signature on an 1343 OCSP request and verifies the signer certificate. The signer 1344 certificate is just checked for a generic purpose and OCSP request 1345 trust settings. 1346 [Steve Henson] 1347 1348 *) Add OCSP_check_validity() function to check the validity of OCSP 1349 responses. OCSP responses are prepared in real time and may only 1350 be a few seconds old. Simply checking that the current time lies 1351 between thisUpdate and nextUpdate max reject otherwise valid responses 1352 caused by either OCSP responder or client clock inaccuracy. Instead 1353 we allow thisUpdate and nextUpdate to fall within a certain period of 1354 the current time. The age of the response can also optionally be 1355 checked. Two new options -validity_period and -status_age added to 1356 ocsp utility. 1357 [Steve Henson] 1358 1359 *) If signature or public key algorithm is unrecognized print out its 1360 OID rather that just UNKNOWN. 1361 [Steve Henson] 1362 1363 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 1364 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 1365 ID to be generated from the issuer certificate alone which can then be 1366 passed to OCSP_id_issuer_cmp(). 1367 [Steve Henson] 1368 1369 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 1370 ASN1 modules to export functions returning ASN1_ITEM pointers 1371 instead of the ASN1_ITEM structures themselves. This adds several 1372 new macros which allow the underlying ASN1 function/structure to 1373 be accessed transparently. As a result code should not use ASN1_ITEM 1374 references directly (such as &X509_it) but instead use the relevant 1375 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 1376 use of the new ASN1 code on platforms where exporting structures 1377 is problematical (for example in shared libraries) but exporting 1378 functions returning pointers to structures is not. 1379 [Steve Henson] 1380 1381 *) Add support for overriding the generation of SSL/TLS session IDs. 1382 These callbacks can be registered either in an SSL_CTX or per SSL. 1383 The purpose of this is to allow applications to control, if they wish, 1384 the arbitrary values chosen for use as session IDs, particularly as it 1385 can be useful for session caching in multiple-server environments. A 1386 command-line switch for testing this (and any client code that wishes 1387 to use such a feature) has been added to "s_server". 1388 [Geoff Thorpe, Lutz Jaenicke] 1389 1390 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 1391 of the form '#if defined(...) || defined(...) || ...' and 1392 '#if !defined(...) && !defined(...) && ...'. This also avoids 1393 the growing number of special cases it was previously handling. 1394 [Richard Levitte] 1395 1396 *) Make all configuration macros available for application by making 1397 sure they are available in opensslconf.h, by giving them names starting 1398 with "OPENSSL_" to avoid conflicts with other packages and by making 1399 sure e_os2.h will cover all platform-specific cases together with 1400 opensslconf.h. 1401 Additionally, it is now possible to define configuration/platform- 1402 specific names (called "system identities"). In the C code, these 1403 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 1404 macro with the name beginning with "OPENSSL_SYS_", which is determined 1405 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 1406 what is available. 1407 [Richard Levitte] 1408 1409 *) New option -set_serial to 'req' and 'x509' this allows the serial 1410 number to use to be specified on the command line. Previously self 1411 signed certificates were hard coded with serial number 0 and the 1412 CA options of 'x509' had to use a serial number in a file which was 1413 auto incremented. 1414 [Steve Henson] 1415 1416 *) New options to 'ca' utility to support V2 CRL entry extensions. 1417 Currently CRL reason, invalidity date and hold instruction are 1418 supported. Add new CRL extensions to V3 code and some new objects. 1419 [Steve Henson] 1420 1421 *) New function EVP_CIPHER_CTX_set_padding() this is used to 1422 disable standard block padding (aka PKCS#5 padding) in the EVP 1423 API, which was previously mandatory. This means that the data is 1424 not padded in any way and so the total length much be a multiple 1425 of the block size, otherwise an error occurs. 1426 [Steve Henson] 1427 1428 *) Initial (incomplete) OCSP SSL support. 1429 [Steve Henson] 1430 1431 *) New function OCSP_parse_url(). This splits up a URL into its host, 1432 port and path components: primarily to parse OCSP URLs. New -url 1433 option to ocsp utility. 1434 [Steve Henson] 1435 1436 *) New nonce behavior. The return value of OCSP_check_nonce() now 1437 reflects the various checks performed. Applications can decide 1438 whether to tolerate certain situations such as an absent nonce 1439 in a response when one was present in a request: the ocsp application 1440 just prints out a warning. New function OCSP_add1_basic_nonce() 1441 this is to allow responders to include a nonce in a response even if 1442 the request is nonce-less. 1443 [Steve Henson] 1444 1445 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 1446 skipped when using openssl x509 multiple times on a single input file, 1447 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 1448 [Bodo Moeller] 1449 1450 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 1451 set string type: to handle setting ASN1_TIME structures. Fix ca 1452 utility to correctly initialize revocation date of CRLs. 1453 [Steve Henson] 1454 1455 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 1456 the clients preferred ciphersuites and rather use its own preferences. 1457 Should help to work around M$ SGC (Server Gated Cryptography) bug in 1458 Internet Explorer by ensuring unchanged hash method during stepup. 1459 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 1460 [Lutz Jaenicke] 1461 1462 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 1463 to aes and add a new 'exist' option to print out symbols that don't 1464 appear to exist. 1465 [Steve Henson] 1466 1467 *) Additional options to ocsp utility to allow flags to be set and 1468 additional certificates supplied. 1469 [Steve Henson] 1470 1471 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 1472 OCSP client a number of certificate to only verify the response 1473 signature against. 1474 [Richard Levitte] 1475 1476 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 1477 handle the new API. Currently only ECB, CBC modes supported. Add new 1478 AES OIDs. 1479 1480 Add TLS AES ciphersuites as described in RFC3268, "Advanced 1481 Encryption Standard (AES) Ciphersuites for Transport Layer 1482 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 1483 not enabled by default and were not part of the "ALL" ciphersuite 1484 alias because they were not yet official; they could be 1485 explicitly requested by specifying the "AESdraft" ciphersuite 1486 group alias. In the final release of OpenSSL 0.9.7, the group 1487 alias is called "AES" and is part of "ALL".) 1488 [Ben Laurie, Steve Henson, Bodo Moeller] 1489 1490 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 1491 request to response. 1492 [Steve Henson] 1493 1494 *) Functions for OCSP responders. OCSP_request_onereq_count(), 1495 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 1496 extract information from a certificate request. OCSP_response_create() 1497 creates a response and optionally adds a basic response structure. 1498 OCSP_basic_add1_status() adds a complete single response to a basic 1499 response and returns the OCSP_SINGLERESP structure just added (to allow 1500 extensions to be included for example). OCSP_basic_add1_cert() adds a 1501 certificate to a basic response and OCSP_basic_sign() signs a basic 1502 response with various flags. New helper functions ASN1_TIME_check() 1503 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 1504 (converts ASN1_TIME to GeneralizedTime). 1505 [Steve Henson] 1506 1507 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 1508 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 1509 structure from a certificate. X509_pubkey_digest() digests the public_key 1510 contents: this is used in various key identifiers. 1511 [Steve Henson] 1512 1513 *) Make sk_sort() tolerate a NULL argument. 1514 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 1515 1516 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 1517 passed by the function are trusted implicitly. If any of them signed the 1518 response then it is assumed to be valid and is not verified. 1519 [Steve Henson] 1520 1521 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 1522 to data. This was previously part of the PKCS7 ASN1 code. This 1523 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 1524 [Steve Henson, reported by Kenneth R. Robinette 1525 <support@securenetterm.com>] 1526 1527 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 1528 routines: without these tracing memory leaks is very painful. 1529 Fix leaks in PKCS12 and PKCS7 routines. 1530 [Steve Henson] 1531 1532 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 1533 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 1534 effectively meant GeneralizedTime would never be used. Now it 1535 is initialised to -1 but X509_time_adj() now has to check the value 1536 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 1537 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 1538 [Steve Henson, reported by Kenneth R. Robinette 1539 <support@securenetterm.com>] 1540 1541 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 1542 result in a zero length in the ASN1_INTEGER structure which was 1543 not consistent with the structure when d2i_ASN1_INTEGER() was used 1544 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 1545 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 1546 where it did not print out a minus for negative ASN1_INTEGER. 1547 [Steve Henson] 1548 1549 *) Add summary printout to ocsp utility. The various functions which 1550 convert status values to strings have been renamed to: 1551 OCSP_response_status_str(), OCSP_cert_status_str() and 1552 OCSP_crl_reason_str() and are no longer static. New options 1553 to verify nonce values and to disable verification. OCSP response 1554 printout format cleaned up. 1555 [Steve Henson] 1556 1557 *) Add additional OCSP certificate checks. These are those specified 1558 in RFC2560. This consists of two separate checks: the CA of the 1559 certificate being checked must either be the OCSP signer certificate 1560 or the issuer of the OCSP signer certificate. In the latter case the 1561 OCSP signer certificate must contain the OCSP signing extended key 1562 usage. This check is performed by attempting to match the OCSP 1563 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 1564 in the OCSP_CERTID structures of the response. 1565 [Steve Henson] 1566 1567 *) Initial OCSP certificate verification added to OCSP_basic_verify() 1568 and related routines. This uses the standard OpenSSL certificate 1569 verify routines to perform initial checks (just CA validity) and 1570 to obtain the certificate chain. Then additional checks will be 1571 performed on the chain. Currently the root CA is checked to see 1572 if it is explicitly trusted for OCSP signing. This is used to set 1573 a root CA as a global signing root: that is any certificate that 1574 chains to that CA is an acceptable OCSP signing certificate. 1575 [Steve Henson] 1576 1577 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 1578 extensions from a separate configuration file. 1579 As when reading extensions from the main configuration file, 1580 the '-extensions ...' option may be used for specifying the 1581 section to use. 1582 [Massimiliano Pala <madwolf@comune.modena.it>] 1583 1584 *) New OCSP utility. Allows OCSP requests to be generated or 1585 read. The request can be sent to a responder and the output 1586 parsed, outputed or printed in text form. Not complete yet: 1587 still needs to check the OCSP response validity. 1588 [Steve Henson] 1589 1590 *) New subcommands for 'openssl ca': 1591 'openssl ca -status <serial>' prints the status of the cert with 1592 the given serial number (according to the index file). 1593 'openssl ca -updatedb' updates the expiry status of certificates 1594 in the index file. 1595 [Massimiliano Pala <madwolf@comune.modena.it>] 1596 1597 *) New '-newreq-nodes' command option to CA.pl. This is like 1598 '-newreq', but calls 'openssl req' with the '-nodes' option 1599 so that the resulting key is not encrypted. 1600 [Damien Miller <djm@mindrot.org>] 1601 1602 *) New configuration for the GNU Hurd. 1603 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 1604 1605 *) Initial code to implement OCSP basic response verify. This 1606 is currently incomplete. Currently just finds the signer's 1607 certificate and verifies the signature on the response. 1608 [Steve Henson] 1609 1610 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 1611 value of OPENSSLDIR. This is available via the new '-d' option 1612 to 'openssl version', and is also included in 'openssl version -a'. 1613 [Bodo Moeller] 1614 1615 *) Allowing defining memory allocation callbacks that will be given 1616 file name and line number information in additional arguments 1617 (a const char* and an int). The basic functionality remains, as 1618 well as the original possibility to just replace malloc(), 1619 realloc() and free() by functions that do not know about these 1620 additional arguments. To register and find out the current 1621 settings for extended allocation functions, the following 1622 functions are provided: 1623 1624 CRYPTO_set_mem_ex_functions 1625 CRYPTO_set_locked_mem_ex_functions 1626 CRYPTO_get_mem_ex_functions 1627 CRYPTO_get_locked_mem_ex_functions 1628 1629 These work the same way as CRYPTO_set_mem_functions and friends. 1630 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 1631 extended allocation function is enabled. 1632 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 1633 a conventional allocation function is enabled. 1634 [Richard Levitte, Bodo Moeller] 1635 1636 *) Finish off removing the remaining LHASH function pointer casts. 1637 There should no longer be any prototype-casting required when using 1638 the LHASH abstraction, and any casts that remain are "bugs". See 1639 the callback types and macros at the head of lhash.h for details 1640 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 1641 [Geoff Thorpe] 1642 1643 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 1644 If /dev/[u]random devices are not available or do not return enough 1645 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 1646 be queried. 1647 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 1648 /etc/entropy will be queried once each in this sequence, quering stops 1649 when enough entropy was collected without querying more sockets. 1650 [Lutz Jaenicke] 1651 1652 *) Change the Unix RAND_poll() variant to be able to poll several 1653 random devices, as specified by DEVRANDOM, until a sufficient amount 1654 of data has been collected. We spend at most 10 ms on each file 1655 (select timeout) and read in non-blocking mode. DEVRANDOM now 1656 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 1657 (previously it was just the string "/dev/urandom"), so on typical 1658 platforms the 10 ms delay will never occur. 1659 Also separate out the Unix variant to its own file, rand_unix.c. 1660 For VMS, there's a currently-empty rand_vms.c. 1661 [Richard Levitte] 1662 1663 *) Move OCSP client related routines to ocsp_cl.c. These 1664 provide utility functions which an application needing 1665 to issue a request to an OCSP responder and analyse the 1666 response will typically need: as opposed to those which an 1667 OCSP responder itself would need which will be added later. 1668 1669 OCSP_request_sign() signs an OCSP request with an API similar 1670 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 1671 response. OCSP_response_get1_basic() extracts basic response 1672 from response. OCSP_resp_find_status(): finds and extracts status 1673 information from an OCSP_CERTID structure (which will be created 1674 when the request structure is built). These are built from lower 1675 level functions which work on OCSP_SINGLERESP structures but 1676 wont normally be used unless the application wishes to examine 1677 extensions in the OCSP response for example. 1678 1679 Replace nonce routines with a pair of functions. 1680 OCSP_request_add1_nonce() adds a nonce value and optionally 1681 generates a random value. OCSP_check_nonce() checks the 1682 validity of the nonce in an OCSP response. 1683 [Steve Henson] 1684 1685 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 1686 This doesn't copy the supplied OCSP_CERTID and avoids the 1687 need to free up the newly created id. Change return type 1688 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 1689 This can then be used to add extensions to the request. 1690 Deleted OCSP_request_new(), since most of its functionality 1691 is now in OCSP_REQUEST_new() (and the case insensitive name 1692 clash) apart from the ability to set the request name which 1693 will be added elsewhere. 1694 [Steve Henson] 1695 1696 *) Update OCSP API. Remove obsolete extensions argument from 1697 various functions. Extensions are now handled using the new 1698 OCSP extension code. New simple OCSP HTTP function which 1699 can be used to send requests and parse the response. 1700 [Steve Henson] 1701 1702 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 1703 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 1704 uses the special reorder version of SET OF to sort the attributes 1705 and reorder them to match the encoded order. This resolves a long 1706 standing problem: a verify on a PKCS7 structure just after signing 1707 it used to fail because the attribute order did not match the 1708 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 1709 it uses the received order. This is necessary to tolerate some broken 1710 software that does not order SET OF. This is handled by encoding 1711 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 1712 to produce the required SET OF. 1713 [Steve Henson] 1714 1715 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 1716 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 1717 files to get correct declarations of the ASN.1 item variables. 1718 [Richard Levitte] 1719 1720 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 1721 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 1722 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 1723 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 1724 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 1725 ASN1_ITEM and no wrapper functions. 1726 [Steve Henson] 1727 1728 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 1729 replace the old function pointer based I/O routines. Change most of 1730 the *_d2i_bio() and *_d2i_fp() functions to use these. 1731 [Steve Henson] 1732 1733 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 1734 lines, recognice more "algorithms" that can be deselected, and make 1735 it complain about algorithm deselection that isn't recognised. 1736 [Richard Levitte] 1737 1738 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 1739 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 1740 to use new functions. Add NO_ASN1_OLD which can be set to remove 1741 some old style ASN1 functions: this can be used to determine if old 1742 code will still work when these eventually go away. 1743 [Steve Henson] 1744 1745 *) New extension functions for OCSP structures, these follow the 1746 same conventions as certificates and CRLs. 1747 [Steve Henson] 1748 1749 *) New function X509V3_add1_i2d(). This automatically encodes and 1750 adds an extension. Its behaviour can be customised with various 1751 flags to append, replace or delete. Various wrappers added for 1752 certifcates and CRLs. 1753 [Steve Henson] 1754 1755 *) Fix to avoid calling the underlying ASN1 print routine when 1756 an extension cannot be parsed. Correct a typo in the 1757 OCSP_SERVICELOC extension. Tidy up print OCSP format. 1758 [Steve Henson] 1759 1760 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 1761 entries for variables. 1762 [Steve Henson] 1763 1764 *) Add functionality to apps/openssl.c for detecting locking 1765 problems: As the program is single-threaded, all we have 1766 to do is register a locking callback using an array for 1767 storing which locks are currently held by the program. 1768 [Bodo Moeller] 1769 1770 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 1771 SSL_get_ex_data_X509_STORE_idx(), which is used in 1772 ssl_verify_cert_chain() and thus can be called at any time 1773 during TLS/SSL handshakes so that thread-safety is essential. 1774 Unfortunately, the ex_data design is not at all suited 1775 for multi-threaded use, so it probably should be abolished. 1776 [Bodo Moeller] 1777 1778 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 1779 [Broadcom, tweaked and integrated by Geoff Thorpe] 1780 1781 *) Move common extension printing code to new function 1782 X509V3_print_extensions(). Reorganise OCSP print routines and 1783 implement some needed OCSP ASN1 functions. Add OCSP extensions. 1784 [Steve Henson] 1785 1786 *) New function X509_signature_print() to remove duplication in some 1787 print routines. 1788 [Steve Henson] 1789 1790 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 1791 set (this was treated exactly the same as SET OF previously). This 1792 is used to reorder the STACK representing the structure to match the 1793 encoding. This will be used to get round a problem where a PKCS7 1794 structure which was signed could not be verified because the STACK 1795 order did not reflect the encoded order. 1796 [Steve Henson] 1797 1798 *) Reimplement the OCSP ASN1 module using the new code. 1799 [Steve Henson] 1800 1801 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 1802 for its ASN1 operations. The old style function pointers still exist 1803 for now but they will eventually go away. 1804 [Steve Henson] 1805 1806 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 1807 completely replaces the old ASN1 functionality with a table driven 1808 encoder and decoder which interprets an ASN1_ITEM structure describing 1809 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 1810 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 1811 has also been converted to the new form. 1812 [Steve Henson] 1813 1814 *) Change BN_mod_exp_recp so that negative moduli are tolerated 1815 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 1816 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 1817 for negative moduli. 1818 [Bodo Moeller] 1819 1820 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 1821 of not touching the result's sign bit. 1822 [Bodo Moeller] 1823 1824 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 1825 set. 1826 [Bodo Moeller] 1827 1828 *) Changed the LHASH code to use prototypes for callbacks, and created 1829 macros to declare and implement thin (optionally static) functions 1830 that provide type-safety and avoid function pointer casting for the 1831 type-specific callbacks. 1832 [Geoff Thorpe] 1833 1834 *) Added Kerberos Cipher Suites to be used with TLS, as written in 1835 RFC 2712. 1836 [Veers Staats <staatsvr@asc.hpc.mil>, 1837 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 1838 1839 *) Reformat the FAQ so the different questions and answers can be divided 1840 in sections depending on the subject. 1841 [Richard Levitte] 1842 1843 *) Have the zlib compression code load ZLIB.DLL dynamically under 1844 Windows. 1845 [Richard Levitte] 1846 1847 *) New function BN_mod_sqrt for computing square roots modulo a prime 1848 (using the probabilistic Tonelli-Shanks algorithm unless 1849 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 1850 be handled deterministically). 1851 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 1852 1853 *) Make BN_mod_inverse faster by explicitly handling small quotients 1854 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 1855 512 bits], about 30% for larger ones [1024 or 2048 bits].) 1856 [Bodo Moeller] 1857 1858 *) New function BN_kronecker. 1859 [Bodo Moeller] 1860 1861 *) Fix BN_gcd so that it works on negative inputs; the result is 1862 positive unless both parameters are zero. 1863 Previously something reasonably close to an infinite loop was 1864 possible because numbers could be growing instead of shrinking 1865 in the implementation of Euclid's algorithm. 1866 [Bodo Moeller] 1867 1868 *) Fix BN_is_word() and BN_is_one() macros to take into account the 1869 sign of the number in question. 1870 1871 Fix BN_is_word(a,w) to work correctly for w == 0. 1872 1873 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 1874 because its test if the absolute value of 'a' equals 'w'. 1875 Note that BN_abs_is_word does *not* handle w == 0 reliably; 1876 it exists mostly for use in the implementations of BN_is_zero(), 1877 BN_is_one(), and BN_is_word(). 1878 [Bodo Moeller] 1879 1880 *) New function BN_swap. 1881 [Bodo Moeller] 1882 1883 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 1884 the exponentiation functions are more likely to produce reasonable 1885 results on negative inputs. 1886 [Bodo Moeller] 1887 1888 *) Change BN_mod_mul so that the result is always non-negative. 1889 Previously, it could be negative if one of the factors was negative; 1890 I don't think anyone really wanted that behaviour. 1891 [Bodo Moeller] 1892 1893 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 1894 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 1895 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 1896 and add new functions: 1897 1898 BN_nnmod 1899 BN_mod_sqr 1900 BN_mod_add 1901 BN_mod_add_quick 1902 BN_mod_sub 1903 BN_mod_sub_quick 1904 BN_mod_lshift1 1905 BN_mod_lshift1_quick 1906 BN_mod_lshift 1907 BN_mod_lshift_quick 1908 1909 These functions always generate non-negative results. 1910 1911 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 1912 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 1913 1914 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 1915 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 1916 be reduced modulo m. 1917 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 1918 1919#if 0 1920 The following entry accidentily appeared in the CHANGES file 1921 distributed with OpenSSL 0.9.7. The modifications described in 1922 it do *not* apply to OpenSSL 0.9.7. 1923 1924 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 1925 was actually never needed) and in BN_mul(). The removal in BN_mul() 1926 required a small change in bn_mul_part_recursive() and the addition 1927 of the functions bn_cmp_part_words(), bn_sub_part_words() and 1928 bn_add_part_words(), which do the same thing as bn_cmp_words(), 1929 bn_sub_words() and bn_add_words() except they take arrays with 1930 differing sizes. 1931 [Richard Levitte] 1932#endif 1933 1934 *) In 'openssl passwd', verify passwords read from the terminal 1935 unless the '-salt' option is used (which usually means that 1936 verification would just waste user's time since the resulting 1937 hash is going to be compared with some given password hash) 1938 or the new '-noverify' option is used. 1939 1940 This is an incompatible change, but it does not affect 1941 non-interactive use of 'openssl passwd' (passwords on the command 1942 line, '-stdin' option, '-in ...' option) and thus should not 1943 cause any problems. 1944 [Bodo Moeller] 1945 1946 *) Remove all references to RSAref, since there's no more need for it. 1947 [Richard Levitte] 1948 1949 *) Make DSO load along a path given through an environment variable 1950 (SHLIB_PATH) with shl_load(). 1951 [Richard Levitte] 1952 1953 *) Constify the ENGINE code as a result of BIGNUM constification. 