CHANGES revision 348343
1 2 OpenSSL CHANGES 3 _______________ 4 5 This is a high-level summary of the most important changes. 6 For a full list of changes, see the git commit log; for example, 7 https://github.com/openssl/openssl/commits/ and pick the appropriate 8 release branch. 9 10 Changes between 1.0.2r and 1.0.2s [28 May 2019] 11 12 *) Change the default RSA, DSA and DH size to 2048 bit instead of 1024. 13 This changes the size when using the genpkey app when no size is given. It 14 fixes an omission in earlier changes that changed all RSA, DSA and DH 15 generation apps to use 2048 bits by default. 16 [Kurt Roeckx] 17 18 *) Add FIPS support for Android Arm 64-bit 19 20 Support for Android Arm 64-bit was added to the OpenSSL FIPS Object 21 Module in Version 2.0.10. For some reason, the corresponding target 22 'android64-aarch64' was missing OpenSSL 1.0.2, whence it could not be 23 built with FIPS support on Android Arm 64-bit. This omission has been 24 fixed. 25 [Matthias St. Pierre] 26 27 Changes between 1.0.2q and 1.0.2r [26 Feb 2019] 28 29 *) 0-byte record padding oracle 30 31 If an application encounters a fatal protocol error and then calls 32 SSL_shutdown() twice (once to send a close_notify, and once to receive one) 33 then OpenSSL can respond differently to the calling application if a 0 byte 34 record is received with invalid padding compared to if a 0 byte record is 35 received with an invalid MAC. If the application then behaves differently 36 based on that in a way that is detectable to the remote peer, then this 37 amounts to a padding oracle that could be used to decrypt data. 38 39 In order for this to be exploitable "non-stitched" ciphersuites must be in 40 use. Stitched ciphersuites are optimised implementations of certain 41 commonly used ciphersuites. Also the application must call SSL_shutdown() 42 twice even if a protocol error has occurred (applications should not do 43 this but some do anyway). 44 45 This issue was discovered by Juraj Somorovsky, Robert Merget and Nimrod 46 Aviram, with additional investigation by Steven Collison and Andrew 47 Hourselt. It was reported to OpenSSL on 10th December 2018. 48 (CVE-2019-1559) 49 [Matt Caswell] 50 51 *) Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0(). 52 [Richard Levitte] 53 54 Changes between 1.0.2p and 1.0.2q [20 Nov 2018] 55 56 *) Microarchitecture timing vulnerability in ECC scalar multiplication 57 58 OpenSSL ECC scalar multiplication, used in e.g. ECDSA and ECDH, has been 59 shown to be vulnerable to a microarchitecture timing side channel attack. 60 An attacker with sufficient access to mount local timing attacks during 61 ECDSA signature generation could recover the private key. 62 63 This issue was reported to OpenSSL on 26th October 2018 by Alejandro 64 Cabrera Aldaya, Billy Brumley, Sohaib ul Hassan, Cesar Pereida Garcia and 65 Nicola Tuveri. 66 (CVE-2018-5407) 67 [Billy Brumley] 68 69 *) Timing vulnerability in DSA signature generation 70 71 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a 72 timing side channel attack. An attacker could use variations in the signing 73 algorithm to recover the private key. 74 75 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser. 76 (CVE-2018-0734) 77 [Paul Dale] 78 79 *) Resolve a compatibility issue in EC_GROUP handling with the FIPS Object 80 Module, accidentally introduced while backporting security fixes from the 81 development branch and hindering the use of ECC in FIPS mode. 82 [Nicola Tuveri] 83 84 Changes between 1.0.2o and 1.0.2p [14 Aug 2018] 85 86 *) Client DoS due to large DH parameter 87 88 During key agreement in a TLS handshake using a DH(E) based ciphersuite a 89 malicious server can send a very large prime value to the client. This will 90 cause the client to spend an unreasonably long period of time generating a 91 key for this prime resulting in a hang until the client has finished. This 92 could be exploited in a Denial Of Service attack. 93 94 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken 95 (CVE-2018-0732) 96 [Guido Vranken] 97 98 *) Cache timing vulnerability in RSA Key Generation 99 100 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to 101 a cache timing side channel attack. An attacker with sufficient access to 102 mount cache timing attacks during the RSA key generation process could 103 recover the private key. 104 105 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera 106 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia. 107 (CVE-2018-0737) 108 [Billy Brumley] 109 110 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 111 parameter is no longer accepted, as it leads to a corrupt table. NULL 112 pem_str is reserved for alias entries only. 113 [Richard Levitte] 114 115 *) Revert blinding in ECDSA sign and instead make problematic addition 116 length-invariant. Switch even to fixed-length Montgomery multiplication. 117 [Andy Polyakov] 118 119 *) Change generating and checking of primes so that the error rate of not 120 being prime depends on the intended use based on the size of the input. 121 For larger primes this will result in more rounds of Miller-Rabin. 122 The maximal error rate for primes with more than 1080 bits is lowered 123 to 2^-128. 124 [Kurt Roeckx, Annie Yousar] 125 126 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 127 [Kurt Roeckx] 128 129 *) Add blinding to ECDSA and DSA signatures to protect against side channel 130 attacks discovered by Keegan Ryan (NCC Group). 131 [Matt Caswell] 132 133 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 134 now allow empty (zero character) pass phrases. 135 [Richard Levitte] 136 137 *) Certificate time validation (X509_cmp_time) enforces stricter 138 compliance with RFC 5280. Fractional seconds and timezone offsets 139 are no longer allowed. 140 [Emilia K��sper] 141 142 Changes between 1.0.2n and 1.0.2o [27 Mar 2018] 143 144 *) Constructed ASN.1 types with a recursive definition could exceed the stack 145 146 Constructed ASN.1 types with a recursive definition (such as can be found 147 in PKCS7) could eventually exceed the stack given malicious input with 148 excessive recursion. This could result in a Denial Of Service attack. There 149 are no such structures used within SSL/TLS that come from untrusted sources 150 so this is considered safe. 151 152 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz 153 project. 154 (CVE-2018-0739) 155 [Matt Caswell] 156 157 Changes between 1.0.2m and 1.0.2n [7 Dec 2017] 158 159 *) Read/write after SSL object in error state 160 161 OpenSSL 1.0.2 (starting from version 1.0.2b) introduced an "error state" 162 mechanism. The intent was that if a fatal error occurred during a handshake 163 then OpenSSL would move into the error state and would immediately fail if 164 you attempted to continue the handshake. This works as designed for the 165 explicit handshake functions (SSL_do_handshake(), SSL_accept() and 166 SSL_connect()), however due to a bug it does not work correctly if 167 SSL_read() or SSL_write() is called directly. In that scenario, if the 168 handshake fails then a fatal error will be returned in the initial function 169 call. If SSL_read()/SSL_write() is subsequently called by the application 170 for the same SSL object then it will succeed and the data is passed without 171 being decrypted/encrypted directly from the SSL/TLS record layer. 172 173 In order to exploit this issue an application bug would have to be present 174 that resulted in a call to SSL_read()/SSL_write() being issued after having 175 already received a fatal error. 176 177 This issue was reported to OpenSSL by David Benjamin (Google). 178 (CVE-2017-3737) 179 [Matt Caswell] 180 181 *) rsaz_1024_mul_avx2 overflow bug on x86_64 182 183 There is an overflow bug in the AVX2 Montgomery multiplication procedure 184 used in exponentiation with 1024-bit moduli. No EC algorithms are affected. 185 Analysis suggests that attacks against RSA and DSA as a result of this 186 defect would be very difficult to perform and are not believed likely. 187 Attacks against DH1024 are considered just feasible, because most of the 188 work necessary to deduce information about a private key may be performed 189 offline. The amount of resources required for such an attack would be 190 significant. However, for an attack on TLS to be meaningful, the server 191 would have to share the DH1024 private key among multiple clients, which is 192 no longer an option since CVE-2016-0701. 193 194 This only affects processors that support the AVX2 but not ADX extensions 195 like Intel Haswell (4th generation). 196 197 This issue was reported to OpenSSL by David Benjamin (Google). The issue 198 was originally found via the OSS-Fuzz project. 199 (CVE-2017-3738) 200 [Andy Polyakov] 201 202 Changes between 1.0.2l and 1.0.2m [2 Nov 2017] 203 204 *) bn_sqrx8x_internal carry bug on x86_64 205 206 There is a carry propagating bug in the x86_64 Montgomery squaring 207 procedure. No EC algorithms are affected. Analysis suggests that attacks 208 against RSA and DSA as a result of this defect would be very difficult to 209 perform and are not believed likely. Attacks against DH are considered just 210 feasible (although very difficult) because most of the work necessary to 211 deduce information about a private key may be performed offline. The amount 212 of resources required for such an attack would be very significant and 213 likely only accessible to a limited number of attackers. An attacker would 214 additionally need online access to an unpatched system using the target 215 private key in a scenario with persistent DH parameters and a private 216 key that is shared between multiple clients. 217 218 This only affects processors that support the BMI1, BMI2 and ADX extensions 219 like Intel Broadwell (5th generation) and later or AMD Ryzen. 220 221 This issue was reported to OpenSSL by the OSS-Fuzz project. 222 (CVE-2017-3736) 223 [Andy Polyakov] 224 225 *) Malformed X.509 IPAddressFamily could cause OOB read 226 227 If an X.509 certificate has a malformed IPAddressFamily extension, 228 OpenSSL could do a one-byte buffer overread. The most likely result 229 would be an erroneous display of the certificate in text format. 230 231 This issue was reported to OpenSSL by the OSS-Fuzz project. 232 (CVE-2017-3735) 233 [Rich Salz] 234 235 Changes between 1.0.2k and 1.0.2l [25 May 2017] 236 237 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 238 platform rather than 'mingw'. 239 [Richard Levitte] 240 241 Changes between 1.0.2j and 1.0.2k [26 Jan 2017] 242 243 *) Truncated packet could crash via OOB read 244 245 If one side of an SSL/TLS path is running on a 32-bit host and a specific 246 cipher is being used, then a truncated packet can cause that host to 247 perform an out-of-bounds read, usually resulting in a crash. 248 249 This issue was reported to OpenSSL by Robert ��wi��cki of Google. 250 (CVE-2017-3731) 251 [Andy Polyakov] 252 253 *) BN_mod_exp may produce incorrect results on x86_64 254 255 There is a carry propagating bug in the x86_64 Montgomery squaring 256 procedure. No EC algorithms are affected. Analysis suggests that attacks 257 against RSA and DSA as a result of this defect would be very difficult to 258 perform and are not believed likely. Attacks against DH are considered just 259 feasible (although very difficult) because most of the work necessary to 260 deduce information about a private key may be performed offline. The amount 261 of resources required for such an attack would be very significant and 262 likely only accessible to a limited number of attackers. An attacker would 263 additionally need online access to an unpatched system using the target 264 private key in a scenario with persistent DH parameters and a private 265 key that is shared between multiple clients. For example this can occur by 266 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 267 similar to CVE-2015-3193 but must be treated as a separate problem. 268 269 This issue was reported to OpenSSL by the OSS-Fuzz project. 270 (CVE-2017-3732) 271 [Andy Polyakov] 272 273 *) Montgomery multiplication may produce incorrect results 274 275 There is a carry propagating bug in the Broadwell-specific Montgomery 276 multiplication procedure that handles input lengths divisible by, but 277 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 278 and DH private keys are impossible. This is because the subroutine in 279 question is not used in operations with the private key itself and an input 280 of the attacker's direct choice. Otherwise the bug can manifest itself as 281 transient authentication and key negotiation failures or reproducible 282 erroneous outcome of public-key operations with specially crafted input. 283 Among EC algorithms only Brainpool P-512 curves are affected and one 284 presumably can attack ECDH key negotiation. Impact was not analyzed in 285 detail, because pre-requisites for attack are considered unlikely. Namely 286 multiple clients have to choose the curve in question and the server has to 287 share the private key among them, neither of which is default behaviour. 288 Even then only clients that chose the curve will be affected. 289 290 This issue was publicly reported as transient failures and was not 291 initially recognized as a security issue. Thanks to Richard Morgan for 292 providing reproducible case. 293 (CVE-2016-7055) 294 [Andy Polyakov] 295 296 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0 297 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 298 prevent issues where no progress is being made and the peer continually 299 sends unrecognised record types, using up resources processing them. 300 [Matt Caswell] 301 302 Changes between 1.0.2i and 1.0.2j [26 Sep 2016] 303 304 *) Missing CRL sanity check 305 306 A bug fix which included a CRL sanity check was added to OpenSSL 1.1.0 307 but was omitted from OpenSSL 1.0.2i. As a result any attempt to use 308 CRLs in OpenSSL 1.0.2i will crash with a null pointer exception. 309 310 This issue only affects the OpenSSL 1.0.2i 311 (CVE-2016-7052) 312 [Matt Caswell] 313 314 Changes between 1.0.2h and 1.0.2i [22 Sep 2016] 315 316 *) OCSP Status Request extension unbounded memory growth 317 318 A malicious client can send an excessively large OCSP Status Request 319 extension. If that client continually requests renegotiation, sending a 320 large OCSP Status Request extension each time, then there will be unbounded 321 memory growth on the server. This will eventually lead to a Denial Of 322 Service attack through memory exhaustion. Servers with a default 323 configuration are vulnerable even if they do not support OCSP. Builds using 324 the "no-ocsp" build time option are not affected. 325 326 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 327 (CVE-2016-6304) 328 [Matt Caswell] 329 330 *) In order to mitigate the SWEET32 attack, the DES ciphers were moved from 331 HIGH to MEDIUM. 332 333 This issue was reported to OpenSSL Karthikeyan Bhargavan and Gaetan 334 Leurent (INRIA) 335 (CVE-2016-2183) 336 [Rich Salz] 337 338 *) OOB write in MDC2_Update() 339 340 An overflow can occur in MDC2_Update() either if called directly or 341 through the EVP_DigestUpdate() function using MDC2. If an attacker 342 is able to supply very large amounts of input data after a previous 343 call to EVP_EncryptUpdate() with a partial block then a length check 344 can overflow resulting in a heap corruption. 345 346 The amount of data needed is comparable to SIZE_MAX which is impractical 347 on most platforms. 348 349 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 350 (CVE-2016-6303) 351 [Stephen Henson] 352 353 *) Malformed SHA512 ticket DoS 354 355 If a server uses SHA512 for TLS session ticket HMAC it is vulnerable to a 356 DoS attack where a malformed ticket will result in an OOB read which will 357 ultimately crash. 358 359 The use of SHA512 in TLS session tickets is comparatively rare as it requires 360 a custom server callback and ticket lookup mechanism. 361 362 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 363 (CVE-2016-6302) 364 [Stephen Henson] 365 366 *) OOB write in BN_bn2dec() 367 368 The function BN_bn2dec() does not check the return value of BN_div_word(). 369 This can cause an OOB write if an application uses this function with an 370 overly large BIGNUM. This could be a problem if an overly large certificate 371 or CRL is printed out from an untrusted source. TLS is not affected because 372 record limits will reject an oversized certificate before it is parsed. 373 374 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 375 (CVE-2016-2182) 376 [Stephen Henson] 377 378 *) OOB read in TS_OBJ_print_bio() 379 380 The function TS_OBJ_print_bio() misuses OBJ_obj2txt(): the return value is 381 the total length the OID text representation would use and not the amount 382 of data written. This will result in OOB reads when large OIDs are 383 presented. 384 385 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 386 (CVE-2016-2180) 387 [Stephen Henson] 388 389 *) Pointer arithmetic undefined behaviour 390 391 Avoid some undefined pointer arithmetic 392 393 A common idiom in the codebase is to check limits in the following manner: 394 "p + len > limit" 395 396 Where "p" points to some malloc'd data of SIZE bytes and 397 limit == p + SIZE 398 399 "len" here could be from some externally supplied data (e.g. from a TLS 400 message). 401 402 The rules of C pointer arithmetic are such that "p + len" is only well 403 defined where len <= SIZE. Therefore the above idiom is actually 404 undefined behaviour. 405 406 For example this could cause problems if some malloc implementation 407 provides an address for "p" such that "p + len" actually overflows for 408 values of len that are too big and therefore p + len < limit. 409 410 This issue was reported to OpenSSL by Guido Vranken 411 (CVE-2016-2177) 412 [Matt Caswell] 413 414 *) Constant time flag not preserved in DSA signing 415 416 Operations in the DSA signing algorithm should run in constant time in 417 order to avoid side channel attacks. A flaw in the OpenSSL DSA 418 implementation means that a non-constant time codepath is followed for 419 certain operations. This has been demonstrated through a cache-timing 420 attack to be sufficient for an attacker to recover the private DSA key. 421 422 This issue was reported by C��sar Pereida (Aalto University), Billy Brumley 423 (Tampere University of Technology), and Yuval Yarom (The University of 424 Adelaide and NICTA). 425 (CVE-2016-2178) 426 [C��sar Pereida] 427 428 *) DTLS buffered message DoS 429 430 In a DTLS connection where handshake messages are delivered out-of-order 431 those messages that OpenSSL is not yet ready to process will be buffered 432 for later use. Under certain circumstances, a flaw in the logic means that 433 those messages do not get removed from the buffer even though the handshake 434 has been completed. An attacker could force up to approx. 15 messages to 435 remain in the buffer when they are no longer required. These messages will 436 be cleared when the DTLS connection is closed. The default maximum size for 437 a message is 100k. Therefore the attacker could force an additional 1500k 438 to be consumed per connection. By opening many simulataneous connections an 439 attacker could cause a DoS attack through memory exhaustion. 440 441 This issue was reported to OpenSSL by Quan Luo. 442 (CVE-2016-2179) 443 [Matt Caswell] 444 445 *) DTLS replay protection DoS 446 447 A flaw in the DTLS replay attack protection mechanism means that records 448 that arrive for future epochs update the replay protection "window" before 449 the MAC for the record has been validated. This could be exploited by an 450 attacker by sending a record for the next epoch (which does not have to 451 decrypt or have a valid MAC), with a very large sequence number. This means 452 that all subsequent legitimate packets are dropped causing a denial of 453 service for a specific DTLS connection. 454 455 This issue was reported to OpenSSL by the OCAP audit team. 456 (CVE-2016-2181) 457 [Matt Caswell] 458 459 *) Certificate message OOB reads 460 461 In OpenSSL 1.0.2 and earlier some missing message length checks can result 462 in OOB reads of up to 2 bytes beyond an allocated buffer. There is a 463 theoretical DoS risk but this has not been observed in practice on common 464 platforms. 465 466 The messages affected are client certificate, client certificate request 467 and server certificate. As a result the attack can only be performed 468 against a client or a server which enables client authentication. 469 470 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 471 (CVE-2016-6306) 472 [Stephen Henson] 473 474 Changes between 1.0.2g and 1.0.2h [3 May 2016] 475 476 *) Prevent padding oracle in AES-NI CBC MAC check 477 478 A MITM attacker can use a padding oracle attack to decrypt traffic 479 when the connection uses an AES CBC cipher and the server support 480 AES-NI. 481 482 This issue was introduced as part of the fix for Lucky 13 padding 483 attack (CVE-2013-0169). The padding check was rewritten to be in 484 constant time by making sure that always the same bytes are read and 485 compared against either the MAC or padding bytes. But it no longer 486 checked that there was enough data to have both the MAC and padding 487 bytes. 488 489 This issue was reported by Juraj Somorovsky using TLS-Attacker. 490 (CVE-2016-2107) 491 [Kurt Roeckx] 492 493 *) Fix EVP_EncodeUpdate overflow 494 495 An overflow can occur in the EVP_EncodeUpdate() function which is used for 496 Base64 encoding of binary data. If an attacker is able to supply very large 497 amounts of input data then a length check can overflow resulting in a heap 498 corruption. 499 500 Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by 501 the PEM_write_bio* family of functions. These are mainly used within the 502 OpenSSL command line applications, so any application which processes data 503 from an untrusted source and outputs it as a PEM file should be considered 504 vulnerable to this issue. User applications that call these APIs directly 505 with large amounts of untrusted data may also be vulnerable. 506 507 This issue was reported by Guido Vranken. 508 (CVE-2016-2105) 509 [Matt Caswell] 510 511 *) Fix EVP_EncryptUpdate overflow 512 513 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 514 is able to supply very large amounts of input data after a previous call to 515 EVP_EncryptUpdate() with a partial block then a length check can overflow 516 resulting in a heap corruption. Following an analysis of all OpenSSL 517 internal usage of the EVP_EncryptUpdate() function all usage is one of two 518 forms. The first form is where the EVP_EncryptUpdate() call is known to be 519 the first called function after an EVP_EncryptInit(), and therefore that 520 specific call must be safe. The second form is where the length passed to 521 EVP_EncryptUpdate() can be seen from the code to be some small value and 522 therefore there is no possibility of an overflow. Since all instances are 523 one of these two forms, it is believed that there can be no overflows in 524 internal code due to this problem. It should be noted that 525 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 526 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 527 of these calls have also been analysed too and it is believed there are no 528 instances in internal usage where an overflow could occur. 529 530 This issue was reported by Guido Vranken. 531 (CVE-2016-2106) 532 [Matt Caswell] 533 534 *) Prevent ASN.1 BIO excessive memory allocation 535 536 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 537 a short invalid encoding can casuse allocation of large amounts of memory 538 potentially consuming excessive resources or exhausting memory. 539 540 Any application parsing untrusted data through d2i BIO functions is 541 affected. The memory based functions such as d2i_X509() are *not* affected. 542 Since the memory based functions are used by the TLS library, TLS 543 applications are not affected. 544 545 This issue was reported by Brian Carpenter. 546 (CVE-2016-2109) 547 [Stephen Henson] 548 549 *) EBCDIC overread 550 551 ASN1 Strings that are over 1024 bytes can cause an overread in applications 552 using the X509_NAME_oneline() function on EBCDIC systems. This could result 553 in arbitrary stack data being returned in the buffer. 554 555 This issue was reported by Guido Vranken. 556 (CVE-2016-2176) 557 [Matt Caswell] 558 559 *) Modify behavior of ALPN to invoke callback after SNI/servername 560 callback, such that updates to the SSL_CTX affect ALPN. 561 [Todd Short] 562 563 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the 564 default. 565 [Kurt Roeckx] 566 567 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the 568 methods are enabled and ssl2 is disabled the methods return NULL. 569 [Kurt Roeckx] 570 571 Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 572 573 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 574 Builds that are not configured with "enable-weak-ssl-ciphers" will not 575 provide any "EXPORT" or "LOW" strength ciphers. 576 [Viktor Dukhovni] 577 578 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 579 is by default disabled at build-time. Builds that are not configured with 580 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 581 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 582 will need to explicitly call either of: 583 584 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 585 or 586 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 587 588 as appropriate. Even if either of those is used, or the application 589 explicitly uses the version-specific SSLv2_method() or its client and 590 server variants, SSLv2 ciphers vulnerable to exhaustive search key 591 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 592 ciphers, and SSLv2 56-bit DES are no longer available. 593 (CVE-2016-0800) 594 [Viktor Dukhovni] 595 596 *) Fix a double-free in DSA code 597 598 A double free bug was discovered when OpenSSL parses malformed DSA private 599 keys and could lead to a DoS attack or memory corruption for applications 600 that receive DSA private keys from untrusted sources. This scenario is 601 considered rare. 602 603 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 604 libFuzzer. 605 (CVE-2016-0705) 606 [Stephen Henson] 607 608 *) Disable SRP fake user seed to address a server memory leak. 609 610 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 611 612 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 613 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 614 was changed to ignore the "fake user" SRP seed, even if the seed 615 is configured. 616 617 Users should use SRP_VBASE_get1_by_user instead. Note that in 618 SRP_VBASE_get1_by_user, caller must free the returned value. Note 619 also that even though configuring the SRP seed attempts to hide 620 invalid usernames by continuing the handshake with fake 621 credentials, this behaviour is not constant time and no strong 622 guarantees are made that the handshake is indistinguishable from 623 that of a valid user. 624 (CVE-2016-0798) 625 [Emilia K��sper] 626 627 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 628 629 In the BN_hex2bn function the number of hex digits is calculated using an 630 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 631 large values of |i| this can result in |bn_expand| not allocating any 632 memory because |i * 4| is negative. This can leave the internal BIGNUM data 633 field as NULL leading to a subsequent NULL ptr deref. For very large values 634 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 635 In this case memory is allocated to the internal BIGNUM data field, but it 636 is insufficiently sized leading to heap corruption. A similar issue exists 637 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 638 is ever called by user applications with very large untrusted hex/dec data. 639 This is anticipated to be a rare occurrence. 640 641 All OpenSSL internal usage of these functions use data that is not expected 642 to be untrusted, e.g. config file data or application command line 643 arguments. If user developed applications generate config file data based 644 on untrusted data then it is possible that this could also lead to security 645 consequences. This is also anticipated to be rare. 646 647 This issue was reported to OpenSSL by Guido Vranken. 648 (CVE-2016-0797) 649 [Matt Caswell] 650 651 *) Fix memory issues in BIO_*printf functions 652 653 The internal |fmtstr| function used in processing a "%s" format string in 654 the BIO_*printf functions could overflow while calculating the length of a 655 string and cause an OOB read when printing very long strings. 656 657 Additionally the internal |doapr_outch| function can attempt to write to an 658 OOB memory location (at an offset from the NULL pointer) in the event of a 659 memory allocation failure. In 1.0.2 and below this could be caused where 660 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 661 could be in processing a very long "%s" format string. Memory leaks can 662 also occur. 663 664 The first issue may mask the second issue dependent on compiler behaviour. 665 These problems could enable attacks where large amounts of untrusted data 666 is passed to the BIO_*printf functions. If applications use these functions 667 in this way then they could be vulnerable. OpenSSL itself uses these 668 functions when printing out human-readable dumps of ASN.1 data. Therefore 669 applications that print this data could be vulnerable if the data is from 670 untrusted sources. OpenSSL command line applications could also be 671 vulnerable where they print out ASN.1 data, or if untrusted data is passed 672 as command line arguments. 673 674 Libssl is not considered directly vulnerable. Additionally certificates etc 675 received via remote connections via libssl are also unlikely to be able to 676 trigger these issues because of message size limits enforced within libssl. 677 678 This issue was reported to OpenSSL Guido Vranken. 679 (CVE-2016-0799) 680 [Matt Caswell] 681 682 *) Side channel attack on modular exponentiation 683 684 A side-channel attack was found which makes use of cache-bank conflicts on 685 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 686 of RSA keys. The ability to exploit this issue is limited as it relies on 687 an attacker who has control of code in a thread running on the same 688 hyper-threaded core as the victim thread which is performing decryptions. 689 690 This issue was reported to OpenSSL by Yuval Yarom, The University of 691 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 692 Nadia Heninger, University of Pennsylvania with more information at 693 http://cachebleed.info. 694 (CVE-2016-0702) 695 [Andy Polyakov] 696 697 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 698 if no keysize is specified with default_bits. This fixes an 699 omission in an earlier change that changed all RSA/DSA key generation 700 apps to use 2048 bits by default. 701 [Emilia K��sper] 702 703 Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 704 705 *) DH small subgroups 706 707 Historically OpenSSL only ever generated DH parameters based on "safe" 708 primes. More recently (in version 1.0.2) support was provided for 709 generating X9.42 style parameter files such as those required for RFC 5114 710 support. The primes used in such files may not be "safe". Where an 711 application is using DH configured with parameters based on primes that are 712 not "safe" then an attacker could use this fact to find a peer's private 713 DH exponent. This attack requires that the attacker complete multiple 714 handshakes in which the peer uses the same private DH exponent. For example 715 this could be used to discover a TLS server's private DH exponent if it's 716 reusing the private DH exponent or it's using a static DH ciphersuite. 717 718 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 719 TLS. It is not on by default. If the option is not set then the server 720 reuses the same private DH exponent for the life of the server process and 721 would be vulnerable to this attack. It is believed that many popular 722 applications do set this option and would therefore not be at risk. 723 724 The fix for this issue adds an additional check where a "q" parameter is 725 available (as is the case in X9.42 based parameters). This detects the 726 only known attack, and is the only possible defense for static DH 727 ciphersuites. This could have some performance impact. 728 729 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 730 default and cannot be disabled. This could have some performance impact. 731 732 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 733 (CVE-2016-0701) 734 [Matt Caswell] 735 736 *) SSLv2 doesn't block disabled ciphers 737 738 A malicious client can negotiate SSLv2 ciphers that have been disabled on 739 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 740 been disabled, provided that the SSLv2 protocol was not also disabled via 741 SSL_OP_NO_SSLv2. 742 743 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 744 and Sebastian Schinzel. 745 (CVE-2015-3197) 746 [Viktor Dukhovni] 747 748 *) Reject DH handshakes with parameters shorter than 1024 bits. 749 [Kurt Roeckx] 750 751 Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 752 753 *) BN_mod_exp may produce incorrect results on x86_64 754 755 There is a carry propagating bug in the x86_64 Montgomery squaring 756 procedure. No EC algorithms are affected. Analysis suggests that attacks 757 against RSA and DSA as a result of this defect would be very difficult to 758 perform and are not believed likely. Attacks against DH are considered just 759 feasible (although very difficult) because most of the work necessary to 760 deduce information about a private key may be performed offline. The amount 761 of resources required for such an attack would be very significant and 762 likely only accessible to a limited number of attackers. An attacker would 763 additionally need online access to an unpatched system using the target 764 private key in a scenario with persistent DH parameters and a private 765 key that is shared between multiple clients. For example this can occur by 766 default in OpenSSL DHE based SSL/TLS ciphersuites. 767 768 This issue was reported to OpenSSL by Hanno B��ck. 769 (CVE-2015-3193) 770 [Andy Polyakov] 771 772 *) Certificate verify crash with missing PSS parameter 773 774 The signature verification routines will crash with a NULL pointer 775 dereference if presented with an ASN.1 signature using the RSA PSS 776 algorithm and absent mask generation function parameter. Since these 777 routines are used to verify certificate signature algorithms this can be 778 used to crash any certificate verification operation and exploited in a 779 DoS attack. Any application which performs certificate verification is 780 vulnerable including OpenSSL clients and servers which enable client 781 authentication. 782 783 This issue was reported to OpenSSL by Lo��c Jonas Etienne (Qnective AG). 784 (CVE-2015-3194) 785 [Stephen Henson] 786 787 *) X509_ATTRIBUTE memory leak 788 789 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 790 memory. This structure is used by the PKCS#7 and CMS routines so any 791 application which reads PKCS#7 or CMS data from untrusted sources is 792 affected. SSL/TLS is not affected. 793 794 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 795 libFuzzer. 796 (CVE-2015-3195) 797 [Stephen Henson] 798 799 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 800 This changes the decoding behaviour for some invalid messages, 801 though the change is mostly in the more lenient direction, and 802 legacy behaviour is preserved as much as possible. 803 [Emilia K��sper] 804 805 *) In DSA_generate_parameters_ex, if the provided seed is too short, 806 use a random seed, as already documented. 807 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 808 809 Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 810 811 *) Alternate chains certificate forgery 812 813 During certificate verfification, OpenSSL will attempt to find an 814 alternative certificate chain if the first attempt to build such a chain 815 fails. An error in the implementation of this logic can mean that an 816 attacker could cause certain checks on untrusted certificates to be 817 bypassed, such as the CA flag, enabling them to use a valid leaf 818 certificate to act as a CA and "issue" an invalid certificate. 819 820 This issue was reported to OpenSSL by Adam Langley/David Benjamin 821 (Google/BoringSSL). 822 (CVE-2015-1793) 823 [Matt Caswell] 824 825 *) Race condition handling PSK identify hint 826 827 If PSK identity hints are received by a multi-threaded client then 828 the values are wrongly updated in the parent SSL_CTX structure. This can 829 result in a race condition potentially leading to a double free of the 830 identify hint data. 831 (CVE-2015-3196) 832 [Stephen Henson] 833 834 Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 835 836 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 837 incompatibility in the handling of HMAC. The previous ABI has now been 838 restored. 839 840 Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 841 842 *) Malformed ECParameters causes infinite loop 843 844 When processing an ECParameters structure OpenSSL enters an infinite loop 845 if the curve specified is over a specially malformed binary polynomial 846 field. 847 848 This can be used to perform denial of service against any 849 system which processes public keys, certificate requests or 850 certificates. This includes TLS clients and TLS servers with 851 client authentication enabled. 852 853 This issue was reported to OpenSSL by Joseph Barr-Pixton. 854 (CVE-2015-1788) 855 [Andy Polyakov] 856 857 *) Exploitable out-of-bounds read in X509_cmp_time 858 859 X509_cmp_time does not properly check the length of the ASN1_TIME 860 string and can read a few bytes out of bounds. In addition, 861 X509_cmp_time accepts an arbitrary number of fractional seconds in the 862 time string. 863 864 An attacker can use this to craft malformed certificates and CRLs of 865 various sizes and potentially cause a segmentation fault, resulting in 866 a DoS on applications that verify certificates or CRLs. TLS clients 867 that verify CRLs are affected. TLS clients and servers with client 868 authentication enabled may be affected if they use custom verification 869 callbacks. 870 871 This issue was reported to OpenSSL by Robert Swiecki (Google), and 872 independently by Hanno B��ck. 873 (CVE-2015-1789) 874 [Emilia K��sper] 875 876 *) PKCS7 crash with missing EnvelopedContent 877 878 The PKCS#7 parsing code does not handle missing inner EncryptedContent 879 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 880 with missing content and trigger a NULL pointer dereference on parsing. 881 882 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 883 structures from untrusted sources are affected. OpenSSL clients and 884 servers are not affected. 885 886 This issue was reported to OpenSSL by Michal Zalewski (Google). 887 (CVE-2015-1790) 888 [Emilia K��sper] 889 890 *) CMS verify infinite loop with unknown hash function 891 892 When verifying a signedData message the CMS code can enter an infinite loop 893 if presented with an unknown hash function OID. This can be used to perform 894 denial of service against any system which verifies signedData messages using 895 the CMS code. 896 This issue was reported to OpenSSL by Johannes Bauer. 897 (CVE-2015-1792) 898 [Stephen Henson] 899 900 *) Race condition handling NewSessionTicket 901 902 If a NewSessionTicket is received by a multi-threaded client when attempting to 903 reuse a previous ticket then a race condition can occur potentially leading to 904 a double free of the ticket data. 905 (CVE-2015-1791) 906 [Matt Caswell] 907 908 *) Removed support for the two export grade static DH ciphersuites 909 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 910 were newly added (along with a number of other static DH ciphersuites) to 911 1.0.2. However the two export ones have *never* worked since they were 912 introduced. It seems strange in any case to be adding new export 913 ciphersuites, and given "logjam" it also does not seem correct to fix them. 914 [Matt Caswell] 915 916 *) Only support 256-bit or stronger elliptic curves with the 917 'ecdh_auto' setting (server) or by default (client). Of supported 918 curves, prefer P-256 (both). 919 [Emilia Kasper] 920 921 *) Reject DH handshakes with parameters shorter than 768 bits. 922 [Kurt Roeckx and Emilia Kasper] 923 924 Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 925 926 *) ClientHello sigalgs DoS fix 927 928 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 929 invalid signature algorithms extension a NULL pointer dereference will 930 occur. This can be exploited in a DoS attack against the server. 931 932 This issue was was reported to OpenSSL by David Ramos of Stanford 933 University. 934 (CVE-2015-0291) 935 [Stephen Henson and Matt Caswell] 936 937 *) Multiblock corrupted pointer fix 938 939 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 940 feature only applies on 64 bit x86 architecture platforms that support AES 941 NI instructions. A defect in the implementation of "multiblock" can cause 942 OpenSSL's internal write buffer to become incorrectly set to NULL when 943 using non-blocking IO. Typically, when the user application is using a 944 socket BIO for writing, this will only result in a failed connection. 945 However if some other BIO is used then it is likely that a segmentation 946 fault will be triggered, thus enabling a potential DoS attack. 947 948 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 949 (CVE-2015-0290) 950 [Matt Caswell] 951 952 *) Segmentation fault in DTLSv1_listen fix 953 954 The DTLSv1_listen function is intended to be stateless and processes the 955 initial ClientHello from many peers. It is common for user code to loop 956 over the call to DTLSv1_listen until a valid ClientHello is received with 957 an associated cookie. A defect in the implementation of DTLSv1_listen means 958 that state is preserved in the SSL object from one invocation to the next 959 that can lead to a segmentation fault. Errors processing the initial 960 ClientHello can trigger this scenario. An example of such an error could be 961 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 962 server. 963 964 This issue was reported to OpenSSL by Per Allansson. 