1954 Also constify the RSA code and most things related to it. In a 1955 few places, most notable in the depth of the ASN.1 code, ugly 1956 casts back to non-const were required (to be solved at a later 1957 time) 1958 [Richard Levitte] 1959 1960 *) Make it so the openssl application has all engines loaded by default. 1961 [Richard Levitte] 1962 1963 *) Constify the BIGNUM routines a little more. 1964 [Richard Levitte] 1965 1966 *) Add the following functions: 1967 1968 ENGINE_load_cswift() 1969 ENGINE_load_chil() 1970 ENGINE_load_atalla() 1971 ENGINE_load_nuron() 1972 ENGINE_load_builtin_engines() 1973 1974 That way, an application can itself choose if external engines that 1975 are built-in in OpenSSL shall ever be used or not. The benefit is 1976 that applications won't have to be linked with libdl or other dso 1977 libraries unless it's really needed. 1978 1979 Changed 'openssl engine' to load all engines on demand. 1980 Changed the engine header files to avoid the duplication of some 1981 declarations (they differed!). 1982 [Richard Levitte] 1983 1984 *) 'openssl engine' can now list capabilities. 1985 [Richard Levitte] 1986 1987 *) Better error reporting in 'openssl engine'. 1988 [Richard Levitte] 1989 1990 *) Never call load_dh_param(NULL) in s_server. 1991 [Bodo Moeller] 1992 1993 *) Add engine application. It can currently list engines by name and 1994 identity, and test if they are actually available. 1995 [Richard Levitte] 1996 1997 *) Improve RPM specification file by forcing symbolic linking and making 1998 sure the installed documentation is also owned by root.root. 1999 [Damien Miller <djm@mindrot.org>] 2000 2001 *) Give the OpenSSL applications more possibilities to make use of 2002 keys (public as well as private) handled by engines. 2003 [Richard Levitte] 2004 2005 *) Add OCSP code that comes from CertCo. 2006 [Richard Levitte] 2007 2008 *) Add VMS support for the Rijndael code. 2009 [Richard Levitte] 2010 2011 *) Added untested support for Nuron crypto accelerator. 2012 [Ben Laurie] 2013 2014 *) Add support for external cryptographic devices. This code was 2015 previously distributed separately as the "engine" branch. 2016 [Geoff Thorpe, Richard Levitte] 2017 2018 *) Rework the filename-translation in the DSO code. It is now possible to 2019 have far greater control over how a "name" is turned into a filename 2020 depending on the operating environment and any oddities about the 2021 different shared library filenames on each system. 2022 [Geoff Thorpe] 2023 2024 *) Support threads on FreeBSD-elf in Configure. 2025 [Richard Levitte] 2026 2027 *) Fix for SHA1 assembly problem with MASM: it produces 2028 warnings about corrupt line number information when assembling 2029 with debugging information. This is caused by the overlapping 2030 of two sections. 2031 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 2032 2033 *) NCONF changes. 2034 NCONF_get_number() has no error checking at all. As a replacement, 2035 NCONF_get_number_e() is defined (_e for "error checking") and is 2036 promoted strongly. The old NCONF_get_number is kept around for 2037 binary backward compatibility. 2038 Make it possible for methods to load from something other than a BIO, 2039 by providing a function pointer that is given a name instead of a BIO. 2040 For example, this could be used to load configuration data from an 2041 LDAP server. 2042 [Richard Levitte] 2043 2044 *) Fix for non blocking accept BIOs. Added new I/O special reason 2045 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 2046 with non blocking I/O was not possible because no retry code was 2047 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 2048 this case. 2049 [Steve Henson] 2050 2051 *) Added the beginnings of Rijndael support. 2052 [Ben Laurie] 2053 2054 *) Fix for bug in DirectoryString mask setting. Add support for 2055 X509_NAME_print_ex() in 'req' and X509_print_ex() function 2056 to allow certificate printing to more controllable, additional 2057 'certopt' option to 'x509' to allow new printing options to be 2058 set. 2059 [Steve Henson] 2060 2061 *) Clean old EAY MD5 hack from e_os.h. 2062 [Richard Levitte] 2063 2064 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 2065 2066 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 2067 by using the Codenomicon TLS Test Tool (CAN-2004-0079) 2068 [Joe Orton, Steve Henson] 2069 2070 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 2071 2072 *) Fix additional bug revealed by the NISCC test suite: 2073 2074 Stop bug triggering large recursion when presented with 2075 certain ASN.1 tags (CAN-2003-0851) 2076 [Steve Henson] 2077 2078 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 2079 2080 *) Fix various bugs revealed by running the NISCC test suite: 2081 2082 Stop out of bounds reads in the ASN1 code when presented with 2083 invalid tags (CAN-2003-0543 and CAN-2003-0544). 2084 2085 If verify callback ignores invalid public key errors don't try to check 2086 certificate signature with the NULL public key. 2087 2088 [Steve Henson] 2089 2090 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 2091 if the server requested one: as stated in TLS 1.0 and SSL 3.0 2092 specifications. 2093 [Steve Henson] 2094 2095 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 2096 extra data after the compression methods not only for TLS 1.0 2097 but also for SSL 3.0 (as required by the specification). 2098 [Bodo Moeller; problem pointed out by Matthias Loepfe] 2099 2100 *) Change X509_certificate_type() to mark the key as exported/exportable 2101 when it's 512 *bits* long, not 512 bytes. 2102 [Richard Levitte] 2103 2104 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 2105 2106 *) Countermeasure against the Klima-Pokorny-Rosa extension of 2107 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 2108 a protocol version number mismatch like a decryption error 2109 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 2110 [Bodo Moeller] 2111 2112 *) Turn on RSA blinding by default in the default implementation 2113 to avoid a timing attack. Applications that don't want it can call 2114 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 2115 They would be ill-advised to do so in most cases. 2116 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 2117 2118 *) Change RSA blinding code so that it works when the PRNG is not 2119 seeded (in this case, the secret RSA exponent is abused as 2120 an unpredictable seed -- if it is not unpredictable, there 2121 is no point in blinding anyway). Make RSA blinding thread-safe 2122 by remembering the creator's thread ID in rsa->blinding and 2123 having all other threads use local one-time blinding factors 2124 (this requires more computation than sharing rsa->blinding, but 2125 avoids excessive locking; and if an RSA object is not shared 2126 between threads, blinding will still be very fast). 2127 [Bodo Moeller] 2128 2129 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 2130 2131 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 2132 via timing by performing a MAC computation even if incorrrect 2133 block cipher padding has been found. This is a countermeasure 2134 against active attacks where the attacker has to distinguish 2135 between bad padding and a MAC verification error. (CAN-2003-0078) 2136 2137 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 2138 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 2139 Martin Vuagnoux (EPFL, Ilion)] 2140 2141 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 2142 2143 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 2144 memory from it's contents. This is done with a counter that will 2145 place alternating values in each byte. This can be used to solve 2146 two issues: 1) the removal of calls to memset() by highly optimizing 2147 compilers, and 2) cleansing with other values than 0, since those can 2148 be read through on certain media, for example a swap space on disk. 2149 [Geoff Thorpe] 2150 2151 *) Bugfix: client side session caching did not work with external caching, 2152 because the session->cipher setting was not restored when reloading 2153 from the external cache. This problem was masked, when 2154 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 2155 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 2156 [Lutz Jaenicke] 2157 2158 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 2159 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 2160 [Zeev Lieber <zeev-l@yahoo.com>] 2161 2162 *) Undo an undocumented change introduced in 0.9.6e which caused 2163 repeated calls to OpenSSL_add_all_ciphers() and 2164 OpenSSL_add_all_digests() to be ignored, even after calling 2165 EVP_cleanup(). 2166 [Richard Levitte] 2167 2168 *) Change the default configuration reader to deal with last line not 2169 being properly terminated. 2170 [Richard Levitte] 2171 2172 *) Change X509_NAME_cmp() so it applies the special rules on handling 2173 DN values that are of type PrintableString, as well as RDNs of type 2174 emailAddress where the value has the type ia5String. 2175 [stefank@valicert.com via Richard Levitte] 2176 2177 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 2178 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 2179 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 2180 the bitwise-OR of the two for use by the majority of applications 2181 wanting this behaviour, and update the docs. The documented 2182 behaviour and actual behaviour were inconsistent and had been 2183 changing anyway, so this is more a bug-fix than a behavioural 2184 change. 2185 [Geoff Thorpe, diagnosed by Nadav Har'El] 2186 2187 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 2188 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 2189 [Bodo Moeller] 2190 2191 *) Fix initialization code race conditions in 2192 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 2193 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 2194 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 2195 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 2196 ssl2_get_cipher_by_char(), 2197 ssl3_get_cipher_by_char(). 2198 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 2199 2200 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 2201 the cached sessions are flushed, as the remove_cb() might use ex_data 2202 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 2203 (see [openssl.org #212]). 2204 [Geoff Thorpe, Lutz Jaenicke] 2205 2206 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 2207 length, instead of the encoding length to d2i_ASN1_OBJECT. 2208 [Steve Henson] 2209 2210 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 2211 2212 *) [In 0.9.6g-engine release:] 2213 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 2214 [Lynn Gazis <lgazis@rainbow.com>] 2215 2216 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 2217 2218 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 2219 and get fix the header length calculation. 2220 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 2221 Alon Kantor <alonk@checkpoint.com> (and others), 2222 Steve Henson] 2223 2224 *) Use proper error handling instead of 'assertions' in buffer 2225 overflow checks added in 0.9.6e. This prevents DoS (the 2226 assertions could call abort()). 2227 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 2228 2229 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 2230 2231 *) Add various sanity checks to asn1_get_length() to reject 2232 the ASN1 length bytes if they exceed sizeof(long), will appear 2233 negative or the content length exceeds the length of the 2234 supplied buffer. 2235 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 2236 2237 *) Fix cipher selection routines: ciphers without encryption had no flags 2238 for the cipher strength set and where therefore not handled correctly 2239 by the selection routines (PR #130). 2240 [Lutz Jaenicke] 2241 2242 *) Fix EVP_dsa_sha macro. 2243 [Nils Larsch] 2244 2245 *) New option 2246 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 2247 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 2248 that was added in OpenSSL 0.9.6d. 2249 2250 As the countermeasure turned out to be incompatible with some 2251 broken SSL implementations, the new option is part of SSL_OP_ALL. 2252 SSL_OP_ALL is usually employed when compatibility with weird SSL 2253 implementations is desired (e.g. '-bugs' option to 's_client' and 2254 's_server'), so the new option is automatically set in many 2255 applications. 2256 [Bodo Moeller] 2257 2258 *) Changes in security patch: 2259 2260 Changes marked "(CHATS)" were sponsored by the Defense Advanced 2261 Research Projects Agency (DARPA) and Air Force Research Laboratory, 2262 Air Force Materiel Command, USAF, under agreement number 2263 F30602-01-2-0537. 2264 2265 *) Add various sanity checks to asn1_get_length() to reject 2266 the ASN1 length bytes if they exceed sizeof(long), will appear 2267 negative or the content length exceeds the length of the 2268 supplied buffer. (CAN-2002-0659) 2269 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 2270 2271 *) Assertions for various potential buffer overflows, not known to 2272 happen in practice. 2273 [Ben Laurie (CHATS)] 2274 2275 *) Various temporary buffers to hold ASCII versions of integers were 2276 too small for 64 bit platforms. (CAN-2002-0655) 2277 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 2278 2279 *) Remote buffer overflow in SSL3 protocol - an attacker could 2280 supply an oversized session ID to a client. (CAN-2002-0656) 2281 [Ben Laurie (CHATS)] 2282 2283 *) Remote buffer overflow in SSL2 protocol - an attacker could 2284 supply an oversized client master key. (CAN-2002-0656) 2285 [Ben Laurie (CHATS)] 2286 2287 Changes between 0.9.6c and 0.9.6d [9 May 2002] 2288 2289 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 2290 encoded as NULL) with id-dsa-with-sha1. 2291 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 2292 2293 *) Check various X509_...() return values in apps/req.c. 2294 [Nils Larsch <nla@trustcenter.de>] 2295 2296 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 2297 an end-of-file condition would erronously be flagged, when the CRLF 2298 was just at the end of a processed block. The bug was discovered when 2299 processing data through a buffering memory BIO handing the data to a 2300 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 2301 <ptsekov@syntrex.com> and Nedelcho Stanev. 2302 [Lutz Jaenicke] 2303 2304 *) Implement a countermeasure against a vulnerability recently found 2305 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 2306 before application data chunks to avoid the use of known IVs 2307 with data potentially chosen by the attacker. 2308 [Bodo Moeller] 2309 2310 *) Fix length checks in ssl3_get_client_hello(). 2311 [Bodo Moeller] 2312 2313 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 2314 to prevent ssl3_read_internal() from incorrectly assuming that 2315 ssl3_read_bytes() found application data while handshake 2316 processing was enabled when in fact s->s3->in_read_app_data was 2317 merely automatically cleared during the initial handshake. 2318 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 2319 2320 *) Fix object definitions for Private and Enterprise: they were not 2321 recognized in their shortname (=lowercase) representation. Extend 2322 obj_dat.pl to issue an error when using undefined keywords instead 2323 of silently ignoring the problem (Svenning Sorensen 2324 <sss@sss.dnsalias.net>). 2325 [Lutz Jaenicke] 2326 2327 *) Fix DH_generate_parameters() so that it works for 'non-standard' 2328 generators, i.e. generators other than 2 and 5. (Previously, the 2329 code did not properly initialise the 'add' and 'rem' values to 2330 BN_generate_prime().) 2331 2332 In the new general case, we do not insist that 'generator' is 2333 actually a primitive root: This requirement is rather pointless; 2334 a generator of the order-q subgroup is just as good, if not 2335 better. 2336 [Bodo Moeller] 2337 2338 *) Map new X509 verification errors to alerts. Discovered and submitted by 2339 Tom Wu <tom@arcot.com>. 2340 [Lutz Jaenicke] 2341 2342 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 2343 returning non-zero before the data has been completely received 2344 when using non-blocking I/O. 2345 [Bodo Moeller; problem pointed out by John Hughes] 2346 2347 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 2348 [Ben Laurie, Lutz Jaenicke] 2349 2350 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 2351 Yoram Zahavi <YoramZ@gilian.com>). 2352 [Lutz Jaenicke] 2353 2354 *) Add information about CygWin 1.3 and on, and preserve proper 2355 configuration for the versions before that. 2356 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 2357 2358 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 2359 check whether we deal with a copy of a session and do not delete from 2360 the cache in this case. Problem reported by "Izhar Shoshani Levi" 2361 <izhar@checkpoint.com>. 2362 [Lutz Jaenicke] 2363 2364 *) Do not store session data into the internal session cache, if it 2365 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 2366 flag is set). Proposed by Aslam <aslam@funk.com>. 2367 [Lutz Jaenicke] 2368 2369 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 2370 value is 0. 2371 [Richard Levitte] 2372 2373 *) [In 0.9.6d-engine release:] 2374 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 2375 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 2376 2377 *) Add the configuration target linux-s390x. 2378 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 2379 2380 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 2381 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 2382 variable as an indication that a ClientHello message has been 2383 received. As the flag value will be lost between multiple 2384 invocations of ssl3_accept when using non-blocking I/O, the 2385 function may not be aware that a handshake has actually taken 2386 place, thus preventing a new session from being added to the 2387 session cache. 2388 2389 To avoid this problem, we now set s->new_session to 2 instead of 2390 using a local variable. 2391 [Lutz Jaenicke, Bodo Moeller] 2392 2393 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 2394 if the SSL_R_LENGTH_MISMATCH error is detected. 2395 [Geoff Thorpe, Bodo Moeller] 2396 2397 *) New 'shared_ldflag' column in Configure platform table. 2398 [Richard Levitte] 2399 2400 *) Fix EVP_CIPHER_mode macro. 2401 ["Dan S. Camper" <dan@bti.net>] 2402 2403 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 2404 type, we must throw them away by setting rr->length to 0. 2405 [D P Chang <dpc@qualys.com>] 2406 2407 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 2408 2409 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 2410 <Dominikus.Scherkl@biodata.com>. (The previous implementation 2411 worked incorrectly for those cases where range = 10..._2 and 2412 3*range is two bits longer than range.) 2413 [Bodo Moeller] 2414 2415 *) Only add signing time to PKCS7 structures if it is not already 2416 present. 2417 [Steve Henson] 2418 2419 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 2420 OBJ_ld_ce should be OBJ_id_ce. 2421 Also some ip-pda OIDs in crypto/objects/objects.txt were 2422 incorrect (cf. RFC 3039). 2423 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 2424 2425 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 2426 returns early because it has nothing to do. 2427 [Andy Schneider <andy.schneider@bjss.co.uk>] 2428 2429 *) [In 0.9.6c-engine release:] 2430 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 2431 [Andy Schneider <andy.schneider@bjss.co.uk>] 2432 2433 *) [In 0.9.6c-engine release:] 2434 Add support for Cryptographic Appliance's keyserver technology. 2435 (Use engine 'keyclient') 2436 [Cryptographic Appliances and Geoff Thorpe] 2437 2438 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 2439 is called via tools/c89.sh because arguments have to be 2440 rearranged (all '-L' options must appear before the first object 2441 modules). 2442 [Richard Shapiro <rshapiro@abinitio.com>] 2443 2444 *) [In 0.9.6c-engine release:] 2445 Add support for Broadcom crypto accelerator cards, backported 2446 from 0.9.7. 2447 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 2448 2449 *) [In 0.9.6c-engine release:] 2450 Add support for SureWare crypto accelerator cards from 2451 Baltimore Technologies. (Use engine 'sureware') 2452 [Baltimore Technologies and Mark Cox] 2453 2454 *) [In 0.9.6c-engine release:] 2455 Add support for crypto accelerator cards from Accelerated 2456 Encryption Processing, www.aep.ie. (Use engine 'aep') 2457 [AEP Inc. and Mark Cox] 2458 2459 *) Add a configuration entry for gcc on UnixWare. 2460 [Gary Benson <gbenson@redhat.com>] 2461 2462 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 2463 messages are stored in a single piece (fixed-length part and 2464 variable-length part combined) and fix various bugs found on the way. 2465 [Bodo Moeller] 2466 2467 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 2468 instead. BIO_gethostbyname() does not know what timeouts are 2469 appropriate, so entries would stay in cache even when they have 2470 become invalid. 2471 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 2472 2473 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 2474 faced with a pathologically small ClientHello fragment that does 2475 not contain client_version: Instead of aborting with an error, 2476 simply choose the highest available protocol version (i.e., 2477 TLS 1.0 unless it is disabled). In practice, ClientHello 2478 messages are never sent like this, but this change gives us 2479 strictly correct behaviour at least for TLS. 2480 [Bodo Moeller] 2481 2482 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 2483 never resets s->method to s->ctx->method when called from within 2484 one of the SSL handshake functions. 2485 [Bodo Moeller; problem pointed out by Niko Baric] 2486 2487 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 2488 (sent using the client's version number) if client_version is 2489 smaller than the protocol version in use. Also change 2490 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 2491 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 2492 the client will at least see that alert. 2493 [Bodo Moeller] 2494 2495 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 2496 correctly. 2497 [Bodo Moeller] 2498 2499 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 2500 client receives HelloRequest while in a handshake. 2501 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 2502 2503 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 2504 should end in 'break', not 'goto end' which circuments various 2505 cleanups done in state SSL_ST_OK. But session related stuff 2506 must be disabled for SSL_ST_OK in the case that we just sent a 2507 HelloRequest. 2508 2509 Also avoid some overhead by not calling ssl_init_wbio_buffer() 2510 before just sending a HelloRequest. 2511 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 2512 2513 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 2514 reveal whether illegal block cipher padding was found or a MAC 2515 verification error occured. (Neither SSLerr() codes nor alerts 2516 are directly visible to potential attackers, but the information 2517 may leak via logfiles.) 2518 2519 Similar changes are not required for the SSL 2.0 implementation 2520 because the number of padding bytes is sent in clear for SSL 2.0, 2521 and the extra bytes are just ignored. However ssl/s2_pkt.c 2522 failed to verify that the purported number of padding bytes is in 2523 the legal range. 2524 [Bodo Moeller] 2525 2526 *) Add OpenUNIX-8 support including shared libraries 2527 (Boyd Lynn Gerber <gerberb@zenez.com>). 2528 [Lutz Jaenicke] 2529 2530 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 2531 'wristwatch attack' using huge encoding parameters (cf. 2532 James H. Manger's CRYPTO 2001 paper). Note that the 2533 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 2534 encoding parameters and hence was not vulnerable. 2535 [Bodo Moeller] 2536 2537 *) BN_sqr() bug fix. 2538 [Ulf M�ller, reported by Jim Ellis <jim.ellis@cavium.com>] 2539 2540 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 2541 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 2542 followed by modular reduction. 2543 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 2544 2545 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 2546 equivalent based on BN_pseudo_rand() instead of BN_rand(). 2547 [Bodo Moeller] 2548 2549 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 2550 This function was broken, as the check for a new client hello message 2551 to handle SGC did not allow these large messages. 2552 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 2553 [Lutz Jaenicke] 2554 2555 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 2556 [Lutz Jaenicke] 2557 2558 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 2559 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 2560 [Lutz Jaenicke] 2561 2562 *) Rework the configuration and shared library support for Tru64 Unix. 2563 The configuration part makes use of modern compiler features and 2564 still retains old compiler behavior for those that run older versions 2565 of the OS. The shared library support part includes a variant that 2566 uses the RPATH feature, and is available through the special 2567 configuration target "alpha-cc-rpath", which will never be selected 2568 automatically. 2569 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 2570 2571 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 2572 with the same message size as in ssl3_get_certificate_request(). 2573 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 2574 messages might inadvertently be reject as too long. 2575 [Petr Lampa <lampa@fee.vutbr.cz>] 2576 2577 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 2578 [Andy Polyakov] 2579 2580 *) Modified SSL library such that the verify_callback that has been set 2581 specificly for an SSL object with SSL_set_verify() is actually being 2582 used. Before the change, a verify_callback set with this function was 2583 ignored and the verify_callback() set in the SSL_CTX at the time of 2584 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 2585 to allow the necessary settings. 2586 [Lutz Jaenicke] 2587 2588 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 2589 explicitly to NULL, as at least on Solaris 8 this seems not always to be 2590 done automatically (in contradiction to the requirements of the C 2591 standard). This made problems when used from OpenSSH. 2592 [Lutz Jaenicke] 2593 2594 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 2595 dh->length and always used 2596 2597 BN_rand_range(priv_key, dh->p). 2598 2599 BN_rand_range() is not necessary for Diffie-Hellman, and this 2600 specific range makes Diffie-Hellman unnecessarily inefficient if 2601 dh->length (recommended exponent length) is much smaller than the 2602 length of dh->p. We could use BN_rand_range() if the order of 2603 the subgroup was stored in the DH structure, but we only have 2604 dh->length. 2605 2606 So switch back to 2607 2608 BN_rand(priv_key, l, ...) 2609 2610 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 2611 otherwise. 2612 [Bodo Moeller] 2613 2614 *) In 2615 2616 RSA_eay_public_encrypt 2617 RSA_eay_private_decrypt 2618 RSA_eay_private_encrypt (signing) 2619 RSA_eay_public_decrypt (signature verification) 2620 2621 (default implementations for RSA_public_encrypt, 2622 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 2623 always reject numbers >= n. 2624 [Bodo Moeller] 2625 2626 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 2627 to synchronize access to 'locking_thread'. This is necessary on 2628 systems where access to 'locking_thread' (an 'unsigned long' 2629 variable) is not atomic. 