965 (CVE-2015-0207) 966 [Matt Caswell] 967 968 *) Segmentation fault in ASN1_TYPE_cmp fix 969 970 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 971 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 972 certificate signature algorithm consistency this can be used to crash any 973 certificate verification operation and exploited in a DoS attack. Any 974 application which performs certificate verification is vulnerable including 975 OpenSSL clients and servers which enable client authentication. 976 (CVE-2015-0286) 977 [Stephen Henson] 978 979 *) Segmentation fault for invalid PSS parameters fix 980 981 The signature verification routines will crash with a NULL pointer 982 dereference if presented with an ASN.1 signature using the RSA PSS 983 algorithm and invalid parameters. Since these routines are used to verify 984 certificate signature algorithms this can be used to crash any 985 certificate verification operation and exploited in a DoS attack. Any 986 application which performs certificate verification is vulnerable including 987 OpenSSL clients and servers which enable client authentication. 988 989 This issue was was reported to OpenSSL by Brian Carpenter. 990 (CVE-2015-0208) 991 [Stephen Henson] 992 993 *) ASN.1 structure reuse memory corruption fix 994 995 Reusing a structure in ASN.1 parsing may allow an attacker to cause 996 memory corruption via an invalid write. Such reuse is and has been 997 strongly discouraged and is believed to be rare. 998 999 Applications that parse structures containing CHOICE or ANY DEFINED BY 1000 components may be affected. Certificate parsing (d2i_X509 and related 1001 functions) are however not affected. OpenSSL clients and servers are 1002 not affected. 1003 (CVE-2015-0287) 1004 [Stephen Henson] 1005 1006 *) PKCS7 NULL pointer dereferences fix 1007 1008 The PKCS#7 parsing code does not handle missing outer ContentInfo 1009 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 1010 missing content and trigger a NULL pointer dereference on parsing. 1011 1012 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 1013 otherwise parse PKCS#7 structures from untrusted sources are 1014 affected. OpenSSL clients and servers are not affected. 1015 1016 This issue was reported to OpenSSL by Michal Zalewski (Google). 1017 (CVE-2015-0289) 1018 [Emilia K��sper] 1019 1020 *) DoS via reachable assert in SSLv2 servers fix 1021 1022 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 1023 servers that both support SSLv2 and enable export cipher suites by sending 1024 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 1025 1026 This issue was discovered by Sean Burford (Google) and Emilia K��sper 1027 (OpenSSL development team). 1028 (CVE-2015-0293) 1029 [Emilia K��sper] 1030 1031 *) Empty CKE with client auth and DHE fix 1032 1033 If client auth is used then a server can seg fault in the event of a DHE 1034 ciphersuite being selected and a zero length ClientKeyExchange message 1035 being sent by the client. This could be exploited in a DoS attack. 1036 (CVE-2015-1787) 1037 [Matt Caswell] 1038 1039 *) Handshake with unseeded PRNG fix 1040 1041 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 1042 with an unseeded PRNG. The conditions are: 1043 - The client is on a platform where the PRNG has not been seeded 1044 automatically, and the user has not seeded manually 1045 - A protocol specific client method version has been used (i.e. not 1046 SSL_client_methodv23) 1047 - A ciphersuite is used that does not require additional random data from 1048 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 1049 1050 If the handshake succeeds then the client random that has been used will 1051 have been generated from a PRNG with insufficient entropy and therefore the 1052 output may be predictable. 1053 1054 For example using the following command with an unseeded openssl will 1055 succeed on an unpatched platform: 1056 1057 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 1058 (CVE-2015-0285) 1059 [Matt Caswell] 1060 1061 *) Use After Free following d2i_ECPrivatekey error fix 1062 1063 A malformed EC private key file consumed via the d2i_ECPrivateKey function 1064 could cause a use after free condition. This, in turn, could cause a double 1065 free in several private key parsing functions (such as d2i_PrivateKey 1066 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 1067 for applications that receive EC private keys from untrusted 1068 sources. This scenario is considered rare. 1069 1070 This issue was discovered by the BoringSSL project and fixed in their 1071 commit 517073cd4b. 1072 (CVE-2015-0209) 1073 [Matt Caswell] 1074 1075 *) X509_to_X509_REQ NULL pointer deref fix 1076 1077 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 1078 the certificate key is invalid. This function is rarely used in practice. 1079 1080 This issue was discovered by Brian Carpenter. 1081 (CVE-2015-0288) 1082 [Stephen Henson] 1083 1084 *) Removed the export ciphers from the DEFAULT ciphers 1085 [Kurt Roeckx] 1086 1087 Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 1088 1089 *) Change RSA and DH/DSA key generation apps to generate 2048-bit 1090 keys by default. 1091 [Kurt Roeckx] 1092 1093 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 1094 ARMv5 through ARMv8, as opposite to "locking" it to single one. 1095 So far those who have to target multiple plaforms would compromise 1096 and argue that binary targeting say ARMv5 would still execute on 1097 ARMv8. "Universal" build resolves this compromise by providing 1098 near-optimal performance even on newer platforms. 1099 [Andy Polyakov] 1100 1101 *) Accelerated NIST P-256 elliptic curve implementation for x86_64 1102 (other platforms pending). 1103 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov] 1104 1105 *) Add support for the SignedCertificateTimestampList certificate and 1106 OCSP response extensions from RFC6962. 1107 [Rob Stradling] 1108 1109 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 1110 for corner cases. (Certain input points at infinity could lead to 1111 bogus results, with non-infinity inputs mapped to infinity too.) 1112 [Bodo Moeller] 1113 1114 *) Initial support for PowerISA 2.0.7, first implemented in POWER8. 1115 This covers AES, SHA256/512 and GHASH. "Initial" means that most 1116 common cases are optimized and there still is room for further 1117 improvements. Vector Permutation AES for Altivec is also added. 1118 [Andy Polyakov] 1119 1120 *) Add support for little-endian ppc64 Linux target. 1121 [Marcelo Cerri (IBM)] 1122 1123 *) Initial support for AMRv8 ISA crypto extensions. This covers AES, 1124 SHA1, SHA256 and GHASH. "Initial" means that most common cases 1125 are optimized and there still is room for further improvements. 1126 Both 32- and 64-bit modes are supported. 1127 [Andy Polyakov, Ard Biesheuvel (Linaro)] 1128 1129 *) Improved ARMv7 NEON support. 1130 [Andy Polyakov] 1131 1132 *) Support for SPARC Architecture 2011 crypto extensions, first 1133 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 1134 SHA256/512, MD5, GHASH and modular exponentiation. 1135 [Andy Polyakov, David Miller] 1136 1137 *) Accelerated modular exponentiation for Intel processors, a.k.a. 1138 RSAZ. 1139 [Shay Gueron & Vlad Krasnov (Intel Corp)] 1140 1141 *) Support for new and upcoming Intel processors, including AVX2, 1142 BMI and SHA ISA extensions. This includes additional "stitched" 1143 implementations, AESNI-SHA256 and GCM, and multi-buffer support 1144 for TLS encrypt. 1145 1146 This work was sponsored by Intel Corp. 1147 [Andy Polyakov] 1148 1149 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 1150 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 1151 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 1152 [Steve Henson] 1153 1154 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 1155 this fixes a limiation in previous versions of OpenSSL. 1156 [Steve Henson] 1157 1158 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 1159 MGF1 digest and OAEP label. 1160 [Steve Henson] 1161 1162 *) Add EVP support for key wrapping algorithms, to avoid problems with 1163 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 1164 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 1165 algorithms and include tests cases. 1166 [Steve Henson] 1167 1168 *) Add functions to allocate and set the fields of an ECDSA_METHOD 1169 structure. 1170 [Douglas E. Engert, Steve Henson] 1171 1172 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 1173 difference in days and seconds between two tm or ASN1_TIME structures. 1174 [Steve Henson] 1175 1176 *) Add -rev test option to s_server to just reverse order of characters 1177 received by client and send back to server. Also prints an abbreviated 1178 summary of the connection parameters. 1179 [Steve Henson] 1180 1181 *) New option -brief for s_client and s_server to print out a brief summary 1182 of connection parameters. 1183 [Steve Henson] 1184 1185 *) Add callbacks for arbitrary TLS extensions. 1186 [Trevor Perrin <trevp@trevp.net> and Ben Laurie] 1187 1188 *) New option -crl_download in several openssl utilities to download CRLs 1189 from CRLDP extension in certificates. 1190 [Steve Henson] 1191 1192 *) New options -CRL and -CRLform for s_client and s_server for CRLs. 1193 [Steve Henson] 1194 1195 *) New function X509_CRL_diff to generate a delta CRL from the difference 1196 of two full CRLs. Add support to "crl" utility. 1197 [Steve Henson] 1198 1199 *) New functions to set lookup_crls function and to retrieve 1200 X509_STORE from X509_STORE_CTX. 1201 [Steve Henson] 1202 1203 *) Print out deprecated issuer and subject unique ID fields in 1204 certificates. 1205 [Steve Henson] 1206 1207 *) Extend OCSP I/O functions so they can be used for simple general purpose 1208 HTTP as well as OCSP. New wrapper function which can be used to download 1209 CRLs using the OCSP API. 1210 [Steve Henson] 1211 1212 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs. 1213 [Steve Henson] 1214 1215 *) SSL_CONF* functions. These provide a common framework for application 1216 configuration using configuration files or command lines. 1217 [Steve Henson] 1218 1219 *) SSL/TLS tracing code. This parses out SSL/TLS records using the 1220 message callback and prints the results. Needs compile time option 1221 "enable-ssl-trace". New options to s_client and s_server to enable 1222 tracing. 1223 [Steve Henson] 1224 1225 *) New ctrl and macro to retrieve supported points extensions. 1226 Print out extension in s_server and s_client. 1227 [Steve Henson] 1228 1229 *) New functions to retrieve certificate signature and signature 1230 OID NID. 1231 [Steve Henson] 1232 1233 *) Add functions to retrieve and manipulate the raw cipherlist sent by a 1234 client to OpenSSL. 1235 [Steve Henson] 1236 1237 *) New Suite B modes for TLS code. These use and enforce the requirements 1238 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 1239 only use Suite B curves. The Suite B modes can be set by using the 1240 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 1241 [Steve Henson] 1242 1243 *) New chain verification flags for Suite B levels of security. Check 1244 algorithms are acceptable when flags are set in X509_verify_cert. 1245 [Steve Henson] 1246 1247 *) Make tls1_check_chain return a set of flags indicating checks passed 1248 by a certificate chain. Add additional tests to handle client 1249 certificates: checks for matching certificate type and issuer name 1250 comparison. 1251 [Steve Henson] 1252 1253 *) If an attempt is made to use a signature algorithm not in the peer 1254 preference list abort the handshake. If client has no suitable 1255 signature algorithms in response to a certificate request do not 1256 use the certificate. 1257 [Steve Henson] 1258 1259 *) If server EC tmp key is not in client preference list abort handshake. 1260 [Steve Henson] 1261 1262 *) Add support for certificate stores in CERT structure. This makes it 1263 possible to have different stores per SSL structure or one store in 1264 the parent SSL_CTX. Include distint stores for certificate chain 1265 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 1266 to build and store a certificate chain in CERT structure: returing 1267 an error if the chain cannot be built: this will allow applications 1268 to test if a chain is correctly configured. 1269 1270 Note: if the CERT based stores are not set then the parent SSL_CTX 1271 store is used to retain compatibility with existing behaviour. 1272 1273 [Steve Henson] 1274 1275 *) New function ssl_set_client_disabled to set a ciphersuite disabled 1276 mask based on the current session, check mask when sending client 1277 hello and checking the requested ciphersuite. 1278 [Steve Henson] 1279 1280 *) New ctrls to retrieve and set certificate types in a certificate 1281 request message. Print out received values in s_client. If certificate 1282 types is not set with custom values set sensible values based on 1283 supported signature algorithms. 1284 [Steve Henson] 1285 1286 *) Support for distinct client and server supported signature algorithms. 1287 [Steve Henson] 1288 1289 *) Add certificate callback. If set this is called whenever a certificate 1290 is required by client or server. An application can decide which 1291 certificate chain to present based on arbitrary criteria: for example 1292 supported signature algorithms. Add very simple example to s_server. 1293 This fixes many of the problems and restrictions of the existing client 1294 certificate callback: for example you can now clear an existing 1295 certificate and specify the whole chain. 1296 [Steve Henson] 1297 1298 *) Add new "valid_flags" field to CERT_PKEY structure which determines what 1299 the certificate can be used for (if anything). Set valid_flags field 1300 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 1301 to have similar checks in it. 1302 1303 Add new "cert_flags" field to CERT structure and include a "strict mode". 1304 This enforces some TLS certificate requirements (such as only permitting 1305 certificate signature algorithms contained in the supported algorithms 1306 extension) which some implementations ignore: this option should be used 1307 with caution as it could cause interoperability issues. 1308 [Steve Henson] 1309 1310 *) Update and tidy signature algorithm extension processing. Work out 1311 shared signature algorithms based on preferences and peer algorithms 1312 and print them out in s_client and s_server. Abort handshake if no 1313 shared signature algorithms. 1314 [Steve Henson] 1315 1316 *) Add new functions to allow customised supported signature algorithms 1317 for SSL and SSL_CTX structures. Add options to s_client and s_server 1318 to support them. 1319 [Steve Henson] 1320 1321 *) New function SSL_certs_clear() to delete all references to certificates 1322 from an SSL structure. Before this once a certificate had been added 1323 it couldn't be removed. 1324 [Steve Henson] 1325 1326 *) Integrate hostname, email address and IP address checking with certificate 1327 verification. New verify options supporting checking in opensl utility. 1328 [Steve Henson] 1329 1330 *) Fixes and wildcard matching support to hostname and email checking 1331 functions. Add manual page. 1332 [Florian Weimer (Red Hat Product Security Team)] 1333 1334 *) New functions to check a hostname email or IP address against a 1335 certificate. Add options x509 utility to print results of checks against 1336 a certificate. 1337 [Steve Henson] 1338 1339 *) Fix OCSP checking. 1340 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie] 1341 1342 *) Initial experimental support for explicitly trusted non-root CAs. 1343 OpenSSL still tries to build a complete chain to a root but if an 1344 intermediate CA has a trust setting included that is used. The first 1345 setting is used: whether to trust (e.g., -addtrust option to the x509 1346 utility) or reject. 1347 [Steve Henson] 1348 1349 *) Add -trusted_first option which attempts to find certificates in the 1350 trusted store even if an untrusted chain is also supplied. 1351 [Steve Henson] 1352 1353 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 1354 platform support for Linux and Android. 1355 [Andy Polyakov] 1356 1357 *) Support for linux-x32, ILP32 environment in x86_64 framework. 1358 [Andy Polyakov] 1359 1360 *) Experimental multi-implementation support for FIPS capable OpenSSL. 1361 When in FIPS mode the approved implementations are used as normal, 1362 when not in FIPS mode the internal unapproved versions are used instead. 1363 This means that the FIPS capable OpenSSL isn't forced to use the 1364 (often lower perfomance) FIPS implementations outside FIPS mode. 1365 [Steve Henson] 1366 1367 *) Transparently support X9.42 DH parameters when calling 1368 PEM_read_bio_DHparameters. This means existing applications can handle 1369 the new parameter format automatically. 1370 [Steve Henson] 1371 1372 *) Initial experimental support for X9.42 DH parameter format: mainly 1373 to support use of 'q' parameter for RFC5114 parameters. 1374 [Steve Henson] 1375 1376 *) Add DH parameters from RFC5114 including test data to dhtest. 1377 [Steve Henson] 1378 1379 *) Support for automatic EC temporary key parameter selection. If enabled 1380 the most preferred EC parameters are automatically used instead of 1381 hardcoded fixed parameters. Now a server just has to call: 1382 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 1383 support ECDH and use the most appropriate parameters. 1384 [Steve Henson] 1385 1386 *) Enhance and tidy EC curve and point format TLS extension code. Use 1387 static structures instead of allocation if default values are used. 1388 New ctrls to set curves we wish to support and to retrieve shared curves. 1389 Print out shared curves in s_server. New options to s_server and s_client 1390 to set list of supported curves. 1391 [Steve Henson] 1392 1393 *) New ctrls to retrieve supported signature algorithms and 1394 supported curve values as an array of NIDs. Extend openssl utility 1395 to print out received values. 1396 [Steve Henson] 1397 1398 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 1399 between NIDs and the more common NIST names such as "P-256". Enhance 1400 ecparam utility and ECC method to recognise the NIST names for curves. 1401 [Steve Henson] 1402 1403 *) Enhance SSL/TLS certificate chain handling to support different 1404 chains for each certificate instead of one chain in the parent SSL_CTX. 1405 [Steve Henson] 1406 1407 *) Support for fixed DH ciphersuite client authentication: where both 1408 server and client use DH certificates with common parameters. 1409 [Steve Henson] 1410 1411 *) Support for fixed DH ciphersuites: those requiring DH server 1412 certificates. 1413 [Steve Henson] 1414 1415 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of 1416 the certificate. 1417 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 1418 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 1419 X509_CINF_get_signature were reverted post internal team review. 1420 1421 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 1422 1423 *) Build fixes for the Windows and OpenVMS platforms 1424 [Matt Caswell and Richard Levitte] 1425 1426 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 1427 1428 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 1429 message can cause a segmentation fault in OpenSSL due to a NULL pointer 1430 dereference. This could lead to a Denial Of Service attack. Thanks to 1431 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 1432 (CVE-2014-3571) 1433 [Steve Henson] 1434 1435 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 1436 dtls1_buffer_record function under certain conditions. In particular this 1437 could occur if an attacker sent repeated DTLS records with the same 1438 sequence number but for the next epoch. The memory leak could be exploited 1439 by an attacker in a Denial of Service attack through memory exhaustion. 1440 Thanks to Chris Mueller for reporting this issue. 1441 (CVE-2015-0206) 1442 [Matt Caswell] 1443 1444 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 1445 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 1446 method would be set to NULL which could later result in a NULL pointer 1447 dereference. Thanks to Frank Schmirler for reporting this issue. 1448 (CVE-2014-3569) 1449 [Kurt Roeckx] 1450 1451 *) Abort handshake if server key exchange message is omitted for ephemeral 1452 ECDH ciphersuites. 1453 1454 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 1455 reporting this issue. 1456 (CVE-2014-3572) 1457 [Steve Henson] 1458 1459 *) Remove non-export ephemeral RSA code on client and server. This code 1460 violated the TLS standard by allowing the use of temporary RSA keys in 1461 non-export ciphersuites and could be used by a server to effectively 1462 downgrade the RSA key length used to a value smaller than the server 1463 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 1464 INRIA or reporting this issue. 1465 (CVE-2015-0204) 1466 [Steve Henson] 1467 1468 *) Fixed issue where DH client certificates are accepted without verification. 1469 An OpenSSL server will accept a DH certificate for client authentication 1470 without the certificate verify message. This effectively allows a client to 1471 authenticate without the use of a private key. This only affects servers 1472 which trust a client certificate authority which issues certificates 1473 containing DH keys: these are extremely rare and hardly ever encountered. 1474 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 1475 this issue. 1476 (CVE-2015-0205) 1477 [Steve Henson] 1478 1479 *) Ensure that the session ID context of an SSL is updated when its 1480 SSL_CTX is updated via SSL_set_SSL_CTX. 1481 1482 The session ID context is typically set from the parent SSL_CTX, 1483 and can vary with the CTX. 1484 [Adam Langley] 1485 1486 *) Fix various certificate fingerprint issues. 1487 1488 By using non-DER or invalid encodings outside the signed portion of a 1489 certificate the fingerprint can be changed without breaking the signature. 1490 Although no details of the signed portion of the certificate can be changed 1491 this can cause problems with some applications: e.g. those using the 1492 certificate fingerprint for blacklists. 1493 1494 1. Reject signatures with non zero unused bits. 1495 1496 If the BIT STRING containing the signature has non zero unused bits reject 1497 the signature. All current signature algorithms require zero unused bits. 1498 1499 2. Check certificate algorithm consistency. 1500 1501 Check the AlgorithmIdentifier inside TBS matches the one in the 1502 certificate signature. NB: this will result in signature failure 1503 errors for some broken certificates. 1504 1505 Thanks to Konrad Kraszewski from Google for reporting this issue. 1506 1507 3. Check DSA/ECDSA signatures use DER. 1508 1509 Reencode DSA/ECDSA signatures and compare with the original received 1510 signature. Return an error if there is a mismatch. 1511 1512 This will reject various cases including garbage after signature 1513 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 1514 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 1515 (negative or with leading zeroes). 1516 1517 Further analysis was conducted and fixes were developed by Stephen Henson 1518 of the OpenSSL core team. 1519 1520 (CVE-2014-8275) 1521 [Steve Henson] 1522 1523 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 1524 results on some platforms, including x86_64. This bug occurs at random 1525 with a very low probability, and is not known to be exploitable in any 1526 way, though its exact impact is difficult to determine. Thanks to Pieter 1527 Wuille (Blockstream) who reported this issue and also suggested an initial 1528 fix. Further analysis was conducted by the OpenSSL development team and 1529 Adam Langley of Google. The final fix was developed by Andy Polyakov of 1530 the OpenSSL core team. 1531 (CVE-2014-3570) 1532 [Andy Polyakov] 1533 1534 *) Do not resume sessions on the server if the negotiated protocol 1535 version does not match the session's version. Resuming with a different 1536 version, while not strictly forbidden by the RFC, is of questionable 1537 sanity and breaks all known clients. 1538 [David Benjamin, Emilia K��sper] 1539 1540 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 1541 early CCS messages during renegotiation. (Note that because 1542 renegotiation is encrypted, this early CCS was not exploitable.) 1543 [Emilia K��sper] 1544 1545 *) Tighten client-side session ticket handling during renegotiation: 1546 ensure that the client only accepts a session ticket if the server sends 1547 the extension anew in the ServerHello. Previously, a TLS client would 1548 reuse the old extension state and thus accept a session ticket if one was 1549 announced in the initial ServerHello. 1550 1551 Similarly, ensure that the client requires a session ticket if one 1552 was advertised in the ServerHello. Previously, a TLS client would 1553 ignore a missing NewSessionTicket message. 1554 [Emilia K��sper] 1555 1556 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 1557 1558 *) SRTP Memory Leak. 1559 1560 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 1561 sends a carefully crafted handshake message, to cause OpenSSL to fail 1562 to free up to 64k of memory causing a memory leak. This could be 1563 exploited in a Denial Of Service attack. This issue affects OpenSSL 1564 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 1565 whether SRTP is used or configured. Implementations of OpenSSL that 1566 have been compiled with OPENSSL_NO_SRTP defined are not affected. 1567 1568 The fix was developed by the OpenSSL team. 1569 (CVE-2014-3513) 1570 [OpenSSL team] 1571 1572 *) Session Ticket Memory Leak. 1573 1574 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 1575 integrity of that ticket is first verified. In the event of a session 1576 ticket integrity check failing, OpenSSL will fail to free memory 1577 causing a memory leak. By sending a large number of invalid session 1578 tickets an attacker could exploit this issue in a Denial Of Service 1579 attack. 1580 (CVE-2014-3567) 1581 [Steve Henson] 1582 1583 *) Build option no-ssl3 is incomplete. 1584 1585 When OpenSSL is configured with "no-ssl3" as a build option, servers 1586 could accept and complete a SSL 3.0 handshake, and clients could be 1587 configured to send them. 1588 (CVE-2014-3568) 1589 [Akamai and the OpenSSL team] 1590 1591 *) Add support for TLS_FALLBACK_SCSV. 1592 Client applications doing fallback retries should call 1593 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 1594 (CVE-2014-3566) 1595 [Adam Langley, Bodo Moeller] 1596 1597 *) Add additional DigestInfo checks. 1598 1599 Reencode DigestInto in DER and check against the original when 1600 verifying RSA signature: this will reject any improperly encoded 1601 DigestInfo structures. 1602 1603 Note: this is a precautionary measure and no attacks are currently known. 1604 1605 [Steve Henson] 1606 1607 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 1608 1609 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 1610 SRP code can be overrun an internal buffer. Add sanity check that 1611 g, A, B < N to SRP code. 1612 1613 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 1614 Group for discovering this issue. 1615 (CVE-2014-3512) 1616 [Steve Henson] 1617 1618 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 1619 TLS 1.0 instead of higher protocol versions when the ClientHello message 1620 is badly fragmented. This allows a man-in-the-middle attacker to force a 1621 downgrade to TLS 1.0 even if both the server and the client support a 1622 higher protocol version, by modifying the client's TLS records. 1623 1624 Thanks to David Benjamin and Adam Langley (Google) for discovering and 1625 researching this issue. 1626 (CVE-2014-3511) 1627 [David Benjamin] 1628 1629 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 1630 to a denial of service attack. A malicious server can crash the client 1631 with a null pointer dereference (read) by specifying an anonymous (EC)DH 1632 ciphersuite and sending carefully crafted handshake messages. 1633 1634 Thanks to Felix Gr��bert (Google) for discovering and researching this 1635 issue. 1636 (CVE-2014-3510) 1637 [Emilia K��sper] 1638 1639 *) By sending carefully crafted DTLS packets an attacker could cause openssl 1640 to leak memory. This can be exploited through a Denial of Service attack. 1641 Thanks to Adam Langley for discovering and researching this issue. 1642 (CVE-2014-3507) 1643 [Adam Langley] 1644 1645 *) An attacker can force openssl to consume large amounts of memory whilst 1646 processing DTLS handshake messages. This can be exploited through a 1647 Denial of Service attack. 1648 Thanks to Adam Langley for discovering and researching this issue. 1649 (CVE-2014-3506) 1650 [Adam Langley] 1651 1652 *) An attacker can force an error condition which causes openssl to crash 1653 whilst processing DTLS packets due to memory being freed twice. This 1654 can be exploited through a Denial of Service attack. 1655 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 1656 this issue. 1657 (CVE-2014-3505) 1658 [Adam Langley] 1659 1660 *) If a multithreaded client connects to a malicious server using a resumed 1661 session and the server sends an ec point format extension it could write 1662 up to 255 bytes to freed memory. 1663 1664 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 1665 issue. 1666 (CVE-2014-3509) 1667 [Gabor Tyukasz] 1668 1669 *) A malicious server can crash an OpenSSL client with a null pointer 1670 dereference (read) by specifying an SRP ciphersuite even though it was not 1671 properly negotiated with the client. This can be exploited through a 1672 Denial of Service attack. 1673 1674 Thanks to Joonas Kuorilehto and Riku Hietam��ki (Codenomicon) for 1675 discovering and researching this issue. 1676 (CVE-2014-5139) 1677 [Steve Henson] 1678 1679 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 1680 X509_name_oneline, X509_name_print_ex et al. to leak some information 1681 from the stack. Applications may be affected if they echo pretty printing 1682 output to the attacker. 1683 1684 Thanks to Ivan Fratric (Google) for discovering this issue. 1685 (CVE-2014-3508) 1686 [Emilia K��sper, and Steve Henson] 1687 1688 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 1689 for corner cases. (Certain input points at infinity could lead to 1690 bogus results, with non-infinity inputs mapped to infinity too.) 1691 [Bodo Moeller] 1692 1693 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 1694 1695 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 1696 handshake can force the use of weak keying material in OpenSSL 1697 SSL/TLS clients and servers. 1698 1699 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 1700 researching this issue. (CVE-2014-0224) 1701 [KIKUCHI Masashi, Steve Henson] 1702 1703 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 1704 OpenSSL DTLS client the code can be made to recurse eventually crashing 1705 in a DoS attack. 1706 1707 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 1708 (CVE-2014-0221) 1709 [Imre Rad, Steve Henson] 1710 1711 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 1712 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 1713 client or server. This is potentially exploitable to run arbitrary 1714 code on a vulnerable client or server. 1715 1716 Thanks to J��ri Aedla for reporting this issue. (CVE-2014-0195) 1717 [J��ri Aedla, Steve Henson] 1718 1719 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 1720 are subject to a denial of service attack. 1721 1722 Thanks to Felix Gr��bert and Ivan Fratric at Google for discovering 1723 this issue. (CVE-2014-3470) 1724 [Felix Gr��bert, Ivan Fratric, Steve Henson] 1725 1726 *) Harmonize version and its documentation. -f flag is used to display 1727 compilation flags. 1728 [mancha <mancha1@zoho.com>] 1729 1730 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 1731 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 1732 [mancha <mancha1@zoho.com>] 1733 1734 *) Fix some double frees. These are not thought to be exploitable. 1735 [mancha <mancha1@zoho.com>] 1736 1737 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 1738 1739 *) A missing bounds check in the handling of the TLS heartbeat extension 1740 can be used to reveal up to 64k of memory to a connected client or 1741 server. 1742 1743 Thanks for Neel Mehta of Google Security for discovering this bug and to 1744 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 1745 preparing the fix (CVE-2014-0160) 1746 [Adam Langley, Bodo Moeller] 1747 1748 *) Fix for the attack described in the paper "Recovering OpenSSL 1749 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 1750 by Yuval Yarom and Naomi Benger. Details can be obtained from: 1751 http://eprint.iacr.org/2014/140 1752 1753 Thanks to Yuval Yarom and Naomi Benger for discovering this 1754 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 1755 [Yuval Yarom and Naomi Benger] 1756 1757 *) TLS pad extension: draft-agl-tls-padding-03 1758 1759 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 1760 TLS client Hello record length value would otherwise be > 255 and 1761 less that 512 pad with a dummy extension containing zeroes so it 1762 is at least 512 bytes long. 1763 1764 [Adam Langley, Steve Henson] 1765 1766 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 1767 1768 *) Fix for TLS record tampering bug. A carefully crafted invalid 1769 handshake could crash OpenSSL with a NULL pointer exception. 1770 Thanks to Anton Johansson for reporting this issues. 1771 (CVE-2013-4353) 1772 1773 *) Keep original DTLS digest and encryption contexts in retransmission 1774 structures so we can use the previous session parameters if they need 1775 to be resent. (CVE-2013-6450) 1776 [Steve Henson] 1777 1778 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 1779 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 1780 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 1781 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 1782 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 1783 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 1784 [Rob Stradling, Adam Langley] 1785 1786 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 1787 1788 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 1789 supporting platforms or when small records were transferred. 1790 [Andy Polyakov, Steve Henson] 1791 1792 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 1793 1794 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 1795 1796 This addresses the flaw in CBC record processing discovered by 1797 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 1798 at: http://www.isg.rhul.ac.uk/tls/ 1799 1800 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 1801 Security Group at Royal Holloway, University of London 1802 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 1803 Emilia K��sper for the initial patch. 1804 (CVE-2013-0169) 1805 [Emilia K��sper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 1806 1807 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 1808 ciphersuites which can be exploited in a denial of service attack. 1809 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 1810 and detecting this bug and to Wolfgang Ettlinger 1811 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 1812 (CVE-2012-2686) 1813 [Adam Langley] 1814 1815 *) Return an error when checking OCSP signatures when key is NULL. 1816 This fixes a DoS attack. (CVE-2013-0166) 1817 [Steve Henson] 1818 1819 *) Make openssl verify return errors. 1820 [Chris Palmer <palmer@google.com> and Ben Laurie] 1821 1822 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 1823 the right response is stapled. Also change SSL_get_certificate() 1824 so it returns the certificate actually sent. 1825 See http://rt.openssl.org/Ticket/Display.html?id=2836. 1826 [Rob Stradling <rob.stradling@comodo.com>] 1827 1828 *) Fix possible deadlock when decoding public keys. 1829 [Steve Henson] 1830 1831 *) Don't use TLS 1.0 record version number in initial client hello 1832 if renegotiating. 1833 [Steve Henson] 1834 1835 Changes between 1.0.1b and 1.0.1c [10 May 2012] 1836 1837 *) Sanity check record length before skipping explicit IV in TLS 1838 1.2, 1.1 and DTLS to fix DoS attack. 1839 1840 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 1841 fuzzing as a service testing platform. 1842 (CVE-2012-2333) 1843 [Steve Henson] 1844 1845 *) Initialise tkeylen properly when encrypting CMS messages. 1846 Thanks to Solar Designer of Openwall for reporting this issue. 1847 [Steve Henson] 1848 1849 *) In FIPS mode don't try to use composite ciphers as they are not 1850 approved. 1851 [Steve Henson] 1852 1853 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 1854 1855 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 1856 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 1857 mean any application compiled against OpenSSL 1.0.0 headers setting 1858 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng 1859 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 1860 0x10000000L Any application which was previously compiled against 1861 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 1862 will need to be recompiled as a result. Letting be results in 1863 inability to disable specifically TLS 1.1 and in client context, 1864 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 1865 [Steve Henson] 1866 1867 *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not 1868 disable just protocol X, but all protocols above X *if* there are 1869 protocols *below* X still enabled. In more practical terms it means 1870 that if application wants to disable TLS1.0 in favor of TLS1.1 and 1871 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 1872 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 1873 client side. 1874 [Andy Polyakov] 1875 1876 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 1877 1878 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 1879 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 1880 in CRYPTO_realloc_clean. 1881 1882 Thanks to Tavis Ormandy, Google Security Team, for discovering this 1883 issue and to Adam Langley <agl@chromium.org> for fixing it. 1884 (CVE-2012-2110) 1885 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 1886 1887 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 1888 [Adam Langley] 1889 1890 *) Workarounds for some broken servers that "hang" if a client hello 1891 record length exceeds 255 bytes. 1892 1893 1. Do not use record version number > TLS 1.0 in initial client 1894 hello: some (but not all) hanging servers will now work. 1895 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 1896 the number of ciphers sent in the client hello. This should be 1897 set to an even number, such as 50, for example by passing: 1898 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 1899 Most broken servers should now work. 1900 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 1901 TLS 1.2 client support entirely. 1902 [Steve Henson] 1903 1904 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 1905 [Andy Polyakov] 1906 1907 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 1908 1909 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 1910 STRING form instead of a DigestInfo. 1911 [Steve Henson] 1912 1913 *) The format used for MDC2 RSA signatures is inconsistent between EVP 1914 and the RSA_sign/RSA_verify functions. This was made more apparent when 1915 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 1916 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 1917 the correct format in RSA_verify so both forms transparently work. 1918 [Steve Henson] 1919 1920 *) Some servers which support TLS 1.0 can choke if we initially indicate 1921 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 1922 encrypted premaster secret. As a workaround use the maximum pemitted 1923 client version in client hello, this should keep such servers happy 1924 and still work with previous versions of OpenSSL. 1925 [Steve Henson] 1926 1927 *) Add support for TLS/DTLS heartbeats. 1928 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1929 1930 *) Add support for SCTP. 1931 [Robin Seggelmann <seggelmann@fh-muenster.de>] 1932 1933 *) Improved PRNG seeding for VOS. 1934 [Paul Green <Paul.Green@stratus.com>] 1935 1936 *) Extensive assembler packs updates, most notably: 1937 1938 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 1939 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 1940 - x86_64: bit-sliced AES implementation; 1941 - ARM: NEON support, contemporary platforms optimizations; 1942 - s390x: z196 support; 1943 - *: GHASH and GF(2^m) multiplication implementations; 1944 1945 [Andy Polyakov] 1946 1947 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 1948 (removal of unnecessary code) 1949 [Peter Sylvester <peter.sylvester@edelweb.fr>] 1950 1951 *) Add TLS key material exporter from RFC 5705. 1952 [Eric Rescorla] 1953 1954 *) Add DTLS-SRTP negotiation from RFC 5764. 1955 [Eric Rescorla] 1956 1957 *) Add Next Protocol Negotiation, 1958 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 1959 disabled with a no-npn flag to config or Configure. Code donated 1960 by Google. 1961 [Adam Langley <agl@google.com> and Ben Laurie] 1962 1963 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 1964 NIST-P256, NIST-P521, with constant-time single point multiplication on 1965 typical inputs. Compiler support for the nonstandard type __uint128_t is 1966 required to use this (present in gcc 4.4 and later, for 64-bit builds). 1967 Code made available under Apache License version 2.0. 1968 1969 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 1970 line to include this in your build of OpenSSL, and run "make depend" (or 1971 "make update"). This enables the following EC_METHODs: 1972 1973 EC_GFp_nistp224_method() 1974 EC_GFp_nistp256_method() 1975 EC_GFp_nistp521_method() 1976 1977 EC_GROUP_new_by_curve_name() will automatically use these (while 1978 EC_GROUP_new_curve_GFp() currently prefers the more flexible 1979 implementations). 1980 [Emilia K��sper, Adam Langley, Bodo Moeller (Google)] 1981 1982 *) Use type ossl_ssize_t instad of ssize_t which isn't available on 1983 all platforms. Move ssize_t definition from e_os.h to the public 1984 header file e_os2.h as it now appears in public header file cms.h 1985 [Steve Henson] 1986 1987 *) New -sigopt option to the ca, req and x509 utilities. Additional 1988 signature parameters can be passed using this option and in 1989 particular PSS. 1990 [Steve Henson] 1991 1992 *) Add RSA PSS signing function. This will generate and set the 1993 appropriate AlgorithmIdentifiers for PSS based on those in the 1994 corresponding EVP_MD_CTX structure. No application support yet. 1995 [Steve Henson] 1996 1997 *) Support for companion algorithm specific ASN1 signing routines. 1998 New function ASN1_item_sign_ctx() signs a pre-initialised 1999 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 2000 the appropriate parameters. 2001 [Steve Henson] 2002 2003 *) Add new algorithm specific ASN1 verification initialisation function 2004 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 2005 handling will be the same no matter what EVP_PKEY_METHOD is used. 2006 Add a PSS handler to support verification of PSS signatures: checked 2007 against a number of sample certificates. 2008 [Steve Henson] 2009 2010 *) Add signature printing for PSS. Add PSS OIDs. 2011 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 2012 2013 *) Add algorithm specific signature printing. An individual ASN1 method 2014 can now print out signatures instead of the standard hex dump. 2015 2016 More complex signatures (e.g. PSS) can print out more meaningful 2017 information. Include DSA version that prints out the signature 2018 parameters r, s. 2019 [Steve Henson] 2020 2021 *) Password based recipient info support for CMS library: implementing 2022 RFC3211. 2023 [Steve Henson] 2024 2025 *) Split password based encryption into PBES2 and PBKDF2 functions. This 2026 neatly separates the code into cipher and PBE sections and is required 2027 for some algorithms that split PBES2 into separate pieces (such as 2028 password based CMS). 2029 [Steve Henson] 2030 2031 *) Session-handling fixes: 2032 - Fix handling of connections that are resuming with a session ID, 2033 but also support Session Tickets. 2034 - Fix a bug that suppressed issuing of a new ticket if the client 2035 presented a ticket with an expired session. 2036 - Try to set the ticket lifetime hint to something reasonable. 2037 - Make tickets shorter by excluding irrelevant information. 2038 - On the client side, don't ignore renewed tickets. 2039 [Adam Langley, Bodo Moeller (Google)] 2040 2041 *) Fix PSK session representation. 2042 [Bodo Moeller] 2043 2044 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 2045 2046 This work was sponsored by Intel. 2047 [Andy Polyakov] 2048 2049 *) Add GCM support to TLS library. Some custom code is needed to split 2050 the IV between the fixed (from PRF) and explicit (from TLS record) 2051 portions. This adds all GCM ciphersuites supported by RFC5288 and 2052 RFC5289. Generalise some AES* cipherstrings to inlclude GCM and 2053 add a special AESGCM string for GCM only. 2054 [Steve Henson] 2055 2056 *) Expand range of ctrls for AES GCM. Permit setting invocation 2057 field on decrypt and retrieval of invocation field only on encrypt. 2058 [Steve Henson] 2059 2060 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 2061 As required by RFC5289 these ciphersuites cannot be used if for 2062 versions of TLS earlier than 1.2. 2063 [Steve Henson] 2064 2065 *) For FIPS capable OpenSSL interpret a NULL default public key method 2066 as unset and return the appopriate default but do *not* set the default. 2067 This means we can return the appopriate method in applications that 2068 swicth between FIPS and non-FIPS modes. 2069 [Steve Henson] 2070 2071 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 2072 ENGINE is used then we cannot handle that in the FIPS module so we 2073 keep original code iff non-FIPS operations are allowed. 2074 [Steve Henson] 2075 2076 *) Add -attime option to openssl utilities. 2077 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 2078 2079 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 2080 [Steve Henson] 2081 2082 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 2083 FIPS EC methods unconditionally for now. 2084 [Steve Henson] 2085 2086 *) New build option no-ec2m to disable characteristic 2 code. 2087 [Steve Henson] 2088 2089 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 2090 all cases can be covered as some introduce binary incompatibilities. 2091 [Steve Henson] 2092 2093 *) Redirect RSA operations to FIPS module including keygen, 2094 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 2095 [Steve Henson] 2096 2097 *) Add similar low level API blocking to ciphers. 2098 [Steve Henson] 2099 2100 *) Low level digest APIs are not approved in FIPS mode: any attempt 2101 to use these will cause a fatal error. Applications that *really* want 2102 to use them can use the private_* version instead. 2103 [Steve Henson] 2104 2105 *) Redirect cipher operations to FIPS module for FIPS builds. 2106 [Steve Henson] 2107 2108 *) Redirect digest operations to FIPS module for FIPS builds. 2109 [Steve Henson] 2110 2111 *) Update build system to add "fips" flag which will link in fipscanister.o 2112 for static and shared library builds embedding a signature if needed. 2113 [Steve Henson] 2114 2115 *) Output TLS supported curves in preference order instead of numerical 2116 order. This is currently hardcoded for the highest order curves first. 2117 This should be configurable so applications can judge speed vs strength. 2118 [Steve Henson] 2119 2120 *) Add TLS v1.2 server support for client authentication. 2121 [Steve Henson] 2122 2123 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 2124 and enable MD5. 2125 [Steve Henson] 2126 2127 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 2128 FIPS modules versions. 2129 [Steve Henson] 2130 2131 *) Add TLS v1.2 client side support for client authentication. Keep cache 2132 of handshake records longer as we don't know the hash algorithm to use 2133 until after the certificate request message is received. 2134 [Steve Henson] 2135 2136 *) Initial TLS v1.2 client support. Add a default signature algorithms 2137 extension including all the algorithms we support. Parse new signature 2138 format in client key exchange. Relax some ECC signing restrictions for 2139 TLS v1.2 as indicated in RFC5246. 2140 [Steve Henson] 2141 2142 *) Add server support for TLS v1.2 signature algorithms extension. Switch 2143 to new signature format when needed using client digest preference. 2144 All server ciphersuites should now work correctly in TLS v1.2. No client 2145 support yet and no support for client certificates. 2146 [Steve Henson] 2147 2148 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 2149 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 2150 ciphersuites. At present only RSA key exchange ciphersuites work with 2151 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 2152 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 2153 and version checking. 2154 [Steve Henson] 2155 2156 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 2157 with this defined it will not be affected by any changes to ssl internal 2158 structures. Add several utility functions to allow openssl application 2159 to work with OPENSSL_NO_SSL_INTERN defined. 2160 [Steve Henson] 2161 2162 *) A long standing patch to add support for SRP from EdelWeb (Peter 2163 Sylvester and Christophe Renou) was integrated. 2164 [Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester 2165 <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and 2166 Ben Laurie] 2167 2168 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 2169 [Steve Henson] 2170 2171 *) Permit abbreviated handshakes when renegotiating using the function 2172 SSL_renegotiate_abbreviated(). 2173 [Robin Seggelmann <seggelmann@fh-muenster.de>] 2174 2175 *) Add call to ENGINE_register_all_complete() to 2176 ENGINE_load_builtin_engines(), so some implementations get used 2177 automatically instead of needing explicit application support. 2178 [Steve Henson] 2179 2180 *) Add support for TLS key exporter as described in RFC5705. 2181 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 2182 2183 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 2184 a few changes are required: 2185 2186 Add SSL_OP_NO_TLSv1_1 flag. 2187 Add TLSv1_1 methods. 2188 Update version checking logic to handle version 1.1. 2189 Add explicit IV handling (ported from DTLS code). 2190 Add command line options to s_client/s_server. 2191 [Steve Henson] 2192 2193 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 2194 2195 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 2196 in CMS and PKCS7 code. When RSA decryption fails use a random key for 2197 content decryption and always return the same error. Note: this attack 2198 needs on average 2^20 messages so it only affects automated senders. The 2199 old behaviour can be reenabled in the CMS code by setting the 2200 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 2201 an MMA defence is not necessary. 2202 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 2203 this issue. (CVE-2012-0884) 2204 [Steve Henson] 2205 2206 *) Fix CVE-2011-4619: make sure we really are receiving a 2207 client hello before rejecting multiple SGC restarts. Thanks to 2208 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 2209 [Steve Henson] 2210 2211 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 2212 2213 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 2214 Thanks to Antonio Martin, Enterprise Secure Access Research and 2215 Development, Cisco Systems, Inc. for discovering this bug and 2216 preparing a fix. (CVE-2012-0050) 2217 [Antonio Martin] 2218 2219 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 2220 2221 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 2222 of the Vaudenay padding oracle attack on CBC mode encryption 2223 which enables an efficient plaintext recovery attack against 2224 the OpenSSL implementation of DTLS. Their attack exploits timing 2225 differences arising during decryption processing. A research 2226 paper describing this attack can be found at: 2227 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 2228 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 2229 Security Group at Royal Holloway, University of London 2230 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 2231 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 2232 for preparing the fix. (CVE-2011-4108) 2233 [Robin Seggelmann, Michael Tuexen] 2234 2235 *) Clear bytes used for block padding of SSL 3.0 records. 2236 (CVE-2011-4576) 2237 [Adam Langley (Google)] 2238 2239 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 2240 Kadianakis <desnacked@gmail.com> for discovering this issue and 2241 Adam Langley for preparing the fix. (CVE-2011-4619) 2242 [Adam Langley (Google)] 2243 2244 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 2245 [Andrey Kulikov <amdeich@gmail.com>] 2246 2247 *) Prevent malformed RFC3779 data triggering an assertion failure. 2248 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 2249 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 2250 [Rob Austein <sra@hactrn.net>] 2251 2252 *) Improved PRNG seeding for VOS. 2253 [Paul Green <Paul.Green@stratus.com>] 2254 2255 *) Fix ssl_ciph.c set-up race. 2256 [Adam Langley (Google)] 2257 2258 *) Fix spurious failures in ecdsatest.c. 2259 [Emilia K��sper (Google)] 2260 2261 *) Fix the BIO_f_buffer() implementation (which was mixing different 2262 interpretations of the '..._len' fields). 2263 [Adam Langley (Google)] 2264 2265 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 2266 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 2267 threads won't reuse the same blinding coefficients. 2268 2269 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 2270 lock to call BN_BLINDING_invert_ex, and avoids one use of 2271 BN_BLINDING_update for each BN_BLINDING structure (previously, 2272 the last update always remained unused). 2273 [Emilia K��sper (Google)] 2274 2275 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 2276 [Bob Buckholz (Google)] 2277 2278 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 2279 2280 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 2281 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 2282 [Kaspar Brand <ossl@velox.ch>] 2283 2284 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 2285 for multi-threaded use of ECDH. (CVE-2011-3210) 2286 [Adam Langley (Google)] 2287 2288 *) Fix x509_name_ex_d2i memory leak on bad inputs. 2289 [Bodo Moeller] 2290 2291 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 2292 signature public key algorithm by using OID xref utilities instead. 2293 Before this you could only use some ECC ciphersuites with SHA1 only. 2294 [Steve Henson] 2295 2296 *) Add protection against ECDSA timing attacks as mentioned in the paper 2297 by Billy Bob Brumley and Nicola Tuveri, see: 2298 2299 http://eprint.iacr.org/2011/232.pdf 2300 2301 [Billy Bob Brumley and Nicola Tuveri] 2302 2303 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 2304 2305 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 2306 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 2307 2308 *) Fix bug in string printing code: if *any* escaping is enabled we must 2309 escape the escape character (backslash) or the resulting string is 2310 ambiguous. 2311 [Steve Henson] 2312 2313 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 2314 2315 *) Disable code workaround for ancient and obsolete Netscape browsers 2316 and servers: an attacker can use it in a ciphersuite downgrade attack. 2317 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 2318 [Steve Henson] 2319 2320 *) Fixed J-PAKE implementation error, originally discovered by 2321 Sebastien Martini, further info and confirmation from Stefan 2322 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 2323 [Ben Laurie] 2324 2325 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 2326 2327 *) Fix extension code to avoid race conditions which can result in a buffer 2328 overrun vulnerability: resumed sessions must not be modified as they can 2329 be shared by multiple threads. CVE-2010-3864 2330 [Steve Henson] 2331 2332 *) Fix WIN32 build system to correctly link an ENGINE directory into 2333 a DLL. 2334 [Steve Henson] 2335 2336 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 2337 2338 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 2339 (CVE-2010-1633) 2340 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 2341 2342 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 2343 2344 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 2345 context. The operation can be customised via the ctrl mechanism in 2346 case ENGINEs want to include additional functionality. 2347 [Steve Henson] 2348 2349 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 2350 [Steve Henson] 2351 2352 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 2353 output hashes compatible with older versions of OpenSSL. 2354 [Willy Weisz <weisz@vcpc.univie.ac.at>] 2355 2356 *) Fix compression algorithm handling: if resuming a session use the 2357 compression algorithm of the resumed session instead of determining 2358 it from client hello again. Don't allow server to change algorithm. 2359 [Steve Henson] 2360 2361 *) Add load_crls() function to apps tidying load_certs() too. Add option 2362 to verify utility to allow additional CRLs to be included. 2363 [Steve Henson] 2364 2365 *) Update OCSP request code to permit adding custom headers to the request: 2366 some responders need this. 2367 [Steve Henson] 2368 2369 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 2370 correctly. 2371 [Julia Lawall <julia@diku.dk>] 2372 2373 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 2374 needlessly dereferenced structures, used obsolete functions and 2375 didn't handle all updated verify codes correctly. 2376 [Steve Henson] 2377 2378 *) Disable MD2 in the default configuration. 2379 [Steve Henson] 2380 2381 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 2382 indicate the initial BIO being pushed or popped. This makes it possible 2383 to determine whether the BIO is the one explicitly called or as a result 2384 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 2385 it handles reference counts correctly and doesn't zero out the I/O bio 2386 when it is not being explicitly popped. WARNING: applications which 2387 included workarounds for the old buggy behaviour will need to be modified 2388 or they could free up already freed BIOs. 2389 [Steve Henson] 2390 2391 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 2392 renaming to all platforms (within the 0.9.8 branch, this was 2393 done conditionally on Netware platforms to avoid a name clash). 2394 [Guenter <lists@gknw.net>] 2395 2396 *) Add ECDHE and PSK support to DTLS. 2397 [Michael Tuexen <tuexen@fh-muenster.de>] 2398 2399 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 2400 be used on C++. 2401 [Steve Henson] 2402 2403 *) Add "missing" function EVP_MD_flags() (without this the only way to 2404 retrieve a digest flags is by accessing the structure directly. Update 2405 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 2406 or cipher is registered as in the "from" argument. Print out all 2407 registered digests in the dgst usage message instead of manually 2408 attempting to work them out. 2409 [Steve Henson] 2410 2411 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 2412 this allows the use of compression and extensions. Change default cipher 2413 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 2414 by default unless an application cipher string requests it. 2415 [Steve Henson] 2416 2417 *) Alter match criteria in PKCS12_parse(). It used to try to use local 2418 key ids to find matching certificates and keys but some PKCS#12 files 2419 don't follow the (somewhat unwritten) rules and this strategy fails. 2420 Now just gather all certificates together and the first private key 2421 then look for the first certificate that matches the key. 2422 [Steve Henson] 2423 2424 *) Support use of registered digest and cipher names for dgst and cipher 2425 commands instead of having to add each one as a special case. So now 2426 you can do: 2427 2428 openssl sha256 foo 2429 2430 as well as: 2431 2432 openssl dgst -sha256 foo 2433 2434 and this works for ENGINE based algorithms too. 2435 2436 [Steve Henson] 2437 2438 *) Update Gost ENGINE to support parameter files. 2439 [Victor B. Wagner <vitus@cryptocom.ru>] 2440 2441 *) Support GeneralizedTime in ca utility. 2442 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 2443 2444 *) Enhance the hash format used for certificate directory links. The new 2445 form uses the canonical encoding (meaning equivalent names will work 2446 even if they aren't identical) and uses SHA1 instead of MD5. This form 2447 is incompatible with the older format and as a result c_rehash should 2448 be used to rebuild symbolic links. 2449 [Steve Henson] 2450 2451 *) Make PKCS#8 the default write format for private keys, replacing the 2452 traditional format. This form is standardised, more secure and doesn't 2453 include an implicit MD5 dependency. 2454 [Steve Henson] 2455 2456 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 2457 committed to OpenSSL should pass this lot as a minimum. 2458 [Steve Henson] 2459 2460 *) Add session ticket override functionality for use by EAP-FAST. 2461 [Jouni Malinen <j@w1.fi>] 2462 2463 *) Modify HMAC functions to return a value. Since these can be implemented 2464 in an ENGINE errors can occur. 2465 [Steve Henson] 2466 2467 *) Type-checked OBJ_bsearch_ex. 2468 [Ben Laurie] 2469 2470 *) Type-checked OBJ_bsearch. Also some constification necessitated 2471 by type-checking. Still to come: TXT_DB, bsearch(?), 2472 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 2473 CONF_VALUE. 2474 [Ben Laurie] 2475 2476 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 2477 seconds to a tm structure directly, instead of going through OS 2478 specific date routines. This avoids any issues with OS routines such 2479 as the year 2038 bug. New *_adj() functions for ASN1 time structures 2480 and X509_time_adj_ex() to cover the extended range. The existing 2481 X509_time_adj() is still usable and will no longer have any date issues. 2482 [Steve Henson] 2483 2484 *) Delta CRL support. New use deltas option which will attempt to locate 2485 and search any appropriate delta CRLs available. 2486 2487 This work was sponsored by Google. 2488 [Steve Henson] 2489 2490 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 2491 code and add additional score elements. Validate alternate CRL paths 2492 as part of the CRL checking and indicate a new error "CRL path validation 2493 error" in this case. Applications wanting additional details can use 2494 the verify callback and check the new "parent" field. If this is not 2495 NULL CRL path validation is taking place. Existing applications wont 2496 see this because it requires extended CRL support which is off by 2497 default. 2498 2499 This work was sponsored by Google. 2500 [Steve Henson] 2501 2502 *) Support for freshest CRL extension. 2503 2504 This work was sponsored by Google. 2505 [Steve Henson] 2506 2507 *) Initial indirect CRL support. Currently only supported in the CRLs 2508 passed directly and not via lookup. Process certificate issuer 2509 CRL entry extension and lookup CRL entries by bother issuer name 2510 and serial number. Check and process CRL issuer entry in IDP extension. 2511 2512 This work was sponsored by Google. 2513 [Steve Henson] 2514 2515 *) Add support for distinct certificate and CRL paths. The CRL issuer 2516 certificate is validated separately in this case. Only enabled if 2517 an extended CRL support flag is set: this flag will enable additional 2518 CRL functionality in future. 2519 2520 This work was sponsored by Google. 2521 [Steve Henson] 2522 2523 *) Add support for policy mappings extension. 2524 2525 This work was sponsored by Google. 2526 [Steve Henson] 2527 2528 *) Fixes to pathlength constraint, self issued certificate handling, 2529 policy processing to align with RFC3280 and PKITS tests. 2530 2531 This work was sponsored by Google. 2532 [Steve Henson] 2533 2534 *) Support for name constraints certificate extension. DN, email, DNS 2535 and URI types are currently supported. 2536 2537 This work was sponsored by Google. 2538 [Steve Henson] 2539 2540 *) To cater for systems that provide a pointer-based thread ID rather 2541 than numeric, deprecate the current numeric thread ID mechanism and 2542 replace it with a structure and associated callback type. This 2543 mechanism allows a numeric "hash" to be extracted from a thread ID in 2544 either case, and on platforms where pointers are larger than 'long', 2545 mixing is done to help ensure the numeric 'hash' is usable even if it 2546 can't be guaranteed unique. The default mechanism is to use "&errno" 2547 as a pointer-based thread ID to distinguish between threads. 2548 2549 Applications that want to provide their own thread IDs should now use 2550 CRYPTO_THREADID_set_callback() to register a callback that will call 2551 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 2552 2553 Note that ERR_remove_state() is now deprecated, because it is tied 2554 to the assumption that thread IDs are numeric. ERR_remove_state(0) 2555 to free the current thread's error state should be replaced by 2556 ERR_remove_thread_state(NULL). 2557 2558 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 2559 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 2560 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 2561 application was previously providing a numeric thread callback that 2562 was inappropriate for distinguishing threads, then uniqueness might 2563 have been obtained with &errno that happened immediately in the 2564 intermediate development versions of OpenSSL; this is no longer the 2565 case, the numeric thread callback will now override the automatic use 2566 of &errno.) 2567 [Geoff Thorpe, with help from Bodo Moeller] 2568 2569 *) Initial support for different CRL issuing certificates. This covers a 2570 simple case where the self issued certificates in the chain exist and 2571 the real CRL issuer is higher in the existing chain. 2572 2573 This work was sponsored by Google. 2574 [Steve Henson] 2575 2576 *) Removed effectively defunct crypto/store from the build. 2577 [Ben Laurie] 2578 2579 *) Revamp of STACK to provide stronger type-checking. Still to come: 2580 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 2581 ASN1_STRING, CONF_VALUE. 2582 [Ben Laurie] 2583 2584 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 2585 RAM on SSL connections. This option can save about 34k per idle SSL. 2586 [Nick Mathewson] 2587 2588 *) Revamp of LHASH to provide stronger type-checking. Still to come: 2589 STACK, TXT_DB, bsearch, qsort. 2590 [Ben Laurie] 2591 2592 *) Initial support for Cryptographic Message Syntax (aka CMS) based 2593 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 2594 support for data, signedData, compressedData, digestedData and 2595 encryptedData, envelopedData types included. Scripts to check against 2596 RFC4134 examples draft and interop and consistency checks of many 2597 content types and variants. 2598 [Steve Henson] 2599 2600 *) Add options to enc utility to support use of zlib compression BIO. 2601 [Steve Henson] 2602 2603 *) Extend mk1mf to support importing of options and assembly language 2604 files from Configure script, currently only included in VC-WIN32. 2605 The assembly language rules can now optionally generate the source 2606 files from the associated perl scripts. 2607 [Steve Henson] 2608 2609 *) Implement remaining functionality needed to support GOST ciphersuites. 2610 Interop testing has been performed using CryptoPro implementations. 2611 [Victor B. Wagner <vitus@cryptocom.ru>] 2612 2613 *) s390x assembler pack. 2614 [Andy Polyakov] 2615 2616 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 2617 "family." 2618 [Andy Polyakov] 2619 2620 *) Implement Opaque PRF Input TLS extension as specified in 2621 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 2622 official specification yet and no extension type assignment by 2623 IANA exists, this extension (for now) will have to be explicitly 2624 enabled when building OpenSSL by providing the extension number 2625 to use. For example, specify an option 2626 2627 -DTLSEXT_TYPE_opaque_prf_input=0x9527 2628 2629 to the "config" or "Configure" script to enable the extension, 2630 assuming extension number 0x9527 (which is a completely arbitrary 2631 and unofficial assignment based on the MD5 hash of the Internet 2632 Draft). Note that by doing so, you potentially lose 2633 interoperability with other TLS implementations since these might 2634 be using the same extension number for other purposes. 2635 2636 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 2637 opaque PRF input value to use in the handshake. This will create 2638 an interal copy of the length-'len' string at 'src', and will 2639 return non-zero for success. 2640 2641 To get more control and flexibility, provide a callback function 2642 by using 2643 2644 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 2645 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 2646 2647 where 2648 2649 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 2650 void *arg; 2651 2652 Callback function 'cb' will be called in handshakes, and is 2653 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 2654 Argument 'arg' is for application purposes (the value as given to 2655 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 2656 be provided to the callback function). The callback function 2657 has to return non-zero to report success: usually 1 to use opaque 2658 PRF input just if possible, or 2 to enforce use of the opaque PRF 2659 input. In the latter case, the library will abort the handshake 2660 if opaque PRF input is not successfully negotiated. 2661 2662 Arguments 'peerinput' and 'len' given to the callback function 2663 will always be NULL and 0 in the case of a client. A server will 2664 see the client's opaque PRF input through these variables if 2665 available (NULL and 0 otherwise). Note that if the server 2666 provides an opaque PRF input, the length must be the same as the 2667 length of the client's opaque PRF input. 2668 2669 Note that the callback function will only be called when creating 2670 a new session (session resumption can resume whatever was 2671 previously negotiated), and will not be called in SSL 2.0 2672 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 2673 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 2674 for applications that need to enforce opaque PRF input. 2675 2676 [Bodo Moeller] 2677 2678 *) Update ssl code to support digests other than SHA1+MD5 for handshake 2679 MAC. 2680 2681 [Victor B. Wagner <vitus@cryptocom.ru>] 2682 2683 *) Add RFC4507 support to OpenSSL. This includes the corrections in 2684 RFC4507bis. The encrypted ticket format is an encrypted encoded 2685 SSL_SESSION structure, that way new session features are automatically 2686 supported. 2687 2688 If a client application caches session in an SSL_SESSION structure 2689 support is transparent because tickets are now stored in the encoded 2690 SSL_SESSION. 2691 2692 The SSL_CTX structure automatically generates keys for ticket 2693 protection in servers so again support should be possible 2694 with no application modification. 2695 2696 If a client or server wishes to disable RFC4507 support then the option 2697 SSL_OP_NO_TICKET can be set. 2698 2699 Add a TLS extension debugging callback to allow the contents of any client 2700 or server extensions to be examined. 2701 2702 This work was sponsored by Google. 2703 [Steve Henson] 2704 2705 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 2706 OpenSSL should now compile cleanly on gcc 4.2 2707 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 2708 2709 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 2710 support including streaming MAC support: this is required for GOST 2711 ciphersuite support. 2712 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 2713 2714 *) Add option -stream to use PKCS#7 streaming in smime utility. New 2715 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 2716 to output in BER and PEM format. 2717 [Steve Henson] 2718 2719 *) Experimental support for use of HMAC via EVP_PKEY interface. This 2720 allows HMAC to be handled via the EVP_DigestSign*() interface. The 2721 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 2722 ENGINE support for HMAC keys which are unextractable. New -mac and 2723 -macopt options to dgst utility. 2724 [Steve Henson] 2725 2726 *) New option -sigopt to dgst utility. Update dgst to use 2727 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 2728 alternative signing paramaters such as X9.31 or PSS in the dgst 2729 utility. 2730 [Steve Henson] 2731 2732 *) Change ssl_cipher_apply_rule(), the internal function that does 2733 the work each time a ciphersuite string requests enabling 2734 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 2735 removing ("!foo+bar") a class of ciphersuites: Now it maintains 2736 the order of disabled ciphersuites such that those ciphersuites 2737 that most recently went from enabled to disabled not only stay 2738 in order with respect to each other, but also have higher priority 2739 than other disabled ciphersuites the next time ciphersuites are 2740 enabled again. 2741 2742 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 2743 the same ciphersuites as with "HIGH" alone, but in a specific 2744 order where the PSK ciphersuites come first (since they are the 2745 most recently disabled ciphersuites when "HIGH" is parsed). 2746 2747 Also, change ssl_create_cipher_list() (using this new 2748 funcionality) such that between otherwise identical 2749 cihpersuites, ephemeral ECDH is preferred over ephemeral DH in 2750 the default order. 2751 [Bodo Moeller] 2752 2753 *) Change ssl_create_cipher_list() so that it automatically 2754 arranges the ciphersuites in reasonable order before starting 2755 to process the rule string. Thus, the definition for "DEFAULT" 2756 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 2757 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 2758 This makes it much easier to arrive at a reasonable default order 2759 in applications for which anonymous ciphers are OK (meaning 2760 that you can't actually use DEFAULT). 2761 [Bodo Moeller; suggested by Victor Duchovni] 2762 2763 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 2764 processing) into multiple integers instead of setting 2765 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 2766 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 2767 (These masks as well as the individual bit definitions are hidden 2768 away into the non-exported interface ssl/ssl_locl.h, so this 2769 change to the definition of the SSL_CIPHER structure shouldn't 2770 affect applications.) This give us more bits for each of these 2771 categories, so there is no longer a need to coagulate AES128 and 2772 AES256 into a single algorithm bit, and to coagulate Camellia128 2773 and Camellia256 into a single algorithm bit, which has led to all 2774 kinds of kludges. 2775 2776 Thus, among other things, the kludge introduced in 0.9.7m and 2777 0.9.8e for masking out AES256 independently of AES128 or masking 2778 out Camellia256 independently of AES256 is not needed here in 0.9.9. 2779 2780 With the change, we also introduce new ciphersuite aliases that 2781 so far were missing: "AES128", "AES256", "CAMELLIA128", and 2782 "CAMELLIA256". 2783 [Bodo Moeller] 2784 2785 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 2786 Use the leftmost N bytes of the signature input if the input is 2787 larger than the prime q (with N being the size in bytes of q). 2788 [Nils Larsch] 2789 2790 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 2791 it yet and it is largely untested. 2792 [Steve Henson] 2793 2794 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 2795 [Nils Larsch] 2796 2797 *) Initial incomplete changes to avoid need for function casts in OpenSSL 2798 some compilers (gcc 4.2 and later) reject their use. Safestack is 2799 reimplemented. Update ASN1 to avoid use of legacy functions. 2800 [Steve Henson] 2801 2802 *) Win32/64 targets are linked with Winsock2. 2803 [Andy Polyakov] 2804 2805 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 2806 to external functions. This can be used to increase CRL handling 2807 efficiency especially when CRLs are very large by (for example) storing 2808 the CRL revoked certificates in a database. 2809 [Steve Henson] 2810 2811 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 2812 new CRLs added to a directory can be used. New command line option 2813 -verify_return_error to s_client and s_server. This causes real errors 2814 to be returned by the verify callback instead of carrying on no matter 2815 what. This reflects the way a "real world" verify callback would behave. 2816 [Steve Henson] 2817 2818 *) GOST engine, supporting several GOST algorithms and public key formats. 2819 Kindly donated by Cryptocom. 2820 [Cryptocom] 2821 2822 *) Partial support for Issuing Distribution Point CRL extension. CRLs 2823 partitioned by DP are handled but no indirect CRL or reason partitioning 2824 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 2825 selected via a scoring technique which handles IDP and AKID in CRLs. 2826 [Steve Henson] 2827 2828 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 2829 will ultimately be used for all verify operations: this will remove the 2830 X509_STORE dependency on certificate verification and allow alternative 2831 lookup methods. X509_STORE based implementations of these two callbacks. 2832 [Steve Henson] 2833 2834 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 2835 Modify get_crl() to find a valid (unexpired) CRL if possible. 2836 [Steve Henson] 2837 2838 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 2839 this would be called X509_CRL_cmp() but that name is already used by 2840 a function that just compares CRL issuer names. Cache several CRL 2841 extensions in X509_CRL structure and cache CRLDP in X509. 2842 [Steve Henson] 2843 2844 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 2845 this maps equivalent X509_NAME structures into a consistent structure. 2846 Name comparison can then be performed rapidly using memcmp(). 2847 [Steve Henson] 2848 2849 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 2850 utility. 2851 [Steve Henson] 2852 2853 *) Allow digests to supply their own micalg string for S/MIME type using 2854 the ctrl EVP_MD_CTRL_MICALG. 2855 [Steve Henson] 2856 2857 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 2858 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 2859 ctrl. It can then customise the structure before and/or after signing 2860 if necessary. 2861 [Steve Henson] 2862 2863 *) New function OBJ_add_sigid() to allow application defined signature OIDs 2864 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 2865 to free up any added signature OIDs. 2866 [Steve Henson] 2867 2868 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 2869 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 2870 digest and cipher tables. New options added to openssl utility: 2871 list-message-digest-algorithms and list-cipher-algorithms. 2872 [Steve Henson] 2873 2874 *) Change the array representation of binary polynomials: the list 2875 of degrees of non-zero coefficients is now terminated with -1. 2876 Previously it was terminated with 0, which was also part of the 2877 value; thus, the array representation was not applicable to 2878 polynomials where t^0 has coefficient zero. This change makes 2879 the array representation useful in a more general context. 2880 [Douglas Stebila] 2881 2882 *) Various modifications and fixes to SSL/TLS cipher string 2883 handling. For ECC, the code now distinguishes between fixed ECDH 2884 with RSA certificates on the one hand and with ECDSA certificates 2885 on the other hand, since these are separate ciphersuites. The 2886 unused code for Fortezza ciphersuites has been removed. 2887 2888 For consistency with EDH, ephemeral ECDH is now called "EECDH" 2889 (not "ECDHE"). For consistency with the code for DH 2890 certificates, use of ECDH certificates is now considered ECDH 2891 authentication, not RSA or ECDSA authentication (the latter is 2892 merely the CA's signing algorithm and not actively used in the 2893 protocol). 2894 2895 The temporary ciphersuite alias "ECCdraft" is no longer 2896 available, and ECC ciphersuites are no longer excluded from "ALL" 2897 and "DEFAULT". The following aliases now exist for RFC 4492 2898 ciphersuites, most of these by analogy with the DH case: 2899 2900 kECDHr - ECDH cert, signed with RSA 2901 kECDHe - ECDH cert, signed with ECDSA 2902 kECDH - ECDH cert (signed with either RSA or ECDSA) 2903 kEECDH - ephemeral ECDH 2904 ECDH - ECDH cert or ephemeral ECDH 2905 2906 aECDH - ECDH cert 2907 aECDSA - ECDSA cert 2908 ECDSA - ECDSA cert 2909 2910 AECDH - anonymous ECDH 2911 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 2912 2913 [Bodo Moeller] 2914 2915 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 2916 Use correct micalg parameters depending on digest(s) in signed message. 2917 [Steve Henson] 2918 2919 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 2920 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 2921 [Steve Henson] 2922 2923 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 2924 an engine to register a method. Add ENGINE lookups for methods and 2925 functional reference processing. 2926 [Steve Henson] 2927 2928 *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of 2929 EVP_{Sign,Verify}* which allow an application to customise the signature 2930 process. 2931 [Steve Henson] 2932 2933 *) New -resign option to smime utility. This adds one or more signers 2934 to an existing PKCS#7 signedData structure. Also -md option to use an 2935 alternative message digest algorithm for signing. 2936 [Steve Henson] 2937 2938 *) Tidy up PKCS#7 routines and add new functions to make it easier to 2939 create PKCS7 structures containing multiple signers. Update smime 2940 application to support multiple signers. 2941 [Steve Henson] 2942 2943 *) New -macalg option to pkcs12 utility to allow setting of an alternative 2944 digest MAC. 2945 [Steve Henson] 2946 2947 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 2948 Reorganize PBE internals to lookup from a static table using NIDs, 2949 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 2950 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 2951 PRF which will be automatically used with PBES2. 2952 [Steve Henson] 2953 2954 *) Replace the algorithm specific calls to generate keys in "req" with the 2955 new API. 2956 [Steve Henson] 2957 2958 *) Update PKCS#7 enveloped data routines to use new API. This is now 2959 supported by any public key method supporting the encrypt operation. A 2960 ctrl is added to allow the public key algorithm to examine or modify 2961 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 2962 a no op. 2963 [Steve Henson] 2964 2965 *) Add a ctrl to asn1 method to allow a public key algorithm to express 2966 a default digest type to use. In most cases this will be SHA1 but some 2967 algorithms (such as GOST) need to specify an alternative digest. The 2968 return value indicates how strong the prefernce is 1 means optional and 2969 2 is mandatory (that is it is the only supported type). Modify 2970 ASN1_item_sign() to accept a NULL digest argument to indicate it should 2971 use the default md. Update openssl utilities to use the default digest 2972 type for signing if it is not explicitly indicated. 2973 [Steve Henson] 2974 2975 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 2976 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 2977 signing method from the key type. This effectively removes the link 2978 between digests and public key types. 2979 [Steve Henson] 2980 2981 *) Add an OID cross reference table and utility functions. Its purpose is to 2982 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 2983 rsaEncryption. This will allow some of the algorithm specific hackery 2984 needed to use the correct OID to be removed. 2985 [Steve Henson] 2986 2987 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 2988 structures for PKCS7_sign(). They are now set up by the relevant public 2989 key ASN1 method. 2990 [Steve Henson] 2991 2992 *) Add provisional EC pkey method with support for ECDSA and ECDH. 2993 [Steve Henson] 2994 2995 *) Add support for key derivation (agreement) in the API, DH method and 2996 pkeyutl. 2997 [Steve Henson] 2998 2999 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 3000 public and private key formats. As a side effect these add additional 3001 command line functionality not previously available: DSA signatures can be 3002 generated and verified using pkeyutl and DH key support and generation in 3003 pkey, genpkey. 