2630 [Bodo Moeller] 2631 2632 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 2633 *before* setting the 'crypto_lock_rand' flag. The previous code had 2634 a race condition if 0 is a valid thread ID. 2635 [Travis Vitek <vitek@roguewave.com>] 2636 2637 *) Add support for shared libraries under Irix. 2638 [Albert Chin-A-Young <china@thewrittenword.com>] 2639 2640 *) Add configuration option to build on Linux on both big-endian and 2641 little-endian MIPS. 2642 [Ralf Baechle <ralf@uni-koblenz.de>] 2643 2644 *) Add the possibility to create shared libraries on HP-UX. 2645 [Richard Levitte] 2646 2647 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 2648 2649 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 2650 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 2651 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 2652 PRNG state recovery was possible based on the output of 2653 one PRNG request appropriately sized to gain knowledge on 2654 'md' followed by enough consecutive 1-byte PRNG requests 2655 to traverse all of 'state'. 2656 2657 1. When updating 'md_local' (the current thread's copy of 'md') 2658 during PRNG output generation, hash all of the previous 2659 'md_local' value, not just the half used for PRNG output. 2660 2661 2. Make the number of bytes from 'state' included into the hash 2662 independent from the number of PRNG bytes requested. 2663 2664 The first measure alone would be sufficient to avoid 2665 Markku-Juhani's attack. (Actually it had never occurred 2666 to me that the half of 'md_local' used for chaining was the 2667 half from which PRNG output bytes were taken -- I had always 2668 assumed that the secret half would be used.) The second 2669 measure makes sure that additional data from 'state' is never 2670 mixed into 'md_local' in small portions; this heuristically 2671 further strengthens the PRNG. 2672 [Bodo Moeller] 2673 2674 *) Fix crypto/bn/asm/mips3.s. 2675 [Andy Polyakov] 2676 2677 *) When only the key is given to "enc", the IV is undefined. Print out 2678 an error message in this case. 2679 [Lutz Jaenicke] 2680 2681 *) Handle special case when X509_NAME is empty in X509 printing routines. 2682 [Steve Henson] 2683 2684 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 2685 positive and less than q. 2686 [Bodo Moeller] 2687 2688 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 2689 used: it isn't thread safe and the add_lock_callback should handle 2690 that itself. 2691 [Paul Rose <Paul.Rose@bridge.com>] 2692 2693 *) Verify that incoming data obeys the block size in 2694 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 2695 [Bodo Moeller] 2696 2697 *) Fix OAEP check. 2698 [Ulf M�ller, Bodo M�ller] 2699 2700 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 2701 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 2702 when fixing the server behaviour for backwards-compatible 'client 2703 hello' messages. (Note that the attack is impractical against 2704 SSL 3.0 and TLS 1.0 anyway because length and version checking 2705 means that the probability of guessing a valid ciphertext is 2706 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 2707 paper.) 2708 2709 Before 0.9.5, the countermeasure (hide the error by generating a 2710 random 'decryption result') did not work properly because 2711 ERR_clear_error() was missing, meaning that SSL_get_error() would 2712 detect the supposedly ignored error. 2713 2714 Both problems are now fixed. 2715 [Bodo Moeller] 2716 2717 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 2718 (previously it was 1024). 2719 [Bodo Moeller] 2720 2721 *) Fix for compatibility mode trust settings: ignore trust settings 2722 unless some valid trust or reject settings are present. 2723 [Steve Henson] 2724 2725 *) Fix for blowfish EVP: its a variable length cipher. 2726 [Steve Henson] 2727 2728 *) Fix various bugs related to DSA S/MIME verification. Handle missing 2729 parameters in DSA public key structures and return an error in the 2730 DSA routines if parameters are absent. 2731 [Steve Henson] 2732 2733 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 2734 in the current directory if neither $RANDFILE nor $HOME was set. 2735 RAND_file_name() in 0.9.6a returned NULL in this case. This has 2736 caused some confusion to Windows users who haven't defined $HOME. 2737 Thus RAND_file_name() is changed again: e_os.h can define a 2738 DEFAULT_HOME, which will be used if $HOME is not set. 2739 For Windows, we use "C:"; on other platforms, we still require 2740 environment variables. 2741 2742 *) Move 'if (!initialized) RAND_poll()' into regions protected by 2743 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 2744 having multiple threads call RAND_poll() concurrently. 2745 [Bodo Moeller] 2746 2747 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 2748 combination of a flag and a thread ID variable. 2749 Otherwise while one thread is in ssleay_rand_bytes (which sets the 2750 flag), *other* threads can enter ssleay_add_bytes without obeying 2751 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 2752 that they do not hold after the first thread unsets add_do_not_lock). 2753 [Bodo Moeller] 2754 2755 *) Change bctest again: '-x' expressions are not available in all 2756 versions of 'test'. 2757 [Bodo Moeller] 2758 2759 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 2760 2761 *) Fix a couple of memory leaks in PKCS7_dataDecode() 2762 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 2763 2764 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 2765 the default extension for executables, if any. Also, make the perl 2766 scripts that use symlink() to test if it really exists and use "cp" 2767 if it doesn't. All this made OpenSSL compilable and installable in 2768 CygWin. 2769 [Richard Levitte] 2770 2771 *) Fix for asn1_GetSequence() for indefinite length constructed data. 2772 If SEQUENCE is length is indefinite just set c->slen to the total 2773 amount of data available. 2774 [Steve Henson, reported by shige@FreeBSD.org] 2775 [This change does not apply to 0.9.7.] 2776 2777 *) Change bctest to avoid here-documents inside command substitution 2778 (workaround for FreeBSD /bin/sh bug). 2779 For compatibility with Ultrix, avoid shell functions (introduced 2780 in the bctest version that searches along $PATH). 2781 [Bodo Moeller] 2782 2783 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 2784 with des_encrypt() defined on some operating systems, like Solaris 2785 and UnixWare. 2786 [Richard Levitte] 2787 2788 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 2789 On the Importance of Eliminating Errors in Cryptographic 2790 Computations, J. Cryptology 14 (2001) 2, 101-119, 2791 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 2792 [Ulf Moeller] 2793 2794 *) MIPS assembler BIGNUM division bug fix. 2795 [Andy Polyakov] 2796 2797 *) Disabled incorrect Alpha assembler code. 2798 [Richard Levitte] 2799 2800 *) Fix PKCS#7 decode routines so they correctly update the length 2801 after reading an EOC for the EXPLICIT tag. 2802 [Steve Henson] 2803 [This change does not apply to 0.9.7.] 2804 2805 *) Fix bug in PKCS#12 key generation routines. This was triggered 2806 if a 3DES key was generated with a 0 initial byte. Include 2807 PKCS12_BROKEN_KEYGEN compilation option to retain the old 2808 (but broken) behaviour. 2809 [Steve Henson] 2810 2811 *) Enhance bctest to search for a working bc along $PATH and print 2812 it when found. 2813 [Tim Rice <tim@multitalents.net> via Richard Levitte] 2814 2815 *) Fix memory leaks in err.c: free err_data string if necessary; 2816 don't write to the wrong index in ERR_set_error_data. 2817 [Bodo Moeller] 2818 2819 *) Implement ssl23_peek (analogous to ssl23_read), which previously 2820 did not exist. 2821 [Bodo Moeller] 2822 2823 *) Replace rdtsc with _emit statements for VC++ version 5. 2824 [Jeremy Cooper <jeremy@baymoo.org>] 2825 2826 *) Make it possible to reuse SSLv2 sessions. 2827 [Richard Levitte] 2828 2829 *) In copy_email() check for >= 0 as a return value for 2830 X509_NAME_get_index_by_NID() since 0 is a valid index. 2831 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 2832 2833 *) Avoid coredump with unsupported or invalid public keys by checking if 2834 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 2835 PKCS7_verify() fails with non detached data. 2836 [Steve Henson] 2837 2838 *) Don't use getenv in library functions when run as setuid/setgid. 2839 New function OPENSSL_issetugid(). 2840 [Ulf Moeller] 2841 2842 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 2843 due to incorrect handling of multi-threading: 2844 2845 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 2846 2847 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 2848 2849 3. Count how many times MemCheck_off() has been called so that 2850 nested use can be treated correctly. This also avoids 2851 inband-signalling in the previous code (which relied on the 2852 assumption that thread ID 0 is impossible). 2853 [Bodo Moeller] 2854 2855 *) Add "-rand" option also to s_client and s_server. 2856 [Lutz Jaenicke] 2857 2858 *) Fix CPU detection on Irix 6.x. 2859 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 2860 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 2861 2862 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 2863 was empty. 2864 [Steve Henson] 2865 [This change does not apply to 0.9.7.] 2866 2867 *) Use the cached encoding of an X509_NAME structure rather than 2868 copying it. This is apparently the reason for the libsafe "errors" 2869 but the code is actually correct. 2870 [Steve Henson] 2871 2872 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 2873 Bleichenbacher's DSA attack. 2874 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 2875 to be set and top=0 forces the highest bit to be set; top=-1 is new 2876 and leaves the highest bit random. 2877 [Ulf Moeller, Bodo Moeller] 2878 2879 *) In the NCONF_...-based implementations for CONF_... queries 2880 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 2881 a temporary CONF structure with the data component set to NULL 2882 (which gives segmentation faults in lh_retrieve). 2883 Instead, use NULL for the CONF pointer in CONF_get_string and 2884 CONF_get_number (which may use environment variables) and directly 2885 return NULL from CONF_get_section. 2886 [Bodo Moeller] 2887 2888 *) Fix potential buffer overrun for EBCDIC. 2889 [Ulf Moeller] 2890 2891 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 2892 keyUsage if basicConstraints absent for a CA. 2893 [Steve Henson] 2894 2895 *) Make SMIME_write_PKCS7() write mail header values with a format that 2896 is more generally accepted (no spaces before the semicolon), since 2897 some programs can't parse those values properly otherwise. Also make 2898 sure BIO's that break lines after each write do not create invalid 2899 headers. 2900 [Richard Levitte] 2901 2902 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 2903 macros previously used would not encode an empty SEQUENCE OF 2904 and break the signature. 2905 [Steve Henson] 2906 [This change does not apply to 0.9.7.] 2907 2908 *) Zero the premaster secret after deriving the master secret in 2909 DH ciphersuites. 2910 [Steve Henson] 2911 2912 *) Add some EVP_add_digest_alias registrations (as found in 2913 OpenSSL_add_all_digests()) to SSL_library_init() 2914 aka OpenSSL_add_ssl_algorithms(). This provides improved 2915 compatibility with peers using X.509 certificates 2916 with unconventional AlgorithmIdentifier OIDs. 2917 [Bodo Moeller] 2918 2919 *) Fix for Irix with NO_ASM. 2920 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 2921 2922 *) ./config script fixes. 2923 [Ulf Moeller, Richard Levitte] 2924 2925 *) Fix 'openssl passwd -1'. 2926 [Bodo Moeller] 2927 2928 *) Change PKCS12_key_gen_asc() so it can cope with non null 2929 terminated strings whose length is passed in the passlen 2930 parameter, for example from PEM callbacks. This was done 2931 by adding an extra length parameter to asc2uni(). 2932 [Steve Henson, reported by <oddissey@samsung.co.kr>] 2933 2934 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 2935 call failed, free the DSA structure. 2936 [Bodo Moeller] 2937 2938 *) Fix to uni2asc() to cope with zero length Unicode strings. 2939 These are present in some PKCS#12 files. 2940 [Steve Henson] 2941 2942 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 2943 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 2944 when writing a 32767 byte record. 2945 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 2946 2947 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 2948 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 2949 2950 (RSA objects have a reference count access to which is protected 2951 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 2952 so they are meant to be shared between threads.) 2953 [Bodo Moeller, Geoff Thorpe; original patch submitted by 2954 "Reddie, Steven" <Steven.Reddie@ca.com>] 2955 2956 *) Fix a deadlock in CRYPTO_mem_leaks(). 2957 [Bodo Moeller] 2958 2959 *) Use better test patterns in bntest. 2960 [Ulf M�ller] 2961 2962 *) rand_win.c fix for Borland C. 2963 [Ulf M�ller] 2964 2965 *) BN_rshift bugfix for n == 0. 2966 [Bodo Moeller] 2967 2968 *) Add a 'bctest' script that checks for some known 'bc' bugs 2969 so that 'make test' does not abort just because 'bc' is broken. 2970 [Bodo Moeller] 2971 2972 *) Store verify_result within SSL_SESSION also for client side to 2973 avoid potential security hole. (Re-used sessions on the client side 2974 always resulted in verify_result==X509_V_OK, not using the original 2975 result of the server certificate verification.) 2976 [Lutz Jaenicke] 2977 2978 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 2979 SSL3_RT_APPLICATION_DATA, return 0. 2980 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 2981 [Bodo Moeller] 2982 2983 *) Fix SSL_peek: 2984 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 2985 releases, have been re-implemented by renaming the previous 2986 implementations of ssl2_read and ssl3_read to ssl2_read_internal 2987 and ssl3_read_internal, respectively, and adding 'peek' parameters 2988 to them. The new ssl[23]_{read,peek} functions are calls to 2989 ssl[23]_read_internal with the 'peek' flag set appropriately. 2990 A 'peek' parameter has also been added to ssl3_read_bytes, which 2991 does the actual work for ssl3_read_internal. 2992 [Bodo Moeller] 2993 2994 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 2995 the method-specific "init()" handler. Also clean up ex_data after 2996 calling the method-specific "finish()" handler. Previously, this was 2997 happening the other way round. 2998 [Geoff Thorpe] 2999 3000 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 3001 The previous value, 12, was not always sufficient for BN_mod_exp(). 3002 [Bodo Moeller] 3003 3004 *) Make sure that shared libraries get the internal name engine with 3005 the full version number and not just 0. This should mark the 3006 shared libraries as not backward compatible. Of course, this should 3007 be changed again when we can guarantee backward binary compatibility. 3008 [Richard Levitte] 3009 3010 *) Fix typo in get_cert_by_subject() in by_dir.c 3011 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 3012 3013 *) Rework the system to generate shared libraries: 3014 3015 - Make note of the expected extension for the shared libraries and 3016 if there is a need for symbolic links from for example libcrypto.so.0 3017 to libcrypto.so.0.9.7. There is extended info in Configure for 3018 that. 3019 3020 - Make as few rebuilds of the shared libraries as possible. 3021 3022 - Still avoid linking the OpenSSL programs with the shared libraries. 3023 3024 - When installing, install the shared libraries separately from the 3025 static ones. 3026 [Richard Levitte] 3027 3028 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 3029 3030 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 3031 and not in SSL_clear because the latter is also used by the 3032 accept/connect functions; previously, the settings made by 3033 SSL_set_read_ahead would be lost during the handshake. 3034 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 3035 3036 *) Correct util/mkdef.pl to be selective about disabled algorithms. 3037 Previously, it would create entries for disableed algorithms no 3038 matter what. 3039 [Richard Levitte] 3040 3041 *) Added several new manual pages for SSL_* function. 3042 [Lutz Jaenicke] 3043 3044 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 3045 3046 *) In ssl23_get_client_hello, generate an error message when faced 3047 with an initial SSL 3.0/TLS record that is too small to contain the 3048 first two bytes of the ClientHello message, i.e. client_version. 3049 (Note that this is a pathologic case that probably has never happened 3050 in real life.) The previous approach was to use the version number 3051 from the record header as a substitute; but our protocol choice 3052 should not depend on that one because it is not authenticated 3053 by the Finished messages. 3054 [Bodo Moeller] 3055 3056 *) More robust randomness gathering functions for Windows. 3057 [Jeffrey Altman <jaltman@columbia.edu>] 3058 3059 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 3060 not set then we don't setup the error code for issuer check errors 3061 to avoid possibly overwriting other errors which the callback does 3062 handle. If an application does set the flag then we assume it knows 3063 what it is doing and can handle the new informational codes 3064 appropriately. 3065 [Steve Henson] 3066 3067 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 3068 a general "ANY" type, as such it should be able to decode anything 3069 including tagged types. However it didn't check the class so it would 3070 wrongly interpret tagged types in the same way as their universal 3071 counterpart and unknown types were just rejected. Changed so that the 3072 tagged and unknown types are handled in the same way as a SEQUENCE: 3073 that is the encoding is stored intact. There is also a new type 3074 "V_ASN1_OTHER" which is used when the class is not universal, in this 3075 case we have no idea what the actual type is so we just lump them all 3076 together. 3077 [Steve Henson] 3078 3079 *) On VMS, stdout may very well lead to a file that is written to 3080 in a record-oriented fashion. That means that every write() will 3081 write a separate record, which will be read separately by the 3082 programs trying to read from it. This can be very confusing. 3083 3084 The solution is to put a BIO filter in the way that will buffer 3085 text until a linefeed is reached, and then write everything a 3086 line at a time, so every record written will be an actual line, 3087 not chunks of lines and not (usually doesn't happen, but I've 3088 seen it once) several lines in one record. BIO_f_linebuffer() is 3089 the answer. 3090 3091 Currently, it's a VMS-only method, because that's where it has 3092 been tested well enough. 3093 [Richard Levitte] 3094 3095 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 3096 it can return incorrect results. 3097 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 3098 but it was in 0.9.6-beta[12].) 3099 [Bodo Moeller] 3100 3101 *) Disable the check for content being present when verifying detached 3102 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 3103 include zero length content when signing messages. 3104 [Steve Henson] 3105 3106 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 3107 BIO_ctrl (for BIO pairs). 3108 [Bodo M�ller] 3109 3110 *) Add DSO method for VMS. 3111 [Richard Levitte] 3112 3113 *) Bug fix: Montgomery multiplication could produce results with the 3114 wrong sign. 3115 [Ulf M�ller] 3116 3117 *) Add RPM specification openssl.spec and modify it to build three 3118 packages. The default package contains applications, application 3119 documentation and run-time libraries. The devel package contains 3120 include files, static libraries and function documentation. The 3121 doc package contains the contents of the doc directory. The original 3122 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 3123 [Richard Levitte] 3124 3125 *) Add a large number of documentation files for many SSL routines. 3126 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 3127 3128 *) Add a configuration entry for Sony News 4. 3129 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 3130 3131 *) Don't set the two most significant bits to one when generating a 3132 random number < q in the DSA library. 3133 [Ulf M�ller] 3134 3135 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 3136 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 3137 the underlying transport is blocking) if a handshake took place. 3138 (The default behaviour is needed by applications such as s_client 3139 and s_server that use select() to determine when to use SSL_read; 3140 but for applications that know in advance when to expect data, it 3141 just makes things more complicated.) 3142 [Bodo Moeller] 3143 3144 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 3145 from EGD. 3146 [Ben Laurie] 3147 3148 *) Add a few more EBCDIC conditionals that make `req' and `x509' 3149 work better on such systems. 3150 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 3151 3152 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 3153 Update PKCS12_parse() so it copies the friendlyName and the 3154 keyid to the certificates aux info. 3155 [Steve Henson] 3156 3157 *) Fix bug in PKCS7_verify() which caused an infinite loop 3158 if there was more than one signature. 3159 [Sven Uszpelkat <su@celocom.de>] 3160 3161 *) Major change in util/mkdef.pl to include extra information 3162 about each symbol, as well as presentig variables as well 3163 as functions. This change means that there's n more need 3164 to rebuild the .num files when some algorithms are excluded. 3165 [Richard Levitte] 3166 3167 *) Allow the verify time to be set by an application, 3168 rather than always using the current time. 3169 [Steve Henson] 3170 3171 *) Phase 2 verify code reorganisation. The certificate 3172 verify code now looks up an issuer certificate by a 3173 number of criteria: subject name, authority key id 3174 and key usage. It also verifies self signed certificates 3175 by the same criteria. The main comparison function is 3176 X509_check_issued() which performs these checks. 3177 3178 Lot of changes were necessary in order to support this 3179 without completely rewriting the lookup code. 3180 3181 Authority and subject key identifier are now cached. 3182 3183 The LHASH 'certs' is X509_STORE has now been replaced 3184 by a STACK_OF(X509_OBJECT). This is mainly because an 3185 LHASH can't store or retrieve multiple objects with 3186 the same hash value. 3187 3188 As a result various functions (which were all internal 3189 use only) have changed to handle the new X509_STORE 3190 structure. This will break anything that messed round 3191 with X509_STORE internally. 3192 3193 The functions X509_STORE_add_cert() now checks for an 3194 exact match, rather than just subject name. 3195 3196 The X509_STORE API doesn't directly support the retrieval 3197 of multiple certificates matching a given criteria, however 3198 this can be worked round by performing a lookup first 3199 (which will fill the cache with candidate certificates) 3200 and then examining the cache for matches. This is probably 3201 the best we can do without throwing out X509_LOOKUP 3202 entirely (maybe later...). 3203 3204 The X509_VERIFY_CTX structure has been enhanced considerably. 3205 3206 All certificate lookup operations now go via a get_issuer() 3207 callback. Although this currently uses an X509_STORE it 3208 can be replaced by custom lookups. This is a simple way 3209 to bypass the X509_STORE hackery necessary to make this 3210 work and makes it possible to use more efficient techniques 3211 in future. A very simple version which uses a simple 3212 STACK for its trusted certificate store is also provided 3213 using X509_STORE_CTX_trusted_stack(). 3214 3215 The verify_cb() and verify() callbacks now have equivalents 3216 in the X509_STORE_CTX structure. 3217 3218 X509_STORE_CTX also has a 'flags' field which can be used 3219 to customise the verify behaviour. 3220 [Steve Henson] 3221 3222 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 3223 excludes S/MIME capabilities. 3224 [Steve Henson] 3225 3226 *) When a certificate request is read in keep a copy of the 3227 original encoding of the signed data and use it when outputing 3228 again. Signatures then use the original encoding rather than 3229 a decoded, encoded version which may cause problems if the 3230 request is improperly encoded. 3231 [Steve Henson] 3232 3233 *) For consistency with other BIO_puts implementations, call 3234 buffer_write(b, ...) directly in buffer_puts instead of calling 3235 BIO_write(b, ...). 3236 3237 In BIO_puts, increment b->num_write as in BIO_write. 3238 [Peter.Sylvester@EdelWeb.fr] 3239 3240 *) Fix BN_mul_word for the case where the word is 0. (We have to use 3241 BN_zero, we may not return a BIGNUM with an array consisting of 3242 words set to zero.) 3243 [Bodo Moeller] 3244 3245 *) Avoid calling abort() from within the library when problems are 3246 detected, except if preprocessor symbols have been defined 3247 (such as REF_CHECK, BN_DEBUG etc.). 3248 [Bodo Moeller] 3249 3250 *) New openssl application 'rsautl'. This utility can be 3251 used for low level RSA operations. DER public key 3252 BIO/fp routines also added. 3253 [Steve Henson] 3254 3255 *) New Configure entry and patches for compiling on QNX 4. 3256 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 3257 3258 *) A demo state-machine implementation was sponsored by 3259 Nuron (http://www.nuron.com/) and is now available in 3260 demos/state_machine. 3261 [Ben Laurie] 3262 3263 *) New options added to the 'dgst' utility for signature 3264 generation and verification. 3265 [Steve Henson] 3266 3267 *) Unrecognized PKCS#7 content types are now handled via a 3268 catch all ASN1_TYPE structure. This allows unsupported 3269 types to be stored as a "blob" and an application can 3270 encode and decode it manually. 3271 [Steve Henson] 3272 3273 *) Fix various signed/unsigned issues to make a_strex.c 3274 compile under VC++. 3275 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 3276 3277 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 3278 length if passed a buffer. ASN1_INTEGER_to_BN failed 3279 if passed a NULL BN and its argument was negative. 3280 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 3281 3282 *) Modification to PKCS#7 encoding routines to output definite 3283 length encoding. Since currently the whole structures are in 3284 memory there's not real point in using indefinite length 3285 constructed encoding. However if OpenSSL is compiled with 3286 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 3287 [Steve Henson] 3288 3289 *) Added BIO_vprintf() and BIO_vsnprintf(). 3290 [Richard Levitte] 3291 3292 *) Added more prefixes to parse for in the the strings written 3293 through a logging bio, to cover all the levels that are available 3294 through syslog. The prefixes are now: 3295 3296 PANIC, EMERG, EMR => LOG_EMERG 3297 ALERT, ALR => LOG_ALERT 3298 CRIT, CRI => LOG_CRIT 3299 ERROR, ERR => LOG_ERR 3300 WARNING, WARN, WAR => LOG_WARNING 3301 NOTICE, NOTE, NOT => LOG_NOTICE 3302 INFO, INF => LOG_INFO 3303 DEBUG, DBG => LOG_DEBUG 3304 3305 and as before, if none of those prefixes are present at the 3306 beginning of the string, LOG_ERR is chosen. 