3004 [Steve Henson] 3005 3006 *) BeOS support. 3007 [Oliver Tappe <zooey@hirschkaefer.de>] 3008 3009 *) New make target "install_html_docs" installs HTML renditions of the 3010 manual pages. 3011 [Oliver Tappe <zooey@hirschkaefer.de>] 3012 3013 *) New utility "genpkey" this is analagous to "genrsa" etc except it can 3014 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 3015 support key and parameter generation and add initial key generation 3016 functionality for RSA. 3017 [Steve Henson] 3018 3019 *) Add functions for main EVP_PKEY_method operations. The undocumented 3020 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 3021 EVP_PKEY_{encrypt,decrypt}_old. 3022 [Steve Henson] 3023 3024 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 3025 key API, doesn't do much yet. 3026 [Steve Henson] 3027 3028 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 3029 public key algorithms. New option to openssl utility: 3030 "list-public-key-algorithms" to print out info. 3031 [Steve Henson] 3032 3033 *) Implement the Supported Elliptic Curves Extension for 3034 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 3035 [Douglas Stebila] 3036 3037 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 3038 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 3039 [Steve Henson] 3040 3041 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 3042 utilities such as rsa, dsa, dsaparam etc except they process any key 3043 type. 3044 [Steve Henson] 3045 3046 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 3047 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 3048 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 3049 structure. 3050 [Steve Henson] 3051 3052 *) Initial support for pluggable public key ASN1. 3053 De-spaghettify the public key ASN1 handling. Move public and private 3054 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 3055 algorithm specific handling to a single module within the relevant 3056 algorithm directory. Add functions to allow (near) opaque processing 3057 of public and private key structures. 3058 [Steve Henson] 3059 3060 *) Implement the Supported Point Formats Extension for 3061 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 3062 [Douglas Stebila] 3063 3064 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 3065 for the psk identity [hint] and the psk callback functions to the 3066 SSL_SESSION, SSL and SSL_CTX structure. 3067 3068 New ciphersuites: 3069 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 3070 PSK-AES256-CBC-SHA 3071 3072 New functions: 3073 SSL_CTX_use_psk_identity_hint 3074 SSL_get_psk_identity_hint 3075 SSL_get_psk_identity 3076 SSL_use_psk_identity_hint 3077 3078 [Mika Kousa and Pasi Eronen of Nokia Corporation] 3079 3080 *) Add RFC 3161 compliant time stamp request creation, response generation 3081 and response verification functionality. 3082 [Zolt��n Gl��zik <zglozik@opentsa.org>, The OpenTSA Project] 3083 3084 *) Add initial support for TLS extensions, specifically for the server_name 3085 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 3086 have new members for a host name. The SSL data structure has an 3087 additional member SSL_CTX *initial_ctx so that new sessions can be 3088 stored in that context to allow for session resumption, even after the 3089 SSL has been switched to a new SSL_CTX in reaction to a client's 3090 server_name extension. 3091 3092 New functions (subject to change): 3093 3094 SSL_get_servername() 3095 SSL_get_servername_type() 3096 SSL_set_SSL_CTX() 3097 3098 New CTRL codes and macros (subject to change): 3099 3100 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 3101 - SSL_CTX_set_tlsext_servername_callback() 3102 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 3103 - SSL_CTX_set_tlsext_servername_arg() 3104 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 3105 3106 openssl s_client has a new '-servername ...' option. 3107 3108 openssl s_server has new options '-servername_host ...', '-cert2 ...', 3109 '-key2 ...', '-servername_fatal' (subject to change). This allows 3110 testing the HostName extension for a specific single host name ('-cert' 3111 and '-key' remain fallbacks for handshakes without HostName 3112 negotiation). If the unrecogninzed_name alert has to be sent, this by 3113 default is a warning; it becomes fatal with the '-servername_fatal' 3114 option. 3115 3116 [Peter Sylvester, Remy Allais, Christophe Renou] 3117 3118 *) Whirlpool hash implementation is added. 3119 [Andy Polyakov] 3120 3121 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 3122 bn(64,32). Because of instruction set limitations it doesn't have 3123 any negative impact on performance. This was done mostly in order 3124 to make it possible to share assembler modules, such as bn_mul_mont 3125 implementations, between 32- and 64-bit builds without hassle. 3126 [Andy Polyakov] 3127 3128 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 3129 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 3130 macro. 3131 [Bodo Moeller] 3132 3133 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 3134 dedicated Montgomery multiplication procedure, is introduced. 3135 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 3136 "64-bit" performance on certain 32-bit targets. 3137 [Andy Polyakov] 3138 3139 *) New option SSL_OP_NO_COMP to disable use of compression selectively 3140 in SSL structures. New SSL ctrl to set maximum send fragment size. 3141 Save memory by seeting the I/O buffer sizes dynamically instead of 3142 using the maximum available value. 3143 [Steve Henson] 3144 3145 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 3146 in addition to the text details. 3147 [Bodo Moeller] 3148 3149 *) Very, very preliminary EXPERIMENTAL support for printing of general 3150 ASN1 structures. This currently produces rather ugly output and doesn't 3151 handle several customised structures at all. 3152 [Steve Henson] 3153 3154 *) Integrated support for PVK file format and some related formats such 3155 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 3156 these in the 'rsa' and 'dsa' utilities. 3157 [Steve Henson] 3158 3159 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 3160 [Steve Henson] 3161 3162 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 3163 place for the (very old) "NETSCAPE" format certificates which are now 3164 handled using new ASN1 code equivalents. 3165 [Steve Henson] 3166 3167 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 3168 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 3169 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 3170 [Nils Larsch] 3171 3172 *) Modify CRL distribution points extension code to print out previously 3173 unsupported fields. Enhance extension setting code to allow setting of 3174 all fields. 3175 [Steve Henson] 3176 3177 *) Add print and set support for Issuing Distribution Point CRL extension. 3178 [Steve Henson] 3179 3180 *) Change 'Configure' script to enable Camellia by default. 3181 [NTT] 3182 3183 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 3184 3185 *) When rejecting SSL/TLS records due to an incorrect version number, never 3186 update s->server with a new major version number. As of 3187 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 3188 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 3189 the previous behavior could result in a read attempt at NULL when 3190 receiving specific incorrect SSL/TLS records once record payload 3191 protection is active. (CVE-2010-0740) 3192 [Bodo Moeller, Adam Langley <agl@chromium.org>] 3193 3194 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 3195 could be crashed if the relevant tables were not present (e.g. chrooted). 3196 [Tomas Hoger <thoger@redhat.com>] 3197 3198 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 3199 3200 *) Always check bn_wexpend() return values for failure. (CVE-2009-3245) 3201 [Martin Olsson, Neel Mehta] 3202 3203 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 3204 accommodate for stack sorting, always a write lock!). 3205 [Bodo Moeller] 3206 3207 *) On some versions of WIN32 Heap32Next is very slow. This can cause 3208 excessive delays in the RAND_poll(): over a minute. As a workaround 3209 include a time check in the inner Heap32Next loop too. 3210 [Steve Henson] 3211 3212 *) The code that handled flushing of data in SSL/TLS originally used the 3213 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 3214 the problem outlined in PR#1949. The fix suggested there however can 3215 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 3216 of Apache). So instead simplify the code to flush unconditionally. 3217 This should be fine since flushing with no data to flush is a no op. 3218 [Steve Henson] 3219 3220 *) Handle TLS versions 2.0 and later properly and correctly use the 3221 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 3222 off ancient servers have a habit of sticking around for a while... 3223 [Steve Henson] 3224 3225 *) Modify compression code so it frees up structures without using the 3226 ex_data callbacks. This works around a problem where some applications 3227 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 3228 restarting) then use compression (e.g. SSL with compression) later. 3229 This results in significant per-connection memory leaks and 3230 has caused some security issues including CVE-2008-1678 and 3231 CVE-2009-4355. 3232 [Steve Henson] 3233 3234 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 3235 change when encrypting or decrypting. 3236 [Bodo Moeller] 3237 3238 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 3239 connect and renegotiate with servers which do not support RI. 3240 Until RI is more widely deployed this option is enabled by default. 3241 [Steve Henson] 3242 3243 *) Add "missing" ssl ctrls to clear options and mode. 3244 [Steve Henson] 3245 3246 *) If client attempts to renegotiate and doesn't support RI respond with 3247 a no_renegotiation alert as required by RFC5746. Some renegotiating 3248 TLS clients will continue a connection gracefully when they receive 3249 the alert. Unfortunately OpenSSL mishandled this alert and would hang 3250 waiting for a server hello which it will never receive. Now we treat a 3251 received no_renegotiation alert as a fatal error. This is because 3252 applications requesting a renegotiation might well expect it to succeed 3253 and would have no code in place to handle the server denying it so the 3254 only safe thing to do is to terminate the connection. 3255 [Steve Henson] 3256 3257 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 3258 peer supports secure renegotiation and 0 otherwise. Print out peer 3259 renegotiation support in s_client/s_server. 3260 [Steve Henson] 3261 3262 *) Replace the highly broken and deprecated SPKAC certification method with 3263 the updated NID creation version. This should correctly handle UTF8. 3264 [Steve Henson] 3265 3266 *) Implement RFC5746. Re-enable renegotiation but require the extension 3267 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 3268 turns out to be a bad idea. It has been replaced by 3269 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 3270 SSL_CTX_set_options(). This is really not recommended unless you 3271 know what you are doing. 3272 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 3273 3274 *) Fixes to stateless session resumption handling. Use initial_ctx when 3275 issuing and attempting to decrypt tickets in case it has changed during 3276 servername handling. Use a non-zero length session ID when attempting 3277 stateless session resumption: this makes it possible to determine if 3278 a resumption has occurred immediately after receiving server hello 3279 (several places in OpenSSL subtly assume this) instead of later in 3280 the handshake. 3281 [Steve Henson] 3282 3283 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 3284 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 3285 fixes for a few places where the return code is not checked 3286 correctly. 3287 [Julia Lawall <julia@diku.dk>] 3288 3289 *) Add --strict-warnings option to Configure script to include devteam 3290 warnings in other configurations. 3291 [Steve Henson] 3292 3293 *) Add support for --libdir option and LIBDIR variable in makefiles. This 3294 makes it possible to install openssl libraries in locations which 3295 have names other than "lib", for example "/usr/lib64" which some 3296 systems need. 3297 [Steve Henson, based on patch from Jeremy Utley] 3298 3299 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 3300 X690 8.9.12 and can produce some misleading textual output of OIDs. 3301 [Steve Henson, reported by Dan Kaminsky] 3302 3303 *) Delete MD2 from algorithm tables. This follows the recommendation in 3304 several standards that it is not used in new applications due to 3305 several cryptographic weaknesses. For binary compatibility reasons 3306 the MD2 API is still compiled in by default. 3307 [Steve Henson] 3308 3309 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 3310 and restored. 3311 [Steve Henson] 3312 3313 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 3314 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 3315 clash. 3316 [Guenter <lists@gknw.net>] 3317 3318 *) Fix the server certificate chain building code to use X509_verify_cert(), 3319 it used to have an ad-hoc builder which was unable to cope with anything 3320 other than a simple chain. 3321 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 3322 3323 *) Don't check self signed certificate signatures in X509_verify_cert() 3324 by default (a flag can override this): it just wastes time without 3325 adding any security. As a useful side effect self signed root CAs 3326 with non-FIPS digests are now usable in FIPS mode. 3327 [Steve Henson] 3328 3329 *) In dtls1_process_out_of_seq_message() the check if the current message 3330 is already buffered was missing. For every new message was memory 3331 allocated, allowing an attacker to perform an denial of service attack 3332 with sending out of seq handshake messages until there is no memory 3333 left. Additionally every future messege was buffered, even if the 3334 sequence number made no sense and would be part of another handshake. 3335 So only messages with sequence numbers less than 10 in advance will be 3336 buffered. (CVE-2009-1378) 3337 [Robin Seggelmann, discovered by Daniel Mentz] 3338 3339 *) Records are buffered if they arrive with a future epoch to be 3340 processed after finishing the corresponding handshake. There is 3341 currently no limitation to this buffer allowing an attacker to perform 3342 a DOS attack with sending records with future epochs until there is no 3343 memory left. This patch adds the pqueue_size() function to detemine 3344 the size of a buffer and limits the record buffer to 100 entries. 3345 (CVE-2009-1377) 3346 [Robin Seggelmann, discovered by Daniel Mentz] 3347 3348 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 3349 parent structure is freed. (CVE-2009-1379) 3350 [Daniel Mentz] 3351 3352 *) Handle non-blocking I/O properly in SSL_shutdown() call. 3353 [Darryl Miles <darryl-mailinglists@netbauds.net>] 3354 3355 *) Add 2.5.4.* OIDs 3356 [Ilya O. <vrghost@gmail.com>] 3357 3358 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 3359 3360 *) Disable renegotiation completely - this fixes a severe security 3361 problem (CVE-2009-3555) at the cost of breaking all 3362 renegotiation. Renegotiation can be re-enabled by setting 3363 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 3364 run-time. This is really not recommended unless you know what 3365 you're doing. 3366 [Ben Laurie] 3367 3368 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 3369 3370 *) Don't set val to NULL when freeing up structures, it is freed up by 3371 underlying code. If sizeof(void *) > sizeof(long) this can result in 3372 zeroing past the valid field. (CVE-2009-0789) 3373 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 3374 3375 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 3376 checked correctly. This would allow some invalid signed attributes to 3377 appear to verify correctly. (CVE-2009-0591) 3378 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3379 3380 *) Reject UniversalString and BMPString types with invalid lengths. This 3381 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 3382 a legal length. (CVE-2009-0590) 3383 [Steve Henson] 3384 3385 *) Set S/MIME signing as the default purpose rather than setting it 3386 unconditionally. This allows applications to override it at the store 3387 level. 3388 [Steve Henson] 3389 3390 *) Permit restricted recursion of ASN1 strings. This is needed in practice 3391 to handle some structures. 3392 [Steve Henson] 3393 3394 *) Improve efficiency of mem_gets: don't search whole buffer each time 3395 for a '\n' 3396 [Jeremy Shapiro <jnshapir@us.ibm.com>] 3397 3398 *) New -hex option for openssl rand. 3399 [Matthieu Herrb] 3400 3401 *) Print out UTF8String and NumericString when parsing ASN1. 3402 [Steve Henson] 3403 3404 *) Support NumericString type for name components. 3405 [Steve Henson] 3406 3407 *) Allow CC in the environment to override the automatically chosen 3408 compiler. Note that nothing is done to ensure flags work with the 3409 chosen compiler. 3410 [Ben Laurie] 3411 3412 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 3413 3414 *) Properly check EVP_VerifyFinal() and similar return values 3415 (CVE-2008-5077). 3416 [Ben Laurie, Bodo Moeller, Google Security Team] 3417 3418 *) Enable TLS extensions by default. 3419 [Ben Laurie] 3420 3421 *) Allow the CHIL engine to be loaded, whether the application is 3422 multithreaded or not. (This does not release the developer from the 3423 obligation to set up the dynamic locking callbacks.) 3424 [Sander Temme <sander@temme.net>] 3425 3426 *) Use correct exit code if there is an error in dgst command. 3427 [Steve Henson; problem pointed out by Roland Dirlewanger] 3428 3429 *) Tweak Configure so that you need to say "experimental-jpake" to enable 3430 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 3431 [Bodo Moeller] 3432 3433 *) Add experimental JPAKE support, including demo authentication in 3434 s_client and s_server. 3435 [Ben Laurie] 3436 3437 *) Set the comparison function in v3_addr_canonize(). 3438 [Rob Austein <sra@hactrn.net>] 3439 3440 *) Add support for XMPP STARTTLS in s_client. 3441 [Philip Paeps <philip@freebsd.org>] 3442 3443 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 3444 to ensure that even with this option, only ciphersuites in the 3445 server's preference list will be accepted. (Note that the option 3446 applies only when resuming a session, so the earlier behavior was 3447 just about the algorithm choice for symmetric cryptography.) 3448 [Bodo Moeller] 3449 3450 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 3451 3452 *) Fix NULL pointer dereference if a DTLS server received 3453 ChangeCipherSpec as first record (CVE-2009-1386). 3454 [PR #1679] 3455 3456 *) Fix a state transitition in s3_srvr.c and d1_srvr.c 3457 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 3458 [Nagendra Modadugu] 3459 3460 *) The fix in 0.9.8c that supposedly got rid of unsafe 3461 double-checked locking was incomplete for RSA blinding, 3462 addressing just one layer of what turns out to have been 3463 doubly unsafe triple-checked locking. 3464 3465 So now fix this for real by retiring the MONT_HELPER macro 3466 in crypto/rsa/rsa_eay.c. 3467 3468 [Bodo Moeller; problem pointed out by Marius Schilder] 3469 3470 *) Various precautionary measures: 3471 3472 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 3473 3474 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 3475 (NB: This would require knowledge of the secret session ticket key 3476 to exploit, in which case you'd be SOL either way.) 3477 3478 - Change bn_nist.c so that it will properly handle input BIGNUMs 3479 outside the expected range. 3480 3481 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 3482 builds. 3483 3484 [Neel Mehta, Bodo Moeller] 3485 3486 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 3487 the load fails. Useful for distros. 3488 [Ben Laurie and the FreeBSD team] 3489 3490 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 3491 [Steve Henson] 3492 3493 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 3494 [Huang Ying] 3495 3496 *) Expand ENGINE to support engine supplied SSL client certificate functions. 3497 3498 This work was sponsored by Logica. 3499 [Steve Henson] 3500 3501 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 3502 keystores. Support for SSL/TLS client authentication too. 3503 Not compiled unless enable-capieng specified to Configure. 3504 3505 This work was sponsored by Logica. 3506 [Steve Henson] 3507 3508 *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using 3509 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 3510 attribute creation routines such as certifcate requests and PKCS#12 3511 files. 3512 [Steve Henson] 3513 3514 Changes between 0.9.8g and 0.9.8h [28 May 2008] 3515 3516 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 3517 handshake which could lead to a cilent crash as found using the 3518 Codenomicon TLS test suite (CVE-2008-1672) 3519 [Steve Henson, Mark Cox] 3520 3521 *) Fix double free in TLS server name extensions which could lead to 3522 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 3523 [Joe Orton] 3524 3525 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 3526 3527 Clear the error queue to ensure that error entries left from 3528 older function calls do not interfere with the correct operation. 3529 [Lutz Jaenicke, Erik de Castro Lopo] 3530 3531 *) Remove root CA certificates of commercial CAs: 3532 3533 The OpenSSL project does not recommend any specific CA and does not 3534 have any policy with respect to including or excluding any CA. 3535 Therefore it does not make any sense to ship an arbitrary selection 3536 of root CA certificates with the OpenSSL software. 3537 [Lutz Jaenicke] 3538 3539 *) RSA OAEP patches to fix two separate invalid memory reads. 3540 The first one involves inputs when 'lzero' is greater than 3541 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 3542 before the beginning of from). The second one involves inputs where 3543 the 'db' section contains nothing but zeroes (there is a one-byte 3544 invalid read after the end of 'db'). 3545 [Ivan Nestlerode <inestlerode@us.ibm.com>] 3546 3547 *) Partial backport from 0.9.9-dev: 3548 3549 Introduce bn_mul_mont (dedicated Montgomery multiplication 3550 procedure) as a candidate for BIGNUM assembler implementation. 3551 While 0.9.9-dev uses assembler for various architectures, only 3552 x86_64 is available by default here in the 0.9.8 branch, and 3553 32-bit x86 is available through a compile-time setting. 3554 3555 To try the 32-bit x86 assembler implementation, use Configure 3556 option "enable-montasm" (which exists only for this backport). 3557 3558 As "enable-montasm" for 32-bit x86 disclaims code stability 3559 anyway, in this constellation we activate additional code 3560 backported from 0.9.9-dev for further performance improvements, 3561 namely BN_from_montgomery_word. (To enable this otherwise, 3562 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 3563 3564 [Andy Polyakov (backport partially by Bodo Moeller)] 3565 3566 *) Add TLS session ticket callback. This allows an application to set 3567 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 3568 values. This is useful for key rollover for example where several key 3569 sets may exist with different names. 3570 [Steve Henson] 3571 3572 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 3573 This was broken until now in 0.9.8 releases, such that the only way 3574 a registered ENGINE could be used (assuming it initialises 3575 successfully on the host) was to explicitly set it as the default 3576 for the relevant algorithms. This is in contradiction with 0.9.7 3577 behaviour and the documentation. With this fix, when an ENGINE is 3578 registered into a given algorithm's table of implementations, the 3579 'uptodate' flag is reset so that auto-discovery will be used next 3580 time a new context for that algorithm attempts to select an 3581 implementation. 3582 [Ian Lister (tweaked by Geoff Thorpe)] 3583 3584 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 3585 implemention in the following ways: 3586 3587 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 3588 hard coded. 3589 3590 Lack of BER streaming support means one pass streaming processing is 3591 only supported if data is detached: setting the streaming flag is 3592 ignored for embedded content. 3593 3594 CMS support is disabled by default and must be explicitly enabled 3595 with the enable-cms configuration option. 3596 [Steve Henson] 3597 3598 *) Update the GMP engine glue to do direct copies between BIGNUM and 3599 mpz_t when openssl and GMP use the same limb size. Otherwise the 3600 existing "conversion via a text string export" trick is still used. 3601 [Paul Sheer <paulsheer@gmail.com>] 3602 3603 *) Zlib compression BIO. This is a filter BIO which compressed and 3604 uncompresses any data passed through it. 3605 [Steve Henson] 3606 3607 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 3608 RFC3394 compatible AES key wrapping. 3609 [Steve Henson] 3610 3611 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 3612 sets string data without copying. X509_ALGOR_set0() and 3613 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 3614 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 3615 from an X509_ATTRIBUTE structure optionally checking it occurs only 3616 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 3617 data. 3618 [Steve Henson] 3619 3620 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 3621 to get the expected BN_FLG_CONSTTIME behavior. 3622 [Bodo Moeller (Google)] 3623 3624 *) Netware support: 3625 3626 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 3627 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 3628 - added some more tests to do_tests.pl 3629 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 3630 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 3631 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 3632 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 3633 - various changes to netware.pl to enable gcc-cross builds on Win32 3634 platform 3635 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 3636 - various changes to fix missing prototype warnings 3637 - fixed x86nasm.pl to create correct asm files for NASM COFF output 3638 - added AES, WHIRLPOOL and CPUID assembler code to build files 3639 - added missing AES assembler make rules to mk1mf.pl 3640 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 3641 [Guenter Knauf <eflash@gmx.net>] 3642 3643 *) Implement certificate status request TLS extension defined in RFC3546. 3644 A client can set the appropriate parameters and receive the encoded 3645 OCSP response via a callback. A server can query the supplied parameters 3646 and set the encoded OCSP response in the callback. Add simplified examples 3647 to s_client and s_server. 3648 [Steve Henson] 3649 3650 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 3651 3652 *) Fix various bugs: 3653 + Binary incompatibility of ssl_ctx_st structure 3654 + DTLS interoperation with non-compliant servers 3655 + Don't call get_session_cb() without proposed session 3656 + Fix ia64 assembler code 3657 [Andy Polyakov, Steve Henson] 3658 3659 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 3660 3661 *) DTLS Handshake overhaul. There were longstanding issues with 3662 OpenSSL DTLS implementation, which were making it impossible for 3663 RFC 4347 compliant client to communicate with OpenSSL server. 3664 Unfortunately just fixing these incompatibilities would "cut off" 3665 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 3666 server keeps tolerating non RFC compliant syntax. The opposite is 3667 not true, 0.9.8f client can not communicate with earlier server. 3668 This update even addresses CVE-2007-4995. 3669 [Andy Polyakov] 3670 3671 *) Changes to avoid need for function casts in OpenSSL: some compilers 3672 (gcc 4.2 and later) reject their use. 3673 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 3674 Steve Henson] 3675 3676 *) Add RFC4507 support to OpenSSL. This includes the corrections in 3677 RFC4507bis. The encrypted ticket format is an encrypted encoded 3678 SSL_SESSION structure, that way new session features are automatically 3679 supported. 3680 3681 If a client application caches session in an SSL_SESSION structure 3682 support is transparent because tickets are now stored in the encoded 3683 SSL_SESSION. 3684 3685 The SSL_CTX structure automatically generates keys for ticket 3686 protection in servers so again support should be possible 3687 with no application modification. 3688 3689 If a client or server wishes to disable RFC4507 support then the option 3690 SSL_OP_NO_TICKET can be set. 3691 3692 Add a TLS extension debugging callback to allow the contents of any client 3693 or server extensions to be examined. 3694 3695 This work was sponsored by Google. 3696 [Steve Henson] 3697 3698 *) Add initial support for TLS extensions, specifically for the server_name 3699 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 3700 have new members for a host name. The SSL data structure has an 3701 additional member SSL_CTX *initial_ctx so that new sessions can be 3702 stored in that context to allow for session resumption, even after the 3703 SSL has been switched to a new SSL_CTX in reaction to a client's 3704 server_name extension. 3705 3706 New functions (subject to change): 3707 3708 SSL_get_servername() 3709 SSL_get_servername_type() 3710 SSL_set_SSL_CTX() 3711 3712 New CTRL codes and macros (subject to change): 3713 3714 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 3715 - SSL_CTX_set_tlsext_servername_callback() 3716 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 3717 - SSL_CTX_set_tlsext_servername_arg() 3718 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 3719 3720 openssl s_client has a new '-servername ...' option. 3721 3722 openssl s_server has new options '-servername_host ...', '-cert2 ...', 3723 '-key2 ...', '-servername_fatal' (subject to change). This allows 3724 testing the HostName extension for a specific single host name ('-cert' 3725 and '-key' remain fallbacks for handshakes without HostName 3726 negotiation). If the unrecogninzed_name alert has to be sent, this by 3727 default is a warning; it becomes fatal with the '-servername_fatal' 3728 option. 3729 3730 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 3731 3732 *) Add AES and SSE2 assembly language support to VC++ build. 3733 [Steve Henson] 3734 3735 *) Mitigate attack on final subtraction in Montgomery reduction. 3736 [Andy Polyakov] 3737 3738 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 3739 (which previously caused an internal error). 3740 [Bodo Moeller] 3741 3742 *) Squeeze another 10% out of IGE mode when in != out. 3743 [Ben Laurie] 3744 3745 *) AES IGE mode speedup. 3746 [Dean Gaudet (Google)] 3747 3748 *) Add the Korean symmetric 128-bit cipher SEED (see 3749 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 3750 add SEED ciphersuites from RFC 4162: 3751 3752 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 3753 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 3754 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 3755 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 3756 3757 To minimize changes between patchlevels in the OpenSSL 0.9.8 3758 series, SEED remains excluded from compilation unless OpenSSL 3759 is configured with 'enable-seed'. 3760 [KISA, Bodo Moeller] 3761 3762 *) Mitigate branch prediction attacks, which can be practical if a 3763 single processor is shared, allowing a spy process to extract 3764 information. For detailed background information, see 3765 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 3766 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 3767 and Necessary Software Countermeasures"). The core of the change 3768 are new versions BN_div_no_branch() and 3769 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 3770 respectively, which are slower, but avoid the security-relevant 3771 conditional branches. These are automatically called by BN_div() 3772 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 3773 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 3774 remove a conditional branch. 3775 3776 BN_FLG_CONSTTIME is the new name for the previous 3777 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 3778 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 3779 in the exponent causes BN_mod_exp_mont() to use the alternative 3780 implementation in BN_mod_exp_mont_consttime().) The old name 3781 remains as a deprecated alias. 3782 3783 Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 3784 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 3785 constant-time implementations for more than just exponentiation. 3786 Here too the old name is kept as a deprecated alias. 3787 3788 BN_BLINDING_new() will now use BN_dup() for the modulus so that 3789 the BN_BLINDING structure gets an independent copy of the 3790 modulus. This means that the previous "BIGNUM *m" argument to 3791 BN_BLINDING_new() and to BN_BLINDING_create_param() now 3792 essentially becomes "const BIGNUM *m", although we can't actually 3793 change this in the header file before 0.9.9. It allows 3794 RSA_setup_blinding() to use BN_with_flags() on the modulus to 3795 enable BN_FLG_CONSTTIME. 3796 3797 [Matthew D Wood (Intel Corp)] 3798 3799 *) In the SSL/TLS server implementation, be strict about session ID 3800 context matching (which matters if an application uses a single 3801 external cache for different purposes). Previously, 3802 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 3803 set. This did ensure strict client verification, but meant that, 3804 with applications using a single external cache for quite 3805 different requirements, clients could circumvent ciphersuite 3806 restrictions for a given session ID context by starting a session 3807 in a different context. 3808 [Bodo Moeller] 3809 3810 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 3811 a ciphersuite string such as "DEFAULT:RSA" cannot enable 3812 authentication-only ciphersuites. 3813 [Bodo Moeller] 3814 3815 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 3816 not complete and could lead to a possible single byte overflow 3817 (CVE-2007-5135) [Ben Laurie] 3818 3819 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 3820 3821 *) Since AES128 and AES256 (and similarly Camellia128 and 3822 Camellia256) share a single mask bit in the logic of 3823 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 3824 kludge to work properly if AES128 is available and AES256 isn't 3825 (or if Camellia128 is available and Camellia256 isn't). 3826 [Victor Duchovni] 3827 3828 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 3829 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 3830 When a point or a seed is encoded in a BIT STRING, we need to 3831 prevent the removal of trailing zero bits to get the proper DER 3832 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 3833 of a NamedBitList, for which trailing 0 bits need to be removed.) 3834 [Bodo Moeller] 3835 3836 *) Have SSL/TLS server implementation tolerate "mismatched" record 3837 protocol version while receiving ClientHello even if the 3838 ClientHello is fragmented. (The server can't insist on the 3839 particular protocol version it has chosen before the ServerHello 3840 message has informed the client about his choice.) 3841 [Bodo Moeller] 3842 3843 *) Add RFC 3779 support. 3844 [Rob Austein for ARIN, Ben Laurie] 3845 3846 *) Load error codes if they are not already present instead of using a 3847 static variable. This allows them to be cleanly unloaded and reloaded. 3848 Improve header file function name parsing. 3849 [Steve Henson] 3850 3851 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 3852 or CAPABILITY handshake as required by RFCs. 3853 [Goetz Babin-Ebell] 3854 3855 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 3856 3857 *) Introduce limits to prevent malicious keys being able to 3858 cause a denial of service. (CVE-2006-2940) 3859 [Steve Henson, Bodo Moeller] 3860 3861 *) Fix ASN.1 parsing of certain invalid structures that can result 3862 in a denial of service. (CVE-2006-2937) [Steve Henson] 3863 3864 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 3865 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 3866 3867 *) Fix SSL client code which could crash if connecting to a 3868 malicious SSLv2 server. (CVE-2006-4343) 3869 [Tavis Ormandy and Will Drewry, Google Security Team] 3870 3871 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 3872 match only those. Before that, "AES256-SHA" would be interpreted 3873 as a pattern and match "AES128-SHA" too (since AES128-SHA got 3874 the same strength classification in 0.9.7h) as we currently only 3875 have a single AES bit in the ciphersuite description bitmap. 3876 That change, however, also applied to ciphersuite strings such as 3877 "RC4-MD5" that intentionally matched multiple ciphersuites -- 3878 namely, SSL 2.0 ciphersuites in addition to the more common ones 3879 from SSL 3.0/TLS 1.0. 3880 3881 So we change the selection algorithm again: Naming an explicit 3882 ciphersuite selects this one ciphersuite, and any other similar 3883 ciphersuite (same bitmap) from *other* protocol versions. 3884 Thus, "RC4-MD5" again will properly select both the SSL 2.0 3885 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 3886 3887 Since SSL 2.0 does not have any ciphersuites for which the 3888 128/256 bit distinction would be relevant, this works for now. 3889 The proper fix will be to use different bits for AES128 and 3890 AES256, which would have avoided the problems from the beginning; 3891 however, bits are scarce, so we can only do this in a new release 3892 (not just a patchlevel) when we can change the SSL_CIPHER 3893 definition to split the single 'unsigned long mask' bitmap into 3894 multiple values to extend the available space. 3895 3896 [Bodo Moeller] 3897 3898 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 3899 3900 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 3901 (CVE-2006-4339) [Ben Laurie and Google Security Team] 3902 3903 *) Add AES IGE and biIGE modes. 3904 [Ben Laurie] 3905 3906 *) Change the Unix randomness entropy gathering to use poll() when 3907 possible instead of select(), since the latter has some 3908 undesirable limitations. 3909 [Darryl Miles via Richard Levitte and Bodo Moeller] 3910 3911 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 3912 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 3913 cannot be implicitly activated as part of, e.g., the "AES" alias. 3914 However, please upgrade to OpenSSL 0.9.9[-dev] for 3915 non-experimental use of the ECC ciphersuites to get TLS extension 3916 support, which is required for curve and point format negotiation 3917 to avoid potential handshake problems. 3918 [Bodo Moeller] 3919 3920 *) Disable rogue ciphersuites: 3921 3922 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 3923 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 3924 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 3925 3926 The latter two were purportedly from 3927 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 3928 appear there. 3929 3930 Also deactivate the remaining ciphersuites from 3931 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 3932 unofficial, and the ID has long expired. 3933 [Bodo Moeller] 3934 3935 *) Fix RSA blinding Heisenbug (problems sometimes occured on 3936 dual-core machines) and other potential thread-safety issues. 3937 [Bodo Moeller] 3938 3939 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 3940 versions), which is now available for royalty-free use 3941 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 3942 Also, add Camellia TLS ciphersuites from RFC 4132. 3943 3944 To minimize changes between patchlevels in the OpenSSL 0.9.8 3945 series, Camellia remains excluded from compilation unless OpenSSL 3946 is configured with 'enable-camellia'. 3947 [NTT] 3948 3949 *) Disable the padding bug check when compression is in use. The padding 3950 bug check assumes the first packet is of even length, this is not 3951 necessarily true if compresssion is enabled and can result in false 3952 positives causing handshake failure. The actual bug test is ancient 3953 code so it is hoped that implementations will either have fixed it by 3954 now or any which still have the bug do not support compression. 3955 [Steve Henson] 3956 3957 Changes between 0.9.8a and 0.9.8b [04 May 2006] 3958 3959 *) When applying a cipher rule check to see if string match is an explicit 3960 cipher suite and only match that one cipher suite if it is. 3961 [Steve Henson] 3962 3963 *) Link in manifests for VC++ if needed. 3964 [Austin Ziegler <halostatue@gmail.com>] 3965 3966 *) Update support for ECC-based TLS ciphersuites according to 3967 draft-ietf-tls-ecc-12.txt with proposed changes (but without 3968 TLS extensions, which are supported starting with the 0.9.9 3969 branch, not in the OpenSSL 0.9.8 branch). 3970 [Douglas Stebila] 3971 3972 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 3973 opaque EVP_CIPHER_CTX handling. 3974 [Steve Henson] 3975 3976 *) Fixes and enhancements to zlib compression code. We now only use 3977 "zlib1.dll" and use the default __cdecl calling convention on Win32 3978 to conform with the standards mentioned here: 3979 http://www.zlib.net/DLL_FAQ.txt 3980 Static zlib linking now works on Windows and the new --with-zlib-include 3981 --with-zlib-lib options to Configure can be used to supply the location 3982 of the headers and library. Gracefully handle case where zlib library 3983 can't be loaded. 3984 [Steve Henson] 3985 3986 *) Several fixes and enhancements to the OID generation code. The old code 3987 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 3988 handle numbers larger than ULONG_MAX, truncated printing and had a 3989 non standard OBJ_obj2txt() behaviour. 3990 [Steve Henson] 3991 3992 *) Add support for building of engines under engine/ as shared libraries 3993 under VC++ build system. 