3307 3308 On Win32, the LOG_* levels are mapped according to this: 3309 3310 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 3311 LOG_WARNING => EVENTLOG_WARNING_TYPE 3312 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 3313 3314 [Richard Levitte] 3315 3316 *) Made it possible to reconfigure with just the configuration 3317 argument "reconf" or "reconfigure". The command line arguments 3318 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 3319 and are retrieved from there when reconfiguring. 3320 [Richard Levitte] 3321 3322 *) MD4 implemented. 3323 [Assar Westerlund <assar@sics.se>, Richard Levitte] 3324 3325 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 3326 [Richard Levitte] 3327 3328 *) The obj_dat.pl script was messing up the sorting of object 3329 names. The reason was that it compared the quoted version 3330 of strings as a result "OCSP" > "OCSP Signing" because 3331 " > SPACE. Changed script to store unquoted versions of 3332 names and add quotes on output. It was also omitting some 3333 names from the lookup table if they were given a default 3334 value (that is if SN is missing it is given the same 3335 value as LN and vice versa), these are now added on the 3336 grounds that if an object has a name we should be able to 3337 look it up. Finally added warning output when duplicate 3338 short or long names are found. 3339 [Steve Henson] 3340 3341 *) Changes needed for Tandem NSK. 3342 [Scott Uroff <scott@xypro.com>] 3343 3344 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 3345 RSA_padding_check_SSLv23(), special padding was never detected 3346 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 3347 version rollback attacks was not effective. 3348 3349 In s23_clnt.c, don't use special rollback-attack detection padding 3350 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 3351 client; similarly, in s23_srvr.c, don't do the rollback check if 3352 SSL 2.0 is the only protocol enabled in the server. 3353 [Bodo Moeller] 3354 3355 *) Make it possible to get hexdumps of unprintable data with 'openssl 3356 asn1parse'. By implication, the functions ASN1_parse_dump() and 3357 BIO_dump_indent() are added. 3358 [Richard Levitte] 3359 3360 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 3361 these print out strings and name structures based on various 3362 flags including RFC2253 support and proper handling of 3363 multibyte characters. Added options to the 'x509' utility 3364 to allow the various flags to be set. 3365 [Steve Henson] 3366 3367 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 3368 Also change the functions X509_cmp_current_time() and 3369 X509_gmtime_adj() work with an ASN1_TIME structure, 3370 this will enable certificates using GeneralizedTime in validity 3371 dates to be checked. 3372 [Steve Henson] 3373 3374 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 3375 negative public key encodings) on by default, 3376 NO_NEG_PUBKEY_BUG can be set to disable it. 3377 [Steve Henson] 3378 3379 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 3380 content octets. An i2c_ASN1_OBJECT is unnecessary because 3381 the encoding can be trivially obtained from the structure. 3382 [Steve Henson] 3383 3384 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 3385 not read locks (CRYPTO_r_[un]lock). 3386 [Bodo Moeller] 3387 3388 *) A first attempt at creating official support for shared 3389 libraries through configuration. I've kept it so the 3390 default is static libraries only, and the OpenSSL programs 3391 are always statically linked for now, but there are 3392 preparations for dynamic linking in place. 3393 This has been tested on Linux and Tru64. 3394 [Richard Levitte] 3395 3396 *) Randomness polling function for Win9x, as described in: 3397 Peter Gutmann, Software Generation of Practically Strong 3398 Random Numbers. 3399 [Ulf M�ller] 3400 3401 *) Fix so PRNG is seeded in req if using an already existing 3402 DSA key. 3403 [Steve Henson] 3404 3405 *) New options to smime application. -inform and -outform 3406 allow alternative formats for the S/MIME message including 3407 PEM and DER. The -content option allows the content to be 3408 specified separately. This should allow things like Netscape 3409 form signing output easier to verify. 3410 [Steve Henson] 3411 3412 *) Fix the ASN1 encoding of tags using the 'long form'. 3413 [Steve Henson] 3414 3415 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 3416 STRING types. These convert content octets to and from the 3417 underlying type. The actual tag and length octets are 3418 already assumed to have been read in and checked. These 3419 are needed because all other string types have virtually 3420 identical handling apart from the tag. By having versions 3421 of the ASN1 functions that just operate on content octets 3422 IMPLICIT tagging can be handled properly. It also allows 3423 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 3424 and ASN1_INTEGER are identical apart from the tag. 3425 [Steve Henson] 3426 3427 *) Change the handling of OID objects as follows: 3428 3429 - New object identifiers are inserted in objects.txt, following 3430 the syntax given in objects.README. 3431 - objects.pl is used to process obj_mac.num and create a new 3432 obj_mac.h. 3433 - obj_dat.pl is used to create a new obj_dat.h, using the data in 3434 obj_mac.h. 3435 3436 This is currently kind of a hack, and the perl code in objects.pl 3437 isn't very elegant, but it works as I intended. The simplest way 3438 to check that it worked correctly is to look in obj_dat.h and 3439 check the array nid_objs and make sure the objects haven't moved 3440 around (this is important!). Additions are OK, as well as 3441 consistent name changes. 3442 [Richard Levitte] 3443 3444 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 3445 [Bodo Moeller] 3446 3447 *) Addition of the command line parameter '-rand file' to 'openssl req'. 3448 The given file adds to whatever has already been seeded into the 3449 random pool through the RANDFILE configuration file option or 3450 environment variable, or the default random state file. 3451 [Richard Levitte] 3452 3453 *) mkstack.pl now sorts each macro group into lexical order. 3454 Previously the output order depended on the order the files 3455 appeared in the directory, resulting in needless rewriting 3456 of safestack.h . 3457 [Steve Henson] 3458 3459 *) Patches to make OpenSSL compile under Win32 again. Mostly 3460 work arounds for the VC++ problem that it treats func() as 3461 func(void). Also stripped out the parts of mkdef.pl that 3462 added extra typesafe functions: these no longer exist. 3463 [Steve Henson] 3464 3465 *) Reorganisation of the stack code. The macros are now all 3466 collected in safestack.h . Each macro is defined in terms of 3467 a "stack macro" of the form SKM_<name>(type, a, b). The 3468 DEBUG_SAFESTACK is now handled in terms of function casts, 3469 this has the advantage of retaining type safety without the 3470 use of additional functions. If DEBUG_SAFESTACK is not defined 3471 then the non typesafe macros are used instead. Also modified the 3472 mkstack.pl script to handle the new form. Needs testing to see 3473 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 3474 the default if no major problems. Similar behaviour for ASN1_SET_OF 3475 and PKCS12_STACK_OF. 3476 [Steve Henson] 3477 3478 *) When some versions of IIS use the 'NET' form of private key the 3479 key derivation algorithm is different. Normally MD5(password) is 3480 used as a 128 bit RC4 key. In the modified case 3481 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 3482 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 3483 as the old Netscape_RSA functions except they have an additional 3484 'sgckey' parameter which uses the modified algorithm. Also added 3485 an -sgckey command line option to the rsa utility. Thanks to 3486 Adrian Peck <bertie@ncipher.com> for posting details of the modified 3487 algorithm to openssl-dev. 3488 [Steve Henson] 3489 3490 *) The evp_local.h macros were using 'c.##kname' which resulted in 3491 invalid expansion on some systems (SCO 5.0.5 for example). 3492 Corrected to 'c.kname'. 3493 [Phillip Porch <root@theporch.com>] 3494 3495 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 3496 a STACK of email addresses from a certificate or request, these look 3497 in the subject name and the subject alternative name extensions and 3498 omit any duplicate addresses. 3499 [Steve Henson] 3500 3501 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 3502 This makes DSA verification about 2 % faster. 3503 [Bodo Moeller] 3504 3505 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 3506 (meaning that now 2^5 values will be precomputed, which is only 4 KB 3507 plus overhead for 1024 bit moduli). 3508 This makes exponentiations about 0.5 % faster for 1024 bit 3509 exponents (as measured by "openssl speed rsa2048"). 3510 [Bodo Moeller] 3511 3512 *) Rename memory handling macros to avoid conflicts with other 3513 software: 3514 Malloc => OPENSSL_malloc 3515 Malloc_locked => OPENSSL_malloc_locked 3516 Realloc => OPENSSL_realloc 3517 Free => OPENSSL_free 3518 [Richard Levitte] 3519 3520 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 3521 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 3522 [Bodo Moeller] 3523 3524 *) CygWin32 support. 3525 [John Jarvie <jjarvie@newsguy.com>] 3526 3527 *) The type-safe stack code has been rejigged. It is now only compiled 3528 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 3529 by default all type-specific stack functions are "#define"d back to 3530 standard stack functions. This results in more streamlined output 3531 but retains the type-safety checking possibilities of the original 3532 approach. 3533 [Geoff Thorpe] 3534 3535 *) The STACK code has been cleaned up, and certain type declarations 3536 that didn't make a lot of sense have been brought in line. This has 3537 also involved a cleanup of sorts in safestack.h to more correctly 3538 map type-safe stack functions onto their plain stack counterparts. 3539 This work has also resulted in a variety of "const"ifications of 3540 lots of the code, especially "_cmp" operations which should normally 3541 be prototyped with "const" parameters anyway. 3542 [Geoff Thorpe] 3543 3544 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 3545 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 3546 (The PRNG state consists of two parts, the large pool 'state' and 'md', 3547 where all of 'md' is used each time the PRNG is used, but 'state' 3548 is used only indexed by a cyclic counter. As entropy may not be 3549 well distributed from the beginning, 'md' is important as a 3550 chaining variable. However, the output function chains only half 3551 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 3552 all of 'md', and seeding with STATE_SIZE dummy bytes will result 3553 in all of 'state' being rewritten, with the new values depending 3554 on virtually all of 'md'. This overcomes the 80 bit limitation.) 3555 [Bodo Moeller] 3556 3557 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 3558 the handshake is continued after ssl_verify_cert_chain(); 3559 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 3560 can lead to 'unexplainable' connection aborts later. 3561 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 3562 3563 *) Major EVP API cipher revision. 3564 Add hooks for extra EVP features. This allows various cipher 3565 parameters to be set in the EVP interface. Support added for variable 3566 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 3567 setting of RC2 and RC5 parameters. 3568 3569 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 3570 ciphers. 3571 3572 Remove lots of duplicated code from the EVP library. For example *every* 3573 cipher init() function handles the 'iv' in the same way according to the 3574 cipher mode. They also all do nothing if the 'key' parameter is NULL and 3575 for CFB and OFB modes they zero ctx->num. 3576 3577 New functionality allows removal of S/MIME code RC2 hack. 3578 3579 Most of the routines have the same form and so can be declared in terms 3580 of macros. 3581 3582 By shifting this to the top level EVP_CipherInit() it can be removed from 3583 all individual ciphers. If the cipher wants to handle IVs or keys 3584 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 3585 flags. 3586 3587 Change lots of functions like EVP_EncryptUpdate() to now return a 3588 value: although software versions of the algorithms cannot fail 3589 any installed hardware versions can. 3590 [Steve Henson] 3591 3592 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 3593 this option is set, tolerate broken clients that send the negotiated 3594 protocol version number instead of the requested protocol version 3595 number. 3596 [Bodo Moeller] 3597 3598 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 3599 i.e. non-zero for export ciphersuites, zero otherwise. 3600 Previous versions had this flag inverted, inconsistent with 3601 rsa_tmp_cb (..._TMP_RSA_CB). 3602 [Bodo Moeller; problem reported by Amit Chopra] 3603 3604 *) Add missing DSA library text string. Work around for some IIS 3605 key files with invalid SEQUENCE encoding. 3606 [Steve Henson] 3607 3608 *) Add a document (doc/standards.txt) that list all kinds of standards 3609 and so on that are implemented in OpenSSL. 3610 [Richard Levitte] 3611 3612 *) Enhance c_rehash script. Old version would mishandle certificates 3613 with the same subject name hash and wouldn't handle CRLs at all. 3614 Added -fingerprint option to crl utility, to support new c_rehash 3615 features. 3616 [Steve Henson] 3617 3618 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 3619 [Ulf M�ller] 3620 3621 *) Fix for SSL server purpose checking. Server checking was 3622 rejecting certificates which had extended key usage present 3623 but no ssl client purpose. 3624 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 3625 3626 *) Make PKCS#12 code work with no password. The PKCS#12 spec 3627 is a little unclear about how a blank password is handled. 3628 Since the password in encoded as a BMPString with terminating 3629 double NULL a zero length password would end up as just the 3630 double NULL. However no password at all is different and is 3631 handled differently in the PKCS#12 key generation code. NS 3632 treats a blank password as zero length. MSIE treats it as no 3633 password on export: but it will try both on import. We now do 3634 the same: PKCS12_parse() tries zero length and no password if 3635 the password is set to "" or NULL (NULL is now a valid password: 3636 it wasn't before) as does the pkcs12 application. 3637 [Steve Henson] 3638 3639 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 3640 perror when PEM_read_bio_X509_REQ fails, the error message must 3641 be obtained from the error queue. 3642 [Bodo Moeller] 3643 3644 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 3645 it in ERR_remove_state if appropriate, and change ERR_get_state 3646 accordingly to avoid race conditions (this is necessary because 3647 thread_hash is no longer constant once set). 3648 [Bodo Moeller] 3649 3650 *) Bugfix for linux-elf makefile.one. 3651 [Ulf M�ller] 3652 3653 *) RSA_get_default_method() will now cause a default 3654 RSA_METHOD to be chosen if one doesn't exist already. 3655 Previously this was only set during a call to RSA_new() 3656 or RSA_new_method(NULL) meaning it was possible for 3657 RSA_get_default_method() to return NULL. 3658 [Geoff Thorpe] 3659 3660 *) Added native name translation to the existing DSO code 3661 that will convert (if the flag to do so is set) filenames 3662 that are sufficiently small and have no path information 3663 into a canonical native form. Eg. "blah" converted to 3664 "libblah.so" or "blah.dll" etc. 3665 [Geoff Thorpe] 3666 3667 *) New function ERR_error_string_n(e, buf, len) which is like 3668 ERR_error_string(e, buf), but writes at most 'len' bytes 3669 including the 0 terminator. For ERR_error_string_n, 'buf' 3670 may not be NULL. 3671 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 3672 3673 *) CONF library reworked to become more general. A new CONF 3674 configuration file reader "class" is implemented as well as a 3675 new functions (NCONF_*, for "New CONF") to handle it. The now 3676 old CONF_* functions are still there, but are reimplemented to 3677 work in terms of the new functions. Also, a set of functions 3678 to handle the internal storage of the configuration data is 3679 provided to make it easier to write new configuration file 3680 reader "classes" (I can definitely see something reading a 3681 configuration file in XML format, for example), called _CONF_*, 3682 or "the configuration storage API"... 3683 3684 The new configuration file reading functions are: 3685 3686 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 3687 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 3688 3689 NCONF_default, NCONF_WIN32 3690 3691 NCONF_dump_fp, NCONF_dump_bio 3692 3693 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 3694 NCONF_new creates a new CONF object. This works in the same way 3695 as other interfaces in OpenSSL, like the BIO interface. 3696 NCONF_dump_* dump the internal storage of the configuration file, 3697 which is useful for debugging. All other functions take the same 3698 arguments as the old CONF_* functions wth the exception of the 3699 first that must be a `CONF *' instead of a `LHASH *'. 3700 3701 To make it easer to use the new classes with the old CONF_* functions, 3702 the function CONF_set_default_method is provided. 3703 [Richard Levitte] 3704 3705 *) Add '-tls1' option to 'openssl ciphers', which was already 3706 mentioned in the documentation but had not been implemented. 3707 (This option is not yet really useful because even the additional 3708 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 3709 [Bodo Moeller] 3710 3711 *) Initial DSO code added into libcrypto for letting OpenSSL (and 3712 OpenSSL-based applications) load shared libraries and bind to 3713 them in a portable way. 3714 [Geoff Thorpe, with contributions from Richard Levitte] 3715 3716 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 3717 3718 *) Make sure _lrotl and _lrotr are only used with MSVC. 3719 3720 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 3721 (the default implementation of RAND_status). 3722 3723 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 3724 to '-clrext' (= clear extensions), as intended and documented. 3725 [Bodo Moeller; inconsistency pointed out by Michael Attili 3726 <attili@amaxo.com>] 3727 3728 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 3729 was larger than the MD block size. 3730 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 3731 3732 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 3733 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 3734 using the passed key: if the passed key was a private key the result 3735 of X509_print(), for example, would be to print out all the private key 3736 components. 3737 [Steve Henson] 3738 3739 *) des_quad_cksum() byte order bug fix. 3740 [Ulf M�ller, using the problem description in krb4-0.9.7, where 3741 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 3742 3743 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 3744 discouraged. 3745 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 3746 3747 *) For easily testing in shell scripts whether some command 3748 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 3749 returns with exit code 0 iff no command of the given name is available. 3750 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 3751 the output goes to stdout and nothing is printed to stderr. 3752 Additional arguments are always ignored. 3753 3754 Since for each cipher there is a command of the same name, 3755 the 'no-cipher' compilation switches can be tested this way. 3756 3757 ('openssl no-XXX' is not able to detect pseudo-commands such 3758 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 3759 [Bodo Moeller] 3760 3761 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 3762 [Bodo Moeller] 3763 3764 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 3765 is set; it will be thrown away anyway because each handshake creates 3766 its own key. 3767 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 3768 to parameters -- in previous versions (since OpenSSL 0.9.3) the 3769 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 3770 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 3771 [Bodo Moeller] 3772 3773 *) New s_client option -ign_eof: EOF at stdin is ignored, and 3774 'Q' and 'R' lose their special meanings (quit/renegotiate). 3775 This is part of what -quiet does; unlike -quiet, -ign_eof 3776 does not suppress any output. 3777 [Richard Levitte] 3778 3779 *) Add compatibility options to the purpose and trust code. The 3780 purpose X509_PURPOSE_ANY is "any purpose" which automatically 3781 accepts a certificate or CA, this was the previous behaviour, 3782 with all the associated security issues. 3783 3784 X509_TRUST_COMPAT is the old trust behaviour: only and 3785 automatically trust self signed roots in certificate store. A 3786 new trust setting X509_TRUST_DEFAULT is used to specify that 3787 a purpose has no associated trust setting and it should instead 3788 use the value in the default purpose. 3789 [Steve Henson] 3790 3791 *) Fix the PKCS#8 DSA private key code so it decodes keys again 3792 and fix a memory leak. 3793 [Steve Henson] 3794 3795 *) In util/mkerr.pl (which implements 'make errors'), preserve 3796 reason strings from the previous version of the .c file, as 3797 the default to have only downcase letters (and digits) in 3798 automatically generated reasons codes is not always appropriate. 3799 [Bodo Moeller] 3800 3801 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 3802 using strerror. Previously, ERR_reason_error_string() returned 3803 library names as reason strings for SYSerr; but SYSerr is a special 3804 case where small numbers are errno values, not library numbers. 3805 [Bodo Moeller] 3806 3807 *) Add '-dsaparam' option to 'openssl dhparam' application. This 3808 converts DSA parameters into DH parameters. (When creating parameters, 3809 DSA_generate_parameters is used.) 3810 [Bodo Moeller] 3811 3812 *) Include 'length' (recommended exponent length) in C code generated 3813 by 'openssl dhparam -C'. 3814 [Bodo Moeller] 3815 3816 *) The second argument to set_label in perlasm was already being used 3817 so couldn't be used as a "file scope" flag. Moved to third argument 3818 which was free. 3819 [Steve Henson] 3820 3821 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 3822 instead of RAND_bytes for encryption IVs and salts. 3823 [Bodo Moeller] 3824 3825 *) Include RAND_status() into RAND_METHOD instead of implementing 3826 it only for md_rand.c Otherwise replacing the PRNG by calling 3827 RAND_set_rand_method would be impossible. 3828 [Bodo Moeller] 3829 3830 *) Don't let DSA_generate_key() enter an infinite loop if the random 3831 number generation fails. 3832 [Bodo Moeller] 3833 3834 *) New 'rand' application for creating pseudo-random output. 3835 [Bodo Moeller] 3836 3837 *) Added configuration support for Linux/IA64 3838 [Rolf Haberrecker <rolf@suse.de>] 3839 3840 *) Assembler module support for Mingw32. 3841 [Ulf M�ller] 3842 3843 *) Shared library support for HPUX (in shlib/). 3844 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 3845 3846 *) Shared library support for Solaris gcc. 3847 [Lutz Behnke <behnke@trustcenter.de>] 3848 3849 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 3850 3851 *) PKCS7_encrypt() was adding text MIME headers twice because they 3852 were added manually and by SMIME_crlf_copy(). 3853 [Steve Henson] 3854 3855 *) In bntest.c don't call BN_rand with zero bits argument. 3856 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 3857 3858 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 3859 case was implemented. This caused BN_div_recp() to fail occasionally. 3860 [Ulf M�ller] 3861 3862 *) Add an optional second argument to the set_label() in the perl 3863 assembly language builder. If this argument exists and is set 3864 to 1 it signals that the assembler should use a symbol whose 3865 scope is the entire file, not just the current function. This 3866 is needed with MASM which uses the format label:: for this scope. 3867 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 3868 3869 *) Change the ASN1 types so they are typedefs by default. Before 3870 almost all types were #define'd to ASN1_STRING which was causing 3871 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 3872 for example. 3873 [Steve Henson] 3874 3875 *) Change names of new functions to the new get1/get0 naming 3876 convention: After 'get1', the caller owns a reference count 3877 and has to call ..._free; 'get0' returns a pointer to some 3878 data structure without incrementing reference counters. 3879 (Some of the existing 'get' functions increment a reference 3880 counter, some don't.) 3881 Similarly, 'set1' and 'add1' functions increase reference 3882 counters or duplicate objects. 3883 [Steve Henson] 3884 3885 *) Allow for the possibility of temp RSA key generation failure: 3886 the code used to assume it always worked and crashed on failure. 3887 [Steve Henson] 3888 3889 *) Fix potential buffer overrun problem in BIO_printf(). 3890 [Ulf M�ller, using public domain code by Patrick Powell; problem 3891 pointed out by David Sacerdote <das33@cornell.edu>] 3892 3893 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 3894 RAND_egd() and RAND_status(). In the command line application, 3895 the EGD socket can be specified like a seed file using RANDFILE 3896 or -rand. 3897 [Ulf M�ller] 3898 3899 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 3900 Some CAs (e.g. Verisign) distribute certificates in this form. 3901 [Steve Henson] 3902 3903 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 3904 list to exclude them. This means that no special compilation option 3905 is needed to use anonymous DH: it just needs to be included in the 3906 cipher list. 3907 [Steve Henson] 3908 3909 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 3910 EVP_MD_type. The old functionality is available in a new macro called 3911 EVP_MD_md(). Change code that uses it and update docs. 3912 [Steve Henson] 3913 3914 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 3915 where the 'void *' argument is replaced by a function pointer argument. 3916 Previously 'void *' was abused to point to functions, which works on 3917 many platforms, but is not correct. As these functions are usually 3918 called by macros defined in OpenSSL header files, most source code 3919 should work without changes. 3920 [Richard Levitte] 3921 3922 *) <openssl/opensslconf.h> (which is created by Configure) now contains 3923 sections with information on -D... compiler switches used for 3924 compiling the library so that applications can see them. To enable 3925 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 3926 must be defined. E.g., 3927 #define OPENSSL_ALGORITHM_DEFINES 3928 #include <openssl/opensslconf.h> 3929 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 3930 [Richard Levitte, Ulf and Bodo M�ller] 3931 3932 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 3933 record layer. 3934 [Bodo Moeller] 3935 3936 *) Change the 'other' type in certificate aux info to a STACK_OF 3937 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 3938 the required ASN1 format: arbitrary types determined by an OID. 3939 [Steve Henson] 3940 3941 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 3942 argument to 'req'. This is not because the function is newer or 3943 better than others it just uses the work 'NEW' in the certificate 3944 request header lines. Some software needs this. 3945 [Steve Henson] 3946 3947 *) Reorganise password command line arguments: now passwords can be 3948 obtained from various sources. Delete the PEM_cb function and make 3949 it the default behaviour: i.e. if the callback is NULL and the 3950 usrdata argument is not NULL interpret it as a null terminated pass 3951 phrase. If usrdata and the callback are NULL then the pass phrase 3952 is prompted for as usual. 