3994 [Steve Henson] 3995 3996 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 3997 Hopefully, we will not see any false combination of paths any more. 3998 [Richard Levitte] 3999 4000 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 4001 4002 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 4003 (part of SSL_OP_ALL). This option used to disable the 4004 countermeasure against man-in-the-middle protocol-version 4005 rollback in the SSL 2.0 server implementation, which is a bad 4006 idea. (CVE-2005-2969) 4007 4008 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 4009 for Information Security, National Institute of Advanced Industrial 4010 Science and Technology [AIST], Japan)] 4011 4012 *) Add two function to clear and return the verify parameter flags. 4013 [Steve Henson] 4014 4015 *) Keep cipherlists sorted in the source instead of sorting them at 4016 runtime, thus removing the need for a lock. 4017 [Nils Larsch] 4018 4019 *) Avoid some small subgroup attacks in Diffie-Hellman. 4020 [Nick Mathewson and Ben Laurie] 4021 4022 *) Add functions for well-known primes. 4023 [Nick Mathewson] 4024 4025 *) Extended Windows CE support. 4026 [Satoshi Nakamura and Andy Polyakov] 4027 4028 *) Initialize SSL_METHOD structures at compile time instead of during 4029 runtime, thus removing the need for a lock. 4030 [Steve Henson] 4031 4032 *) Make PKCS7_decrypt() work even if no certificate is supplied by 4033 attempting to decrypt each encrypted key in turn. Add support to 4034 smime utility. 4035 [Steve Henson] 4036 4037 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 4038 4039 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 4040 OpenSSL 0.9.8.] 4041 4042 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 4043 [Richard Levitte] 4044 4045 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 4046 key into the same file any more. 4047 [Richard Levitte] 4048 4049 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 4050 [Andy Polyakov] 4051 4052 *) Add -utf8 command line and config file option to 'ca'. 4053 [Stefan <stf@udoma.org] 4054 4055 *) Removed the macro des_crypt(), as it seems to conflict with some 4056 libraries. Use DES_crypt(). 4057 [Richard Levitte] 4058 4059 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 4060 involves renaming the source and generated shared-libs for 4061 both. The engines will accept the corrected or legacy ids 4062 ('ncipher' and '4758_cca' respectively) when binding. NB, 4063 this only applies when building 'shared'. 4064 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 4065 4066 *) Add attribute functions to EVP_PKEY structure. Modify 4067 PKCS12_create() to recognize a CSP name attribute and 4068 use it. Make -CSP option work again in pkcs12 utility. 4069 [Steve Henson] 4070 4071 *) Add new functionality to the bn blinding code: 4072 - automatic re-creation of the BN_BLINDING parameters after 4073 a fixed number of uses (currently 32) 4074 - add new function for parameter creation 4075 - introduce flags to control the update behaviour of the 4076 BN_BLINDING parameters 4077 - hide BN_BLINDING structure 4078 Add a second BN_BLINDING slot to the RSA structure to improve 4079 performance when a single RSA object is shared among several 4080 threads. 4081 [Nils Larsch] 4082 4083 *) Add support for DTLS. 4084 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 4085 4086 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 4087 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 4088 [Walter Goulet] 4089 4090 *) Remove buggy and incompletet DH cert support from 4091 ssl/ssl_rsa.c and ssl/s3_both.c 4092 [Nils Larsch] 4093 4094 *) Use SHA-1 instead of MD5 as the default digest algorithm for 4095 the apps/openssl applications. 4096 [Nils Larsch] 4097 4098 *) Compile clean with "-Wall -Wmissing-prototypes 4099 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 4100 DEBUG_SAFESTACK must also be set. 4101 [Ben Laurie] 4102 4103 *) Change ./Configure so that certain algorithms can be disabled by default. 4104 The new counterpiece to "no-xxx" is "enable-xxx". 4105 4106 The patented RC5 and MDC2 algorithms will now be disabled unless 4107 "enable-rc5" and "enable-mdc2", respectively, are specified. 4108 4109 (IDEA remains enabled despite being patented. This is because IDEA 4110 is frequently required for interoperability, and there is no license 4111 fee for non-commercial use. As before, "no-idea" can be used to 4112 avoid this algorithm.) 4113 4114 [Bodo Moeller] 4115 4116 *) Add processing of proxy certificates (see RFC 3820). This work was 4117 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 4118 EGEE (Enabling Grids for E-science in Europe). 4119 [Richard Levitte] 4120 4121 *) RC4 performance overhaul on modern architectures/implementations, such 4122 as Intel P4, IA-64 and AMD64. 4123 [Andy Polyakov] 4124 4125 *) New utility extract-section.pl. This can be used specify an alternative 4126 section number in a pod file instead of having to treat each file as 4127 a separate case in Makefile. This can be done by adding two lines to the 4128 pod file: 4129 4130 =for comment openssl_section:XXX 4131 4132 The blank line is mandatory. 4133 4134 [Steve Henson] 4135 4136 *) New arguments -certform, -keyform and -pass for s_client and s_server 4137 to allow alternative format key and certificate files and passphrase 4138 sources. 4139 [Steve Henson] 4140 4141 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 4142 update associated structures and add various utility functions. 4143 4144 Add new policy related verify parameters, include policy checking in 4145 standard verify code. Enhance 'smime' application with extra parameters 4146 to support policy checking and print out. 4147 [Steve Henson] 4148 4149 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 4150 Nehemiah processors. These extensions support AES encryption in hardware 4151 as well as RNG (though RNG support is currently disabled). 4152 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 4153 4154 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 4155 [Geoff Thorpe] 4156 4157 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 4158 [Andy Polyakov and a number of other people] 4159 4160 *) Improved PowerPC platform support. Most notably BIGNUM assembler 4161 implementation contributed by IBM. 4162 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 4163 4164 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 4165 exponent rather than 'unsigned long'. There is a corresponding change to 4166 the new 'rsa_keygen' element of the RSA_METHOD structure. 4167 [Jelte Jansen, Geoff Thorpe] 4168 4169 *) Functionality for creating the initial serial number file is now 4170 moved from CA.pl to the 'ca' utility with a new option -create_serial. 4171 4172 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 4173 number file to 1, which is bound to cause problems. To avoid 4174 the problems while respecting compatibility between different 0.9.7 4175 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 4176 CA.pl for serial number initialization. With the new release 0.9.8, 4177 we can fix the problem directly in the 'ca' utility.) 4178 [Steve Henson] 4179 4180 *) Reduced header interdepencies by declaring more opaque objects in 4181 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 4182 give fewer recursive includes, which could break lazy source code - so 4183 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 4184 developers should define this symbol when building and using openssl to 4185 ensure they track the recommended behaviour, interfaces, [etc], but 4186 backwards-compatible behaviour prevails when this isn't defined. 4187 [Geoff Thorpe] 4188 4189 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 4190 [Steve Henson] 4191 4192 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 4193 This will generate a random key of the appropriate length based on the 4194 cipher context. The EVP_CIPHER can provide its own random key generation 4195 routine to support keys of a specific form. This is used in the des and 4196 3des routines to generate a key of the correct parity. Update S/MIME 4197 code to use new functions and hence generate correct parity DES keys. 4198 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 4199 valid (weak or incorrect parity). 4200 [Steve Henson] 4201 4202 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 4203 as looking them up. This is useful when the verified structure may contain 4204 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 4205 present unless the new PKCS7_NO_CRL flag is asserted. 4206 [Steve Henson] 4207 4208 *) Extend ASN1 oid configuration module. It now additionally accepts the 4209 syntax: 4210 4211 shortName = some long name, 1.2.3.4 4212 [Steve Henson] 4213 4214 *) Reimplemented the BN_CTX implementation. There is now no more static 4215 limitation on the number of variables it can handle nor the depth of the 4216 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 4217 information can now expand as required, and rather than having a single 4218 static array of bignums, BN_CTX now uses a linked-list of such arrays 4219 allowing it to expand on demand whilst maintaining the usefulness of 4220 BN_CTX's "bundling". 4221 [Geoff Thorpe] 4222 4223 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 4224 to allow all RSA operations to function using a single BN_CTX. 4225 [Geoff Thorpe] 4226 4227 *) Preliminary support for certificate policy evaluation and checking. This 4228 is initially intended to pass the tests outlined in "Conformance Testing 4229 of Relying Party Client Certificate Path Processing Logic" v1.07. 4230 [Steve Henson] 4231 4232 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 4233 remained unused and not that useful. A variety of other little bignum 4234 tweaks and fixes have also been made continuing on from the audit (see 4235 below). 4236 [Geoff Thorpe] 4237 4238 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 4239 associated ASN1, EVP and SSL functions and old ASN1 macros. 4240 [Richard Levitte] 4241 4242 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 4243 and this should never fail. So the return value from the use of 4244 BN_set_word() (which can fail due to needless expansion) is now deprecated; 4245 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 4246 [Geoff Thorpe] 4247 4248 *) BN_CTX_get() should return zero-valued bignums, providing the same 4249 initialised value as BN_new(). 4250 [Geoff Thorpe, suggested by Ulf M��ller] 4251 4252 *) Support for inhibitAnyPolicy certificate extension. 4253 [Steve Henson] 4254 4255 *) An audit of the BIGNUM code is underway, for which debugging code is 4256 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 4257 is considered valid when processing BIGNUMs, and causes execution to 4258 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 4259 further steps are taken to deliberately pollute unused data in BIGNUM 4260 structures to try and expose faulty code further on. For now, openssl will 4261 (in its default mode of operation) continue to tolerate the inconsistent 4262 forms that it has tolerated in the past, but authors and packagers should 4263 consider trying openssl and their own applications when compiled with 4264 these debugging symbols defined. It will help highlight potential bugs in 4265 their own code, and will improve the test coverage for OpenSSL itself. At 4266 some point, these tighter rules will become openssl's default to improve 4267 maintainability, though the assert()s and other overheads will remain only 4268 in debugging configurations. See bn.h for more details. 4269 [Geoff Thorpe, Nils Larsch, Ulf M��ller] 4270 4271 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 4272 that can only be obtained through BN_CTX_new() (which implicitly 4273 initialises it). The presence of this function only made it possible 4274 to overwrite an existing structure (and cause memory leaks). 4275 [Geoff Thorpe] 4276 4277 *) Because of the callback-based approach for implementing LHASH as a 4278 template type, lh_insert() adds opaque objects to hash-tables and 4279 lh_doall() or lh_doall_arg() are typically used with a destructor callback 4280 to clean up those corresponding objects before destroying the hash table 4281 (and losing the object pointers). So some over-zealous constifications in 4282 LHASH have been relaxed so that lh_insert() does not take (nor store) the 4283 objects as "const" and the lh_doall[_arg] callback wrappers are not 4284 prototyped to have "const" restrictions on the object pointers they are 4285 given (and so aren't required to cast them away any more). 4286 [Geoff Thorpe] 4287 4288 *) The tmdiff.h API was so ugly and minimal that our own timing utility 4289 (speed) prefers to use its own implementation. The two implementations 4290 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 4291 its object type properly exposed (MS_TM) instead of casting to/from "char 4292 *". This may still change yet if someone realises MS_TM and "ms_time_***" 4293 aren't necessarily the greatest nomenclatures - but this is what was used 4294 internally to the implementation so I've used that for now. 4295 [Geoff Thorpe] 4296 4297 *) Ensure that deprecated functions do not get compiled when 4298 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 4299 the self-tests were still using deprecated key-generation functions so 4300 these have been updated also. 4301 [Geoff Thorpe] 4302 4303 *) Reorganise PKCS#7 code to separate the digest location functionality 4304 into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest(). 4305 New function PKCS7_set_digest() to set the digest type for PKCS#7 4306 digestedData type. Add additional code to correctly generate the 4307 digestedData type and add support for this type in PKCS7 initialization 4308 functions. 4309 [Steve Henson] 4310 4311 *) New function PKCS7_set0_type_other() this initializes a PKCS7 4312 structure of type "other". 4313 [Steve Henson] 4314 4315 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 4316 sure the loop does correctly stop and breaking ("division by zero") 4317 modulus operations are not performed. The (pre-generated) prime 4318 table crypto/bn/bn_prime.h was already correct, but it could not be 4319 re-generated on some platforms because of the "division by zero" 4320 situation in the script. 4321 [Ralf S. Engelschall] 4322 4323 *) Update support for ECC-based TLS ciphersuites according to 4324 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 4325 SHA-1 now is only used for "small" curves (where the 4326 representation of a field element takes up to 24 bytes); for 4327 larger curves, the field element resulting from ECDH is directly 4328 used as premaster secret. 4329 [Douglas Stebila (Sun Microsystems Laboratories)] 4330 4331 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 4332 curve secp160r1 to the tests. 4333 [Douglas Stebila (Sun Microsystems Laboratories)] 4334 4335 *) Add the possibility to load symbols globally with DSO. 4336 [G��tz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 4337 4338 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 4339 control of the error stack. 4340 [Richard Levitte] 4341 4342 *) Add support for STORE in ENGINE. 4343 [Richard Levitte] 4344 4345 *) Add the STORE type. The intention is to provide a common interface 4346 to certificate and key stores, be they simple file-based stores, or 4347 HSM-type store, or LDAP stores, or... 4348 NOTE: The code is currently UNTESTED and isn't really used anywhere. 4349 [Richard Levitte] 4350 4351 *) Add a generic structure called OPENSSL_ITEM. This can be used to 4352 pass a list of arguments to any function as well as provide a way 4353 for a function to pass data back to the caller. 4354 [Richard Levitte] 4355 4356 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 4357 works like BUF_strdup() but can be used to duplicate a portion of 4358 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 4359 a memory area. 4360 [Richard Levitte] 4361 4362 *) Add the function sk_find_ex() which works like sk_find(), but will 4363 return an index to an element even if an exact match couldn't be 4364 found. The index is guaranteed to point at the element where the 4365 searched-for key would be inserted to preserve sorting order. 4366 [Richard Levitte] 4367 4368 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 4369 takes an extra flags argument for optional functionality. Currently, 4370 the following flags are defined: 4371 4372 OBJ_BSEARCH_VALUE_ON_NOMATCH 4373 This one gets OBJ_bsearch_ex() to return a pointer to the first 4374 element where the comparing function returns a negative or zero 4375 number. 4376 4377 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 4378 This one gets OBJ_bsearch_ex() to return a pointer to the first 4379 element where the comparing function returns zero. This is useful 4380 if there are more than one element where the comparing function 4381 returns zero. 4382 [Richard Levitte] 4383 4384 *) Make it possible to create self-signed certificates with 'openssl ca' 4385 in such a way that the self-signed certificate becomes part of the 4386 CA database and uses the same mechanisms for serial number generation 4387 as all other certificate signing. The new flag '-selfsign' enables 4388 this functionality. Adapt CA.sh and CA.pl.in. 4389 [Richard Levitte] 4390 4391 *) Add functionality to check the public key of a certificate request 4392 against a given private. This is useful to check that a certificate 4393 request can be signed by that key (self-signing). 4394 [Richard Levitte] 4395 4396 *) Make it possible to have multiple active certificates with the same 4397 subject in the CA index file. This is done only if the keyword 4398 'unique_subject' is set to 'no' in the main CA section (default 4399 if 'CA_default') of the configuration file. The value is saved 4400 with the database itself in a separate index attribute file, 4401 named like the index file with '.attr' appended to the name. 4402 [Richard Levitte] 4403 4404 *) Generate muti valued AVAs using '+' notation in config files for 4405 req and dirName. 4406 [Steve Henson] 4407 4408 *) Support for nameConstraints certificate extension. 4409 [Steve Henson] 4410 4411 *) Support for policyConstraints certificate extension. 4412 [Steve Henson] 4413 4414 *) Support for policyMappings certificate extension. 4415 [Steve Henson] 4416 4417 *) Make sure the default DSA_METHOD implementation only uses its 4418 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 4419 and change its own handlers to be NULL so as to remove unnecessary 4420 indirection. This lets alternative implementations fallback to the 4421 default implementation more easily. 4422 [Geoff Thorpe] 4423 4424 *) Support for directoryName in GeneralName related extensions 4425 in config files. 4426 [Steve Henson] 4427 4428 *) Make it possible to link applications using Makefile.shared. 4429 Make that possible even when linking against static libraries! 4430 [Richard Levitte] 4431 4432 *) Support for single pass processing for S/MIME signing. This now 4433 means that S/MIME signing can be done from a pipe, in addition 4434 cleartext signing (multipart/signed type) is effectively streaming 4435 and the signed data does not need to be all held in memory. 4436 4437 This is done with a new flag PKCS7_STREAM. When this flag is set 4438 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 4439 is done after the data is output (and digests calculated) in 4440 SMIME_write_PKCS7(). 4441 [Steve Henson] 4442 4443 *) Add full support for -rpath/-R, both in shared libraries and 4444 applications, at least on the platforms where it's known how 4445 to do it. 4446 [Richard Levitte] 4447 4448 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 4449 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 4450 will now compute a table of multiples of the generator that 4451 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 4452 faster (notably in the case of a single point multiplication, 4453 scalar * generator). 4454 [Nils Larsch, Bodo Moeller] 4455 4456 *) IPv6 support for certificate extensions. The various extensions 4457 which use the IP:a.b.c.d can now take IPv6 addresses using the 4458 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 4459 correctly. 4460 [Steve Henson] 4461 4462 *) Added an ENGINE that implements RSA by performing private key 4463 exponentiations with the GMP library. The conversions to and from 4464 GMP's mpz_t format aren't optimised nor are any montgomery forms 4465 cached, and on x86 it appears OpenSSL's own performance has caught up. 4466 However there are likely to be other architectures where GMP could 4467 provide a boost. This ENGINE is not built in by default, but it can be 4468 specified at Configure time and should be accompanied by the necessary 4469 linker additions, eg; 4470 ./config -DOPENSSL_USE_GMP -lgmp 4471 [Geoff Thorpe] 4472 4473 *) "openssl engine" will not display ENGINE/DSO load failure errors when 4474 testing availability of engines with "-t" - the old behaviour is 4475 produced by increasing the feature's verbosity with "-tt". 4476 [Geoff Thorpe] 4477 4478 *) ECDSA routines: under certain error conditions uninitialized BN objects 4479 could be freed. Solution: make sure initialization is performed early 4480 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 4481 via PR#459) 4482 [Lutz Jaenicke] 4483 4484 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 4485 and DH_METHOD (eg. by ENGINE implementations) to override the normal 4486 software implementations. For DSA and DH, parameter generation can 4487 also be overriden by providing the appropriate method callbacks. 4488 [Geoff Thorpe] 4489 4490 *) Change the "progress" mechanism used in key-generation and 4491 primality testing to functions that take a new BN_GENCB pointer in 4492 place of callback/argument pairs. The new API functions have "_ex" 4493 postfixes and the older functions are reimplemented as wrappers for 4494 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 4495 declarations of the old functions to help (graceful) attempts to 4496 migrate to the new functions. Also, the new key-generation API 4497 functions operate on a caller-supplied key-structure and return 4498 success/failure rather than returning a key or NULL - this is to 4499 help make "keygen" another member function of RSA_METHOD etc. 4500 4501 Example for using the new callback interface: 4502 4503 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 4504 void *my_arg = ...; 4505 BN_GENCB my_cb; 4506 4507 BN_GENCB_set(&my_cb, my_callback, my_arg); 4508 4509 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 4510 /* For the meaning of a, b in calls to my_callback(), see the 4511 * documentation of the function that calls the callback. 4512 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 4513 * my_callback should return 1 if it wants BN_is_prime_ex() 4514 * to continue, or 0 to stop. 4515 */ 4516 4517 [Geoff Thorpe] 4518 4519 *) Change the ZLIB compression method to be stateful, and make it 4520 available to TLS with the number defined in 4521 draft-ietf-tls-compression-04.txt. 4522 [Richard Levitte] 4523 4524 *) Add the ASN.1 structures and functions for CertificatePair, which 4525 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 4526 4527 CertificatePair ::= SEQUENCE { 4528 forward [0] Certificate OPTIONAL, 4529 reverse [1] Certificate OPTIONAL, 4530 -- at least one of the pair shall be present -- } 4531 4532 Also implement the PEM functions to read and write certificate 4533 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 4534 4535 This needed to be defined, mostly for the sake of the LDAP 4536 attribute crossCertificatePair, but may prove useful elsewhere as 4537 well. 4538 [Richard Levitte] 4539 4540 *) Make it possible to inhibit symlinking of shared libraries in 4541 Makefile.shared, for Cygwin's sake. 4542 [Richard Levitte] 4543 4544 *) Extend the BIGNUM API by creating a function 4545 void BN_set_negative(BIGNUM *a, int neg); 4546 and a macro that behave like 4547 int BN_is_negative(const BIGNUM *a); 4548 4549 to avoid the need to access 'a->neg' directly in applications. 4550 [Nils Larsch] 4551 4552 *) Implement fast modular reduction for pseudo-Mersenne primes 4553 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 4554 EC_GROUP_new_curve_GFp() will now automatically use this 4555 if applicable. 4556 [Nils Larsch <nla@trustcenter.de>] 4557 4558 *) Add new lock type (CRYPTO_LOCK_BN). 4559 [Bodo Moeller] 4560 4561 *) Change the ENGINE framework to automatically load engines 4562 dynamically from specific directories unless they could be 4563 found to already be built in or loaded. Move all the 4564 current engines except for the cryptodev one to a new 4565 directory engines/. 4566 The engines in engines/ are built as shared libraries if 4567 the "shared" options was given to ./Configure or ./config. 4568 Otherwise, they are inserted in libcrypto.a. 4569 /usr/local/ssl/engines is the default directory for dynamic 4570 engines, but that can be overriden at configure time through 4571 the usual use of --prefix and/or --openssldir, and at run 4572 time with the environment variable OPENSSL_ENGINES. 4573 [Geoff Thorpe and Richard Levitte] 4574 4575 *) Add Makefile.shared, a helper makefile to build shared 4576 libraries. Addapt Makefile.org. 4577 [Richard Levitte] 4578 4579 *) Add version info to Win32 DLLs. 4580 [Peter 'Luna' Runestig" <peter@runestig.com>] 4581 4582 *) Add new 'medium level' PKCS#12 API. Certificates and keys 4583 can be added using this API to created arbitrary PKCS#12 4584 files while avoiding the low level API. 4585 4586 New options to PKCS12_create(), key or cert can be NULL and 4587 will then be omitted from the output file. The encryption 4588 algorithm NIDs can be set to -1 for no encryption, the mac 4589 iteration count can be set to 0 to omit the mac. 4590 4591 Enhance pkcs12 utility by making the -nokeys and -nocerts 4592 options work when creating a PKCS#12 file. New option -nomac 4593 to omit the mac, NONE can be set for an encryption algorithm. 4594 New code is modified to use the enhanced PKCS12_create() 4595 instead of the low level API. 4596 [Steve Henson] 4597 4598 *) Extend ASN1 encoder to support indefinite length constructed 4599 encoding. This can output sequences tags and octet strings in 4600 this form. Modify pk7_asn1.c to support indefinite length 4601 encoding. This is experimental and needs additional code to 4602 be useful, such as an ASN1 bio and some enhanced streaming 4603 PKCS#7 code. 4604 4605 Extend template encode functionality so that tagging is passed 4606 down to the template encoder. 4607 [Steve Henson] 4608 4609 *) Let 'openssl req' fail if an argument to '-newkey' is not 4610 recognized instead of using RSA as a default. 4611 [Bodo Moeller] 4612 4613 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 4614 As these are not official, they are not included in "ALL"; 4615 the "ECCdraft" ciphersuite group alias can be used to select them. 4616 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 4617 4618 *) Add ECDH engine support. 4619 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 4620 4621 *) Add ECDH in new directory crypto/ecdh/. 4622 [Douglas Stebila (Sun Microsystems Laboratories)] 4623 4624 *) Let BN_rand_range() abort with an error after 100 iterations 4625 without success (which indicates a broken PRNG). 4626 [Bodo Moeller] 4627 4628 *) Change BN_mod_sqrt() so that it verifies that the input value 4629 is really the square of the return value. (Previously, 4630 BN_mod_sqrt would show GIGO behaviour.) 4631 [Bodo Moeller] 4632 4633 *) Add named elliptic curves over binary fields from X9.62, SECG, 4634 and WAP/WTLS; add OIDs that were still missing. 4635 4636 [Sheueling Chang Shantz and Douglas Stebila 4637 (Sun Microsystems Laboratories)] 4638 4639 *) Extend the EC library for elliptic curves over binary fields 4640 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 4641 New EC_METHOD: 4642 4643 EC_GF2m_simple_method 4644 4645 New API functions: 4646 4647 EC_GROUP_new_curve_GF2m 4648 EC_GROUP_set_curve_GF2m 4649 EC_GROUP_get_curve_GF2m 4650 EC_POINT_set_affine_coordinates_GF2m 4651 EC_POINT_get_affine_coordinates_GF2m 4652 EC_POINT_set_compressed_coordinates_GF2m 4653 4654 Point compression for binary fields is disabled by default for 4655 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 4656 enable it). 4657 4658 As binary polynomials are represented as BIGNUMs, various members 4659 of the EC_GROUP and EC_POINT data structures can be shared 4660 between the implementations for prime fields and binary fields; 4661 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 4662 are essentially identical to their ..._GFp counterparts. 4663 (For simplicity, the '..._GFp' prefix has been dropped from 4664 various internal method names.) 4665 4666 An internal 'field_div' method (similar to 'field_mul' and 4667 'field_sqr') has been added; this is used only for binary fields. 4668 4669 [Sheueling Chang Shantz and Douglas Stebila 4670 (Sun Microsystems Laboratories)] 4671 4672 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 4673 through methods ('mul', 'precompute_mult'). 4674 4675 The generic implementations (now internally called 'ec_wNAF_mul' 4676 and 'ec_wNAF_precomputed_mult') remain the default if these 4677 methods are undefined. 4678 4679 [Sheueling Chang Shantz and Douglas Stebila 4680 (Sun Microsystems Laboratories)] 4681 4682 *) New function EC_GROUP_get_degree, which is defined through 4683 EC_METHOD. For curves over prime fields, this returns the bit 4684 length of the modulus. 4685 4686 [Sheueling Chang Shantz and Douglas Stebila 4687 (Sun Microsystems Laboratories)] 4688 4689 *) New functions EC_GROUP_dup, EC_POINT_dup. 4690 (These simply call ..._new and ..._copy). 4691 4692 [Sheueling Chang Shantz and Douglas Stebila 4693 (Sun Microsystems Laboratories)] 4694 4695 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 4696 Polynomials are represented as BIGNUMs (where the sign bit is not 4697 used) in the following functions [macros]: 4698 4699 BN_GF2m_add 4700 BN_GF2m_sub [= BN_GF2m_add] 4701 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 4702 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 4703 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 4704 BN_GF2m_mod_inv 4705 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 4706 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 4707 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 4708 BN_GF2m_cmp [= BN_ucmp] 4709 4710 (Note that only the 'mod' functions are actually for fields GF(2^m). 4711 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 4712 4713 For some functions, an the irreducible polynomial defining a 4714 field can be given as an 'unsigned int[]' with strictly 4715 decreasing elements giving the indices of those bits that are set; 4716 i.e., p[] represents the polynomial 4717 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 4718 where 4719 p[0] > p[1] > ... > p[k] = 0. 4720 This applies to the following functions: 4721 4722 BN_GF2m_mod_arr 4723 BN_GF2m_mod_mul_arr 4724 BN_GF2m_mod_sqr_arr 4725 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 4726 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 4727 BN_GF2m_mod_exp_arr 4728 BN_GF2m_mod_sqrt_arr 4729 BN_GF2m_mod_solve_quad_arr 4730 BN_GF2m_poly2arr 4731 BN_GF2m_arr2poly 4732 4733 Conversion can be performed by the following functions: 4734 4735 BN_GF2m_poly2arr 4736 BN_GF2m_arr2poly 4737 4738 bntest.c has additional tests for binary polynomial arithmetic. 4739 4740 Two implementations for BN_GF2m_mod_div() are available. 4741 The default algorithm simply uses BN_GF2m_mod_inv() and 4742 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 4743 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 4744 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 4745 4746 [Sheueling Chang Shantz and Douglas Stebila 4747 (Sun Microsystems Laboratories)] 4748 4749 *) Add new error code 'ERR_R_DISABLED' that can be used when some 4750 functionality is disabled at compile-time. 4751 [Douglas Stebila <douglas.stebila@sun.com>] 4752 4753 *) Change default behaviour of 'openssl asn1parse' so that more 4754 information is visible when viewing, e.g., a certificate: 4755 4756 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 4757 mode the content of non-printable OCTET STRINGs is output in a 4758 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 4759 avoid the appearance of a printable string. 4760 [Nils Larsch <nla@trustcenter.de>] 4761 4762 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 4763 functions 4764 EC_GROUP_set_asn1_flag() 4765 EC_GROUP_get_asn1_flag() 4766 EC_GROUP_set_point_conversion_form() 4767 EC_GROUP_get_point_conversion_form() 4768 These control ASN1 encoding details: 4769 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 4770 has been set to OPENSSL_EC_NAMED_CURVE. 4771 - Points are encoded in uncompressed form by default; options for 4772 asn1_for are as for point2oct, namely 4773 POINT_CONVERSION_COMPRESSED 4774 POINT_CONVERSION_UNCOMPRESSED 4775 POINT_CONVERSION_HYBRID 4776 4777 Also add 'seed' and 'seed_len' members to EC_GROUP with access 4778 functions 4779 EC_GROUP_set_seed() 4780 EC_GROUP_get0_seed() 4781 EC_GROUP_get_seed_len() 4782 This is used only for ASN1 purposes (so far). 4783 [Nils Larsch <nla@trustcenter.de>] 4784 4785 *) Add 'field_type' member to EC_METHOD, which holds the NID 4786 of the appropriate field type OID. The new function 4787 EC_METHOD_get_field_type() returns this value. 4788 [Nils Larsch <nla@trustcenter.de>] 4789 4790 *) Add functions 4791 EC_POINT_point2bn() 4792 EC_POINT_bn2point() 4793 EC_POINT_point2hex() 4794 EC_POINT_hex2point() 4795 providing useful interfaces to EC_POINT_point2oct() and 4796 EC_POINT_oct2point(). 4797 [Nils Larsch <nla@trustcenter.de>] 4798 4799 *) Change internals of the EC library so that the functions 4800 EC_GROUP_set_generator() 4801 EC_GROUP_get_generator() 4802 EC_GROUP_get_order() 4803 EC_GROUP_get_cofactor() 4804 are implemented directly in crypto/ec/ec_lib.c and not dispatched 4805 to methods, which would lead to unnecessary code duplication when 4806 adding different types of curves. 4807 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 4808 4809 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 4810 arithmetic, and such that modified wNAFs are generated 4811 (which avoid length expansion in many cases). 4812 [Bodo Moeller] 4813 4814 *) Add a function EC_GROUP_check_discriminant() (defined via 4815 EC_METHOD) that verifies that the curve discriminant is non-zero. 4816 4817 Add a function EC_GROUP_check() that makes some sanity tests 4818 on a EC_GROUP, its generator and order. This includes 4819 EC_GROUP_check_discriminant(). 4820 [Nils Larsch <nla@trustcenter.de>] 4821 4822 *) Add ECDSA in new directory crypto/ecdsa/. 4823 4824 Add applications 'openssl ecparam' and 'openssl ecdsa' 4825 (these are based on 'openssl dsaparam' and 'openssl dsa'). 4826 4827 ECDSA support is also included in various other files across the 4828 library. Most notably, 4829 - 'openssl req' now has a '-newkey ecdsa:file' option; 4830 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 4831 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 4832 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 4833 them suitable for ECDSA where domain parameters must be 4834 extracted before the specific public key; 4835 - ECDSA engine support has been added. 4836 [Nils Larsch <nla@trustcenter.de>] 4837 4838 *) Include some named elliptic curves, and add OIDs from X9.62, 4839 SECG, and WAP/WTLS. Each curve can be obtained from the new 4840 function 4841 EC_GROUP_new_by_curve_name(), 4842 and the list of available named curves can be obtained with 4843 EC_get_builtin_curves(). 4844 Also add a 'curve_name' member to EC_GROUP objects, which can be 4845 accessed via 4846 EC_GROUP_set_curve_name() 4847 EC_GROUP_get_curve_name() 4848 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 4849 4850 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 4851 was actually never needed) and in BN_mul(). The removal in BN_mul() 4852 required a small change in bn_mul_part_recursive() and the addition 4853 of the functions bn_cmp_part_words(), bn_sub_part_words() and 4854 bn_add_part_words(), which do the same thing as bn_cmp_words(), 4855 bn_sub_words() and bn_add_words() except they take arrays with 4856 differing sizes. 4857 [Richard Levitte] 4858 4859 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 4860 4861 *) Cleanse PEM buffers before freeing them since they may contain 4862 sensitive data. 4863 [Benjamin Bennett <ben@psc.edu>] 4864 4865 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 4866 a ciphersuite string such as "DEFAULT:RSA" cannot enable 4867 authentication-only ciphersuites. 4868 [Bodo Moeller] 4869 4870 *) Since AES128 and AES256 share a single mask bit in the logic of 4871 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 4872 kludge to work properly if AES128 is available and AES256 isn't. 4873 [Victor Duchovni] 4874 4875 *) Expand security boundary to match 1.1.1 module. 4876 [Steve Henson] 4877 4878 *) Remove redundant features: hash file source, editing of test vectors 4879 modify fipsld to use external fips_premain.c signature. 4880 [Steve Henson] 4881 4882 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 4883 run algorithm test programs. 4884 [Steve Henson] 4885 4886 *) Make algorithm test programs more tolerant of whitespace. 4887 [Steve Henson] 4888 4889 *) Have SSL/TLS server implementation tolerate "mismatched" record 4890 protocol version while receiving ClientHello even if the 4891 ClientHello is fragmented. (The server can't insist on the 4892 particular protocol version it has chosen before the ServerHello 4893 message has informed the client about his choice.) 4894 [Bodo Moeller] 4895 4896 *) Load error codes if they are not already present instead of using a 4897 static variable. This allows them to be cleanly unloaded and reloaded. 4898 [Steve Henson] 4899 4900 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 4901 4902 *) Introduce limits to prevent malicious keys being able to 4903 cause a denial of service. (CVE-2006-2940) 4904 [Steve Henson, Bodo Moeller] 4905 4906 *) Fix ASN.1 parsing of certain invalid structures that can result 4907 in a denial of service. (CVE-2006-2937) [Steve Henson] 4908 4909 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 4910 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 4911 4912 *) Fix SSL client code which could crash if connecting to a 4913 malicious SSLv2 server. (CVE-2006-4343) 4914 [Tavis Ormandy and Will Drewry, Google Security Team] 4915 4916 *) Change ciphersuite string processing so that an explicit 4917 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 4918 will no longer include "AES128-SHA"), and any other similar 4919 ciphersuite (same bitmap) from *other* protocol versions (so that 4920 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 4921 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 4922 changes from 0.9.8b and 0.9.8d. 4923 [Bodo Moeller] 4924 4925 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 4926 4927 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 4928 (CVE-2006-4339) [Ben Laurie and Google Security Team] 4929 4930 *) Change the Unix randomness entropy gathering to use poll() when 4931 possible instead of select(), since the latter has some 4932 undesirable limitations. 4933 [Darryl Miles via Richard Levitte and Bodo Moeller] 4934 4935 *) Disable rogue ciphersuites: 4936 4937 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 4938 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 4939 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 4940 4941 The latter two were purportedly from 4942 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 4943 appear there. 4944 4945 Also deactive the remaining ciphersuites from 4946 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 4947 unofficial, and the ID has long expired. 4948 [Bodo Moeller] 4949 4950 *) Fix RSA blinding Heisenbug (problems sometimes occured on 4951 dual-core machines) and other potential thread-safety issues. 4952 [Bodo Moeller] 4953 4954 Changes between 0.9.7i and 0.9.7j [04 May 2006] 4955 4956 *) Adapt fipsld and the build system to link against the validated FIPS 4957 module in FIPS mode. 4958 [Steve Henson] 4959 4960 *) Fixes for VC++ 2005 build under Windows. 4961 [Steve Henson] 4962 4963 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 4964 from a Windows bash shell such as MSYS. It is autodetected from the 4965 "config" script when run from a VC++ environment. Modify standard VC++ 4966 build to use fipscanister.o from the GNU make build. 4967 [Steve Henson] 4968 4969 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 4970 4971 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 4972 The value now differs depending on if you build for FIPS or not. 4973 BEWARE! A program linked with a shared FIPSed libcrypto can't be 4974 safely run with a non-FIPSed libcrypto, as it may crash because of 4975 the difference induced by this change. 4976 [Andy Polyakov] 4977 4978 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 4979 4980 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 4981 (part of SSL_OP_ALL). This option used to disable the 4982 countermeasure against man-in-the-middle protocol-version 4983 rollback in the SSL 2.0 server implementation, which is a bad 4984 idea. (CVE-2005-2969) 4985 4986 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 4987 for Information Security, National Institute of Advanced Industrial 4988 Science and Technology [AIST], Japan)] 4989 4990 *) Minimal support for X9.31 signatures and PSS padding modes. This is 4991 mainly for FIPS compliance and not fully integrated at this stage. 4992 [Steve Henson] 4993 4994 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 4995 the exponentiation using a fixed-length exponent. (Otherwise, 4996 the information leaked through timing could expose the secret key 4997 after many signatures; cf. Bleichenbacher's attack on DSA with 4998 biased k.) 4999 [Bodo Moeller] 5000 5001 *) Make a new fixed-window mod_exp implementation the default for 5002 RSA, DSA, and DH private-key operations so that the sequence of 5003 squares and multiplies and the memory access pattern are 5004 independent of the particular secret key. This will mitigate 5005 cache-timing and potential related attacks. 5006 5007 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 5008 and this is automatically used by BN_mod_exp_mont() if the new flag 5009 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 5010 will use this BN flag for private exponents unless the flag 5011 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 5012 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 5013 5014 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 5015 5016 *) Change the client implementation for SSLv23_method() and 5017 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 5018 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 5019 (Previously, the SSL 2.0 backwards compatible Client Hello 5020 message format would be used even with SSL_OP_NO_SSLv2.) 