3953 [Steve Henson] 3954 3955 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 3956 the support is automatically enabled. The resulting binaries will 3957 autodetect the card and use it if present. 3958 [Ben Laurie and Compaq Inc.] 3959 3960 *) Work around for Netscape hang bug. This sends certificate request 3961 and server done in one record. Since this is perfectly legal in the 3962 SSL/TLS protocol it isn't a "bug" option and is on by default. See 3963 the bugs/SSLv3 entry for more info. 3964 [Steve Henson] 3965 3966 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 3967 [Andy Polyakov] 3968 3969 *) Add -rand argument to smime and pkcs12 applications and read/write 3970 of seed file. 3971 [Steve Henson] 3972 3973 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 3974 [Bodo Moeller] 3975 3976 *) Add command line password options to the remaining applications. 3977 [Steve Henson] 3978 3979 *) Bug fix for BN_div_recp() for numerators with an even number of 3980 bits. 3981 [Ulf M�ller] 3982 3983 *) More tests in bntest.c, and changed test_bn output. 3984 [Ulf M�ller] 3985 3986 *) ./config recognizes MacOS X now. 3987 [Andy Polyakov] 3988 3989 *) Bug fix for BN_div() when the first words of num and divsor are 3990 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 3991 [Ulf M�ller] 3992 3993 *) Add support for various broken PKCS#8 formats, and command line 3994 options to produce them. 3995 [Steve Henson] 3996 3997 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 3998 get temporary BIGNUMs from a BN_CTX. 3999 [Ulf M�ller] 4000 4001 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 4002 for p == 0. 4003 [Ulf M�ller] 4004 4005 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 4006 include a #define from the old name to the new. The original intent 4007 was that statically linked binaries could for example just call 4008 SSLeay_add_all_ciphers() to just add ciphers to the table and not 4009 link with digests. This never worked becayse SSLeay_add_all_digests() 4010 and SSLeay_add_all_ciphers() were in the same source file so calling 4011 one would link with the other. They are now in separate source files. 4012 [Steve Henson] 4013 4014 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 4015 [Steve Henson] 4016 4017 *) Use a less unusual form of the Miller-Rabin primality test (it used 4018 a binary algorithm for exponentiation integrated into the Miller-Rabin 4019 loop, our standard modexp algorithms are faster). 4020 [Bodo Moeller] 4021 4022 *) Support for the EBCDIC character set completed. 4023 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 4024 4025 *) Source code cleanups: use const where appropriate, eliminate casts, 4026 use void * instead of char * in lhash. 4027 [Ulf M�ller] 4028 4029 *) Bugfix: ssl3_send_server_key_exchange was not restartable 4030 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 4031 this the server could overwrite ephemeral keys that the client 4032 has already seen). 4033 [Bodo Moeller] 4034 4035 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 4036 using 50 iterations of the Rabin-Miller test. 4037 4038 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 4039 iterations of the Rabin-Miller test as required by the appendix 4040 to FIPS PUB 186[-1]) instead of DSA_is_prime. 4041 As BN_is_prime_fasttest includes trial division, DSA parameter 4042 generation becomes much faster. 4043 4044 This implies a change for the callback functions in DSA_is_prime 4045 and DSA_generate_parameters: The callback function is called once 4046 for each positive witness in the Rabin-Miller test, not just 4047 occasionally in the inner loop; and the parameters to the 4048 callback function now provide an iteration count for the outer 4049 loop rather than for the current invocation of the inner loop. 4050 DSA_generate_parameters additionally can call the callback 4051 function with an 'iteration count' of -1, meaning that a 4052 candidate has passed the trial division test (when q is generated 4053 from an application-provided seed, trial division is skipped). 4054 [Bodo Moeller] 4055 4056 *) New function BN_is_prime_fasttest that optionally does trial 4057 division before starting the Rabin-Miller test and has 4058 an additional BN_CTX * argument (whereas BN_is_prime always 4059 has to allocate at least one BN_CTX). 4060 'callback(1, -1, cb_arg)' is called when a number has passed the 4061 trial division stage. 4062 [Bodo Moeller] 4063 4064 *) Fix for bug in CRL encoding. The validity dates weren't being handled 4065 as ASN1_TIME. 4066 [Steve Henson] 4067 4068 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 4069 [Steve Henson] 4070 4071 *) New function BN_pseudo_rand(). 4072 [Ulf M�ller] 4073 4074 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 4075 bignum version of BN_from_montgomery() with the working code from 4076 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 4077 the comments. 4078 [Ulf M�ller] 4079 4080 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 4081 made it impossible to use the same SSL_SESSION data structure in 4082 SSL2 clients in multiple threads. 4083 [Bodo Moeller] 4084 4085 *) The return value of RAND_load_file() no longer counts bytes obtained 4086 by stat(). RAND_load_file(..., -1) is new and uses the complete file 4087 to seed the PRNG (previously an explicit byte count was required). 4088 [Ulf M�ller, Bodo M�ller] 4089 4090 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 4091 used (char *) instead of (void *) and had casts all over the place. 4092 [Steve Henson] 4093 4094 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 4095 [Ulf M�ller] 4096 4097 *) Retain source code compatibility for BN_prime_checks macro: 4098 BN_is_prime(..., BN_prime_checks, ...) now uses 4099 BN_prime_checks_for_size to determine the appropriate number of 4100 Rabin-Miller iterations. 4101 [Ulf M�ller] 4102 4103 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 4104 DH_CHECK_P_NOT_SAFE_PRIME. 4105 (Check if this is true? OpenPGP calls them "strong".) 4106 [Ulf M�ller] 4107 4108 *) Merge the functionality of "dh" and "gendh" programs into a new program 4109 "dhparam". The old programs are retained for now but will handle DH keys 4110 (instead of parameters) in future. 4111 [Steve Henson] 4112 4113 *) Make the ciphers, s_server and s_client programs check the return values 4114 when a new cipher list is set. 4115 [Steve Henson] 4116 4117 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 4118 ciphers. Before when the 56bit ciphers were enabled the sorting was 4119 wrong. 4120 4121 The syntax for the cipher sorting has been extended to support sorting by 4122 cipher-strength (using the strength_bits hard coded in the tables). 4123 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 4124 4125 Fix a bug in the cipher-command parser: when supplying a cipher command 4126 string with an "undefined" symbol (neither command nor alphanumeric 4127 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 4128 an error is flagged. 4129 4130 Due to the strength-sorting extension, the code of the 4131 ssl_create_cipher_list() function was completely rearranged. I hope that 4132 the readability was also increased :-) 4133 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 4134 4135 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 4136 for the first serial number and places 2 in the serial number file. This 4137 avoids problems when the root CA is created with serial number zero and 4138 the first user certificate has the same issuer name and serial number 4139 as the root CA. 4140 [Steve Henson] 4141 4142 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 4143 the new code. Add documentation for this stuff. 4144 [Steve Henson] 4145 4146 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 4147 X509_*() to X509at_*() on the grounds that they don't handle X509 4148 structures and behave in an analagous way to the X509v3 functions: 4149 they shouldn't be called directly but wrapper functions should be used 4150 instead. 4151 4152 So we also now have some wrapper functions that call the X509at functions 4153 when passed certificate requests. (TO DO: similar things can be done with 4154 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 4155 things. Some of these need some d2i or i2d and print functionality 4156 because they handle more complex structures.) 4157 [Steve Henson] 4158 4159 *) Add missing #ifndefs that caused missing symbols when building libssl 4160 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 4161 NO_RSA in ssl/s2*.c. 4162 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M�ller] 4163 4164 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 4165 has a return value which indicates the quality of the random data 4166 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 4167 error queue. New function RAND_pseudo_bytes() generates output that is 4168 guaranteed to be unique but not unpredictable. RAND_add is like 4169 RAND_seed, but takes an extra argument for an entropy estimate 4170 (RAND_seed always assumes full entropy). 4171 [Ulf M�ller] 4172 4173 *) Do more iterations of Rabin-Miller probable prime test (specifically, 4174 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 4175 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 4176 in crypto/bn/bn_prime.c for the complete table). This guarantees a 4177 false-positive rate of at most 2^-80 for random input. 4178 [Bodo Moeller] 4179 4180 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 4181 [Bodo Moeller] 4182 4183 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 4184 in the 0.9.5 release), this returns the chain 4185 from an X509_CTX structure with a dup of the stack and all 4186 the X509 reference counts upped: so the stack will exist 4187 after X509_CTX_cleanup() has been called. Modify pkcs12.c 4188 to use this. 4189 4190 Also make SSL_SESSION_print() print out the verify return 4191 code. 4192 [Steve Henson] 4193 4194 *) Add manpage for the pkcs12 command. Also change the default 4195 behaviour so MAC iteration counts are used unless the new 4196 -nomaciter option is used. This improves file security and 4197 only older versions of MSIE (4.0 for example) need it. 4198 [Steve Henson] 4199 4200 *) Honor the no-xxx Configure options when creating .DEF files. 4201 [Ulf M�ller] 4202 4203 *) Add PKCS#10 attributes to field table: challengePassword, 4204 unstructuredName and unstructuredAddress. These are taken from 4205 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 4206 international characters are used. 4207 4208 More changes to X509_ATTRIBUTE code: allow the setting of types 4209 based on strings. Remove the 'loc' parameter when adding 4210 attributes because these will be a SET OF encoding which is sorted 4211 in ASN1 order. 4212 [Steve Henson] 4213 4214 *) Initial changes to the 'req' utility to allow request generation 4215 automation. This will allow an application to just generate a template 4216 file containing all the field values and have req construct the 4217 request. 4218 4219 Initial support for X509_ATTRIBUTE handling. Stacks of these are 4220 used all over the place including certificate requests and PKCS#7 4221 structures. They are currently handled manually where necessary with 4222 some primitive wrappers for PKCS#7. The new functions behave in a 4223 manner analogous to the X509 extension functions: they allow 4224 attributes to be looked up by NID and added. 4225 4226 Later something similar to the X509V3 code would be desirable to 4227 automatically handle the encoding, decoding and printing of the 4228 more complex types. The string types like challengePassword can 4229 be handled by the string table functions. 4230 4231 Also modified the multi byte string table handling. Now there is 4232 a 'global mask' which masks out certain types. The table itself 4233 can use the flag STABLE_NO_MASK to ignore the mask setting: this 4234 is useful when for example there is only one permissible type 4235 (as in countryName) and using the mask might result in no valid 4236 types at all. 4237 [Steve Henson] 4238 4239 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 4240 SSL_get_peer_finished to allow applications to obtain the latest 4241 Finished messages sent to the peer or expected from the peer, 4242 respectively. (SSL_get_peer_finished is usually the Finished message 4243 actually received from the peer, otherwise the protocol will be aborted.) 4244 4245 As the Finished message are message digests of the complete handshake 4246 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 4247 be used for external authentication procedures when the authentication 4248 provided by SSL/TLS is not desired or is not enough. 4249 [Bodo Moeller] 4250 4251 *) Enhanced support for Alpha Linux is added. Now ./config checks if 4252 the host supports BWX extension and if Compaq C is present on the 4253 $PATH. Just exploiting of the BWX extension results in 20-30% 4254 performance kick for some algorithms, e.g. DES and RC4 to mention 4255 a couple. Compaq C in turn generates ~20% faster code for MD5 and 4256 SHA1. 4257 [Andy Polyakov] 4258 4259 *) Add support for MS "fast SGC". This is arguably a violation of the 4260 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 4261 weak crypto and after checking the certificate is SGC a second one 4262 with strong crypto. MS SGC stops the first handshake after receiving 4263 the server certificate message and sends a second client hello. Since 4264 a server will typically do all the time consuming operations before 4265 expecting any further messages from the client (server key exchange 4266 is the most expensive) there is little difference between the two. 4267 4268 To get OpenSSL to support MS SGC we have to permit a second client 4269 hello message after we have sent server done. In addition we have to 4270 reset the MAC if we do get this second client hello. 4271 [Steve Henson] 4272 4273 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 4274 if a DER encoded private key is RSA or DSA traditional format. Changed 4275 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 4276 format DER encoded private key. Newer code should use PKCS#8 format which 4277 has the key type encoded in the ASN1 structure. Added DER private key 4278 support to pkcs8 application. 4279 [Steve Henson] 4280 4281 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 4282 ciphersuites has been selected (as required by the SSL 3/TLS 1 4283 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 4284 is set, we interpret this as a request to violate the specification 4285 (the worst that can happen is a handshake failure, and 'correct' 4286 behaviour would result in a handshake failure anyway). 4287 [Bodo Moeller] 4288 4289 *) In SSL_CTX_add_session, take into account that there might be multiple 4290 SSL_SESSION structures with the same session ID (e.g. when two threads 4291 concurrently obtain them from an external cache). 4292 The internal cache can handle only one SSL_SESSION with a given ID, 4293 so if there's a conflict, we now throw out the old one to achieve 4294 consistency. 4295 [Bodo Moeller] 4296 4297 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 4298 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 4299 some routines that use cipher OIDs: some ciphers do not have OIDs 4300 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 4301 example. 4302 [Steve Henson] 4303 4304 *) Simplify the trust setting structure and code. Now we just have 4305 two sequences of OIDs for trusted and rejected settings. These will 4306 typically have values the same as the extended key usage extension 4307 and any application specific purposes. 4308 4309 The trust checking code now has a default behaviour: it will just 4310 check for an object with the same NID as the passed id. Functions can 4311 be provided to override either the default behaviour or the behaviour 4312 for a given id. SSL client, server and email already have functions 4313 in place for compatibility: they check the NID and also return "trusted" 4314 if the certificate is self signed. 4315 [Steve Henson] 4316 4317 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 4318 traditional format into an EVP_PKEY structure. 4319 [Steve Henson] 4320 4321 *) Add a password callback function PEM_cb() which either prompts for 4322 a password if usr_data is NULL or otherwise assumes it is a null 4323 terminated password. Allow passwords to be passed on command line 4324 environment or config files in a few more utilities. 4325 [Steve Henson] 4326 4327 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 4328 keys. Add some short names for PKCS#8 PBE algorithms and allow them 4329 to be specified on the command line for the pkcs8 and pkcs12 utilities. 4330 Update documentation. 4331 [Steve Henson] 4332 4333 *) Support for ASN1 "NULL" type. This could be handled before by using 4334 ASN1_TYPE but there wasn't any function that would try to read a NULL 4335 and produce an error if it couldn't. For compatibility we also have 4336 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 4337 don't allocate anything because they don't need to. 4338 [Steve Henson] 4339 4340 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 4341 for details. 4342 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 4343 4344 *) Rebuild of the memory allocation routines used by OpenSSL code and 4345 possibly others as well. The purpose is to make an interface that 4346 provide hooks so anyone can build a separate set of allocation and 4347 deallocation routines to be used by OpenSSL, for example memory 4348 pool implementations, or something else, which was previously hard 4349 since Malloc(), Realloc() and Free() were defined as macros having 4350 the values malloc, realloc and free, respectively (except for Win32 4351 compilations). The same is provided for memory debugging code. 4352 OpenSSL already comes with functionality to find memory leaks, but 4353 this gives people a chance to debug other memory problems. 4354 4355 With these changes, a new set of functions and macros have appeared: 4356 4357 CRYPTO_set_mem_debug_functions() [F] 4358 CRYPTO_get_mem_debug_functions() [F] 4359 CRYPTO_dbg_set_options() [F] 4360 CRYPTO_dbg_get_options() [F] 4361 CRYPTO_malloc_debug_init() [M] 4362 4363 The memory debug functions are NULL by default, unless the library 4364 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 4365 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 4366 gives the standard debugging functions that come with OpenSSL) or 4367 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 4368 provided by the library user) must be used. When the standard 4369 debugging functions are used, CRYPTO_dbg_set_options can be used to 4370 request additional information: 4371 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 4372 the CRYPTO_MDEBUG_xxx macro when compiling the library. 4373 4374 Also, things like CRYPTO_set_mem_functions will always give the 4375 expected result (the new set of functions is used for allocation 4376 and deallocation) at all times, regardless of platform and compiler 4377 options. 4378 4379 To finish it up, some functions that were never use in any other 4380 way than through macros have a new API and new semantic: 4381 4382 CRYPTO_dbg_malloc() 4383 CRYPTO_dbg_realloc() 4384 CRYPTO_dbg_free() 4385 4386 All macros of value have retained their old syntax. 4387 [Richard Levitte and Bodo Moeller] 4388 4389 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 4390 ordering of SMIMECapabilities wasn't in "strength order" and there 4391 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 4392 algorithm. 4393 [Steve Henson] 4394 4395 *) Some ASN1 types with illegal zero length encoding (INTEGER, 4396 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 4397 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 4398 4399 *) Merge in my S/MIME library for OpenSSL. This provides a simple 4400 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 4401 functionality to handle multipart/signed properly) and a utility 4402 called 'smime' to call all this stuff. This is based on code I 4403 originally wrote for Celo who have kindly allowed it to be 4404 included in OpenSSL. 4405 [Steve Henson] 4406 4407 *) Add variants des_set_key_checked and des_set_key_unchecked of 4408 des_set_key (aka des_key_sched). Global variable des_check_key 4409 decides which of these is called by des_set_key; this way 4410 des_check_key behaves as it always did, but applications and 4411 the library itself, which was buggy for des_check_key == 1, 4412 have a cleaner way to pick the version they need. 4413 [Bodo Moeller] 4414 4415 *) New function PKCS12_newpass() which changes the password of a 4416 PKCS12 structure. 4417 [Steve Henson] 4418 4419 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 4420 dynamic mix. In both cases the ids can be used as an index into the 4421 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 4422 functions so they accept a list of the field values and the 4423 application doesn't need to directly manipulate the X509_TRUST 4424 structure. 4425 [Steve Henson] 4426 4427 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 4428 need initialising. 4429 [Steve Henson] 4430 4431 *) Modify the way the V3 extension code looks up extensions. This now 4432 works in a similar way to the object code: we have some "standard" 4433 extensions in a static table which is searched with OBJ_bsearch() 4434 and the application can add dynamic ones if needed. The file 4435 crypto/x509v3/ext_dat.h now has the info: this file needs to be 4436 updated whenever a new extension is added to the core code and kept 4437 in ext_nid order. There is a simple program 'tabtest.c' which checks 4438 this. New extensions are not added too often so this file can readily 4439 be maintained manually. 4440 4441 There are two big advantages in doing things this way. The extensions 4442 can be looked up immediately and no longer need to be "added" using 4443 X509V3_add_standard_extensions(): this function now does nothing. 4444 [Side note: I get *lots* of email saying the extension code doesn't 4445 work because people forget to call this function] 4446 Also no dynamic allocation is done unless new extensions are added: 4447 so if we don't add custom extensions there is no need to call 4448 X509V3_EXT_cleanup(). 4449 [Steve Henson] 4450 4451 *) Modify enc utility's salting as follows: make salting the default. Add a 4452 magic header, so unsalted files fail gracefully instead of just decrypting 4453 to garbage. This is because not salting is a big security hole, so people 4454 should be discouraged from doing it. 4455 [Ben Laurie] 4456 4457 *) Fixes and enhancements to the 'x509' utility. It allowed a message 4458 digest to be passed on the command line but it only used this 4459 parameter when signing a certificate. Modified so all relevant 4460 operations are affected by the digest parameter including the 4461 -fingerprint and -x509toreq options. Also -x509toreq choked if a 4462 DSA key was used because it didn't fix the digest. 4463 [Steve Henson] 4464 4465 *) Initial certificate chain verify code. Currently tests the untrusted 4466 certificates for consistency with the verify purpose (which is set 4467 when the X509_STORE_CTX structure is set up) and checks the pathlength. 4468 4469 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 4470 this is because it will reject chains with invalid extensions whereas 4471 every previous version of OpenSSL and SSLeay made no checks at all. 4472 4473 Trust code: checks the root CA for the relevant trust settings. Trust 4474 settings have an initial value consistent with the verify purpose: e.g. 4475 if the verify purpose is for SSL client use it expects the CA to be 4476 trusted for SSL client use. However the default value can be changed to 4477 permit custom trust settings: one example of this would be to only trust 4478 certificates from a specific "secure" set of CAs. 4479 4480 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 4481 which should be used for version portability: especially since the 4482 verify structure is likely to change more often now. 4483 4484 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 4485 to set them. If not set then assume SSL clients will verify SSL servers 4486 and vice versa. 4487 4488 Two new options to the verify program: -untrusted allows a set of 4489 untrusted certificates to be passed in and -purpose which sets the 4490 intended purpose of the certificate. If a purpose is set then the 4491 new chain verify code is used to check extension consistency. 4492 [Steve Henson] 4493 4494 *) Support for the authority information access extension. 4495 [Steve Henson] 4496 4497 *) Modify RSA and DSA PEM read routines to transparently handle 4498 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 4499 public keys in a format compatible with certificate 4500 SubjectPublicKeyInfo structures. Unfortunately there were already 4501 functions called *_PublicKey_* which used various odd formats so 4502 these are retained for compatibility: however the DSA variants were 4503 never in a public release so they have been deleted. Changed dsa/rsa 4504 utilities to handle the new format: note no releases ever handled public 4505 keys so we should be OK. 4506 4507 The primary motivation for this change is to avoid the same fiasco 4508 that dogs private keys: there are several incompatible private key 4509 formats some of which are standard and some OpenSSL specific and 4510 require various evil hacks to allow partial transparent handling and 4511 even then it doesn't work with DER formats. Given the option anything 4512 other than PKCS#8 should be dumped: but the other formats have to 4513 stay in the name of compatibility. 4514 4515 With public keys and the benefit of hindsight one standard format 4516 is used which works with EVP_PKEY, RSA or DSA structures: though 4517 it clearly returns an error if you try to read the wrong kind of key. 4518 4519 Added a -pubkey option to the 'x509' utility to output the public key. 4520 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 4521 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 4522 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 4523 that do the same as the EVP_PKEY_assign_*() except they up the 4524 reference count of the added key (they don't "swallow" the 4525 supplied key). 4526 [Steve Henson] 4527 4528 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 4529 CRLs would fail if the file contained no certificates or no CRLs: 4530 added a new function to read in both types and return the number 4531 read: this means that if none are read it will be an error. The 4532 DER versions of the certificate and CRL reader would always fail 4533 because it isn't possible to mix certificates and CRLs in DER format 4534 without choking one or the other routine. Changed this to just read 4535 a certificate: this is the best we can do. Also modified the code 4536 in apps/verify.c to take notice of return codes: it was previously 4537 attempting to read in certificates from NULL pointers and ignoring 4538 any errors: this is one reason why the cert and CRL reader seemed 4539 to work. It doesn't check return codes from the default certificate 4540 routines: these may well fail if the certificates aren't installed. 4541 [Steve Henson] 4542 4543 *) Code to support otherName option in GeneralName. 