5021 [Bodo Moeller] 5022 5023 *) Add support for smime-type MIME parameter in S/MIME messages which some 5024 clients need. 5025 [Steve Henson] 5026 5027 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 5028 a threadsafe manner. Modify rsa code to use new function and add calls 5029 to dsa and dh code (which had race conditions before). 5030 [Steve Henson] 5031 5032 *) Include the fixed error library code in the C error file definitions 5033 instead of fixing them up at runtime. This keeps the error code 5034 structures constant. 5035 [Steve Henson] 5036 5037 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 5038 5039 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 5040 OpenSSL 0.9.8.] 5041 5042 *) Fixes for newer kerberos headers. NB: the casts are needed because 5043 the 'length' field is signed on one version and unsigned on another 5044 with no (?) obvious way to tell the difference, without these VC++ 5045 complains. Also the "definition" of FAR (blank) is no longer included 5046 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 5047 some needed definitions. 5048 [Steve Henson] 5049 5050 *) Undo Cygwin change. 5051 [Ulf M��ller] 5052 5053 *) Added support for proxy certificates according to RFC 3820. 5054 Because they may be a security thread to unaware applications, 5055 they must be explicitely allowed in run-time. See 5056 docs/HOWTO/proxy_certificates.txt for further information. 5057 [Richard Levitte] 5058 5059 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 5060 5061 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 5062 server and client random values. Previously 5063 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 5064 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 5065 5066 This change has negligible security impact because: 5067 5068 1. Server and client random values still have 24 bytes of pseudo random 5069 data. 5070 5071 2. Server and client random values are sent in the clear in the initial 5072 handshake. 5073 5074 3. The master secret is derived using the premaster secret (48 bytes in 5075 size for static RSA ciphersuites) as well as client server and random 5076 values. 5077 5078 The OpenSSL team would like to thank the UK NISCC for bringing this issue 5079 to our attention. 5080 5081 [Stephen Henson, reported by UK NISCC] 5082 5083 *) Use Windows randomness collection on Cygwin. 5084 [Ulf M��ller] 5085 5086 *) Fix hang in EGD/PRNGD query when communication socket is closed 5087 prematurely by EGD/PRNGD. 5088 [Darren Tucker <dtucker@zip.com.au> via Lutz J��nicke, resolves #1014] 5089 5090 *) Prompt for pass phrases when appropriate for PKCS12 input format. 5091 [Steve Henson] 5092 5093 *) Back-port of selected performance improvements from development 5094 branch, as well as improved support for PowerPC platforms. 5095 [Andy Polyakov] 5096 5097 *) Add lots of checks for memory allocation failure, error codes to indicate 5098 failure and freeing up memory if a failure occurs. 5099 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 5100 5101 *) Add new -passin argument to dgst. 5102 [Steve Henson] 5103 5104 *) Perform some character comparisons of different types in X509_NAME_cmp: 5105 this is needed for some certificates that reencode DNs into UTF8Strings 5106 (in violation of RFC3280) and can't or wont issue name rollover 5107 certificates. 5108 [Steve Henson] 5109 5110 *) Make an explicit check during certificate validation to see that 5111 the CA setting in each certificate on the chain is correct. As a 5112 side effect always do the following basic checks on extensions, 5113 not just when there's an associated purpose to the check: 5114 5115 - if there is an unhandled critical extension (unless the user 5116 has chosen to ignore this fault) 5117 - if the path length has been exceeded (if one is set at all) 5118 - that certain extensions fit the associated purpose (if one has 5119 been given) 5120 [Richard Levitte] 5121 5122 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 5123 5124 *) Avoid a race condition when CRLs are checked in a multi threaded 5125 environment. This would happen due to the reordering of the revoked 5126 entries during signature checking and serial number lookup. Now the 5127 encoding is cached and the serial number sort performed under a lock. 5128 Add new STACK function sk_is_sorted(). 5129 [Steve Henson] 5130 5131 *) Add Delta CRL to the extension code. 5132 [Steve Henson] 5133 5134 *) Various fixes to s3_pkt.c so alerts are sent properly. 5135 [David Holmes <d.holmes@f5.com>] 5136 5137 *) Reduce the chances of duplicate issuer name and serial numbers (in 5138 violation of RFC3280) using the OpenSSL certificate creation utilities. 5139 This is done by creating a random 64 bit value for the initial serial 5140 number when a serial number file is created or when a self signed 5141 certificate is created using 'openssl req -x509'. The initial serial 5142 number file is created using 'openssl x509 -next_serial' in CA.pl 5143 rather than being initialized to 1. 5144 [Steve Henson] 5145 5146 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 5147 5148 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 5149 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 5150 [Joe Orton, Steve Henson] 5151 5152 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 5153 (CVE-2004-0112) 5154 [Joe Orton, Steve Henson] 5155 5156 *) Make it possible to have multiple active certificates with the same 5157 subject in the CA index file. This is done only if the keyword 5158 'unique_subject' is set to 'no' in the main CA section (default 5159 if 'CA_default') of the configuration file. The value is saved 5160 with the database itself in a separate index attribute file, 5161 named like the index file with '.attr' appended to the name. 5162 [Richard Levitte] 5163 5164 *) X509 verify fixes. Disable broken certificate workarounds when 5165 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 5166 keyUsage extension present. Don't accept CRLs with unhandled critical 5167 extensions: since verify currently doesn't process CRL extensions this 5168 rejects a CRL with *any* critical extensions. Add new verify error codes 5169 for these cases. 5170 [Steve Henson] 5171 5172 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 5173 A clarification of RFC2560 will require the use of OCTET STRINGs and 5174 some implementations cannot handle the current raw format. Since OpenSSL 5175 copies and compares OCSP nonces as opaque blobs without any attempt at 5176 parsing them this should not create any compatibility issues. 5177 [Steve Henson] 5178 5179 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 5180 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 5181 this HMAC (and other) operations are several times slower than OpenSSL 5182 < 0.9.7. 5183 [Steve Henson] 5184 5185 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 5186 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 5187 5188 *) Use the correct content when signing type "other". 5189 [Steve Henson] 5190 5191 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 5192 5193 *) Fix various bugs revealed by running the NISCC test suite: 5194 5195 Stop out of bounds reads in the ASN1 code when presented with 5196 invalid tags (CVE-2003-0543 and CVE-2003-0544). 5197 5198 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 5199 5200 If verify callback ignores invalid public key errors don't try to check 5201 certificate signature with the NULL public key. 5202 5203 [Steve Henson] 5204 5205 *) New -ignore_err option in ocsp application to stop the server 5206 exiting on the first error in a request. 5207 [Steve Henson] 5208 5209 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 5210 if the server requested one: as stated in TLS 1.0 and SSL 3.0 5211 specifications. 5212 [Steve Henson] 5213 5214 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 5215 extra data after the compression methods not only for TLS 1.0 5216 but also for SSL 3.0 (as required by the specification). 5217 [Bodo Moeller; problem pointed out by Matthias Loepfe] 5218 5219 *) Change X509_certificate_type() to mark the key as exported/exportable 5220 when it's 512 *bits* long, not 512 bytes. 5221 [Richard Levitte] 5222 5223 *) Change AES_cbc_encrypt() so it outputs exact multiple of 5224 blocks during encryption. 5225 [Richard Levitte] 5226 5227 *) Various fixes to base64 BIO and non blocking I/O. On write 5228 flushes were not handled properly if the BIO retried. On read 5229 data was not being buffered properly and had various logic bugs. 5230 This also affects blocking I/O when the data being decoded is a 5231 certain size. 5232 [Steve Henson] 5233 5234 *) Various S/MIME bugfixes and compatibility changes: 5235 output correct application/pkcs7 MIME type if 5236 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 5237 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 5238 of files as .eml work). Correctly handle very long lines in MIME 5239 parser. 5240 [Steve Henson] 5241 5242 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 5243 5244 *) Countermeasure against the Klima-Pokorny-Rosa extension of 5245 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 5246 a protocol version number mismatch like a decryption error 5247 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 5248 [Bodo Moeller] 5249 5250 *) Turn on RSA blinding by default in the default implementation 5251 to avoid a timing attack. Applications that don't want it can call 5252 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 5253 They would be ill-advised to do so in most cases. 5254 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 5255 5256 *) Change RSA blinding code so that it works when the PRNG is not 5257 seeded (in this case, the secret RSA exponent is abused as 5258 an unpredictable seed -- if it is not unpredictable, there 5259 is no point in blinding anyway). Make RSA blinding thread-safe 5260 by remembering the creator's thread ID in rsa->blinding and 5261 having all other threads use local one-time blinding factors 5262 (this requires more computation than sharing rsa->blinding, but 5263 avoids excessive locking; and if an RSA object is not shared 5264 between threads, blinding will still be very fast). 5265 [Bodo Moeller] 5266 5267 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 5268 ENGINE as defaults for all supported algorithms irrespective of 5269 the 'flags' parameter. 'flags' is now honoured, so applications 5270 should make sure they are passing it correctly. 5271 [Geoff Thorpe] 5272 5273 *) Target "mingw" now allows native Windows code to be generated in 5274 the Cygwin environment as well as with the MinGW compiler. 5275 [Ulf Moeller] 5276 5277 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 5278 5279 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 5280 via timing by performing a MAC computation even if incorrrect 5281 block cipher padding has been found. This is a countermeasure 5282 against active attacks where the attacker has to distinguish 5283 between bad padding and a MAC verification error. (CVE-2003-0078) 5284 5285 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 5286 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 5287 Martin Vuagnoux (EPFL, Ilion)] 5288 5289 *) Make the no-err option work as intended. The intention with no-err 5290 is not to have the whole error stack handling routines removed from 5291 libcrypto, it's only intended to remove all the function name and 5292 reason texts, thereby removing some of the footprint that may not 5293 be interesting if those errors aren't displayed anyway. 5294 5295 NOTE: it's still possible for any application or module to have it's 5296 own set of error texts inserted. The routines are there, just not 5297 used by default when no-err is given. 5298 [Richard Levitte] 5299 5300 *) Add support for FreeBSD on IA64. 5301 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 5302 5303 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 5304 Kerberos function mit_des_cbc_cksum(). Before this change, 5305 the value returned by DES_cbc_cksum() was like the one from 5306 mit_des_cbc_cksum(), except the bytes were swapped. 5307 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 5308 5309 *) Allow an application to disable the automatic SSL chain building. 5310 Before this a rather primitive chain build was always performed in 5311 ssl3_output_cert_chain(): an application had no way to send the 5312 correct chain if the automatic operation produced an incorrect result. 5313 5314 Now the chain builder is disabled if either: 5315 5316 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 5317 5318 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 5319 5320 The reasoning behind this is that an application would not want the 5321 auto chain building to take place if extra chain certificates are 5322 present and it might also want a means of sending no additional 5323 certificates (for example the chain has two certificates and the 5324 root is omitted). 5325 [Steve Henson] 5326 5327 *) Add the possibility to build without the ENGINE framework. 5328 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5329 5330 *) Under Win32 gmtime() can return NULL: check return value in 5331 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 5332 [Steve Henson] 5333 5334 *) DSA routines: under certain error conditions uninitialized BN objects 5335 could be freed. Solution: make sure initialization is performed early 5336 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 5337 Nils Larsch <nla@trustcenter.de> via PR#459) 5338 [Lutz Jaenicke] 5339 5340 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 5341 checked on reconnect on the client side, therefore session resumption 5342 could still fail with a "ssl session id is different" error. This 5343 behaviour is masked when SSL_OP_ALL is used due to 5344 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 5345 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5346 followup to PR #377. 5347 [Lutz Jaenicke] 5348 5349 *) IA-32 assembler support enhancements: unified ELF targets, support 5350 for SCO/Caldera platforms, fix for Cygwin shared build. 5351 [Andy Polyakov] 5352 5353 *) Add support for FreeBSD on sparc64. As a consequence, support for 5354 FreeBSD on non-x86 processors is separate from x86 processors on 5355 the config script, much like the NetBSD support. 5356 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 5357 5358 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 5359 5360 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 5361 OpenSSL 0.9.7.] 5362 5363 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 5364 code (06) was taken as the first octet of the session ID and the last 5365 octet was ignored consequently. As a result SSLv2 client side session 5366 caching could not have worked due to the session ID mismatch between 5367 client and server. 5368 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 5369 PR #377. 5370 [Lutz Jaenicke] 5371 5372 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 5373 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 5374 removed entirely. 5375 [Richard Levitte] 5376 5377 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 5378 seems that in spite of existing for more than a year, many application 5379 author have done nothing to provide the necessary callbacks, which 5380 means that this particular engine will not work properly anywhere. 5381 This is a very unfortunate situation which forces us, in the name 5382 of usability, to give the hw_ncipher.c a static lock, which is part 5383 of libcrypto. 5384 NOTE: This is for the 0.9.7 series ONLY. This hack will never 5385 appear in 0.9.8 or later. We EXPECT application authors to have 5386 dealt properly with this when 0.9.8 is released (unless we actually 5387 make such changes in the libcrypto locking code that changes will 5388 have to be made anyway). 5389 [Richard Levitte] 5390 5391 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 5392 octets have been read, EOF or an error occurs. Without this change 5393 some truncated ASN1 structures will not produce an error. 5394 [Steve Henson] 5395 5396 *) Disable Heimdal support, since it hasn't been fully implemented. 5397 Still give the possibility to force the use of Heimdal, but with 5398 warnings and a request that patches get sent to openssl-dev. 5399 [Richard Levitte] 5400 5401 *) Add the VC-CE target, introduce the WINCE sysname, and add 5402 INSTALL.WCE and appropriate conditionals to make it build. 5403 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 5404 5405 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 5406 cygssl-x.y.z.dll, where x, y and z are the major, minor and 5407 edit numbers of the version. 5408 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 5409 5410 *) Introduce safe string copy and catenation functions 5411 (BUF_strlcpy() and BUF_strlcat()). 5412 [Ben Laurie (CHATS) and Richard Levitte] 5413 5414 *) Avoid using fixed-size buffers for one-line DNs. 5415 [Ben Laurie (CHATS)] 5416 5417 *) Add BUF_MEM_grow_clean() to avoid information leakage when 5418 resizing buffers containing secrets, and use where appropriate. 5419 [Ben Laurie (CHATS)] 5420 5421 *) Avoid using fixed size buffers for configuration file location. 5422 [Ben Laurie (CHATS)] 5423 5424 *) Avoid filename truncation for various CA files. 5425 [Ben Laurie (CHATS)] 5426 5427 *) Use sizeof in preference to magic numbers. 5428 [Ben Laurie (CHATS)] 5429 5430 *) Avoid filename truncation in cert requests. 5431 [Ben Laurie (CHATS)] 5432 5433 *) Add assertions to check for (supposedly impossible) buffer 5434 overflows. 5435 [Ben Laurie (CHATS)] 5436 5437 *) Don't cache truncated DNS entries in the local cache (this could 5438 potentially lead to a spoofing attack). 5439 [Ben Laurie (CHATS)] 5440 5441 *) Fix various buffers to be large enough for hex/decimal 5442 representations in a platform independent manner. 5443 [Ben Laurie (CHATS)] 5444 5445 *) Add CRYPTO_realloc_clean() to avoid information leakage when 5446 resizing buffers containing secrets, and use where appropriate. 5447 [Ben Laurie (CHATS)] 5448 5449 *) Add BIO_indent() to avoid much slightly worrying code to do 5450 indents. 5451 [Ben Laurie (CHATS)] 5452 5453 *) Convert sprintf()/BIO_puts() to BIO_printf(). 5454 [Ben Laurie (CHATS)] 5455 5456 *) buffer_gets() could terminate with the buffer only half 5457 full. Fixed. 5458 [Ben Laurie (CHATS)] 5459 5460 *) Add assertions to prevent user-supplied crypto functions from 5461 overflowing internal buffers by having large block sizes, etc. 5462 [Ben Laurie (CHATS)] 5463 5464 *) New OPENSSL_assert() macro (similar to assert(), but enabled 5465 unconditionally). 5466 [Ben Laurie (CHATS)] 5467 5468 *) Eliminate unused copy of key in RC4. 5469 [Ben Laurie (CHATS)] 5470 5471 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 5472 [Ben Laurie (CHATS)] 5473 5474 *) Fix off-by-one error in EGD path. 5475 [Ben Laurie (CHATS)] 5476 5477 *) If RANDFILE path is too long, ignore instead of truncating. 5478 [Ben Laurie (CHATS)] 5479 5480 *) Eliminate unused and incorrectly sized X.509 structure 5481 CBCParameter. 5482 [Ben Laurie (CHATS)] 5483 5484 *) Eliminate unused and dangerous function knumber(). 5485 [Ben Laurie (CHATS)] 5486 5487 *) Eliminate unused and dangerous structure, KSSL_ERR. 5488 [Ben Laurie (CHATS)] 5489 5490 *) Protect against overlong session ID context length in an encoded 5491 session object. Since these are local, this does not appear to be 5492 exploitable. 5493 [Ben Laurie (CHATS)] 5494 5495 *) Change from security patch (see 0.9.6e below) that did not affect 5496 the 0.9.6 release series: 5497 5498 Remote buffer overflow in SSL3 protocol - an attacker could 5499 supply an oversized master key in Kerberos-enabled versions. 5500 (CVE-2002-0657) 5501 [Ben Laurie (CHATS)] 5502 5503 *) Change the SSL kerb5 codes to match RFC 2712. 5504 [Richard Levitte] 5505 5506 *) Make -nameopt work fully for req and add -reqopt switch. 5507 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 5508 5509 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 5510 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 5511 5512 *) Make sure tests can be performed even if the corresponding algorithms 5513 have been removed entirely. This was also the last step to make 5514 OpenSSL compilable with DJGPP under all reasonable conditions. 5515 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 5516 5517 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 5518 to allow version independent disabling of normally unselected ciphers, 5519 which may be activated as a side-effect of selecting a single cipher. 5520 5521 (E.g., cipher list string "RSA" enables ciphersuites that are left 5522 out of "ALL" because they do not provide symmetric encryption. 5523 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 5524 [Lutz Jaenicke, Bodo Moeller] 5525 5526 *) Add appropriate support for separate platform-dependent build 5527 directories. The recommended way to make a platform-dependent 5528 build directory is the following (tested on Linux), maybe with 5529 some local tweaks: 5530 5531 # Place yourself outside of the OpenSSL source tree. In 5532 # this example, the environment variable OPENSSL_SOURCE 5533 # is assumed to contain the absolute OpenSSL source directory. 5534 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 5535 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 5536 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 5537 mkdir -p `dirname $F` 5538 ln -s $OPENSSL_SOURCE/$F $F 5539 done 5540 5541 To be absolutely sure not to disturb the source tree, a "make clean" 5542 is a good thing. If it isn't successfull, don't worry about it, 5543 it probably means the source directory is very clean. 5544 [Richard Levitte] 5545 5546 *) Make sure any ENGINE control commands make local copies of string 5547 pointers passed to them whenever necessary. Otherwise it is possible 5548 the caller may have overwritten (or deallocated) the original string 5549 data when a later ENGINE operation tries to use the stored values. 5550 [G��tz Babin-Ebell <babinebell@trustcenter.de>] 5551 5552 *) Improve diagnostics in file reading and command-line digests. 5553 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 5554 5555 *) Add AES modes CFB and OFB to the object database. Correct an 5556 error in AES-CFB decryption. 5557 [Richard Levitte] 5558 5559 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 5560 allows existing EVP_CIPHER_CTX structures to be reused after 5561 calling EVP_*Final(). This behaviour is used by encryption 5562 BIOs and some applications. This has the side effect that 5563 applications must explicitly clean up cipher contexts with 5564 EVP_CIPHER_CTX_cleanup() or they will leak memory. 5565 [Steve Henson] 5566 5567 *) Check the values of dna and dnb in bn_mul_recursive before calling 5568 bn_mul_comba (a non zero value means the a or b arrays do not contain 5569 n2 elements) and fallback to bn_mul_normal if either is not zero. 5570 [Steve Henson] 5571 5572 *) Fix escaping of non-ASCII characters when using the -subj option 5573 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 5574 [Lutz Jaenicke] 5575 5576 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 5577 form for "surname", serialNumber has no short form. 5578 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 5579 therefore remove "mail" short name for "internet 7". 5580 The OID for unique identifiers in X509 certificates is 5581 x500UniqueIdentifier, not uniqueIdentifier. 5582 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 5583 [Lutz Jaenicke] 5584 5585 *) Add an "init" command to the ENGINE config module and auto initialize 5586 ENGINEs. Without any "init" command the ENGINE will be initialized 5587 after all ctrl commands have been executed on it. If init=1 the 5588 ENGINE is initailized at that point (ctrls before that point are run 5589 on the uninitialized ENGINE and after on the initialized one). If 5590 init=0 then the ENGINE will not be iniatialized at all. 5591 [Steve Henson] 5592 5593 *) Fix the 'app_verify_callback' interface so that the user-defined 5594 argument is actually passed to the callback: In the 5595 SSL_CTX_set_cert_verify_callback() prototype, the callback 5596 declaration has been changed from 5597 int (*cb)() 5598 into 5599 int (*cb)(X509_STORE_CTX *,void *); 5600 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 5601 i=s->ctx->app_verify_callback(&ctx) 5602 has been changed into 5603 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 5604 5605 To update applications using SSL_CTX_set_cert_verify_callback(), 5606 a dummy argument can be added to their callback functions. 5607 [D. K. Smetters <smetters@parc.xerox.com>] 5608 5609 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 5610 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 5611 5612 *) Add and OPENSSL_LOAD_CONF define which will cause 5613 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 5614 This allows older applications to transparently support certain 5615 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 5616 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 5617 load the config file and OPENSSL_add_all_algorithms_conf() which will 5618 always load it have also been added. 5619 [Steve Henson] 5620 5621 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 5622 Adjust NIDs and EVP layer. 5623 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5624 5625 *) Config modules support in openssl utility. 5626 5627 Most commands now load modules from the config file, 5628 though in a few (such as version) this isn't done 5629 because it couldn't be used for anything. 5630 5631 In the case of ca and req the config file used is 5632 the same as the utility itself: that is the -config 5633 command line option can be used to specify an 5634 alternative file. 5635 [Steve Henson] 5636 5637 *) Move default behaviour from OPENSSL_config(). If appname is NULL 5638 use "openssl_conf" if filename is NULL use default openssl config file. 5639 [Steve Henson] 5640 5641 *) Add an argument to OPENSSL_config() to allow the use of an alternative 5642 config section name. Add a new flag to tolerate a missing config file 5643 and move code to CONF_modules_load_file(). 5644 [Steve Henson] 5645 5646 *) Support for crypto accelerator cards from Accelerated Encryption 5647 Processing, www.aep.ie. (Use engine 'aep') 5648 The support was copied from 0.9.6c [engine] and adapted/corrected 5649 to work with the new engine framework. 5650 [AEP Inc. and Richard Levitte] 5651 5652 *) Support for SureWare crypto accelerator cards from Baltimore 5653 Technologies. (Use engine 'sureware') 5654 The support was copied from 0.9.6c [engine] and adapted 5655 to work with the new engine framework. 5656 [Richard Levitte] 5657 5658 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 5659 make the newer ENGINE framework commands for the CHIL engine work. 5660 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 5661 5662 *) Make it possible to produce shared libraries on ReliantUNIX. 5663 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 5664 5665 *) Add the configuration target debug-linux-ppro. 5666 Make 'openssl rsa' use the general key loading routines 5667 implemented in apps.c, and make those routines able to 5668 handle the key format FORMAT_NETSCAPE and the variant 5669 FORMAT_IISSGC. 5670 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5671 5672 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 5673 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 5674 5675 *) Add -keyform to rsautl, and document -engine. 5676 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 5677 5678 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 5679 BIO_R_NO_SUCH_FILE error code rather than the generic 5680 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 5681 [Ben Laurie] 5682 5683 *) Add new functions 5684 ERR_peek_last_error 5685 ERR_peek_last_error_line 5686 ERR_peek_last_error_line_data. 5687 These are similar to 5688 ERR_peek_error 5689 ERR_peek_error_line 5690 ERR_peek_error_line_data, 5691 but report on the latest error recorded rather than the first one 5692 still in the error queue. 5693 [Ben Laurie, Bodo Moeller] 5694 5695 *) default_algorithms option in ENGINE config module. This allows things 5696 like: 5697 default_algorithms = ALL 5698 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 5699 [Steve Henson] 5700 5701 *) Prelminary ENGINE config module. 5702 [Steve Henson] 5703 5704 *) New experimental application configuration code. 5705 [Steve Henson] 5706 5707 *) Change the AES code to follow the same name structure as all other 5708 symmetric ciphers, and behave the same way. Move everything to 5709 the directory crypto/aes, thereby obsoleting crypto/rijndael. 5710 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 5711 5712 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 5713 [Ben Laurie and Theo de Raadt] 5714 5715 *) Add option to output public keys in req command. 5716 [Massimiliano Pala madwolf@openca.org] 5717 5718 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 5719 (up to about 10% better than before for P-192 and P-224). 5720 [Bodo Moeller] 5721 5722 *) New functions/macros 5723 5724 SSL_CTX_set_msg_callback(ctx, cb) 5725 SSL_CTX_set_msg_callback_arg(ctx, arg) 5726 SSL_set_msg_callback(ssl, cb) 5727 SSL_set_msg_callback_arg(ssl, arg) 5728 5729 to request calling a callback function 5730 5731 void cb(int write_p, int version, int content_type, 5732 const void *buf, size_t len, SSL *ssl, void *arg) 5733 5734 whenever a protocol message has been completely received 5735 (write_p == 0) or sent (write_p == 1). Here 'version' is the 5736 protocol version according to which the SSL library interprets 5737 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 5738 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 5739 the content type as defined in the SSL 3.0/TLS 1.0 protocol 5740 specification (change_cipher_spec(20), alert(21), handshake(22)). 5741 'buf' and 'len' point to the actual message, 'ssl' to the 5742 SSL object, and 'arg' is the application-defined value set by 5743 SSL[_CTX]_set_msg_callback_arg(). 5744 5745 'openssl s_client' and 'openssl s_server' have new '-msg' options 5746 to enable a callback that displays all protocol messages. 5747 [Bodo Moeller] 5748 5749 *) Change the shared library support so shared libraries are built as 5750 soon as the corresponding static library is finished, and thereby get 5751 openssl and the test programs linked against the shared library. 5752 This still only happens when the keyword "shard" has been given to 5753 the configuration scripts. 5754 5755 NOTE: shared library support is still an experimental thing, and 5756 backward binary compatibility is still not guaranteed. 5757 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 5758 5759 *) Add support for Subject Information Access extension. 5760 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 5761 5762 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 5763 additional bytes when new memory had to be allocated, not just 5764 when reusing an existing buffer. 5765 [Bodo Moeller] 5766 5767 *) New command line and configuration option 'utf8' for the req command. 5768 This allows field values to be specified as UTF8 strings. 5769 [Steve Henson] 5770 5771 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 5772 runs for the former and machine-readable output for the latter. 5773 [Ben Laurie] 5774 5775 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 5776 of the e-mail address in the DN (i.e., it will go into a certificate 5777 extension only). The new configuration file option 'email_in_dn = no' 5778 has the same effect. 5779 [Massimiliano Pala madwolf@openca.org] 5780 5781 *) Change all functions with names starting with des_ to be starting 5782 with DES_ instead. Add wrappers that are compatible with libdes, 5783 but are named _ossl_old_des_*. Finally, add macros that map the 5784 des_* symbols to the corresponding _ossl_old_des_* if libdes 5785 compatibility is desired. If OpenSSL 0.9.6c compatibility is 5786 desired, the des_* symbols will be mapped to DES_*, with one 5787 exception. 5788 5789 Since we provide two compatibility mappings, the user needs to 5790 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 5791 compatibility is desired. The default (i.e., when that macro 5792 isn't defined) is OpenSSL 0.9.6c compatibility. 5793 5794 There are also macros that enable and disable the support of old 5795 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 5796 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 5797 are defined, the default will apply: to support the old des routines. 5798 5799 In either case, one must include openssl/des.h to get the correct 5800 definitions. Do not try to just include openssl/des_old.h, that 5801 won't work. 5802 5803 NOTE: This is a major break of an old API into a new one. Software 5804 authors are encouraged to switch to the DES_ style functions. Some 5805 time in the future, des_old.h and the libdes compatibility functions 5806 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 5807 default), and then completely removed. 5808 [Richard Levitte] 5809 5810 *) Test for certificates which contain unsupported critical extensions. 5811 If such a certificate is found during a verify operation it is 5812 rejected by default: this behaviour can be overridden by either 5813 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 5814 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 5815 X509_supported_extension() has also been added which returns 1 if a 5816 particular extension is supported. 5817 [Steve Henson] 5818 5819 *) Modify the behaviour of EVP cipher functions in similar way to digests 5820 to retain compatibility with existing code. 5821 [Steve Henson] 5822 5823 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 5824 compatibility with existing code. In particular the 'ctx' parameter does 5825 not have to be to be initialized before the call to EVP_DigestInit() and 5826 it is tidied up after a call to EVP_DigestFinal(). New function 5827 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 5828 EVP_MD_CTX_copy() changed to not require the destination to be 5829 initialized valid and new function EVP_MD_CTX_copy_ex() added which 5830 requires the destination to be valid. 5831 5832 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 5833 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 5834 [Steve Henson] 5835 5836 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 5837 so that complete 'Handshake' protocol structures are kept in memory 5838 instead of overwriting 'msg_type' and 'length' with 'body' data. 5839 [Bodo Moeller] 5840 5841 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 5842 [Massimo Santin via Richard Levitte] 5843 5844 *) Major restructuring to the underlying ENGINE code. This includes 5845 reduction of linker bloat, separation of pure "ENGINE" manipulation 5846 (initialisation, etc) from functionality dealing with implementations 5847 of specific crypto iterfaces. This change also introduces integrated 5848 support for symmetric ciphers and digest implementations - so ENGINEs 5849 can now accelerate these by providing EVP_CIPHER and EVP_MD 5850 implementations of their own. This is detailed in crypto/engine/README 5851 as it couldn't be adequately described here. However, there are a few 5852 API changes worth noting - some RSA, DSA, DH, and RAND functions that 5853 were changed in the original introduction of ENGINE code have now 5854 reverted back - the hooking from this code to ENGINE is now a good 5855 deal more passive and at run-time, operations deal directly with 5856 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 5857 dereferencing through an ENGINE pointer any more. Also, the ENGINE 5858 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 5859 they were not being used by the framework as there is no concept of a 5860 BIGNUM_METHOD and they could not be generalised to the new 5861 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 5862 ENGINE_cpy() has been removed as it cannot be consistently defined in 5863 the new code. 5864 [Geoff Thorpe] 5865 5866 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 5867 [Steve Henson] 5868 5869 *) Change mkdef.pl to sort symbols that get the same entry number, 5870 and make sure the automatically generated functions ERR_load_* 5871 become part of libeay.num as well. 5872 [Richard Levitte] 5873 5874 *) New function SSL_renegotiate_pending(). This returns true once 5875 renegotiation has been requested (either SSL_renegotiate() call 5876 or HelloRequest/ClientHello receveived from the peer) and becomes 5877 false once a handshake has been completed. 5878 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 5879 sends a HelloRequest, but does not ensure that a handshake takes 5880 place. SSL_renegotiate_pending() is useful for checking if the 5881 client has followed the request.) 5882 [Bodo Moeller] 5883 5884 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 5885 By default, clients may request session resumption even during 5886 renegotiation (if session ID contexts permit); with this option, 5887 session resumption is possible only in the first handshake. 5888 5889 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 5890 more bits available for options that should not be part of 5891 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 5892 [Bodo Moeller] 5893 5894 *) Add some demos for certificate and certificate request creation. 5895 [Steve Henson] 5896 5897 *) Make maximum certificate chain size accepted from the peer application 5898 settable (SSL*_get/set_max_cert_list()), as proposed by 5899 "Douglas E. Engert" <deengert@anl.gov>. 5900 [Lutz Jaenicke] 5901 5902 *) Add support for shared libraries for Unixware-7 5903 (Boyd Lynn Gerber <gerberb@zenez.com>). 5904 [Lutz Jaenicke] 5905 5906 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 5907 be done prior to destruction. Use this to unload error strings from 5908 ENGINEs that load their own error strings. NB: This adds two new API 5909 functions to "get" and "set" this destroy handler in an ENGINE. 5910 [Geoff Thorpe] 5911 5912 *) Alter all existing ENGINE implementations (except "openssl" and 5913 "openbsd") to dynamically instantiate their own error strings. This 5914 makes them more flexible to be built both as statically-linked ENGINEs 5915 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 5916 Also, add stub code to each that makes building them as self-contained 5917 shared-libraries easier (see README.ENGINE). 5918 [Geoff Thorpe] 5919 5920 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 5921 implementations into applications that are completely implemented in 5922 self-contained shared-libraries. The "dynamic" ENGINE exposes control 5923 commands that can be used to configure what shared-library to load and 5924 to control aspects of the way it is handled. Also, made an update to 5925 the README.ENGINE file that brings its information up-to-date and 5926 provides some information and instructions on the "dynamic" ENGINE 5927 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 5928 [Geoff Thorpe] 5929 5930 *) Make it possible to unload ranges of ERR strings with a new 5931 "ERR_unload_strings" function. 5932 [Geoff Thorpe] 5933 5934 *) Add a copy() function to EVP_MD. 5935 [Ben Laurie] 5936 5937 *) Make EVP_MD routines take a context pointer instead of just the 5938 md_data void pointer. 5939 [Ben Laurie] 5940 5941 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 5942 that the digest can only process a single chunk of data 5943 (typically because it is provided by a piece of 5944 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 5945 is only going to provide a single chunk of data, and hence the 5946 framework needn't accumulate the data for oneshot drivers. 5947 [Ben Laurie] 5948 5949 *) As with "ERR", make it possible to replace the underlying "ex_data" 5950 functions. This change also alters the storage and management of global 5951 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 5952 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 5953 index counters. The API functions that use this state have been changed 5954 to take a "class_index" rather than pointers to the class's local STACK 5955 and counter, and there is now an API function to dynamically create new 5956 classes. This centralisation allows us to (a) plug a lot of the 5957 thread-safety problems that existed, and (b) makes it possible to clean 5958 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 5959 such data would previously have always leaked in application code and 5960 workarounds were in place to make the memory debugging turn a blind eye 5961 to it. Application code that doesn't use this new function will still 5962 leak as before, but their memory debugging output will announce it now 5963 rather than letting it slide. 5964 5965 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 5966 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 5967 has a return value to indicate success or failure. 5968 [Geoff Thorpe] 5969 5970 *) Make it possible to replace the underlying "ERR" functions such that the 5971 global state (2 LHASH tables and 2 locks) is only used by the "default" 5972 implementation. This change also adds two functions to "get" and "set" 5973 the implementation prior to it being automatically set the first time 5974 any other ERR function takes place. Ie. an application can call "get", 5975 pass the return value to a module it has just loaded, and that module 5976 can call its own "set" function using that value. This means the 5977 module's "ERR" operations will use (and modify) the error state in the 5978 application and not in its own statically linked copy of OpenSSL code. 5979 [Geoff Thorpe] 5980 5981 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 5982 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 5983 the operation, and provides a more encapsulated way for external code 5984 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 5985 to use these functions rather than manually incrementing the counts. 5986 5987 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 5988 [Geoff Thorpe] 5989 5990 *) Add EVP test program. 