4544 [Steve Henson] 4545 4546 *) First update to verify code. Change the verify utility 4547 so it warns if it is passed a self signed certificate: 4548 for consistency with the normal behaviour. X509_verify 4549 has been modified to it will now verify a self signed 4550 certificate if *exactly* the same certificate appears 4551 in the store: it was previously impossible to trust a 4552 single self signed certificate. This means that: 4553 openssl verify ss.pem 4554 now gives a warning about a self signed certificate but 4555 openssl verify -CAfile ss.pem ss.pem 4556 is OK. 4557 [Steve Henson] 4558 4559 *) For servers, store verify_result in SSL_SESSION data structure 4560 (and add it to external session representation). 4561 This is needed when client certificate verifications fails, 4562 but an application-provided verification callback (set by 4563 SSL_CTX_set_cert_verify_callback) allows accepting the session 4564 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 4565 but returns 1): When the session is reused, we have to set 4566 ssl->verify_result to the appropriate error code to avoid 4567 security holes. 4568 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 4569 4570 *) Fix a bug in the new PKCS#7 code: it didn't consider the 4571 case in PKCS7_dataInit() where the signed PKCS7 structure 4572 didn't contain any existing data because it was being created. 4573 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 4574 4575 *) Add a salt to the key derivation routines in enc.c. This 4576 forms the first 8 bytes of the encrypted file. Also add a 4577 -S option to allow a salt to be input on the command line. 4578 [Steve Henson] 4579 4580 *) New function X509_cmp(). Oddly enough there wasn't a function 4581 to compare two certificates. We do this by working out the SHA1 4582 hash and comparing that. X509_cmp() will be needed by the trust 4583 code. 4584 [Steve Henson] 4585 4586 *) SSL_get1_session() is like SSL_get_session(), but increments 4587 the reference count in the SSL_SESSION returned. 4588 [Geoff Thorpe <geoff@eu.c2.net>] 4589 4590 *) Fix for 'req': it was adding a null to request attributes. 4591 Also change the X509_LOOKUP and X509_INFO code to handle 4592 certificate auxiliary information. 4593 [Steve Henson] 4594 4595 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 4596 the 'enc' command. 4597 [Steve Henson] 4598 4599 *) Add the possibility to add extra information to the memory leak 4600 detecting output, to form tracebacks, showing from where each 4601 allocation was originated: CRYPTO_push_info("constant string") adds 4602 the string plus current file name and line number to a per-thread 4603 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 4604 is like calling CYRPTO_pop_info() until the stack is empty. 4605 Also updated memory leak detection code to be multi-thread-safe. 4606 [Richard Levitte] 4607 4608 *) Add options -text and -noout to pkcs7 utility and delete the 4609 encryption options which never did anything. Update docs. 4610 [Steve Henson] 4611 4612 *) Add options to some of the utilities to allow the pass phrase 4613 to be included on either the command line (not recommended on 4614 OSes like Unix) or read from the environment. Update the 4615 manpages and fix a few bugs. 4616 [Steve Henson] 4617 4618 *) Add a few manpages for some of the openssl commands. 4619 [Steve Henson] 4620 4621 *) Fix the -revoke option in ca. It was freeing up memory twice, 4622 leaking and not finding already revoked certificates. 4623 [Steve Henson] 4624 4625 *) Extensive changes to support certificate auxiliary information. 4626 This involves the use of X509_CERT_AUX structure and X509_AUX 4627 functions. An X509_AUX function such as PEM_read_X509_AUX() 4628 can still read in a certificate file in the usual way but it 4629 will also read in any additional "auxiliary information". By 4630 doing things this way a fair degree of compatibility can be 4631 retained: existing certificates can have this information added 4632 using the new 'x509' options. 4633 4634 Current auxiliary information includes an "alias" and some trust 4635 settings. The trust settings will ultimately be used in enhanced 4636 certificate chain verification routines: currently a certificate 4637 can only be trusted if it is self signed and then it is trusted 4638 for all purposes. 4639 [Steve Henson] 4640 4641 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 4642 The problem was that one of the replacement routines had not been working 4643 since SSLeay releases. For now the offending routine has been replaced 4644 with non-optimised assembler. Even so, this now gives around 95% 4645 performance improvement for 1024 bit RSA signs. 4646 [Mark Cox] 4647 4648 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 4649 handling. Most clients have the effective key size in bits equal to 4650 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 4651 A few however don't do this and instead use the size of the decrypted key 4652 to determine the RC2 key length and the AlgorithmIdentifier to determine 4653 the effective key length. In this case the effective key length can still 4654 be 40 bits but the key length can be 168 bits for example. This is fixed 4655 by manually forcing an RC2 key into the EVP_PKEY structure because the 4656 EVP code can't currently handle unusual RC2 key sizes: it always assumes 4657 the key length and effective key length are equal. 4658 [Steve Henson] 4659 4660 *) Add a bunch of functions that should simplify the creation of 4661 X509_NAME structures. Now you should be able to do: 4662 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 4663 and have it automatically work out the correct field type and fill in 4664 the structures. The more adventurous can try: 4665 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 4666 and it will (hopefully) work out the correct multibyte encoding. 4667 [Steve Henson] 4668 4669 *) Change the 'req' utility to use the new field handling and multibyte 4670 copy routines. Before the DN field creation was handled in an ad hoc 4671 way in req, ca, and x509 which was rather broken and didn't support 4672 BMPStrings or UTF8Strings. Since some software doesn't implement 4673 BMPStrings or UTF8Strings yet, they can be enabled using the config file 4674 using the dirstring_type option. See the new comment in the default 4675 openssl.cnf for more info. 4676 [Steve Henson] 4677 4678 *) Make crypto/rand/md_rand.c more robust: 4679 - Assure unique random numbers after fork(). 4680 - Make sure that concurrent threads access the global counter and 4681 md serializably so that we never lose entropy in them 4682 or use exactly the same state in multiple threads. 4683 Access to the large state is not always serializable because 4684 the additional locking could be a performance killer, and 4685 md should be large enough anyway. 4686 [Bodo Moeller] 4687 4688 *) New file apps/app_rand.c with commonly needed functionality 4689 for handling the random seed file. 4690 4691 Use the random seed file in some applications that previously did not: 4692 ca, 4693 dsaparam -genkey (which also ignored its '-rand' option), 4694 s_client, 4695 s_server, 4696 x509 (when signing). 4697 Except on systems with /dev/urandom, it is crucial to have a random 4698 seed file at least for key creation, DSA signing, and for DH exchanges; 4699 for RSA signatures we could do without one. 4700 4701 gendh and gendsa (unlike genrsa) used to read only the first byte 4702 of each file listed in the '-rand' option. The function as previously 4703 found in genrsa is now in app_rand.c and is used by all programs 4704 that support '-rand'. 4705 [Bodo Moeller] 4706 4707 *) In RAND_write_file, use mode 0600 for creating files; 4708 don't just chmod when it may be too late. 4709 [Bodo Moeller] 4710 4711 *) Report an error from X509_STORE_load_locations 4712 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 4713 [Bill Perry] 4714 4715 *) New function ASN1_mbstring_copy() this copies a string in either 4716 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 4717 into an ASN1_STRING type. A mask of permissible types is passed 4718 and it chooses the "minimal" type to use or an error if not type 4719 is suitable. 4720 [Steve Henson] 4721 4722 *) Add function equivalents to the various macros in asn1.h. The old 4723 macros are retained with an M_ prefix. Code inside the library can 4724 use the M_ macros. External code (including the openssl utility) 4725 should *NOT* in order to be "shared library friendly". 4726 [Steve Henson] 4727 4728 *) Add various functions that can check a certificate's extensions 4729 to see if it usable for various purposes such as SSL client, 4730 server or S/MIME and CAs of these types. This is currently 4731 VERY EXPERIMENTAL but will ultimately be used for certificate chain 4732 verification. Also added a -purpose flag to x509 utility to 4733 print out all the purposes. 4734 [Steve Henson] 4735 4736 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 4737 functions. 4738 [Steve Henson] 4739 4740 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 4741 for, obtain and decode and extension and obtain its critical flag. 4742 This allows all the necessary extension code to be handled in a 4743 single function call. 4744 [Steve Henson] 4745 4746 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 4747 platforms. See crypto/rc4/rc4_enc.c for further details. 4748 [Andy Polyakov] 4749 4750 *) New -noout option to asn1parse. This causes no output to be produced 4751 its main use is when combined with -strparse and -out to extract data 4752 from a file (which may not be in ASN.1 format). 4753 [Steve Henson] 4754 4755 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 4756 when producing the local key id. 4757 [Richard Levitte <levitte@stacken.kth.se>] 4758 4759 *) New option -dhparam in s_server. This allows a DH parameter file to be 4760 stated explicitly. If it is not stated then it tries the first server 4761 certificate file. The previous behaviour hard coded the filename 4762 "server.pem". 4763 [Steve Henson] 4764 4765 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 4766 a public key to be input or output. For example: 4767 openssl rsa -in key.pem -pubout -out pubkey.pem 4768 Also added necessary DSA public key functions to handle this. 4769 [Steve Henson] 4770 4771 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 4772 in the message. This was handled by allowing 4773 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 4774 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 4775 4776 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 4777 to the end of the strings whereas this didn't. This would cause problems 4778 if strings read with d2i_ASN1_bytes() were later modified. 4779 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 4780 4781 *) Fix for base64 decode bug. When a base64 bio reads only one line of 4782 data and it contains EOF it will end up returning an error. This is 4783 caused by input 46 bytes long. The cause is due to the way base64 4784 BIOs find the start of base64 encoded data. They do this by trying a 4785 trial decode on each line until they find one that works. When they 4786 do a flag is set and it starts again knowing it can pass all the 4787 data directly through the decoder. Unfortunately it doesn't reset 4788 the context it uses. This means that if EOF is reached an attempt 4789 is made to pass two EOFs through the context and this causes the 4790 resulting error. This can also cause other problems as well. As is 4791 usual with these problems it takes *ages* to find and the fix is 4792 trivial: move one line. 4793 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 4794 4795 *) Ugly workaround to get s_client and s_server working under Windows. The 4796 old code wouldn't work because it needed to select() on sockets and the 4797 tty (for keypresses and to see if data could be written). Win32 only 4798 supports select() on sockets so we select() with a 1s timeout on the 4799 sockets and then see if any characters are waiting to be read, if none 4800 are present then we retry, we also assume we can always write data to 4801 the tty. This isn't nice because the code then blocks until we've 4802 received a complete line of data and it is effectively polling the 4803 keyboard at 1s intervals: however it's quite a bit better than not 4804 working at all :-) A dedicated Windows application might handle this 4805 with an event loop for example. 4806 [Steve Henson] 4807 4808 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 4809 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 4810 will be called when RSA_sign() and RSA_verify() are used. This is useful 4811 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 4812 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 4813 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 4814 This necessitated the support of an extra signature type NID_md5_sha1 4815 for SSL signatures and modifications to the SSL library to use it instead 4816 of calling RSA_public_decrypt() and RSA_private_encrypt(). 4817 [Steve Henson] 4818 4819 *) Add new -verify -CAfile and -CApath options to the crl program, these 4820 will lookup a CRL issuers certificate and verify the signature in a 4821 similar way to the verify program. Tidy up the crl program so it 4822 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 4823 less strict. It will now permit CRL extensions even if it is not 4824 a V2 CRL: this will allow it to tolerate some broken CRLs. 4825 [Steve Henson] 4826 4827 *) Initialize all non-automatic variables each time one of the openssl 4828 sub-programs is started (this is necessary as they may be started 4829 multiple times from the "OpenSSL>" prompt). 4830 [Lennart Bang, Bodo Moeller] 4831 4832 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 4833 removing all other RSA functionality (this is what NO_RSA does). This 4834 is so (for example) those in the US can disable those operations covered 4835 by the RSA patent while allowing storage and parsing of RSA keys and RSA 4836 key generation. 4837 [Steve Henson] 4838 4839 *) Non-copying interface to BIO pairs. 4840 (still largely untested) 4841 [Bodo Moeller] 4842 4843 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 4844 ASCII string. This was handled independently in various places before. 4845 [Steve Henson] 4846 4847 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 4848 UTF8 strings a character at a time. 4849 [Steve Henson] 4850 4851 *) Use client_version from client hello to select the protocol 4852 (s23_srvr.c) and for RSA client key exchange verification 4853 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 4854 [Bodo Moeller] 4855 4856 *) Add various utility functions to handle SPKACs, these were previously 4857 handled by poking round in the structure internals. Added new function 4858 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 4859 print, verify and generate SPKACs. Based on an original idea from 4860 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 4861 [Steve Henson] 4862 4863 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 4864 [Andy Polyakov] 4865 4866 *) Allow the config file extension section to be overwritten on the 4867 command line. Based on an original idea from Massimiliano Pala 4868 <madwolf@comune.modena.it>. The new option is called -extensions 4869 and can be applied to ca, req and x509. Also -reqexts to override 4870 the request extensions in req and -crlexts to override the crl extensions 4871 in ca. 4872 [Steve Henson] 4873 4874 *) Add new feature to the SPKAC handling in ca. Now you can include 4875 the same field multiple times by preceding it by "XXXX." for example: 4876 1.OU="Unit name 1" 4877 2.OU="Unit name 2" 4878 this is the same syntax as used in the req config file. 4879 [Steve Henson] 4880 4881 *) Allow certificate extensions to be added to certificate requests. These 4882 are specified in a 'req_extensions' option of the req section of the 4883 config file. They can be printed out with the -text option to req but 4884 are otherwise ignored at present. 4885 [Steve Henson] 4886 4887 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 4888 data read consists of only the final block it would not decrypted because 4889 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 4890 A misplaced 'break' also meant the decrypted final block might not be 4891 copied until the next read. 4892 [Steve Henson] 4893 4894 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 4895 a few extra parameters to the DH structure: these will be useful if 4896 for example we want the value of 'q' or implement X9.42 DH. 4897 [Steve Henson] 4898 4899 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 4900 provides hooks that allow the default DSA functions or functions on a 4901 "per key" basis to be replaced. This allows hardware acceleration and 4902 hardware key storage to be handled without major modification to the 4903 library. Also added low level modexp hooks and CRYPTO_EX structure and 4904 associated functions. 4905 [Steve Henson] 4906 4907 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 4908 as "read only": it can't be written to and the buffer it points to will 4909 not be freed. Reading from a read only BIO is much more efficient than 4910 a normal memory BIO. This was added because there are several times when 4911 an area of memory needs to be read from a BIO. The previous method was 4912 to create a memory BIO and write the data to it, this results in two 4913 copies of the data and an O(n^2) reading algorithm. There is a new 4914 function BIO_new_mem_buf() which creates a read only memory BIO from 4915 an area of memory. Also modified the PKCS#7 routines to use read only 4916 memory BIOs. 4917 [Steve Henson] 4918 4919 *) Bugfix: ssl23_get_client_hello did not work properly when called in 4920 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 4921 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 4922 but a retry condition occured while trying to read the rest. 4923 [Bodo Moeller] 4924 4925 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 4926 NID_pkcs7_encrypted by default: this was wrong since this should almost 4927 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 4928 the encrypted data type: this is a more sensible place to put it and it 4929 allows the PKCS#12 code to be tidied up that duplicated this 4930 functionality. 4931 [Steve Henson] 4932 4933 *) Changed obj_dat.pl script so it takes its input and output files on 4934 the command line. This should avoid shell escape redirection problems 4935 under Win32. 4936 [Steve Henson] 4937 4938 *) Initial support for certificate extension requests, these are included 4939 in things like Xenroll certificate requests. Included functions to allow 4940 extensions to be obtained and added. 4941 [Steve Henson] 4942 4943 *) -crlf option to s_client and s_server for sending newlines as 4944 CRLF (as required by many protocols). 4945 [Bodo Moeller] 4946 4947 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 4948 4949 *) Install libRSAglue.a when OpenSSL is built with RSAref. 4950 [Ralf S. Engelschall] 4951 4952 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 4953 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 4954 4955 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 4956 program. 4957 [Steve Henson] 4958 4959 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 4960 DH parameters/keys (q is lost during that conversion, but the resulting 4961 DH parameters contain its length). 4962 4963 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 4964 much faster than DH_generate_parameters (which creates parameters 4965 where p = 2*q + 1), and also the smaller q makes DH computations 4966 much more efficient (160-bit exponentiation instead of 1024-bit 4967 exponentiation); so this provides a convenient way to support DHE 4968 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 4969 utter importance to use 4970 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 4971 or 4972 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 4973 when such DH parameters are used, because otherwise small subgroup 4974 attacks may become possible! 4975 [Bodo Moeller] 4976 4977 *) Avoid memory leak in i2d_DHparams. 4978 [Bodo Moeller] 4979 4980 *) Allow the -k option to be used more than once in the enc program: 4981 this allows the same encrypted message to be read by multiple recipients. 4982 [Steve Henson] 4983 4984 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 4985 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 4986 it will always use the numerical form of the OID, even if it has a short 4987 or long name. 4988 [Steve Henson] 4989 4990 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 4991 method only got called if p,q,dmp1,dmq1,iqmp components were present, 4992 otherwise bn_mod_exp was called. In the case of hardware keys for example 4993 no private key components need be present and it might store extra data 4994 in the RSA structure, which cannot be accessed from bn_mod_exp. 4995 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 4996 private key operations. 4997 [Steve Henson] 4998 4999 *) Added support for SPARC Linux. 5000 [Andy Polyakov] 5001 5002 *) pem_password_cb function type incompatibly changed from 5003 typedef int pem_password_cb(char *buf, int size, int rwflag); 5004 to 5005 ....(char *buf, int size, int rwflag, void *userdata); 5006 so that applications can pass data to their callbacks: 5007 The PEM[_ASN1]_{read,write}... functions and macros now take an 5008 additional void * argument, which is just handed through whenever 5009 the password callback is called. 5010 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 5011 5012 New function SSL_CTX_set_default_passwd_cb_userdata. 5013 5014 Compatibility note: As many C implementations push function arguments 5015 onto the stack in reverse order, the new library version is likely to 5016 interoperate with programs that have been compiled with the old 5017 pem_password_cb definition (PEM_whatever takes some data that 5018 happens to be on the stack as its last argument, and the callback 5019 just ignores this garbage); but there is no guarantee whatsoever that 5020 this will work. 5021 5022 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 5023 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 5024 problems not only on Windows, but also on some Unix platforms. 5025 To avoid problematic command lines, these definitions are now in an 5026 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 5027 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 5028 [Bodo Moeller] 5029 5030 *) MIPS III/IV assembler module is reimplemented. 5031 [Andy Polyakov] 5032 5033 *) More DES library cleanups: remove references to srand/rand and 5034 delete an unused file. 5035 [Ulf M�ller] 5036 5037 *) Add support for the the free Netwide assembler (NASM) under Win32, 5038 since not many people have MASM (ml) and it can be hard to obtain. 5039 This is currently experimental but it seems to work OK and pass all 5040 the tests. Check out INSTALL.W32 for info. 5041 [Steve Henson] 5042 5043 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 5044 without temporary keys kept an extra copy of the server key, 5045 and connections with temporary keys did not free everything in case 5046 of an error. 5047 [Bodo Moeller] 5048 5049 *) New function RSA_check_key and new openssl rsa option -check 5050 for verifying the consistency of RSA keys. 5051 [Ulf Moeller, Bodo Moeller] 5052 5053 *) Various changes to make Win32 compile work: 5054 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 5055 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 5056 comparison" warnings. 5057 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 5058 [Steve Henson] 5059 5060 *) Add a debugging option to PKCS#5 v2 key generation function: when 5061 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 5062 derived keys are printed to stderr. 5063 [Steve Henson] 5064 5065 *) Copy the flags in ASN1_STRING_dup(). 5066 [Roman E. Pavlov <pre@mo.msk.ru>] 5067 5068 *) The x509 application mishandled signing requests containing DSA 5069 keys when the signing key was also DSA and the parameters didn't match. 5070 5071 It was supposed to omit the parameters when they matched the signing key: 5072 the verifying software was then supposed to automatically use the CA's 5073 parameters if they were absent from the end user certificate. 5074 5075 Omitting parameters is no longer recommended. The test was also 5076 the wrong way round! This was probably due to unusual behaviour in 5077 EVP_cmp_parameters() which returns 1 if the parameters match. 5078 This meant that parameters were omitted when they *didn't* match and 5079 the certificate was useless. Certificates signed with 'ca' didn't have 5080 this bug. 5081 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 5082 5083 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 5084 The interface is as follows: 5085 Applications can use 5086 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 5087 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 5088 "off" is now the default. 5089 The library internally uses 5090 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 5091 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 5092 to disable memory-checking temporarily. 5093 5094 Some inconsistent states that previously were possible (and were 5095 even the default) are now avoided. 5096 5097 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 5098 with each memory chunk allocated; this is occasionally more helpful 5099 than just having a counter. 5100 5101 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 5102 5103 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 5104 extensions. 5105 [Bodo Moeller] 5106 5107 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 5108 which largely parallels "options", but is for changing API behaviour, 5109 whereas "options" are about protocol behaviour. 5110 Initial "mode" flags are: 5111 5112 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 5113 a single record has been written. 5114 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 5115 retries use the same buffer location. 5116 (But all of the contents must be 5117 copied!) 5118 [Bodo Moeller] 5119 5120 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 5121 worked. 5122 5123 *) Fix problems with no-hmac etc. 5124 [Ulf M�ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 5125 5126 *) New functions RSA_get_default_method(), RSA_set_method() and 5127 RSA_get_method(). These allows replacement of RSA_METHODs without having 5128 to mess around with the internals of an RSA structure. 5129 [Steve Henson] 5130 5131 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 5132 Also really enable memory leak checks in openssl.c and in some 5133 test programs. 5134 [Chad C. Mulligan, Bodo Moeller] 5135 5136 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 5137 up the length of negative integers. This has now been simplified to just 5138 store the length when it is first determined and use it later, rather 5139 than trying to keep track of where data is copied and updating it to 5140 point to the end. 5141 [Steve Henson, reported by Brien Wheeler 5142 <bwheeler@authentica-security.com>] 5143 5144 *) Add a new function PKCS7_signatureVerify. This allows the verification 5145 of a PKCS#7 signature but with the signing certificate passed to the 5146 function itself. This contrasts with PKCS7_dataVerify which assumes the 5147 certificate is present in the PKCS#7 structure. This isn't always the 5148 case: certificates can be omitted from a PKCS#7 structure and be 5149 distributed by "out of band" means (such as a certificate database). 5150 [Steve Henson] 5151 5152 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 5153 function prototypes in pem.h, also change util/mkdef.pl to add the 5154 necessary function names. 5155 [Steve Henson] 5156 5157 *) mk1mf.pl (used by Windows builds) did not properly read the 5158 options set by Configure in the top level Makefile, and Configure 5159 was not even able to write more than one option correctly. 5160 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 5161 [Bodo Moeller] 5162 5163 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 5164 file to be loaded from a BIO or FILE pointer. The BIO version will 5165 for example allow memory BIOs to contain config info. 5166 [Steve Henson] 5167 5168 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 5169 Whoever hopes to achieve shared-library compatibility across versions 5170 must use this, not the compile-time macro. 5171 (Exercise 0.9.4: Which is the minimum library version required by 5172 such programs?) 5173 Note: All this applies only to multi-threaded programs, others don't 5174 need locks. 5175 [Bodo Moeller] 5176 5177 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 5178 through a BIO pair triggered the default case, i.e. 5179 SSLerr(...,SSL_R_UNKNOWN_STATE). 5180 [Bodo Moeller] 5181 5182 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 5183 can use the SSL library even if none of the specific BIOs is 5184 appropriate. 5185 [Bodo Moeller] 5186 5187 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 5188 for the encoded length. 5189 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 5190 5191 *) Add initial documentation of the X509V3 functions. 5192 [Steve Henson] 5193 5194 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 5195 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 5196 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 5197 secure PKCS#8 private key format with a high iteration count. 5198 [Steve Henson] 5199 5200 *) Fix determination of Perl interpreter: A perl or perl5 5201 _directory_ in $PATH was also accepted as the interpreter. 5202 [Ralf S. Engelschall] 5203 5204 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 5205 wrong with it but it was very old and did things like calling 5206 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 5207 unusual formatting. 5208 [Steve Henson] 5209 5210 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 5211 to use the new extension code. 5212 [Steve Henson] 5213 5214 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 5215 with macros. This should make it easier to change their form, add extra 5216 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 5217 constant. 5218 [Steve Henson] 5219 5220 *) Add to configuration table a new entry that can specify an alternative 5221 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 5222 according to Mark Crispin <MRC@Panda.COM>. 5223 [Bodo Moeller] 5224 5225#if 0 5226 *) DES CBC did not update the IV. Weird. 5227 [Ben Laurie] 5228#else 5229 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 5230 Changing the behaviour of the former might break existing programs -- 5231 where IV updating is needed, des_ncbc_encrypt can be used. 5232#endif 5233 5234 *) When bntest is run from "make test" it drives bc to check its 5235 calculations, as well as internally checking them. If an internal check 5236 fails, it needs to cause bc to give a non-zero result or make test carries 5237 on without noticing the failure. Fixed. 5238 [Ben Laurie] 5239 5240 *) DES library cleanups. 5241 [Ulf M�ller] 5242 5243 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 5244 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 5245 ciphers. NOTE: although the key derivation function has been verified 5246 against some published test vectors it has not been extensively tested 5247 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 5248 of v2.0. 5249 [Steve Henson] 5250 5251 *) Instead of "mkdir -p", which is not fully portable, use new 5252 Perl script "util/mkdir-p.pl". 5253 [Bodo Moeller] 5254 5255 *) Rewrite the way password based encryption (PBE) is handled. It used to 5256 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 5257 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 5258 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 5259 the 'parameter' field of the AlgorithmIdentifier is passed to the 5260 underlying key generation function so it must do its own ASN1 parsing. 5261 This has also changed the EVP_PBE_CipherInit() function which now has a 5262 'parameter' argument instead of literal salt and iteration count values 5263 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 5264 [Steve Henson] 5265 5266 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 5267 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 5268 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 5269 KEY" because this clashed with PKCS#8 unencrypted string. Since this 5270 value was just used as a "magic string" and not used directly its 5271 value doesn't matter. 5272 [Steve Henson] 5273 5274 *) Introduce some semblance of const correctness to BN. Shame C doesn't 5275 support mutable. 5276 [Ben Laurie] 5277 5278 *) "linux-sparc64" configuration (ultrapenguin). 5279 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 5280 "linux-sparc" configuration. 5281 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 5282 5283 *) config now generates no-xxx options for missing ciphers. 5284 [Ulf M�ller] 5285 5286 *) Support the EBCDIC character set (work in progress). 5287 File ebcdic.c not yet included because it has a different license. 5288 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 5289 5290 *) Support BS2000/OSD-POSIX. 5291 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 5292 5293 *) Make callbacks for key generation use void * instead of char *. 5294 [Ben Laurie] 5295 5296 *) Make S/MIME samples compile (not yet tested). 5297 [Ben Laurie] 5298 5299 *) Additional typesafe stacks. 5300 [Ben Laurie] 5301 5302 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 5303 [Bodo Moeller] 5304 5305 5306 Changes between 0.9.3 and 0.9.3a [29 May 1999] 5307 5308 *) New configuration variant "sco5-gcc". 5309 5310 *) Updated some demos. 5311 [Sean O Riordain, Wade Scholine] 5312 5313 *) Add missing BIO_free at exit of pkcs12 application. 5314 [Wu Zhigang] 5315 5316 *) Fix memory leak in conf.c. 5317 [Steve Henson] 5318 5319 *) Updates for Win32 to assembler version of MD5. 5320 [Steve Henson] 5321 5322 *) Set #! path to perl in apps/der_chop to where we found it 5323 instead of using a fixed path. 5324 [Bodo Moeller] 5325 5326 *) SHA library changes for irix64-mips4-cc. 5327 [Andy Polyakov] 5328 5329 *) Improvements for VMS support. 5330 [Richard Levitte] 5331 5332 5333 Changes between 0.9.2b and 0.9.3 [24 May 1999] 5334 5335 *) Bignum library bug fix. IRIX 6 passes "make test" now! 5336 This also avoids the problems with SC4.2 and unpatched SC5. 5337 [Andy Polyakov <appro@fy.chalmers.se>] 5338 5339 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 5340 These are required because of the typesafe stack would otherwise break 5341 existing code. If old code used a structure member which used to be STACK 5342 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 5343 sk_num or sk_value it would produce an error because the num, data members 5344 are not present in STACK_OF. Now it just produces a warning. sk_set 5345 replaces the old method of assigning a value to sk_value 5346 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 5347 that does this will no longer work (and should use sk_set instead) but 5348 this could be regarded as a "questionable" behaviour anyway. 5349 [Steve Henson] 5350 5351 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 5352 correctly handle encrypted S/MIME data. 5353 [Steve Henson] 5354 5355 *) Change type of various DES function arguments from des_cblock 5356 (which means, in function argument declarations, pointer to char) 5357 to des_cblock * (meaning pointer to array with 8 char elements), 5358 which allows the compiler to do more typechecking; it was like 5359 that back in SSLeay, but with lots of ugly casts. 5360 5361 Introduce new type const_des_cblock. 5362 [Bodo Moeller] 5363 5364 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 5365 problems: find RecipientInfo structure that matches recipient certificate 5366 and initialise the ASN1 structures properly based on passed cipher. 5367 [Steve Henson] 5368 5369 *) Belatedly make the BN tests actually check the results. 5370 [Ben Laurie] 5371 5372 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 5373 to and from BNs: it was completely broken. New compilation option 5374 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 5375 key elements as negative integers. 5376 [Steve Henson] 5377 5378 *) Reorganize and speed up MD5. 5379 [Andy Polyakov <appro@fy.chalmers.se>] 5380 5381 *) VMS support. 5382 [Richard Levitte <richard@levitte.org>] 5383 5384 *) New option -out to asn1parse to allow the parsed structure to be 5385 output to a file. This is most useful when combined with the -strparse 5386 option to examine the output of things like OCTET STRINGS. 5387 [Steve Henson] 5388 5389 *) Make SSL library a little more fool-proof by not requiring any longer 5390 that SSL_set_{accept,connect}_state be called before 5391 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 5392 in many applications because usually everything *appeared* to work as 5393 intended anyway -- now it really works as intended). 5394 [Bodo Moeller] 5395 5396 *) Move openssl.cnf out of lib/. 5397 [Ulf M�ller] 5398 5399 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 5400 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 5401 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 5402 [Ralf S. Engelschall] 5403 5404 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 5405 handle PKCS#7 enveloped data properly. 5406 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 5407 5408 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 5409 copying pointers. The cert_st handling is changed by this in 5410 various ways (and thus what used to be known as ctx->default_cert 5411 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 5412 any longer when s->cert does not give us what we need). 5413 ssl_cert_instantiate becomes obsolete by this change. 5414 As soon as we've got the new code right (possibly it already is?), 5415 we have solved a couple of bugs of the earlier code where s->cert 5416 was used as if it could not have been shared with other SSL structures. 5417 5418 Note that using the SSL API in certain dirty ways now will result 5419 in different behaviour than observed with earlier library versions: 5420 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 5421 does not influence s as it used to. 5422 5423 In order to clean up things more thoroughly, inside SSL_SESSION 5424 we don't use CERT any longer, but a new structure SESS_CERT 5425 that holds per-session data (if available); currently, this is 5426 the peer's certificate chain and, for clients, the server's certificate 5427 and temporary key. CERT holds only those values that can have 5428 meaningful defaults in an SSL_CTX. 5429 [Bodo Moeller] 5430 5431 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 5432 from the internal representation. Various PKCS#7 fixes: remove some 5433 evil casts and set the enc_dig_alg field properly based on the signing 5434 key type. 5435 [Steve Henson] 5436 5437 *) Allow PKCS#12 password to be set from the command line or the 5438 environment. Let 'ca' get its config file name from the environment 5439 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 5440 and 'x509'). 5441 [Steve Henson] 5442 5443 *) Allow certificate policies extension to use an IA5STRING for the 5444 organization field. This is contrary to the PKIX definition but 5445 VeriSign uses it and IE5 only recognises this form. Document 'x509' 5446 extension option. 5447 [Steve Henson] 5448 5449 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 5450 without disallowing inline assembler and the like for non-pedantic builds. 5451 [Ben Laurie] 5452 5453 *) Support Borland C++ builder. 5454 [Janez Jere <jj@void.si>, modified by Ulf M�ller] 5455 5456 *) Support Mingw32. 5457 [Ulf M�ller] 5458 5459 *) SHA-1 cleanups and performance enhancements. 5460 [Andy Polyakov <appro@fy.chalmers.se>] 5461 5462 *) Sparc v8plus assembler for the bignum library. 5463 [Andy Polyakov <appro@fy.chalmers.se>] 5464 5465 *) Accept any -xxx and +xxx compiler options in Configure. 5466 [Ulf M�ller] 5467 5468 *) Update HPUX configuration. 5469 [Anonymous] 5470 5471 *) Add missing sk_<type>_unshift() function to safestack.h 5472 [Ralf S. Engelschall] 5473 5474 *) New function SSL_CTX_use_certificate_chain_file that sets the 5475 "extra_cert"s in addition to the certificate. (This makes sense 5476 only for "PEM" format files, as chains as a whole are not 5477 DER-encoded.) 5478 [Bodo Moeller] 5479 5480 *) Support verify_depth from the SSL API. 5481 x509_vfy.c had what can be considered an off-by-one-error: 5482 Its depth (which was not part of the external interface) 5483 was actually counting the number of certificates in a chain; 5484 now it really counts the depth. 5485 [Bodo Moeller] 5486 5487 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 5488 instead of X509err, which often resulted in confusing error 5489 messages since the error codes are not globally unique 5490 (e.g. an alleged error in ssl3_accept when a certificate 5491 didn't match the private key). 5492 5493 *) New function SSL_CTX_set_session_id_context that allows to set a default 5494 value (so that you don't need SSL_set_session_id_context for each 5495 connection using the SSL_CTX). 5496 [Bodo Moeller] 5497 5498 *) OAEP decoding bug fix. 5499 [Ulf M�ller] 5500 5501 *) Support INSTALL_PREFIX for package builders, as proposed by 5502 David Harris. 5503 [Bodo Moeller] 5504 5505 *) New Configure options "threads" and "no-threads". For systems 5506 where the proper compiler options are known (currently Solaris 5507 and Linux), "threads" is the default. 5508 [Bodo Moeller] 5509 5510 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 5511 [Bodo Moeller] 5512 5513 *) Install various scripts to $(OPENSSLDIR)/misc, not to 5514 $(INSTALLTOP)/bin -- they shouldn't clutter directories 5515 such as /usr/local/bin. 5516 [Bodo Moeller] 5517 5518 *) "make linux-shared" to build shared libraries. 5519 [Niels Poppe <niels@netbox.org>] 5520 5521 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 5522 [Ulf M�ller] 5523 5524 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 5525 extension adding in x509 utility. 5526 [Steve Henson] 5527 5528 *) Remove NOPROTO sections and error code comments. 5529 [Ulf M�ller] 5530 5531 *) Partial rewrite of the DEF file generator to now parse the ANSI 5532 prototypes. 5533 [Steve Henson] 5534 5535 *) New Configure options --prefix=DIR and --openssldir=DIR. 5536 [Ulf M�ller] 5537 5538 *) Complete rewrite of the error code script(s). It is all now handled 5539 by one script at the top level which handles error code gathering, 5540 header rewriting and C source file generation. It should be much better 5541 than the old method: it now uses a modified version of Ulf's parser to 5542 read the ANSI prototypes in all header files (thus the old K&R definitions 5543 aren't needed for error creation any more) and do a better job of 5544 translating function codes into names. The old 'ASN1 error code imbedded 5545 in a comment' is no longer necessary and it doesn't use .err files which 5546 have now been deleted. Also the error code call doesn't have to appear all 5547 on one line (which resulted in some large lines...). 5548 [Steve Henson] 5549 5550 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 5551 [Bodo Moeller] 5552 5553 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 5554 0 (which usually indicates a closed connection), but continue reading. 5555 [Bodo Moeller] 5556 5557 *) Fix some race conditions. 5558 [Bodo Moeller] 5559 5560 *) Add support for CRL distribution points extension. Add Certificate 5561 Policies and CRL distribution points documentation. 5562 [Steve Henson] 5563 5564 *) Move the autogenerated header file parts to crypto/opensslconf.h. 5565 [Ulf M�ller] 5566 5567 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 5568 8 of keying material. Merlin has also confirmed interop with this fix 5569 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 5570 [Merlin Hughes <merlin@baltimore.ie>] 5571 5572 *) Fix lots of warnings. 5573 [Richard Levitte <levitte@stacken.kth.se>] 5574 5575 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 5576 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 5577 [Richard Levitte <levitte@stacken.kth.se>] 5578 5579 *) Fix problems with sizeof(long) == 8. 5580 [Andy Polyakov <appro@fy.chalmers.se>] 5581 5582 *) Change functions to ANSI C. 5583 [Ulf M�ller] 5584 5585 *) Fix typos in error codes. 5586 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M�ller] 5587 5588 *) Remove defunct assembler files from Configure. 5589 [Ulf M�ller] 5590 5591 *) SPARC v8 assembler BIGNUM implementation. 5592 [Andy Polyakov <appro@fy.chalmers.se>] 5593 5594 *) Support for Certificate Policies extension: both print and set. 5595 Various additions to support the r2i method this uses. 5596 [Steve Henson] 5597 5598 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 5599 return a const string when you are expecting an allocated buffer. 5600 [Ben Laurie] 5601 5602 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 5603 types DirectoryString and DisplayText. 5604 [Steve Henson] 5605 5606 *) Add code to allow r2i extensions to access the configuration database, 5607 add an LHASH database driver and add several ctx helper functions. 5608 [Steve Henson] 5609 5610 *) Fix an evil bug in bn_expand2() which caused various BN functions to 5611 fail when they extended the size of a BIGNUM. 5612 [Steve Henson] 5613 5614 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 5615 support typesafe stack. 5616 [Steve Henson] 5617 5618 *) Fix typo in SSL_[gs]et_options(). 5619 [Nils Frostberg <nils@medcom.se>] 5620 5621 *) Delete various functions and files that belonged to the (now obsolete) 5622 old X509V3 handling code. 5623 [Steve Henson] 5624 5625 *) New Configure option "rsaref". 5626 [Ulf M�ller] 5627 5628 *) Don't auto-generate pem.h. 5629 [Bodo Moeller] 5630 5631 *) Introduce type-safe ASN.1 SETs. 5632 [Ben Laurie] 5633 5634 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 5635 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 5636 5637 *) Introduce type-safe STACKs. This will almost certainly break lots of code 5638 that links with OpenSSL (well at least cause lots of warnings), but fear 5639 not: the conversion is trivial, and it eliminates loads of evil casts. A 5640 few STACKed things have been converted already. Feel free to convert more. 5641 In the fullness of time, I'll do away with the STACK type altogether. 5642 [Ben Laurie] 5643 5644 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 5645 specified in <certfile> by updating the entry in the index.txt file. 5646 This way one no longer has to edit the index.txt file manually for 5647 revoking a certificate. The -revoke option does the gory details now. 5648 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 5649 5650 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 5651 `-text' option at all and this way the `-noout -text' combination was 5652 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 5653 [Ralf S. Engelschall] 5654 5655 *) Make sure a corresponding plain text error message exists for the 5656 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 5657 verify callback function determined that a certificate was revoked. 5658 [Ralf S. Engelschall] 5659 5660 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 5661 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 5662 all available cipers including rc5, which was forgotten until now. 5663 In order to let the testing shell script know which algorithms 5664 are available, a new (up to now undocumented) command 5665 "openssl list-cipher-commands" is used. 5666 [Bodo Moeller] 5667 5668 *) Bugfix: s_client occasionally would sleep in select() when 5669 it should have checked SSL_pending() first. 5670 [Bodo Moeller] 5671 5672 *) New functions DSA_do_sign and DSA_do_verify to provide access to 5673 the raw DSA values prior to ASN.1 encoding. 5674 [Ulf M�ller] 5675 5676 *) Tweaks to Configure 5677 [Niels Poppe <niels@netbox.org>] 5678 5679 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 5680 yet... 5681 [Steve Henson] 5682 5683 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 5684 [Ulf M�ller] 5685 5686 *) New config option to avoid instructions that are illegal on the 80386. 5687 The default code is faster, but requires at least a 486. 5688 [Ulf M�ller] 5689 5690 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 5691 SSL2_SERVER_VERSION (not used at all) macros, which are now the 5692 same as SSL2_VERSION anyway. 5693 [Bodo Moeller] 5694 5695 *) New "-showcerts" option for s_client. 5696 [Bodo Moeller] 5697 5698 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 5699 application. Various cleanups and fixes. 5700 [Steve Henson] 5701 5702 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 5703 modify error routines to work internally. Add error codes and PBE init 5704 to library startup routines. 5705 [Steve Henson] 5706 5707 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 5708 packing functions to asn1 and evp. Changed function names and error 5709 codes along the way. 5710 [Steve Henson] 5711 5712 *) PKCS12 integration: and so it begins... First of several patches to 5713 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 5714 objects to objects.h 5715 [Steve Henson] 5716 5717 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 5718 and display support for Thawte strong extranet extension. 5719 [Steve Henson] 5720 5721 *) Add LinuxPPC support. 5722 [Jeff Dubrule <igor@pobox.org>] 5723 5724 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 5725 bn_div_words in alpha.s. 5726 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 5727 5728 *) Make sure the RSA OAEP test is skipped under -DRSAref because 5729 OAEP isn't supported when OpenSSL is built with RSAref. 5730 [Ulf Moeller <ulf@fitug.de>] 5731 5732 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 5733 so they no longer are missing under -DNOPROTO. 5734 [Soren S. Jorvang <soren@t.dk>] 5735 5736 5737 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 5738 5739 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 5740 doesn't work when the session is reused. Coming soon! 5741 [Ben Laurie] 5742 5743 *) Fix a security hole, that allows sessions to be reused in the wrong 5744 context thus bypassing client cert protection! All software that uses 5745 client certs and session caches in multiple contexts NEEDS PATCHING to 5746 allow session reuse! A fuller solution is in the works. 5747 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 5748 5749 *) Some more source tree cleanups (removed obsolete files 5750 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 5751 permission on "config" script to be executable) and a fix for the INSTALL 5752 document. 5753 [Ulf Moeller <ulf@fitug.de>] 5754 5755 *) Remove some legacy and erroneous uses of malloc, free instead of 5756 Malloc, Free. 5757 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 5758 5759 *) Make rsa_oaep_test return non-zero on error. 5760 [Ulf Moeller <ulf@fitug.de>] 5761 5762 *) Add support for native Solaris shared libraries. Configure 5763 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 5764 if someone would make that last step automatic. 5765 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 5766 5767 *) ctx_size was not built with the right compiler during "make links". Fixed. 5768 [Ben Laurie] 5769 5770 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 5771 except NULL ciphers". This means the default cipher list will no longer 5772 enable NULL ciphers. They need to be specifically enabled e.g. with 5773 the string "DEFAULT:eNULL". 5774 [Steve Henson] 5775 5776 *) Fix to RSA private encryption routines: if p < q then it would 5777 occasionally produce an invalid result. This will only happen with 5778 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 5779 [Steve Henson] 5780 5781 *) Be less restrictive and allow also `perl util/perlpath.pl 5782 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 5783 because this way one can also use an interpreter named `perl5' (which is 5784 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 5785 installed as `perl'). 5786 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 5787 5788 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 5789 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 5790 5791 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 5792 advapi32.lib to Win32 build and change the pem test comparision 5793 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 5794 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 5795 and crypto/des/ede_cbcm_enc.c. 5796 [Steve Henson] 5797 5798 *) DES quad checksum was broken on big-endian architectures. Fixed. 5799 [Ben Laurie] 5800 5801 *) Comment out two functions in bio.h that aren't implemented. Fix up the 5802 Win32 test batch file so it (might) work again. The Win32 test batch file 5803 is horrible: I feel ill.... 5804 [Steve Henson] 5805 5806 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 5807 in e_os.h. Audit of header files to check ANSI and non ANSI 5808 sections: 10 functions were absent from non ANSI section and not exported 5809 from Windows DLLs. Fixed up libeay.num for new functions. 5810 [Steve Henson] 5811 5812 *) Make `openssl version' output lines consistent. 5813 [Ralf S. Engelschall] 5814 5815 *) Fix Win32 symbol export lists for BIO functions: Added 5816 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 5817 to ms/libeay{16,32}.def. 5818 [Ralf S. Engelschall] 5819 5820 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 5821 fine under Unix and passes some trivial tests I've now added. But the 5822 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 5823 added to make sure no one expects that this stuff really works in the 5824 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 5825 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 5826 openssl_bio.xs. 5827 [Ralf S. Engelschall] 5828 5829 *) Fix the generation of two part addresses in perl. 5830 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 5831 5832 *) Add config entry for Linux on MIPS. 5833 [John Tobey <jtobey@channel1.com>] 5834 5835 *) Make links whenever Configure is run, unless we are on Windoze. 5836 [Ben Laurie] 5837 5838 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 5839 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 5840 in CRLs. 5841 [Steve Henson] 5842 5843 *) Add a useful kludge to allow package maintainers to specify compiler and 5844 other platforms details on the command line without having to patch the 5845 Configure script everytime: One now can use ``perl Configure 5846 <id>:<details>'', i.e. platform ids are allowed to have details appended 5847 to them (seperated by colons). This is treated as there would be a static 5848 pre-configured entry in Configure's %table under key <id> with value 5849 <details> and ``perl Configure <id>'' is called. So, when you want to 5850 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 5851 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 5852 now, which overrides the FreeBSD-elf entry on-the-fly. 5853 [Ralf S. Engelschall] 5854 5855 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 5856 [Ben Laurie] 5857 5858 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 5859 on the `perl Configure ...' command line. This way one can compile 5860 OpenSSL libraries with Position Independent Code (PIC) which is needed 5861 for linking it into DSOs. 5862 [Ralf S. Engelschall] 5863 5864 *) Remarkably, export ciphers were totally broken and no-one had noticed! 5865 Fixed. 