5991 [Ben Laurie] 5992 5993 *) Add symmetric cipher support to ENGINE. Expect the API to change! 5994 [Ben Laurie] 5995 5996 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 5997 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 5998 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 5999 These allow a CRL to be built without having to access X509_CRL fields 6000 directly. Modify 'ca' application to use new functions. 6001 [Steve Henson] 6002 6003 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 6004 bug workarounds. Rollback attack detection is a security feature. 6005 The problem will only arise on OpenSSL servers when TLSv1 is not 6006 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 6007 Software authors not wanting to support TLSv1 will have special reasons 6008 for their choice and can explicitly enable this option. 6009 [Bodo Moeller, Lutz Jaenicke] 6010 6011 *) Rationalise EVP so it can be extended: don't include a union of 6012 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 6013 (similar to those existing for EVP_CIPHER_CTX). 6014 Usage example: 6015 6016 EVP_MD_CTX md; 6017 6018 EVP_MD_CTX_init(&md); /* new function call */ 6019 EVP_DigestInit(&md, EVP_sha1()); 6020 EVP_DigestUpdate(&md, in, len); 6021 EVP_DigestFinal(&md, out, NULL); 6022 EVP_MD_CTX_cleanup(&md); /* new function call */ 6023 6024 [Ben Laurie] 6025 6026 *) Make DES key schedule conform to the usual scheme, as well as 6027 correcting its structure. This means that calls to DES functions 6028 now have to pass a pointer to a des_key_schedule instead of a 6029 plain des_key_schedule (which was actually always a pointer 6030 anyway): E.g., 6031 6032 des_key_schedule ks; 6033 6034 des_set_key_checked(..., &ks); 6035 des_ncbc_encrypt(..., &ks, ...); 6036 6037 (Note that a later change renames 'des_...' into 'DES_...'.) 6038 [Ben Laurie] 6039 6040 *) Initial reduction of linker bloat: the use of some functions, such as 6041 PEM causes large amounts of unused functions to be linked in due to 6042 poor organisation. For example pem_all.c contains every PEM function 6043 which has a knock on effect of linking in large amounts of (unused) 6044 ASN1 code. Grouping together similar functions and splitting unrelated 6045 functions prevents this. 6046 [Steve Henson] 6047 6048 *) Cleanup of EVP macros. 6049 [Ben Laurie] 6050 6051 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 6052 correct _ecb suffix. 6053 [Ben Laurie] 6054 6055 *) Add initial OCSP responder support to ocsp application. The 6056 revocation information is handled using the text based index 6057 use by the ca application. The responder can either handle 6058 requests generated internally, supplied in files (for example 6059 via a CGI script) or using an internal minimal server. 6060 [Steve Henson] 6061 6062 *) Add configuration choices to get zlib compression for TLS. 6063 [Richard Levitte] 6064 6065 *) Changes to Kerberos SSL for RFC 2712 compliance: 6066 1. Implemented real KerberosWrapper, instead of just using 6067 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 6068 2. Implemented optional authenticator field of KerberosWrapper. 6069 6070 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 6071 and authenticator structs; see crypto/krb5/. 6072 6073 Generalized Kerberos calls to support multiple Kerberos libraries. 6074 [Vern Staats <staatsvr@asc.hpc.mil>, 6075 Jeffrey Altman <jaltman@columbia.edu> 6076 via Richard Levitte] 6077 6078 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 6079 already does with RSA. testdsa.h now has 'priv_key/pub_key' 6080 values for each of the key sizes rather than having just 6081 parameters (and 'speed' generating keys each time). 6082 [Geoff Thorpe] 6083 6084 *) Speed up EVP routines. 6085 Before: 6086encrypt 6087type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 6088des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 6089des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 6090des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 6091decrypt 6092des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 6093des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 6094des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 6095 After: 6096encrypt 6097des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 6098decrypt 6099des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 6100 [Ben Laurie] 6101 6102 *) Added the OS2-EMX target. 6103 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 6104 6105 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 6106 to support NCONF routines in extension code. New function CONF_set_nconf() 6107 to allow functions which take an NCONF to also handle the old LHASH 6108 structure: this means that the old CONF compatible routines can be 6109 retained (in particular wrt extensions) without having to duplicate the 6110 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 6111 [Steve Henson] 6112 6113 *) Enhance the general user interface with mechanisms for inner control 6114 and with possibilities to have yes/no kind of prompts. 6115 [Richard Levitte] 6116 6117 *) Change all calls to low level digest routines in the library and 6118 applications to use EVP. Add missing calls to HMAC_cleanup() and 6119 don't assume HMAC_CTX can be copied using memcpy(). 6120 [Verdon Walker <VWalker@novell.com>, Steve Henson] 6121 6122 *) Add the possibility to control engines through control names but with 6123 arbitrary arguments instead of just a string. 6124 Change the key loaders to take a UI_METHOD instead of a callback 6125 function pointer. NOTE: this breaks binary compatibility with earlier 6126 versions of OpenSSL [engine]. 6127 Adapt the nCipher code for these new conditions and add a card insertion 6128 callback. 6129 [Richard Levitte] 6130 6131 *) Enhance the general user interface with mechanisms to better support 6132 dialog box interfaces, application-defined prompts, the possibility 6133 to use defaults (for example default passwords from somewhere else) 6134 and interrupts/cancellations. 6135 [Richard Levitte] 6136 6137 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 6138 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 6139 [Steve Henson] 6140 6141 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 6142 tidy up some unnecessarily weird code in 'sk_new()'). 6143 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 6144 6145 *) Change the key loading routines for ENGINEs to use the same kind 6146 callback (pem_password_cb) as all other routines that need this 6147 kind of callback. 6148 [Richard Levitte] 6149 6150 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 6151 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 6152 than this minimum value is recommended. 6153 [Lutz Jaenicke] 6154 6155 *) New random seeder for OpenVMS, using the system process statistics 6156 that are easily reachable. 6157 [Richard Levitte] 6158 6159 *) Windows apparently can't transparently handle global 6160 variables defined in DLLs. Initialisations such as: 6161 6162 const ASN1_ITEM *it = &ASN1_INTEGER_it; 6163 6164 wont compile. This is used by the any applications that need to 6165 declare their own ASN1 modules. This was fixed by adding the option 6166 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 6167 needed for static libraries under Win32. 6168 [Steve Henson] 6169 6170 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 6171 setting of purpose and trust fields. New X509_STORE trust and 6172 purpose functions and tidy up setting in other SSL functions. 6173 [Steve Henson] 6174 6175 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 6176 structure. These are inherited by X509_STORE_CTX when it is 6177 initialised. This allows various defaults to be set in the 6178 X509_STORE structure (such as flags for CRL checking and custom 6179 purpose or trust settings) for functions which only use X509_STORE_CTX 6180 internally such as S/MIME. 6181 6182 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 6183 trust settings if they are not set in X509_STORE. This allows X509_STORE 6184 purposes and trust (in S/MIME for example) to override any set by default. 6185 6186 Add command line options for CRL checking to smime, s_client and s_server 6187 applications. 6188 [Steve Henson] 6189 6190 *) Initial CRL based revocation checking. If the CRL checking flag(s) 6191 are set then the CRL is looked up in the X509_STORE structure and 6192 its validity and signature checked, then if the certificate is found 6193 in the CRL the verify fails with a revoked error. 6194 6195 Various new CRL related callbacks added to X509_STORE_CTX structure. 6196 6197 Command line options added to 'verify' application to support this. 6198 6199 This needs some additional work, such as being able to handle multiple 6200 CRLs with different times, extension based lookup (rather than just 6201 by subject name) and ultimately more complete V2 CRL extension 6202 handling. 6203 [Steve Henson] 6204 6205 *) Add a general user interface API (crypto/ui/). This is designed 6206 to replace things like des_read_password and friends (backward 6207 compatibility functions using this new API are provided). 6208 The purpose is to remove prompting functions from the DES code 6209 section as well as provide for prompting through dialog boxes in 6210 a window system and the like. 6211 [Richard Levitte] 6212 6213 *) Add "ex_data" support to ENGINE so implementations can add state at a 6214 per-structure level rather than having to store it globally. 6215 [Geoff] 6216 6217 *) Make it possible for ENGINE structures to be copied when retrieved by 6218 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 6219 This causes the "original" ENGINE structure to act like a template, 6220 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 6221 operational state can be localised to each ENGINE structure, despite the 6222 fact they all share the same "methods". New ENGINE structures returned in 6223 this case have no functional references and the return value is the single 6224 structural reference. This matches the single structural reference returned 6225 by ENGINE_by_id() normally, when it is incremented on the pre-existing 6226 ENGINE structure. 6227 [Geoff] 6228 6229 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 6230 needs to match any other type at all we need to manually clear the 6231 tag cache. 6232 [Steve Henson] 6233 6234 *) Changes to the "openssl engine" utility to include; 6235 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 6236 about an ENGINE's available control commands. 6237 - executing control commands from command line arguments using the 6238 '-pre' and '-post' switches. '-post' is only used if '-t' is 6239 specified and the ENGINE is successfully initialised. The syntax for 6240 the individual commands are colon-separated, for example; 6241 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 6242 [Geoff] 6243 6244 *) New dynamic control command support for ENGINEs. ENGINEs can now 6245 declare their own commands (numbers), names (strings), descriptions, 6246 and input types for run-time discovery by calling applications. A 6247 subset of these commands are implicitly classed as "executable" 6248 depending on their input type, and only these can be invoked through 6249 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 6250 can be based on user input, config files, etc). The distinction is 6251 that "executable" commands cannot return anything other than a boolean 6252 result and can only support numeric or string input, whereas some 6253 discoverable commands may only be for direct use through 6254 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 6255 pointers, or other custom uses. The "executable" commands are to 6256 support parameterisations of ENGINE behaviour that can be 6257 unambiguously defined by ENGINEs and used consistently across any 6258 OpenSSL-based application. Commands have been added to all the 6259 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 6260 control over shared-library paths without source code alterations. 6261 [Geoff] 6262 6263 *) Changed all ENGINE implementations to dynamically allocate their 6264 ENGINEs rather than declaring them statically. Apart from this being 6265 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 6266 this also allows the implementations to compile without using the 6267 internal engine_int.h header. 6268 [Geoff] 6269 6270 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 6271 'const' value. Any code that should be able to modify a RAND_METHOD 6272 should already have non-const pointers to it (ie. they should only 6273 modify their own ones). 6274 [Geoff] 6275 6276 *) Made a variety of little tweaks to the ENGINE code. 6277 - "atalla" and "ubsec" string definitions were moved from header files 6278 to C code. "nuron" string definitions were placed in variables 6279 rather than hard-coded - allowing parameterisation of these values 6280 later on via ctrl() commands. 6281 - Removed unused "#if 0"'d code. 6282 - Fixed engine list iteration code so it uses ENGINE_free() to release 6283 structural references. 6284 - Constified the RAND_METHOD element of ENGINE structures. 6285 - Constified various get/set functions as appropriate and added 6286 missing functions (including a catch-all ENGINE_cpy that duplicates 6287 all ENGINE values onto a new ENGINE except reference counts/state). 6288 - Removed NULL parameter checks in get/set functions. Setting a method 6289 or function to NULL is a way of cancelling out a previously set 6290 value. Passing a NULL ENGINE parameter is just plain stupid anyway 6291 and doesn't justify the extra error symbols and code. 6292 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 6293 flags from engine_int.h to engine.h. 6294 - Changed prototypes for ENGINE handler functions (init(), finish(), 6295 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 6296 [Geoff] 6297 6298 *) Implement binary inversion algorithm for BN_mod_inverse in addition 6299 to the algorithm using long division. The binary algorithm can be 6300 used only if the modulus is odd. On 32-bit systems, it is faster 6301 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 6302 roughly 5-15% for 256-bit moduli), so we use it only for moduli 6303 up to 450 bits. In 64-bit environments, the binary algorithm 6304 appears to be advantageous for much longer moduli; here we use it 6305 for moduli up to 2048 bits. 6306 [Bodo Moeller] 6307 6308 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 6309 could not support the combine flag in choice fields. 6310 [Steve Henson] 6311 6312 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 6313 extensions from a certificate request to the certificate. 6314 [Steve Henson] 6315 6316 *) Allow multiple 'certopt' and 'nameopt' options to be separated 6317 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 6318 file: this allows the display of the certificate about to be 6319 signed to be customised, to allow certain fields to be included 6320 or excluded and extension details. The old system didn't display 6321 multicharacter strings properly, omitted fields not in the policy 6322 and couldn't display additional details such as extensions. 6323 [Steve Henson] 6324 6325 *) Function EC_POINTs_mul for multiple scalar multiplication 6326 of an arbitrary number of elliptic curve points 6327 \sum scalars[i]*points[i], 6328 optionally including the generator defined for the EC_GROUP: 6329 scalar*generator + \sum scalars[i]*points[i]. 6330 6331 EC_POINT_mul is a simple wrapper function for the typical case 6332 that the point list has just one item (besides the optional 6333 generator). 6334 [Bodo Moeller] 6335 6336 *) First EC_METHODs for curves over GF(p): 6337 6338 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 6339 operations and provides various method functions that can also 6340 operate with faster implementations of modular arithmetic. 6341 6342 EC_GFp_mont_method() reuses most functions that are part of 6343 EC_GFp_simple_method, but uses Montgomery arithmetic. 6344 6345 [Bodo Moeller; point addition and point doubling 6346 implementation directly derived from source code provided by 6347 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 6348 6349 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 6350 crypto/ec/ec_lib.c): 6351 6352 Curves are EC_GROUP objects (with an optional group generator) 6353 based on EC_METHODs that are built into the library. 6354 6355 Points are EC_POINT objects based on EC_GROUP objects. 6356 6357 Most of the framework would be able to handle curves over arbitrary 6358 finite fields, but as there are no obvious types for fields other 6359 than GF(p), some functions are limited to that for now. 6360 [Bodo Moeller] 6361 6362 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 6363 that the file contains a complete HTTP response. 6364 [Richard Levitte] 6365 6366 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 6367 change the def and num file printf format specifier from "%-40sXXX" 6368 to "%-39s XXX". The latter will always guarantee a space after the 6369 field while the former will cause them to run together if the field 6370 is 40 of more characters long. 6371 [Steve Henson] 6372 6373 *) Constify the cipher and digest 'method' functions and structures 6374 and modify related functions to take constant EVP_MD and EVP_CIPHER 6375 pointers. 6376 [Steve Henson] 6377 6378 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 6379 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 6380 [Bodo Moeller] 6381 6382 *) Modify EVP_Digest*() routines so they now return values. Although the 6383 internal software routines can never fail additional hardware versions 6384 might. 6385 [Steve Henson] 6386 6387 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 6388 6389 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 6390 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 6391 6392 ASN1 error codes 6393 ERR_R_NESTED_ASN1_ERROR 6394 ... 6395 ERR_R_MISSING_ASN1_EOS 6396 were 4 .. 9, conflicting with 6397 ERR_LIB_RSA (= ERR_R_RSA_LIB) 6398 ... 6399 ERR_LIB_PEM (= ERR_R_PEM_LIB). 6400 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 6401 6402 Add new error code 'ERR_R_INTERNAL_ERROR'. 6403 [Bodo Moeller] 6404 6405 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 6406 suffices. 6407 [Bodo Moeller] 6408 6409 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 6410 sets the subject name for a new request or supersedes the 6411 subject name in a given request. Formats that can be parsed are 6412 'CN=Some Name, OU=myOU, C=IT' 6413 and 6414 'CN=Some Name/OU=myOU/C=IT'. 6415 6416 Add options '-batch' and '-verbose' to 'openssl req'. 6417 [Massimiliano Pala <madwolf@hackmasters.net>] 6418 6419 *) Introduce the possibility to access global variables through 6420 functions on platform were that's the best way to handle exporting 6421 global variables in shared libraries. To enable this functionality, 6422 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 6423 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 6424 is normally done by Configure or something similar). 6425 6426 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 6427 in the source file (foo.c) like this: 6428 6429 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 6430 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 6431 6432 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 6433 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 6434 6435 OPENSSL_DECLARE_GLOBAL(int,foo); 6436 #define foo OPENSSL_GLOBAL_REF(foo) 6437 OPENSSL_DECLARE_GLOBAL(double,bar); 6438 #define bar OPENSSL_GLOBAL_REF(bar) 6439 6440 The #defines are very important, and therefore so is including the 6441 header file everywhere where the defined globals are used. 6442 6443 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 6444 of ASN.1 items, but that structure is a bit different. 6445 6446 The largest change is in util/mkdef.pl which has been enhanced with 6447 better and easier to understand logic to choose which symbols should 6448 go into the Windows .def files as well as a number of fixes and code 6449 cleanup (among others, algorithm keywords are now sorted 6450 lexicographically to avoid constant rewrites). 6451 [Richard Levitte] 6452 6453 *) In BN_div() keep a copy of the sign of 'num' before writing the 6454 result to 'rm' because if rm==num the value will be overwritten 6455 and produce the wrong result if 'num' is negative: this caused 6456 problems with BN_mod() and BN_nnmod(). 6457 [Steve Henson] 6458 6459 *) Function OCSP_request_verify(). This checks the signature on an 6460 OCSP request and verifies the signer certificate. The signer 6461 certificate is just checked for a generic purpose and OCSP request 6462 trust settings. 6463 [Steve Henson] 6464 6465 *) Add OCSP_check_validity() function to check the validity of OCSP 6466 responses. OCSP responses are prepared in real time and may only 6467 be a few seconds old. Simply checking that the current time lies 6468 between thisUpdate and nextUpdate max reject otherwise valid responses 6469 caused by either OCSP responder or client clock inaccuracy. Instead 6470 we allow thisUpdate and nextUpdate to fall within a certain period of 6471 the current time. The age of the response can also optionally be 6472 checked. Two new options -validity_period and -status_age added to 6473 ocsp utility. 6474 [Steve Henson] 6475 6476 *) If signature or public key algorithm is unrecognized print out its 6477 OID rather that just UNKNOWN. 6478 [Steve Henson] 6479 6480 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 6481 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 6482 ID to be generated from the issuer certificate alone which can then be 6483 passed to OCSP_id_issuer_cmp(). 6484 [Steve Henson] 6485 6486 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 6487 ASN1 modules to export functions returning ASN1_ITEM pointers 6488 instead of the ASN1_ITEM structures themselves. This adds several 6489 new macros which allow the underlying ASN1 function/structure to 6490 be accessed transparently. As a result code should not use ASN1_ITEM 6491 references directly (such as &X509_it) but instead use the relevant 6492 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 6493 use of the new ASN1 code on platforms where exporting structures 6494 is problematical (for example in shared libraries) but exporting 6495 functions returning pointers to structures is not. 6496 [Steve Henson] 6497 6498 *) Add support for overriding the generation of SSL/TLS session IDs. 6499 These callbacks can be registered either in an SSL_CTX or per SSL. 6500 The purpose of this is to allow applications to control, if they wish, 6501 the arbitrary values chosen for use as session IDs, particularly as it 6502 can be useful for session caching in multiple-server environments. A 6503 command-line switch for testing this (and any client code that wishes 6504 to use such a feature) has been added to "s_server". 6505 [Geoff Thorpe, Lutz Jaenicke] 6506 6507 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 6508 of the form '#if defined(...) || defined(...) || ...' and 6509 '#if !defined(...) && !defined(...) && ...'. This also avoids 6510 the growing number of special cases it was previously handling. 6511 [Richard Levitte] 6512 6513 *) Make all configuration macros available for application by making 6514 sure they are available in opensslconf.h, by giving them names starting 6515 with "OPENSSL_" to avoid conflicts with other packages and by making 6516 sure e_os2.h will cover all platform-specific cases together with 6517 opensslconf.h. 6518 Additionally, it is now possible to define configuration/platform- 6519 specific names (called "system identities"). In the C code, these 6520 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 6521 macro with the name beginning with "OPENSSL_SYS_", which is determined 6522 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 6523 what is available. 6524 [Richard Levitte] 6525 6526 *) New option -set_serial to 'req' and 'x509' this allows the serial 6527 number to use to be specified on the command line. Previously self 6528 signed certificates were hard coded with serial number 0 and the 6529 CA options of 'x509' had to use a serial number in a file which was 6530 auto incremented. 6531 [Steve Henson] 6532 6533 *) New options to 'ca' utility to support V2 CRL entry extensions. 6534 Currently CRL reason, invalidity date and hold instruction are 6535 supported. Add new CRL extensions to V3 code and some new objects. 6536 [Steve Henson] 6537 6538 *) New function EVP_CIPHER_CTX_set_padding() this is used to 6539 disable standard block padding (aka PKCS#5 padding) in the EVP 6540 API, which was previously mandatory. This means that the data is 6541 not padded in any way and so the total length much be a multiple 6542 of the block size, otherwise an error occurs. 6543 [Steve Henson] 6544 6545 *) Initial (incomplete) OCSP SSL support. 6546 [Steve Henson] 6547 6548 *) New function OCSP_parse_url(). This splits up a URL into its host, 6549 port and path components: primarily to parse OCSP URLs. New -url 6550 option to ocsp utility. 6551 [Steve Henson] 6552 6553 *) New nonce behavior. The return value of OCSP_check_nonce() now 6554 reflects the various checks performed. Applications can decide 6555 whether to tolerate certain situations such as an absent nonce 6556 in a response when one was present in a request: the ocsp application 6557 just prints out a warning. New function OCSP_add1_basic_nonce() 6558 this is to allow responders to include a nonce in a response even if 6559 the request is nonce-less. 6560 [Steve Henson] 6561 6562 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 6563 skipped when using openssl x509 multiple times on a single input file, 6564 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 6565 [Bodo Moeller] 6566 6567 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 6568 set string type: to handle setting ASN1_TIME structures. Fix ca 6569 utility to correctly initialize revocation date of CRLs. 6570 [Steve Henson] 6571 6572 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 6573 the clients preferred ciphersuites and rather use its own preferences. 6574 Should help to work around M$ SGC (Server Gated Cryptography) bug in 6575 Internet Explorer by ensuring unchanged hash method during stepup. 6576 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 6577 [Lutz Jaenicke] 6578 6579 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 6580 to aes and add a new 'exist' option to print out symbols that don't 6581 appear to exist. 6582 [Steve Henson] 6583 6584 *) Additional options to ocsp utility to allow flags to be set and 6585 additional certificates supplied. 6586 [Steve Henson] 6587 6588 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 6589 OCSP client a number of certificate to only verify the response 6590 signature against. 6591 [Richard Levitte] 6592 6593 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 6594 handle the new API. Currently only ECB, CBC modes supported. Add new 6595 AES OIDs. 6596 6597 Add TLS AES ciphersuites as described in RFC3268, "Advanced 6598 Encryption Standard (AES) Ciphersuites for Transport Layer 6599 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 6600 not enabled by default and were not part of the "ALL" ciphersuite 6601 alias because they were not yet official; they could be 6602 explicitly requested by specifying the "AESdraft" ciphersuite 6603 group alias. In the final release of OpenSSL 0.9.7, the group 6604 alias is called "AES" and is part of "ALL".) 6605 [Ben Laurie, Steve Henson, Bodo Moeller] 6606 6607 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 6608 request to response. 6609 [Steve Henson] 6610 6611 *) Functions for OCSP responders. OCSP_request_onereq_count(), 6612 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 6613 extract information from a certificate request. OCSP_response_create() 6614 creates a response and optionally adds a basic response structure. 6615 OCSP_basic_add1_status() adds a complete single response to a basic 6616 response and returns the OCSP_SINGLERESP structure just added (to allow 6617 extensions to be included for example). OCSP_basic_add1_cert() adds a 6618 certificate to a basic response and OCSP_basic_sign() signs a basic 6619 response with various flags. New helper functions ASN1_TIME_check() 6620 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 6621 (converts ASN1_TIME to GeneralizedTime). 6622 [Steve Henson] 6623 6624 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 6625 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 6626 structure from a certificate. X509_pubkey_digest() digests the public_key 6627 contents: this is used in various key identifiers. 6628 [Steve Henson] 6629 6630 *) Make sk_sort() tolerate a NULL argument. 6631 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 6632 6633 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 6634 passed by the function are trusted implicitly. If any of them signed the 6635 response then it is assumed to be valid and is not verified. 6636 [Steve Henson] 6637 6638 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 6639 to data. This was previously part of the PKCS7 ASN1 code. This 6640 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 6641 [Steve Henson, reported by Kenneth R. Robinette 6642 <support@securenetterm.com>] 6643 6644 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 6645 routines: without these tracing memory leaks is very painful. 6646 Fix leaks in PKCS12 and PKCS7 routines. 6647 [Steve Henson] 6648 6649 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 6650 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 6651 effectively meant GeneralizedTime would never be used. Now it 6652 is initialised to -1 but X509_time_adj() now has to check the value 6653 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 6654 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 6655 [Steve Henson, reported by Kenneth R. Robinette 6656 <support@securenetterm.com>] 6657 6658 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 6659 result in a zero length in the ASN1_INTEGER structure which was 6660 not consistent with the structure when d2i_ASN1_INTEGER() was used 6661 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 6662 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 6663 where it did not print out a minus for negative ASN1_INTEGER. 6664 [Steve Henson] 6665 6666 *) Add summary printout to ocsp utility. The various functions which 6667 convert status values to strings have been renamed to: 6668 OCSP_response_status_str(), OCSP_cert_status_str() and 6669 OCSP_crl_reason_str() and are no longer static. New options 6670 to verify nonce values and to disable verification. OCSP response 6671 printout format cleaned up. 6672 [Steve Henson] 6673 6674 *) Add additional OCSP certificate checks. These are those specified 6675 in RFC2560. This consists of two separate checks: the CA of the 6676 certificate being checked must either be the OCSP signer certificate 6677 or the issuer of the OCSP signer certificate. In the latter case the 6678 OCSP signer certificate must contain the OCSP signing extended key 6679 usage. This check is performed by attempting to match the OCSP 6680 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 6681 in the OCSP_CERTID structures of the response. 6682 [Steve Henson] 6683 6684 *) Initial OCSP certificate verification added to OCSP_basic_verify() 6685 and related routines. This uses the standard OpenSSL certificate 6686 verify routines to perform initial checks (just CA validity) and 6687 to obtain the certificate chain. Then additional checks will be 6688 performed on the chain. Currently the root CA is checked to see 6689 if it is explicitly trusted for OCSP signing. This is used to set 6690 a root CA as a global signing root: that is any certificate that 6691 chains to that CA is an acceptable OCSP signing certificate. 6692 [Steve Henson] 6693 6694 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 6695 extensions from a separate configuration file. 6696 As when reading extensions from the main configuration file, 6697 the '-extensions ...' option may be used for specifying the 6698 section to use. 6699 [Massimiliano Pala <madwolf@comune.modena.it>] 6700 6701 *) New OCSP utility. Allows OCSP requests to be generated or 6702 read. The request can be sent to a responder and the output 6703 parsed, outputed or printed in text form. Not complete yet: 6704 still needs to check the OCSP response validity. 6705 [Steve Henson] 6706 6707 *) New subcommands for 'openssl ca': 6708 'openssl ca -status <serial>' prints the status of the cert with 6709 the given serial number (according to the index file). 6710 'openssl ca -updatedb' updates the expiry status of certificates 6711 in the index file. 6712 [Massimiliano Pala <madwolf@comune.modena.it>] 6713 6714 *) New '-newreq-nodes' command option to CA.pl. This is like 6715 '-newreq', but calls 'openssl req' with the '-nodes' option 6716 so that the resulting key is not encrypted. 6717 [Damien Miller <djm@mindrot.org>] 6718 6719 *) New configuration for the GNU Hurd. 6720 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 6721 6722 *) Initial code to implement OCSP basic response verify. This 6723 is currently incomplete. Currently just finds the signer's 6724 certificate and verifies the signature on the response. 6725 [Steve Henson] 6726 6727 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 6728 value of OPENSSLDIR. This is available via the new '-d' option 6729 to 'openssl version', and is also included in 'openssl version -a'. 6730 [Bodo Moeller] 6731 6732 *) Allowing defining memory allocation callbacks that will be given 6733 file name and line number information in additional arguments 6734 (a const char* and an int). The basic functionality remains, as 6735 well as the original possibility to just replace malloc(), 6736 realloc() and free() by functions that do not know about these 6737 additional arguments. To register and find out the current 6738 settings for extended allocation functions, the following 6739 functions are provided: 6740 6741 CRYPTO_set_mem_ex_functions 6742 CRYPTO_set_locked_mem_ex_functions 6743 CRYPTO_get_mem_ex_functions 6744 CRYPTO_get_locked_mem_ex_functions 6745 6746 These work the same way as CRYPTO_set_mem_functions and friends. 6747 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 6748 extended allocation function is enabled. 6749 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 6750 a conventional allocation function is enabled. 6751 [Richard Levitte, Bodo Moeller] 6752 6753 *) Finish off removing the remaining LHASH function pointer casts. 6754 There should no longer be any prototype-casting required when using 6755 the LHASH abstraction, and any casts that remain are "bugs". See 6756 the callback types and macros at the head of lhash.h for details 6757 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 6758 [Geoff Thorpe] 6759 6760 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 6761 If /dev/[u]random devices are not available or do not return enough 6762 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 6763 be queried. 6764 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 6765 /etc/entropy will be queried once each in this sequence, quering stops 6766 when enough entropy was collected without querying more sockets. 6767 [Lutz Jaenicke] 6768 6769 *) Change the Unix RAND_poll() variant to be able to poll several 6770 random devices, as specified by DEVRANDOM, until a sufficient amount 6771 of data has been collected. We spend at most 10 ms on each file 6772 (select timeout) and read in non-blocking mode. DEVRANDOM now 6773 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 6774 (previously it was just the string "/dev/urandom"), so on typical 6775 platforms the 10 ms delay will never occur. 6776 Also separate out the Unix variant to its own file, rand_unix.c. 6777 For VMS, there's a currently-empty rand_vms.c. 6778 [Richard Levitte] 6779 6780 *) Move OCSP client related routines to ocsp_cl.c. These 6781 provide utility functions which an application needing 6782 to issue a request to an OCSP responder and analyse the 6783 response will typically need: as opposed to those which an 6784 OCSP responder itself would need which will be added later. 6785 6786 OCSP_request_sign() signs an OCSP request with an API similar 6787 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 6788 response. OCSP_response_get1_basic() extracts basic response 6789 from response. OCSP_resp_find_status(): finds and extracts status 6790 information from an OCSP_CERTID structure (which will be created 6791 when the request structure is built). These are built from lower 6792 level functions which work on OCSP_SINGLERESP structures but 6793 wont normally be used unless the application wishes to examine 6794 extensions in the OCSP response for example. 6795 6796 Replace nonce routines with a pair of functions. 6797 OCSP_request_add1_nonce() adds a nonce value and optionally 6798 generates a random value. OCSP_check_nonce() checks the 6799 validity of the nonce in an OCSP response. 6800 [Steve Henson] 6801 6802 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 6803 This doesn't copy the supplied OCSP_CERTID and avoids the 6804 need to free up the newly created id. Change return type 6805 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 6806 This can then be used to add extensions to the request. 6807 Deleted OCSP_request_new(), since most of its functionality 6808 is now in OCSP_REQUEST_new() (and the case insensitive name 6809 clash) apart from the ability to set the request name which 6810 will be added elsewhere. 6811 [Steve Henson] 6812 6813 *) Update OCSP API. Remove obsolete extensions argument from 6814 various functions. Extensions are now handled using the new 6815 OCSP extension code. New simple OCSP HTTP function which 6816 can be used to send requests and parse the response. 6817 [Steve Henson] 6818 6819 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 6820 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 6821 uses the special reorder version of SET OF to sort the attributes 6822 and reorder them to match the encoded order. This resolves a long 6823 standing problem: a verify on a PKCS7 structure just after signing 6824 it used to fail because the attribute order did not match the 6825 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 6826 it uses the received order. This is necessary to tolerate some broken 6827 software that does not order SET OF. This is handled by encoding 6828 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 6829 to produce the required SET OF. 6830 [Steve Henson] 6831 6832 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 6833 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 6834 files to get correct declarations of the ASN.1 item variables. 6835 [Richard Levitte] 6836 6837 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 6838 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 6839 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 6840 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 6841 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 6842 ASN1_ITEM and no wrapper functions. 6843 [Steve Henson] 6844 6845 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 6846 replace the old function pointer based I/O routines. Change most of 6847 the *_d2i_bio() and *_d2i_fp() functions to use these. 6848 [Steve Henson] 6849 6850 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 6851 lines, recognice more "algorithms" that can be deselected, and make 6852 it complain about algorithm deselection that isn't recognised. 6853 [Richard Levitte] 6854 6855 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 6856 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 6857 to use new functions. Add NO_ASN1_OLD which can be set to remove 6858 some old style ASN1 functions: this can be used to determine if old 6859 code will still work when these eventually go away. 6860 [Steve Henson] 6861 6862 *) New extension functions for OCSP structures, these follow the 6863 same conventions as certificates and CRLs. 6864 [Steve Henson] 6865 6866 *) New function X509V3_add1_i2d(). This automatically encodes and 6867 adds an extension. Its behaviour can be customised with various 6868 flags to append, replace or delete. Various wrappers added for 6869 certifcates and CRLs. 6870 [Steve Henson] 6871 6872 *) Fix to avoid calling the underlying ASN1 print routine when 6873 an extension cannot be parsed. Correct a typo in the 6874 OCSP_SERVICELOC extension. Tidy up print OCSP format. 6875 [Steve Henson] 6876 6877 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate 6878 entries for variables. 6879 [Steve Henson] 6880 6881 *) Add functionality to apps/openssl.c for detecting locking 6882 problems: As the program is single-threaded, all we have 6883 to do is register a locking callback using an array for 6884 storing which locks are currently held by the program. 6885 [Bodo Moeller] 6886 6887 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 6888 SSL_get_ex_data_X509_STORE_idx(), which is used in 6889 ssl_verify_cert_chain() and thus can be called at any time 6890 during TLS/SSL handshakes so that thread-safety is essential. 6891 Unfortunately, the ex_data design is not at all suited 6892 for multi-threaded use, so it probably should be abolished. 6893 [Bodo Moeller] 6894 6895 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 6896 [Broadcom, tweaked and integrated by Geoff Thorpe] 6897 6898 *) Move common extension printing code to new function 6899 X509V3_print_extensions(). Reorganise OCSP print routines and 6900 implement some needed OCSP ASN1 functions. Add OCSP extensions. 6901 [Steve Henson] 6902 6903 *) New function X509_signature_print() to remove duplication in some 6904 print routines. 6905 [Steve Henson] 6906 6907 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 6908 set (this was treated exactly the same as SET OF previously). This 6909 is used to reorder the STACK representing the structure to match the 6910 encoding. This will be used to get round a problem where a PKCS7 6911 structure which was signed could not be verified because the STACK 6912 order did not reflect the encoded order. 