5866 [Ben Laurie] 5867 5868 *) Cleaned up the LICENSE document: The official contact for any license 5869 questions now is the OpenSSL core team under openssl-core@openssl.org. 5870 And add a paragraph about the dual-license situation to make sure people 5871 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 5872 to the OpenSSL toolkit. 5873 [Ralf S. Engelschall] 5874 5875 *) General source tree makefile cleanups: Made `making xxx in yyy...' 5876 display consistent in the source tree and replaced `/bin/rm' by `rm'. 5877 Additonally cleaned up the `make links' target: Remove unnecessary 5878 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 5879 to speed processing and no longer clutter the display with confusing 5880 stuff. Instead only the actually done links are displayed. 5881 [Ralf S. Engelschall] 5882 5883 *) Permit null encryption ciphersuites, used for authentication only. It used 5884 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 5885 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 5886 encryption. 5887 [Ben Laurie] 5888 5889 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 5890 signed attributes when verifying signatures (this would break them), 5891 the detached data encoding was wrong and public keys obtained using 5892 X509_get_pubkey() weren't freed. 5893 [Steve Henson] 5894 5895 *) Add text documentation for the BUFFER functions. Also added a work around 5896 to a Win95 console bug. This was triggered by the password read stuff: the 5897 last character typed gets carried over to the next fread(). If you were 5898 generating a new cert request using 'req' for example then the last 5899 character of the passphrase would be CR which would then enter the first 5900 field as blank. 5901 [Steve Henson] 5902 5903 *) Added the new `Includes OpenSSL Cryptography Software' button as 5904 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 5905 button and can be used by applications based on OpenSSL to show the 5906 relationship to the OpenSSL project. 5907 [Ralf S. Engelschall] 5908 5909 *) Remove confusing variables in function signatures in files 5910 ssl/ssl_lib.c and ssl/ssl.h. 5911 [Lennart Bong <lob@kulthea.stacken.kth.se>] 5912 5913 *) Don't install bss_file.c under PREFIX/include/ 5914 [Lennart Bong <lob@kulthea.stacken.kth.se>] 5915 5916 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 5917 functions that return function pointers and has support for NT specific 5918 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 5919 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 5920 unsigned to signed types: this was killing the Win32 compile. 5921 [Steve Henson] 5922 5923 *) Add new certificate file to stack functions, 5924 SSL_add_dir_cert_subjects_to_stack() and 5925 SSL_add_file_cert_subjects_to_stack(). These largely supplant 5926 SSL_load_client_CA_file(), and can be used to add multiple certs easily 5927 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 5928 This means that Apache-SSL and similar packages don't have to mess around 5929 to add as many CAs as they want to the preferred list. 5930 [Ben Laurie] 5931 5932 *) Experiment with doxygen documentation. Currently only partially applied to 5933 ssl/ssl_lib.c. 5934 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 5935 openssl.doxy as the configuration file. 5936 [Ben Laurie] 5937 5938 *) Get rid of remaining C++-style comments which strict C compilers hate. 5939 [Ralf S. Engelschall, pointed out by Carlos Amengual] 5940 5941 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 5942 compiled in by default: it has problems with large keys. 5943 [Steve Henson] 5944 5945 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 5946 DH private keys and/or callback functions which directly correspond to 5947 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 5948 is needed for applications which have to configure certificates on a 5949 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 5950 (e.g. s_server). 5951 For the RSA certificate situation is makes no difference, but 5952 for the DSA certificate situation this fixes the "no shared cipher" 5953 problem where the OpenSSL cipher selection procedure failed because the 5954 temporary keys were not overtaken from the context and the API provided 5955 no way to reconfigure them. 5956 The new functions now let applications reconfigure the stuff and they 5957 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 5958 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 5959 non-public-API function ssl_cert_instantiate() is used as a helper 5960 function and also to reduce code redundancy inside ssl_rsa.c. 5961 [Ralf S. Engelschall] 5962 5963 *) Move s_server -dcert and -dkey options out of the undocumented feature 5964 area because they are useful for the DSA situation and should be 5965 recognized by the users. 5966 [Ralf S. Engelschall] 5967 5968 *) Fix the cipher decision scheme for export ciphers: the export bits are 5969 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 5970 SSL_EXP_MASK. So, the original variable has to be used instead of the 5971 already masked variable. 5972 [Richard Levitte <levitte@stacken.kth.se>] 5973 5974 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 5975 [Richard Levitte <levitte@stacken.kth.se>] 5976 5977 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 5978 from `int' to `unsigned int' because it's a length and initialized by 5979 EVP_DigestFinal() which expects an `unsigned int *'. 5980 [Richard Levitte <levitte@stacken.kth.se>] 5981 5982 *) Don't hard-code path to Perl interpreter on shebang line of Configure 5983 script. Instead use the usual Shell->Perl transition trick. 5984 [Ralf S. Engelschall] 5985 5986 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 5987 (in addition to RSA certificates) to match the behaviour of `openssl dsa 5988 -noout -modulus' as it's already the case for `openssl rsa -noout 5989 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 5990 currently the public key is printed (a decision which was already done by 5991 `openssl dsa -modulus' in the past) which serves a similar purpose. 5992 Additionally the NO_RSA no longer completely removes the whole -modulus 5993 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 5994 now, too. 5995 [Ralf S. Engelschall] 5996 5997 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 5998 BIO. See the source (crypto/evp/bio_ok.c) for more info. 5999 [Arne Ansper <arne@ats.cyber.ee>] 6000 6001 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 6002 to be added. Now both 'req' and 'ca' can use new objects defined in the 6003 config file. 6004 [Steve Henson] 6005 6006 *) Add cool BIO that does syslog (or event log on NT). 6007 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 6008 6009 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 6010 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 6011 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 6012 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 6013 [Ben Laurie] 6014 6015 *) Add preliminary config info for new extension code. 6016 [Steve Henson] 6017 6018 *) Make RSA_NO_PADDING really use no padding. 6019 [Ulf Moeller <ulf@fitug.de>] 6020 6021 *) Generate errors when private/public key check is done. 6022 [Ben Laurie] 6023 6024 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 6025 for some CRL extensions and new objects added. 6026 [Steve Henson] 6027 6028 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 6029 key usage extension and fuller support for authority key id. 6030 [Steve Henson] 6031 6032 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 6033 padding method for RSA, which is recommended for new applications in PKCS 6034 #1 v2.0 (RFC 2437, October 1998). 6035 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 6036 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 6037 against Bleichbacher's attack on RSA. 6038 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 6039 Ben Laurie] 6040 6041 *) Updates to the new SSL compression code 6042 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 6043 6044 *) Fix so that the version number in the master secret, when passed 6045 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 6046 (because the server will not accept higher), that the version number 6047 is 0x03,0x01, not 0x03,0x00 6048 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 6049 6050 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 6051 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 6052 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 6053 [Steve Henson] 6054 6055 *) Support for RAW extensions where an arbitrary extension can be 6056 created by including its DER encoding. See apps/openssl.cnf for 6057 an example. 6058 [Steve Henson] 6059 6060 *) Make sure latest Perl versions don't interpret some generated C array 6061 code as Perl array code in the crypto/err/err_genc.pl script. 6062 [Lars Weber <3weber@informatik.uni-hamburg.de>] 6063 6064 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 6065 not many people have the assembler. Various Win32 compilation fixes and 6066 update to the INSTALL.W32 file with (hopefully) more accurate Win32 6067 build instructions. 6068 [Steve Henson] 6069 6070 *) Modify configure script 'Configure' to automatically create crypto/date.h 6071 file under Win32 and also build pem.h from pem.org. New script 6072 util/mkfiles.pl to create the MINFO file on environments that can't do a 6073 'make files': perl util/mkfiles.pl >MINFO should work. 6074 [Steve Henson] 6075 6076 *) Major rework of DES function declarations, in the pursuit of correctness 6077 and purity. As a result, many evil casts evaporated, and some weirdness, 6078 too. You may find this causes warnings in your code. Zapping your evil 6079 casts will probably fix them. Mostly. 6080 [Ben Laurie] 6081 6082 *) Fix for a typo in asn1.h. Bug fix to object creation script 6083 obj_dat.pl. It considered a zero in an object definition to mean 6084 "end of object": none of the objects in objects.h have any zeros 6085 so it wasn't spotted. 6086 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 6087 6088 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 6089 Masking (CBCM). In the absence of test vectors, the best I have been able 6090 to do is check that the decrypt undoes the encrypt, so far. Send me test 6091 vectors if you have them. 6092 [Ben Laurie] 6093 6094 *) Correct calculation of key length for export ciphers (too much space was 6095 allocated for null ciphers). This has not been tested! 6096 [Ben Laurie] 6097 6098 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 6099 message is now correct (it understands "crypto" and "ssl" on its 6100 command line). There is also now an "update" option. This will update 6101 the util/ssleay.num and util/libeay.num files with any new functions. 6102 If you do a: 6103 perl util/mkdef.pl crypto ssl update 6104 it will update them. 6105 [Steve Henson] 6106 6107 *) Overhauled the Perl interface (perl/*): 6108 - ported BN stuff to OpenSSL's different BN library 6109 - made the perl/ source tree CVS-aware 6110 - renamed the package from SSLeay to OpenSSL (the files still contain 6111 their history because I've copied them in the repository) 6112 - removed obsolete files (the test scripts will be replaced 6113 by better Test::Harness variants in the future) 6114 [Ralf S. Engelschall] 6115 6116 *) First cut for a very conservative source tree cleanup: 6117 1. merge various obsolete readme texts into doc/ssleay.txt 6118 where we collect the old documents and readme texts. 6119 2. remove the first part of files where I'm already sure that we no 6120 longer need them because of three reasons: either they are just temporary 6121 files which were left by Eric or they are preserved original files where 6122 I've verified that the diff is also available in the CVS via "cvs diff 6123 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 6124 the crypto/md/ stuff). 6125 [Ralf S. Engelschall] 6126 6127 *) More extension code. Incomplete support for subject and issuer alt 6128 name, issuer and authority key id. Change the i2v function parameters 6129 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 6130 what that's for :-) Fix to ASN1 macro which messed up 6131 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 6132 [Steve Henson] 6133 6134 *) Preliminary support for ENUMERATED type. This is largely copied from the 6135 INTEGER code. 6136 [Steve Henson] 6137 6138 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 6139 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 6140 6141 *) Make sure `make rehash' target really finds the `openssl' program. 6142 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 6143 6144 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 6145 like to hear about it if this slows down other processors. 6146 [Ben Laurie] 6147 6148 *) Add CygWin32 platform information to Configure script. 6149 [Alan Batie <batie@aahz.jf.intel.com>] 6150 6151 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 6152 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 6153 6154 *) New program nseq to manipulate netscape certificate sequences 6155 [Steve Henson] 6156 6157 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 6158 few typos. 6159 [Steve Henson] 6160 6161 *) Fixes to BN code. Previously the default was to define BN_RECURSION 6162 but the BN code had some problems that would cause failures when 6163 doing certificate verification and some other functions. 6164 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 6165 6166 *) Add ASN1 and PEM code to support netscape certificate sequences. 6167 [Steve Henson] 6168 6169 *) Add ASN1 and PEM code to support netscape certificate sequences. 6170 [Steve Henson] 6171 6172 *) Add several PKIX and private extended key usage OIDs. 6173 [Steve Henson] 6174 6175 *) Modify the 'ca' program to handle the new extension code. Modify 6176 openssl.cnf for new extension format, add comments. 6177 [Steve Henson] 6178 6179 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 6180 and add a sample to openssl.cnf so req -x509 now adds appropriate 6181 CA extensions. 6182 [Steve Henson] 6183 6184 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 6185 error code, add initial support to X509_print() and x509 application. 6186 [Steve Henson] 6187 6188 *) Takes a deep breath and start addding X509 V3 extension support code. Add 6189 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 6190 stuff is currently isolated and isn't even compiled yet. 6191 [Steve Henson] 6192 6193 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 6194 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 6195 Removed the versions check from X509 routines when loading extensions: 6196 this allows certain broken certificates that don't set the version 6197 properly to be processed. 6198 [Steve Henson] 6199 6200 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 6201 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 6202 can still be regenerated with "make depend". 6203 [Ben Laurie] 6204 6205 *) Spelling mistake in C version of CAST-128. 6206 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 6207 6208 *) Changes to the error generation code. The perl script err-code.pl 6209 now reads in the old error codes and retains the old numbers, only 6210 adding new ones if necessary. It also only changes the .err files if new 6211 codes are added. The makefiles have been modified to only insert errors 6212 when needed (to avoid needlessly modifying header files). This is done 6213 by only inserting errors if the .err file is newer than the auto generated 6214 C file. To rebuild all the error codes from scratch (the old behaviour) 6215 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 6216 or delete all the .err files. 6217 [Steve Henson] 6218 6219 *) CAST-128 was incorrectly implemented for short keys. The C version has 6220 been fixed, but is untested. The assembler versions are also fixed, but 6221 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 6222 to regenerate it if needed. 6223 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 6224 Hagino <itojun@kame.net>] 6225 6226 *) File was opened incorrectly in randfile.c. 6227 [Ulf M�ller <ulf@fitug.de>] 6228 6229 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 6230 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 6231 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 6232 al: it's just almost always a UTCTime. Note this patch adds new error 6233 codes so do a "make errors" if there are problems. 6234 [Steve Henson] 6235 6236 *) Correct Linux 1 recognition in config. 6237 [Ulf M�ller <ulf@fitug.de>] 6238 6239 *) Remove pointless MD5 hash when using DSA keys in ca. 6240 [Anonymous <nobody@replay.com>] 6241 6242 *) Generate an error if given an empty string as a cert directory. Also 6243 generate an error if handed NULL (previously returned 0 to indicate an 6244 error, but didn't set one). 6245 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 6246 6247 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 6248 [Ben Laurie] 6249 6250 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 6251 parameters. This was causing a warning which killed off the Win32 compile. 6252 [Steve Henson] 6253 6254 *) Remove C++ style comments from crypto/bn/bn_local.h. 6255 [Neil Costigan <neil.costigan@celocom.com>] 6256 6257 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 6258 based on a text string, looking up short and long names and finally 6259 "dot" format. The "dot" format stuff didn't work. Added new function 6260 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 6261 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 6262 OID is not part of the table. 6263 [Steve Henson] 6264 6265 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 6266 X509_LOOKUP_by_alias(). 6267 [Ben Laurie] 6268 6269 *) Sort openssl functions by name. 6270 [Ben Laurie] 6271 6272 *) Get the gendsa program working (hopefully) and add it to app list. Remove 6273 encryption from sample DSA keys (in case anyone is interested the password 6274 was "1234"). 6275 [Steve Henson] 6276 6277 *) Make _all_ *_free functions accept a NULL pointer. 6278 [Frans Heymans <fheymans@isaserver.be>] 6279 6280 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 6281 NULL pointers. 6282 [Anonymous <nobody@replay.com>] 6283 6284 *) s_server should send the CAfile as acceptable CAs, not its own cert. 6285 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 6286 6287 *) Don't blow it for numeric -newkey arguments to apps/req. 6288 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 6289 6290 *) Temp key "for export" tests were wrong in s3_srvr.c. 6291 [Anonymous <nobody@replay.com>] 6292 6293 *) Add prototype for temp key callback functions 6294 SSL_CTX_set_tmp_{rsa,dh}_callback(). 6295 [Ben Laurie] 6296 6297 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 6298 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 6299 [Steve Henson] 6300 6301 *) X509_name_add_entry() freed the wrong thing after an error. 6302 [Arne Ansper <arne@ats.cyber.ee>] 6303 6304 *) rsa_eay.c would attempt to free a NULL context. 6305 [Arne Ansper <arne@ats.cyber.ee>] 6306 6307 *) BIO_s_socket() had a broken should_retry() on Windoze. 6308 [Arne Ansper <arne@ats.cyber.ee>] 6309 6310 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 6311 [Arne Ansper <arne@ats.cyber.ee>] 6312 6313 *) Make sure the already existing X509_STORE->depth variable is initialized 6314 in X509_STORE_new(), but document the fact that this variable is still 6315 unused in the certificate verification process. 6316 [Ralf S. Engelschall] 6317 6318 *) Fix the various library and apps files to free up pkeys obtained from 6319 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 6320 [Steve Henson] 6321 6322 *) Fix reference counting in X509_PUBKEY_get(). This makes 6323 demos/maurice/example2.c work, amongst others, probably. 6324 [Steve Henson and Ben Laurie] 6325 6326 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 6327 `openssl' and second, the shortcut symlinks for the `openssl <command>' 6328 are no longer created. This way we have a single and consistent command 6329 line interface `openssl <command>', similar to `cvs <command>'. 6330 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 6331 6332 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 6333 BIT STRING wrapper always have zero unused bits. 6334 [Steve Henson] 6335 6336 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 6337 [Steve Henson] 6338 6339 *) Make the top-level INSTALL documentation easier to understand. 6340 [Paul Sutton] 6341 6342 *) Makefiles updated to exit if an error occurs in a sub-directory 6343 make (including if user presses ^C) [Paul Sutton] 6344 6345 *) Make Montgomery context stuff explicit in RSA data structure. 6346 [Ben Laurie] 6347 6348 *) Fix build order of pem and err to allow for generated pem.h. 6349 [Ben Laurie] 6350 6351 *) Fix renumbering bug in X509_NAME_delete_entry(). 6352 [Ben Laurie] 6353 6354 *) Enhanced the err-ins.pl script so it makes the error library number 6355 global and can add a library name. This is needed for external ASN1 and 6356 other error libraries. 6357 [Steve Henson] 6358 6359 *) Fixed sk_insert which never worked properly. 6360 [Steve Henson] 6361 6362 *) Fix ASN1 macros so they can handle indefinite length construted 6363 EXPLICIT tags. Some non standard certificates use these: they can now 6364 be read in. 6365 [Steve Henson] 6366 6367 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 6368 into a single doc/ssleay.txt bundle. This way the information is still 6369 preserved but no longer messes up this directory. Now it's new room for 6370 the new set of documenation files. 6371 [Ralf S. Engelschall] 6372 6373 *) SETs were incorrectly DER encoded. This was a major pain, because they 6374 shared code with SEQUENCEs, which aren't coded the same. This means that 6375 almost everything to do with SETs or SEQUENCEs has either changed name or 6376 number of arguments. 6377 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 6378 6379 *) Fix test data to work with the above. 6380 [Ben Laurie] 6381 6382 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 6383 was already fixed by Eric for 0.9.1 it seems. 6384 [Ben Laurie - pointed out by Ulf M�ller <ulf@fitug.de>] 6385 6386 *) Autodetect FreeBSD3. 6387 [Ben Laurie] 6388 6389 *) Fix various bugs in Configure. This affects the following platforms: 6390 nextstep 6391 ncr-scde 6392 unixware-2.0 6393 unixware-2.0-pentium 6394 sco5-cc. 6395 [Ben Laurie] 6396 6397 *) Eliminate generated files from CVS. Reorder tests to regenerate files 6398 before they are needed. 6399 [Ben Laurie] 6400 6401 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 6402 [Ben Laurie] 6403 6404 6405 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 6406 6407 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 6408 changed SSLeay to OpenSSL in version strings. 6409 [Ralf S. Engelschall] 6410 6411 *) Some fixups to the top-level documents. 6412 [Paul Sutton] 6413 6414 *) Fixed the nasty bug where rsaref.h was not found under compile-time 6415 because the symlink to include/ was missing. 6416 [Ralf S. Engelschall] 6417 6418 *) Incorporated the popular no-RSA/DSA-only patches 6419 which allow to compile a RSA-free SSLeay. 6420 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 6421 6422 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 6423 when "ssleay" is still not found. 6424 [Ralf S. Engelschall] 6425 6426 *) Added more platforms to Configure: Cray T3E, HPUX 11, 6427 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 6428 6429 *) Updated the README file. 6430 [Ralf S. Engelschall] 6431 6432 *) Added various .cvsignore files in the CVS repository subdirs 6433 to make a "cvs update" really silent. 6434 [Ralf S. Engelschall] 6435 6436 *) Recompiled the error-definition header files and added 6437 missing symbols to the Win32 linker tables. 6438 [Ralf S. Engelschall] 6439 6440 *) Cleaned up the top-level documents; 6441 o new files: CHANGES and LICENSE 6442 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 6443 o merged COPYRIGHT into LICENSE 6444 o removed obsolete TODO file 6445 o renamed MICROSOFT to INSTALL.W32 6446 [Ralf S. Engelschall] 6447 6448 *) Removed dummy files from the 0.9.1b source tree: 6449 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 6450 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 6451 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 6452 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 6453 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 6454 [Ralf S. Engelschall] 6455 6456 *) Added various platform portability fixes. 6457 [Mark J. Cox] 6458 6459 *) The Genesis of the OpenSSL rpject: 6460 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 6461 Young and Tim J. Hudson created while they were working for C2Net until 6462 summer 1998. 6463 [The OpenSSL Project] 6464 6465 6466 Changes between 0.9.0b and 0.9.1b [not released] 6467 6468 *) Updated a few CA certificates under certs/ 6469 [Eric A. Young] 6470 6471 *) Changed some BIGNUM api stuff. 6472 [Eric A. Young] 6473 6474 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 6475 DGUX x86, Linux Alpha, etc. 6476 [Eric A. Young] 6477 6478 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 6479 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 6480 available). 6481 [Eric A. Young] 6482 6483 *) Add -strparse option to asn1pars program which parses nested 6484 binary structures 6485 [Dr Stephen Henson <shenson@bigfoot.com>] 6486 6487 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 6488 [Eric A. Young] 6489 6490 *) DSA fix for "ca" program. 6491 [Eric A. Young] 6492 6493 *) Added "-genkey" option to "dsaparam" program. 6494 [Eric A. Young] 6495 6496 *) Added RIPE MD160 (rmd160) message digest. 6497 [Eric A. Young] 6498 6499 *) Added -a (all) option to "ssleay version" command. 6500 [Eric A. Young] 6501 6502 *) Added PLATFORM define which is the id given to Configure. 6503 [Eric A. Young] 6504 6505 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 6506 [Eric A. Young] 6507 6508 *) Extended the ASN.1 parser routines. 6509 [Eric A. Young] 6510 6511 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 6512 [Eric A. Young] 6513 6514 *) Added a BN_CTX to the BN library. 6515 [Eric A. Young] 6516 6517 *) Fixed the weak key values in DES library 6518 [Eric A. Young] 6519 6520 *) Changed API in EVP library for cipher aliases. 6521 [Eric A. Young] 6522 6523 *) Added support for RC2/64bit cipher. 6524 [Eric A. Young] 6525 6526 *) Converted the lhash library to the crypto/mem.c functions. 6527 [Eric A. Young] 6528 6529 *) Added more recognized ASN.1 object ids. 6530 [Eric A. Young] 6531 6532 *) Added more RSA padding checks for SSL/TLS. 6533 [Eric A. Young] 6534 6535 *) Added BIO proxy/filter functionality. 6536 [Eric A. Young] 6537 6538 *) Added extra_certs to SSL_CTX which can be used 6539 send extra CA certificates to the client in the CA cert chain sending 6540 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 6541 [Eric A. Young] 6542 6543 *) Now Fortezza is denied in the authentication phase because 6544 this is key exchange mechanism is not supported by SSLeay at all. 6545 [Eric A. Young] 6546 6547 *) Additional PKCS1 checks. 6548 [Eric A. Young] 6549 6550 *) Support the string "TLSv1" for all TLS v1 ciphers. 6551 [Eric A. Young] 6552 6553 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 6554 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 6555 [Eric A. Young] 6556 6557 *) Fixed a few memory leaks. 6558 [Eric A. Young] 6559 6560 *) Fixed various code and comment typos. 6561 [Eric A. Young] 6562 6563 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 6564 bytes sent in the client random. 6565 [Edward Bishop <ebishop@spyglass.com>] 6566 6567