6913 [Steve Henson] 6914 6915 *) Reimplement the OCSP ASN1 module using the new code. 6916 [Steve Henson] 6917 6918 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 6919 for its ASN1 operations. The old style function pointers still exist 6920 for now but they will eventually go away. 6921 [Steve Henson] 6922 6923 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 6924 completely replaces the old ASN1 functionality with a table driven 6925 encoder and decoder which interprets an ASN1_ITEM structure describing 6926 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 6927 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 6928 has also been converted to the new form. 6929 [Steve Henson] 6930 6931 *) Change BN_mod_exp_recp so that negative moduli are tolerated 6932 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 6933 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 6934 for negative moduli. 6935 [Bodo Moeller] 6936 6937 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 6938 of not touching the result's sign bit. 6939 [Bodo Moeller] 6940 6941 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 6942 set. 6943 [Bodo Moeller] 6944 6945 *) Changed the LHASH code to use prototypes for callbacks, and created 6946 macros to declare and implement thin (optionally static) functions 6947 that provide type-safety and avoid function pointer casting for the 6948 type-specific callbacks. 6949 [Geoff Thorpe] 6950 6951 *) Added Kerberos Cipher Suites to be used with TLS, as written in 6952 RFC 2712. 6953 [Veers Staats <staatsvr@asc.hpc.mil>, 6954 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 6955 6956 *) Reformat the FAQ so the different questions and answers can be divided 6957 in sections depending on the subject. 6958 [Richard Levitte] 6959 6960 *) Have the zlib compression code load ZLIB.DLL dynamically under 6961 Windows. 6962 [Richard Levitte] 6963 6964 *) New function BN_mod_sqrt for computing square roots modulo a prime 6965 (using the probabilistic Tonelli-Shanks algorithm unless 6966 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 6967 be handled deterministically). 6968 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 6969 6970 *) Make BN_mod_inverse faster by explicitly handling small quotients 6971 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 6972 512 bits], about 30% for larger ones [1024 or 2048 bits].) 6973 [Bodo Moeller] 6974 6975 *) New function BN_kronecker. 6976 [Bodo Moeller] 6977 6978 *) Fix BN_gcd so that it works on negative inputs; the result is 6979 positive unless both parameters are zero. 6980 Previously something reasonably close to an infinite loop was 6981 possible because numbers could be growing instead of shrinking 6982 in the implementation of Euclid's algorithm. 6983 [Bodo Moeller] 6984 6985 *) Fix BN_is_word() and BN_is_one() macros to take into account the 6986 sign of the number in question. 6987 6988 Fix BN_is_word(a,w) to work correctly for w == 0. 6989 6990 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 6991 because its test if the absolute value of 'a' equals 'w'. 6992 Note that BN_abs_is_word does *not* handle w == 0 reliably; 6993 it exists mostly for use in the implementations of BN_is_zero(), 6994 BN_is_one(), and BN_is_word(). 6995 [Bodo Moeller] 6996 6997 *) New function BN_swap. 6998 [Bodo Moeller] 6999 7000 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 7001 the exponentiation functions are more likely to produce reasonable 7002 results on negative inputs. 7003 [Bodo Moeller] 7004 7005 *) Change BN_mod_mul so that the result is always non-negative. 7006 Previously, it could be negative if one of the factors was negative; 7007 I don't think anyone really wanted that behaviour. 7008 [Bodo Moeller] 7009 7010 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 7011 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 7012 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 7013 and add new functions: 7014 7015 BN_nnmod 7016 BN_mod_sqr 7017 BN_mod_add 7018 BN_mod_add_quick 7019 BN_mod_sub 7020 BN_mod_sub_quick 7021 BN_mod_lshift1 7022 BN_mod_lshift1_quick 7023 BN_mod_lshift 7024 BN_mod_lshift_quick 7025 7026 These functions always generate non-negative results. 7027 7028 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 7029 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 7030 7031 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 7032 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 7033 be reduced modulo m. 7034 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 7035 7036#if 0 7037 The following entry accidentily appeared in the CHANGES file 7038 distributed with OpenSSL 0.9.7. The modifications described in 7039 it do *not* apply to OpenSSL 0.9.7. 7040 7041 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 7042 was actually never needed) and in BN_mul(). The removal in BN_mul() 7043 required a small change in bn_mul_part_recursive() and the addition 7044 of the functions bn_cmp_part_words(), bn_sub_part_words() and 7045 bn_add_part_words(), which do the same thing as bn_cmp_words(), 7046 bn_sub_words() and bn_add_words() except they take arrays with 7047 differing sizes. 7048 [Richard Levitte] 7049#endif 7050 7051 *) In 'openssl passwd', verify passwords read from the terminal 7052 unless the '-salt' option is used (which usually means that 7053 verification would just waste user's time since the resulting 7054 hash is going to be compared with some given password hash) 7055 or the new '-noverify' option is used. 7056 7057 This is an incompatible change, but it does not affect 7058 non-interactive use of 'openssl passwd' (passwords on the command 7059 line, '-stdin' option, '-in ...' option) and thus should not 7060 cause any problems. 7061 [Bodo Moeller] 7062 7063 *) Remove all references to RSAref, since there's no more need for it. 7064 [Richard Levitte] 7065 7066 *) Make DSO load along a path given through an environment variable 7067 (SHLIB_PATH) with shl_load(). 7068 [Richard Levitte] 7069 7070 *) Constify the ENGINE code as a result of BIGNUM constification. 7071 Also constify the RSA code and most things related to it. In a 7072 few places, most notable in the depth of the ASN.1 code, ugly 7073 casts back to non-const were required (to be solved at a later 7074 time) 7075 [Richard Levitte] 7076 7077 *) Make it so the openssl application has all engines loaded by default. 7078 [Richard Levitte] 7079 7080 *) Constify the BIGNUM routines a little more. 7081 [Richard Levitte] 7082 7083 *) Add the following functions: 7084 7085 ENGINE_load_cswift() 7086 ENGINE_load_chil() 7087 ENGINE_load_atalla() 7088 ENGINE_load_nuron() 7089 ENGINE_load_builtin_engines() 7090 7091 That way, an application can itself choose if external engines that 7092 are built-in in OpenSSL shall ever be used or not. The benefit is 7093 that applications won't have to be linked with libdl or other dso 7094 libraries unless it's really needed. 7095 7096 Changed 'openssl engine' to load all engines on demand. 7097 Changed the engine header files to avoid the duplication of some 7098 declarations (they differed!). 7099 [Richard Levitte] 7100 7101 *) 'openssl engine' can now list capabilities. 7102 [Richard Levitte] 7103 7104 *) Better error reporting in 'openssl engine'. 7105 [Richard Levitte] 7106 7107 *) Never call load_dh_param(NULL) in s_server. 7108 [Bodo Moeller] 7109 7110 *) Add engine application. It can currently list engines by name and 7111 identity, and test if they are actually available. 7112 [Richard Levitte] 7113 7114 *) Improve RPM specification file by forcing symbolic linking and making 7115 sure the installed documentation is also owned by root.root. 7116 [Damien Miller <djm@mindrot.org>] 7117 7118 *) Give the OpenSSL applications more possibilities to make use of 7119 keys (public as well as private) handled by engines. 7120 [Richard Levitte] 7121 7122 *) Add OCSP code that comes from CertCo. 7123 [Richard Levitte] 7124 7125 *) Add VMS support for the Rijndael code. 7126 [Richard Levitte] 7127 7128 *) Added untested support for Nuron crypto accelerator. 7129 [Ben Laurie] 7130 7131 *) Add support for external cryptographic devices. This code was 7132 previously distributed separately as the "engine" branch. 7133 [Geoff Thorpe, Richard Levitte] 7134 7135 *) Rework the filename-translation in the DSO code. It is now possible to 7136 have far greater control over how a "name" is turned into a filename 7137 depending on the operating environment and any oddities about the 7138 different shared library filenames on each system. 7139 [Geoff Thorpe] 7140 7141 *) Support threads on FreeBSD-elf in Configure. 7142 [Richard Levitte] 7143 7144 *) Fix for SHA1 assembly problem with MASM: it produces 7145 warnings about corrupt line number information when assembling 7146 with debugging information. This is caused by the overlapping 7147 of two sections. 7148 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 7149 7150 *) NCONF changes. 7151 NCONF_get_number() has no error checking at all. As a replacement, 7152 NCONF_get_number_e() is defined (_e for "error checking") and is 7153 promoted strongly. The old NCONF_get_number is kept around for 7154 binary backward compatibility. 7155 Make it possible for methods to load from something other than a BIO, 7156 by providing a function pointer that is given a name instead of a BIO. 7157 For example, this could be used to load configuration data from an 7158 LDAP server. 7159 [Richard Levitte] 7160 7161 *) Fix for non blocking accept BIOs. Added new I/O special reason 7162 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 7163 with non blocking I/O was not possible because no retry code was 7164 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 7165 this case. 7166 [Steve Henson] 7167 7168 *) Added the beginnings of Rijndael support. 7169 [Ben Laurie] 7170 7171 *) Fix for bug in DirectoryString mask setting. Add support for 7172 X509_NAME_print_ex() in 'req' and X509_print_ex() function 7173 to allow certificate printing to more controllable, additional 7174 'certopt' option to 'x509' to allow new printing options to be 7175 set. 7176 [Steve Henson] 7177 7178 *) Clean old EAY MD5 hack from e_os.h. 7179 [Richard Levitte] 7180 7181 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 7182 7183 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 7184 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 7185 [Joe Orton, Steve Henson] 7186 7187 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 7188 7189 *) Fix additional bug revealed by the NISCC test suite: 7190 7191 Stop bug triggering large recursion when presented with 7192 certain ASN.1 tags (CVE-2003-0851) 7193 [Steve Henson] 7194 7195 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 7196 7197 *) Fix various bugs revealed by running the NISCC test suite: 7198 7199 Stop out of bounds reads in the ASN1 code when presented with 7200 invalid tags (CVE-2003-0543 and CVE-2003-0544). 7201 7202 If verify callback ignores invalid public key errors don't try to check 7203 certificate signature with the NULL public key. 7204 7205 [Steve Henson] 7206 7207 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 7208 if the server requested one: as stated in TLS 1.0 and SSL 3.0 7209 specifications. 7210 [Steve Henson] 7211 7212 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 7213 extra data after the compression methods not only for TLS 1.0 7214 but also for SSL 3.0 (as required by the specification). 7215 [Bodo Moeller; problem pointed out by Matthias Loepfe] 7216 7217 *) Change X509_certificate_type() to mark the key as exported/exportable 7218 when it's 512 *bits* long, not 512 bytes. 7219 [Richard Levitte] 7220 7221 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 7222 7223 *) Countermeasure against the Klima-Pokorny-Rosa extension of 7224 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 7225 a protocol version number mismatch like a decryption error 7226 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 7227 [Bodo Moeller] 7228 7229 *) Turn on RSA blinding by default in the default implementation 7230 to avoid a timing attack. Applications that don't want it can call 7231 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 7232 They would be ill-advised to do so in most cases. 7233 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 7234 7235 *) Change RSA blinding code so that it works when the PRNG is not 7236 seeded (in this case, the secret RSA exponent is abused as 7237 an unpredictable seed -- if it is not unpredictable, there 7238 is no point in blinding anyway). Make RSA blinding thread-safe 7239 by remembering the creator's thread ID in rsa->blinding and 7240 having all other threads use local one-time blinding factors 7241 (this requires more computation than sharing rsa->blinding, but 7242 avoids excessive locking; and if an RSA object is not shared 7243 between threads, blinding will still be very fast). 7244 [Bodo Moeller] 7245 7246 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 7247 7248 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 7249 via timing by performing a MAC computation even if incorrrect 7250 block cipher padding has been found. This is a countermeasure 7251 against active attacks where the attacker has to distinguish 7252 between bad padding and a MAC verification error. (CVE-2003-0078) 7253 7254 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 7255 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 7256 Martin Vuagnoux (EPFL, Ilion)] 7257 7258 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 7259 7260 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 7261 memory from it's contents. This is done with a counter that will 7262 place alternating values in each byte. This can be used to solve 7263 two issues: 1) the removal of calls to memset() by highly optimizing 7264 compilers, and 2) cleansing with other values than 0, since those can 7265 be read through on certain media, for example a swap space on disk. 7266 [Geoff Thorpe] 7267 7268 *) Bugfix: client side session caching did not work with external caching, 7269 because the session->cipher setting was not restored when reloading 7270 from the external cache. This problem was masked, when 7271 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 7272 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 7273 [Lutz Jaenicke] 7274 7275 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 7276 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 7277 [Zeev Lieber <zeev-l@yahoo.com>] 7278 7279 *) Undo an undocumented change introduced in 0.9.6e which caused 7280 repeated calls to OpenSSL_add_all_ciphers() and 7281 OpenSSL_add_all_digests() to be ignored, even after calling 7282 EVP_cleanup(). 7283 [Richard Levitte] 7284 7285 *) Change the default configuration reader to deal with last line not 7286 being properly terminated. 7287 [Richard Levitte] 7288 7289 *) Change X509_NAME_cmp() so it applies the special rules on handling 7290 DN values that are of type PrintableString, as well as RDNs of type 7291 emailAddress where the value has the type ia5String. 7292 [stefank@valicert.com via Richard Levitte] 7293 7294 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 7295 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 7296 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 7297 the bitwise-OR of the two for use by the majority of applications 7298 wanting this behaviour, and update the docs. The documented 7299 behaviour and actual behaviour were inconsistent and had been 7300 changing anyway, so this is more a bug-fix than a behavioural 7301 change. 7302 [Geoff Thorpe, diagnosed by Nadav Har'El] 7303 7304 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 7305 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 7306 [Bodo Moeller] 7307 7308 *) Fix initialization code race conditions in 7309 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 7310 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 7311 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 7312 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 7313 ssl2_get_cipher_by_char(), 7314 ssl3_get_cipher_by_char(). 7315 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 7316 7317 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 7318 the cached sessions are flushed, as the remove_cb() might use ex_data 7319 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 7320 (see [openssl.org #212]). 7321 [Geoff Thorpe, Lutz Jaenicke] 7322 7323 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 7324 length, instead of the encoding length to d2i_ASN1_OBJECT. 7325 [Steve Henson] 7326 7327 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 7328 7329 *) [In 0.9.6g-engine release:] 7330 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 7331 [Lynn Gazis <lgazis@rainbow.com>] 7332 7333 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 7334 7335 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 7336 and get fix the header length calculation. 7337 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 7338 Alon Kantor <alonk@checkpoint.com> (and others), 7339 Steve Henson] 7340 7341 *) Use proper error handling instead of 'assertions' in buffer 7342 overflow checks added in 0.9.6e. This prevents DoS (the 7343 assertions could call abort()). 7344 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 7345 7346 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 7347 7348 *) Add various sanity checks to asn1_get_length() to reject 7349 the ASN1 length bytes if they exceed sizeof(long), will appear 7350 negative or the content length exceeds the length of the 7351 supplied buffer. 7352 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7353 7354 *) Fix cipher selection routines: ciphers without encryption had no flags 7355 for the cipher strength set and where therefore not handled correctly 7356 by the selection routines (PR #130). 7357 [Lutz Jaenicke] 7358 7359 *) Fix EVP_dsa_sha macro. 7360 [Nils Larsch] 7361 7362 *) New option 7363 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 7364 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 7365 that was added in OpenSSL 0.9.6d. 7366 7367 As the countermeasure turned out to be incompatible with some 7368 broken SSL implementations, the new option is part of SSL_OP_ALL. 7369 SSL_OP_ALL is usually employed when compatibility with weird SSL 7370 implementations is desired (e.g. '-bugs' option to 's_client' and 7371 's_server'), so the new option is automatically set in many 7372 applications. 7373 [Bodo Moeller] 7374 7375 *) Changes in security patch: 7376 7377 Changes marked "(CHATS)" were sponsored by the Defense Advanced 7378 Research Projects Agency (DARPA) and Air Force Research Laboratory, 7379 Air Force Materiel Command, USAF, under agreement number 7380 F30602-01-2-0537. 7381 7382 *) Add various sanity checks to asn1_get_length() to reject 7383 the ASN1 length bytes if they exceed sizeof(long), will appear 7384 negative or the content length exceeds the length of the 7385 supplied buffer. (CVE-2002-0659) 7386 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 7387 7388 *) Assertions for various potential buffer overflows, not known to 7389 happen in practice. 7390 [Ben Laurie (CHATS)] 7391 7392 *) Various temporary buffers to hold ASCII versions of integers were 7393 too small for 64 bit platforms. (CVE-2002-0655) 7394 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 7395 7396 *) Remote buffer overflow in SSL3 protocol - an attacker could 7397 supply an oversized session ID to a client. (CVE-2002-0656) 7398 [Ben Laurie (CHATS)] 7399 7400 *) Remote buffer overflow in SSL2 protocol - an attacker could 7401 supply an oversized client master key. (CVE-2002-0656) 7402 [Ben Laurie (CHATS)] 7403 7404 Changes between 0.9.6c and 0.9.6d [9 May 2002] 7405 7406 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 7407 encoded as NULL) with id-dsa-with-sha1. 7408 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 7409 7410 *) Check various X509_...() return values in apps/req.c. 7411 [Nils Larsch <nla@trustcenter.de>] 7412 7413 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 7414 an end-of-file condition would erronously be flagged, when the CRLF 7415 was just at the end of a processed block. The bug was discovered when 7416 processing data through a buffering memory BIO handing the data to a 7417 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 7418 <ptsekov@syntrex.com> and Nedelcho Stanev. 7419 [Lutz Jaenicke] 7420 7421 *) Implement a countermeasure against a vulnerability recently found 7422 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 7423 before application data chunks to avoid the use of known IVs 7424 with data potentially chosen by the attacker. 7425 [Bodo Moeller] 7426 7427 *) Fix length checks in ssl3_get_client_hello(). 7428 [Bodo Moeller] 7429 7430 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 7431 to prevent ssl3_read_internal() from incorrectly assuming that 7432 ssl3_read_bytes() found application data while handshake 7433 processing was enabled when in fact s->s3->in_read_app_data was 7434 merely automatically cleared during the initial handshake. 7435 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 7436 7437 *) Fix object definitions for Private and Enterprise: they were not 7438 recognized in their shortname (=lowercase) representation. Extend 7439 obj_dat.pl to issue an error when using undefined keywords instead 7440 of silently ignoring the problem (Svenning Sorensen 7441 <sss@sss.dnsalias.net>). 7442 [Lutz Jaenicke] 7443 7444 *) Fix DH_generate_parameters() so that it works for 'non-standard' 7445 generators, i.e. generators other than 2 and 5. (Previously, the 7446 code did not properly initialise the 'add' and 'rem' values to 7447 BN_generate_prime().) 7448 7449 In the new general case, we do not insist that 'generator' is 7450 actually a primitive root: This requirement is rather pointless; 7451 a generator of the order-q subgroup is just as good, if not 7452 better. 7453 [Bodo Moeller] 7454 7455 *) Map new X509 verification errors to alerts. Discovered and submitted by 7456 Tom Wu <tom@arcot.com>. 7457 [Lutz Jaenicke] 7458 7459 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 7460 returning non-zero before the data has been completely received 7461 when using non-blocking I/O. 7462 [Bodo Moeller; problem pointed out by John Hughes] 7463 7464 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 7465 [Ben Laurie, Lutz Jaenicke] 7466 7467 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 7468 Yoram Zahavi <YoramZ@gilian.com>). 7469 [Lutz Jaenicke] 7470 7471 *) Add information about CygWin 1.3 and on, and preserve proper 7472 configuration for the versions before that. 7473 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 7474 7475 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 7476 check whether we deal with a copy of a session and do not delete from 7477 the cache in this case. Problem reported by "Izhar Shoshani Levi" 7478 <izhar@checkpoint.com>. 7479 [Lutz Jaenicke] 7480 7481 *) Do not store session data into the internal session cache, if it 7482 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 7483 flag is set). Proposed by Aslam <aslam@funk.com>. 7484 [Lutz Jaenicke] 7485 7486 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 7487 value is 0. 7488 [Richard Levitte] 7489 7490 *) [In 0.9.6d-engine release:] 7491 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 7492 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7493 7494 *) Add the configuration target linux-s390x. 7495 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 7496 7497 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 7498 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 7499 variable as an indication that a ClientHello message has been 7500 received. As the flag value will be lost between multiple 7501 invocations of ssl3_accept when using non-blocking I/O, the 7502 function may not be aware that a handshake has actually taken 7503 place, thus preventing a new session from being added to the 7504 session cache. 7505 7506 To avoid this problem, we now set s->new_session to 2 instead of 7507 using a local variable. 7508 [Lutz Jaenicke, Bodo Moeller] 7509 7510 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 7511 if the SSL_R_LENGTH_MISMATCH error is detected. 7512 [Geoff Thorpe, Bodo Moeller] 7513 7514 *) New 'shared_ldflag' column in Configure platform table. 7515 [Richard Levitte] 7516 7517 *) Fix EVP_CIPHER_mode macro. 7518 ["Dan S. Camper" <dan@bti.net>] 7519 7520 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 7521 type, we must throw them away by setting rr->length to 0. 7522 [D P Chang <dpc@qualys.com>] 7523 7524 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 7525 7526 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 7527 <Dominikus.Scherkl@biodata.com>. (The previous implementation 7528 worked incorrectly for those cases where range = 10..._2 and 7529 3*range is two bits longer than range.) 7530 [Bodo Moeller] 7531 7532 *) Only add signing time to PKCS7 structures if it is not already 7533 present. 7534 [Steve Henson] 7535 7536 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 7537 OBJ_ld_ce should be OBJ_id_ce. 7538 Also some ip-pda OIDs in crypto/objects/objects.txt were 7539 incorrect (cf. RFC 3039). 7540 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 7541 7542 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 7543 returns early because it has nothing to do. 7544 [Andy Schneider <andy.schneider@bjss.co.uk>] 7545 7546 *) [In 0.9.6c-engine release:] 7547 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 7548 [Andy Schneider <andy.schneider@bjss.co.uk>] 7549 7550 *) [In 0.9.6c-engine release:] 7551 Add support for Cryptographic Appliance's keyserver technology. 7552 (Use engine 'keyclient') 7553 [Cryptographic Appliances and Geoff Thorpe] 7554 7555 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 7556 is called via tools/c89.sh because arguments have to be 7557 rearranged (all '-L' options must appear before the first object 7558 modules). 7559 [Richard Shapiro <rshapiro@abinitio.com>] 7560 7561 *) [In 0.9.6c-engine release:] 7562 Add support for Broadcom crypto accelerator cards, backported 7563 from 0.9.7. 7564 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 7565 7566 *) [In 0.9.6c-engine release:] 7567 Add support for SureWare crypto accelerator cards from 7568 Baltimore Technologies. (Use engine 'sureware') 7569 [Baltimore Technologies and Mark Cox] 7570 7571 *) [In 0.9.6c-engine release:] 7572 Add support for crypto accelerator cards from Accelerated 7573 Encryption Processing, www.aep.ie. (Use engine 'aep') 7574 [AEP Inc. and Mark Cox] 7575 7576 *) Add a configuration entry for gcc on UnixWare. 7577 [Gary Benson <gbenson@redhat.com>] 7578 7579 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 7580 messages are stored in a single piece (fixed-length part and 7581 variable-length part combined) and fix various bugs found on the way. 7582 [Bodo Moeller] 7583 7584 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 7585 instead. BIO_gethostbyname() does not know what timeouts are 7586 appropriate, so entries would stay in cache even when they have 7587 become invalid. 7588 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 7589 7590 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 7591 faced with a pathologically small ClientHello fragment that does 7592 not contain client_version: Instead of aborting with an error, 7593 simply choose the highest available protocol version (i.e., 7594 TLS 1.0 unless it is disabled). In practice, ClientHello 7595 messages are never sent like this, but this change gives us 7596 strictly correct behaviour at least for TLS. 7597 [Bodo Moeller] 7598 7599 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 7600 never resets s->method to s->ctx->method when called from within 7601 one of the SSL handshake functions. 7602 [Bodo Moeller; problem pointed out by Niko Baric] 7603 7604 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 7605 (sent using the client's version number) if client_version is 7606 smaller than the protocol version in use. Also change 7607 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 7608 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 7609 the client will at least see that alert. 7610 [Bodo Moeller] 7611 7612 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 7613 correctly. 7614 [Bodo Moeller] 7615 7616 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 7617 client receives HelloRequest while in a handshake. 7618 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 7619 7620 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 7621 should end in 'break', not 'goto end' which circuments various 7622 cleanups done in state SSL_ST_OK. But session related stuff 7623 must be disabled for SSL_ST_OK in the case that we just sent a 7624 HelloRequest. 7625 7626 Also avoid some overhead by not calling ssl_init_wbio_buffer() 7627 before just sending a HelloRequest. 7628 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 7629 7630 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 7631 reveal whether illegal block cipher padding was found or a MAC 7632 verification error occured. (Neither SSLerr() codes nor alerts 7633 are directly visible to potential attackers, but the information 7634 may leak via logfiles.) 7635 7636 Similar changes are not required for the SSL 2.0 implementation 7637 because the number of padding bytes is sent in clear for SSL 2.0, 7638 and the extra bytes are just ignored. However ssl/s2_pkt.c 7639 failed to verify that the purported number of padding bytes is in 7640 the legal range. 7641 [Bodo Moeller] 7642 7643 *) Add OpenUNIX-8 support including shared libraries 7644 (Boyd Lynn Gerber <gerberb@zenez.com>). 7645 [Lutz Jaenicke] 7646 7647 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 7648 'wristwatch attack' using huge encoding parameters (cf. 7649 James H. Manger's CRYPTO 2001 paper). Note that the 7650 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 7651 encoding parameters and hence was not vulnerable. 7652 [Bodo Moeller] 7653 7654 *) BN_sqr() bug fix. 7655 [Ulf M��ller, reported by Jim Ellis <jim.ellis@cavium.com>] 7656 7657 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 7658 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 7659 followed by modular reduction. 7660 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 7661 7662 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 7663 equivalent based on BN_pseudo_rand() instead of BN_rand(). 7664 [Bodo Moeller] 7665 7666 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 7667 This function was broken, as the check for a new client hello message 7668 to handle SGC did not allow these large messages. 7669 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 7670 [Lutz Jaenicke] 7671 7672 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 7673 [Lutz Jaenicke] 7674 7675 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 7676 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 7677 [Lutz Jaenicke] 7678 7679 *) Rework the configuration and shared library support for Tru64 Unix. 7680 The configuration part makes use of modern compiler features and 7681 still retains old compiler behavior for those that run older versions 7682 of the OS. The shared library support part includes a variant that 7683 uses the RPATH feature, and is available through the special 7684 configuration target "alpha-cc-rpath", which will never be selected 7685 automatically. 7686 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 7687 7688 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 7689 with the same message size as in ssl3_get_certificate_request(). 7690 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 7691 messages might inadvertently be reject as too long. 7692 [Petr Lampa <lampa@fee.vutbr.cz>] 7693 7694 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 7695 [Andy Polyakov] 7696 7697 *) Modified SSL library such that the verify_callback that has been set 7698 specificly for an SSL object with SSL_set_verify() is actually being 7699 used. Before the change, a verify_callback set with this function was 7700 ignored and the verify_callback() set in the SSL_CTX at the time of 7701 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 7702 to allow the necessary settings. 7703 [Lutz Jaenicke] 7704 7705 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 7706 explicitly to NULL, as at least on Solaris 8 this seems not always to be 7707 done automatically (in contradiction to the requirements of the C 7708 standard). This made problems when used from OpenSSH. 7709 [Lutz Jaenicke] 7710 7711 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 7712 dh->length and always used 7713 7714 BN_rand_range(priv_key, dh->p). 7715 7716 BN_rand_range() is not necessary for Diffie-Hellman, and this 7717 specific range makes Diffie-Hellman unnecessarily inefficient if 7718 dh->length (recommended exponent length) is much smaller than the 7719 length of dh->p. We could use BN_rand_range() if the order of 7720 the subgroup was stored in the DH structure, but we only have 7721 dh->length. 7722 7723 So switch back to 7724 7725 BN_rand(priv_key, l, ...) 7726 7727 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 7728 otherwise. 7729 [Bodo Moeller] 7730 7731 *) In 7732 7733 RSA_eay_public_encrypt 7734 RSA_eay_private_decrypt 7735 RSA_eay_private_encrypt (signing) 7736 RSA_eay_public_decrypt (signature verification) 7737 7738 (default implementations for RSA_public_encrypt, 7739 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 7740 always reject numbers >= n. 7741 [Bodo Moeller] 7742 7743 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 7744 to synchronize access to 'locking_thread'. This is necessary on 7745 systems where access to 'locking_thread' (an 'unsigned long' 7746 variable) is not atomic. 7747 [Bodo Moeller] 7748 7749 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 7750 *before* setting the 'crypto_lock_rand' flag. The previous code had 7751 a race condition if 0 is a valid thread ID. 7752 [Travis Vitek <vitek@roguewave.com>] 7753 7754 *) Add support for shared libraries under Irix. 7755 [Albert Chin-A-Young <china@thewrittenword.com>] 7756 7757 *) Add configuration option to build on Linux on both big-endian and 7758 little-endian MIPS. 7759 [Ralf Baechle <ralf@uni-koblenz.de>] 7760 7761 *) Add the possibility to create shared libraries on HP-UX. 7762 [Richard Levitte] 7763 7764 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 7765 7766 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 7767 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 7768 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 7769 PRNG state recovery was possible based on the output of 7770 one PRNG request appropriately sized to gain knowledge on 7771 'md' followed by enough consecutive 1-byte PRNG requests 7772 to traverse all of 'state'. 7773 7774 1. When updating 'md_local' (the current thread's copy of 'md') 7775 during PRNG output generation, hash all of the previous 7776 'md_local' value, not just the half used for PRNG output. 7777 7778 2. Make the number of bytes from 'state' included into the hash 7779 independent from the number of PRNG bytes requested. 7780 7781 The first measure alone would be sufficient to avoid 7782 Markku-Juhani's attack. (Actually it had never occurred 7783 to me that the half of 'md_local' used for chaining was the 7784 half from which PRNG output bytes were taken -- I had always 7785 assumed that the secret half would be used.) The second 7786 measure makes sure that additional data from 'state' is never 7787 mixed into 'md_local' in small portions; this heuristically 7788 further strengthens the PRNG. 7789 [Bodo Moeller] 7790 7791 *) Fix crypto/bn/asm/mips3.s. 7792 [Andy Polyakov] 7793 7794 *) When only the key is given to "enc", the IV is undefined. Print out 7795 an error message in this case. 7796 [Lutz Jaenicke] 7797 7798 *) Handle special case when X509_NAME is empty in X509 printing routines. 7799 [Steve Henson] 7800 7801 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 7802 positive and less than q. 7803 [Bodo Moeller] 7804 7805 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 7806 used: it isn't thread safe and the add_lock_callback should handle 7807 that itself. 7808 [Paul Rose <Paul.Rose@bridge.com>] 7809 7810 *) Verify that incoming data obeys the block size in 7811 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 7812 [Bodo Moeller] 7813 7814 *) Fix OAEP check. 7815 [Ulf M��ller, Bodo M��ller] 7816 7817 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 7818 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 7819 when fixing the server behaviour for backwards-compatible 'client 7820 hello' messages. (Note that the attack is impractical against 7821 SSL 3.0 and TLS 1.0 anyway because length and version checking 7822 means that the probability of guessing a valid ciphertext is 7823 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 7824 paper.) 7825 7826 Before 0.9.5, the countermeasure (hide the error by generating a 7827 random 'decryption result') did not work properly because 7828 ERR_clear_error() was missing, meaning that SSL_get_error() would 7829 detect the supposedly ignored error. 7830 7831 Both problems are now fixed. 7832 [Bodo Moeller] 7833 7834 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 7835 (previously it was 1024). 7836 [Bodo Moeller] 7837 7838 *) Fix for compatibility mode trust settings: ignore trust settings 7839 unless some valid trust or reject settings are present. 7840 [Steve Henson] 7841 7842 *) Fix for blowfish EVP: its a variable length cipher. 7843 [Steve Henson] 7844 7845 *) Fix various bugs related to DSA S/MIME verification. Handle missing 7846 parameters in DSA public key structures and return an error in the 7847 DSA routines if parameters are absent. 7848 [Steve Henson] 7849 7850 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 7851 in the current directory if neither $RANDFILE nor $HOME was set. 7852 RAND_file_name() in 0.9.6a returned NULL in this case. This has 7853 caused some confusion to Windows users who haven't defined $HOME. 7854 Thus RAND_file_name() is changed again: e_os.h can define a 7855 DEFAULT_HOME, which will be used if $HOME is not set. 7856 For Windows, we use "C:"; on other platforms, we still require 7857 environment variables. 7858 7859 *) Move 'if (!initialized) RAND_poll()' into regions protected by 7860 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 7861 having multiple threads call RAND_poll() concurrently. 7862 [Bodo Moeller] 7863 7864 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 7865 combination of a flag and a thread ID variable. 7866 Otherwise while one thread is in ssleay_rand_bytes (which sets the 7867 flag), *other* threads can enter ssleay_add_bytes without obeying 7868 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 7869 that they do not hold after the first thread unsets add_do_not_lock). 7870 [Bodo Moeller] 7871 7872 *) Change bctest again: '-x' expressions are not available in all 7873 versions of 'test'. 7874 [Bodo Moeller] 7875 7876 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 7877 7878 *) Fix a couple of memory leaks in PKCS7_dataDecode() 7879 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 7880 7881 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 7882 the default extension for executables, if any. Also, make the perl 7883 scripts that use symlink() to test if it really exists and use "cp" 7884 if it doesn't. All this made OpenSSL compilable and installable in 7885 CygWin. 7886 [Richard Levitte] 7887 7888 *) Fix for asn1_GetSequence() for indefinite length constructed data. 7889 If SEQUENCE is length is indefinite just set c->slen to the total 7890 amount of data available. 7891 [Steve Henson, reported by shige@FreeBSD.org] 7892 [This change does not apply to 0.9.7.] 7893 7894 *) Change bctest to avoid here-documents inside command substitution 7895 (workaround for FreeBSD /bin/sh bug). 7896 For compatibility with Ultrix, avoid shell functions (introduced 7897 in the bctest version that searches along $PATH). 7898 [Bodo Moeller] 7899 7900 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 7901 with des_encrypt() defined on some operating systems, like Solaris 7902 and UnixWare. 7903 [Richard Levitte] 7904 7905 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 7906 On the Importance of Eliminating Errors in Cryptographic 7907 Computations, J. Cryptology 14 (2001) 2, 101-119, 7908 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 7909 [Ulf Moeller] 7910 7911 *) MIPS assembler BIGNUM division bug fix. 7912 [Andy Polyakov] 7913 7914 *) Disabled incorrect Alpha assembler code. 7915 [Richard Levitte] 7916 7917 *) Fix PKCS#7 decode routines so they correctly update the length 7918 after reading an EOC for the EXPLICIT tag. 7919 [Steve Henson] 7920 [This change does not apply to 0.9.7.] 7921 7922 *) Fix bug in PKCS#12 key generation routines. This was triggered 7923 if a 3DES key was generated with a 0 initial byte. Include 7924 PKCS12_BROKEN_KEYGEN compilation option to retain the old 7925 (but broken) behaviour. 7926 [Steve Henson] 7927 7928 *) Enhance bctest to search for a working bc along $PATH and print 7929 it when found. 7930 [Tim Rice <tim@multitalents.net> via Richard Levitte] 7931 7932 *) Fix memory leaks in err.c: free err_data string if necessary; 7933 don't write to the wrong index in ERR_set_error_data. 7934 [Bodo Moeller] 7935 7936 *) Implement ssl23_peek (analogous to ssl23_read), which previously 7937 did not exist. 7938 [Bodo Moeller] 7939 7940 *) Replace rdtsc with _emit statements for VC++ version 5. 7941 [Jeremy Cooper <jeremy@baymoo.org>] 7942 7943 *) Make it possible to reuse SSLv2 sessions. 7944 [Richard Levitte] 7945 7946 *) In copy_email() check for >= 0 as a return value for 7947 X509_NAME_get_index_by_NID() since 0 is a valid index. 7948 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 7949 7950 *) Avoid coredump with unsupported or invalid public keys by checking if 7951 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 7952 PKCS7_verify() fails with non detached data. 7953 [Steve Henson] 7954 7955 *) Don't use getenv in library functions when run as setuid/setgid. 7956 New function OPENSSL_issetugid(). 7957 [Ulf Moeller] 7958 7959 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 7960 due to incorrect handling of multi-threading: 7961 7962 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 7963 7964 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 7965 7966 3. Count how many times MemCheck_off() has been called so that 7967 nested use can be treated correctly. This also avoids 7968 inband-signalling in the previous code (which relied on the 7969 assumption that thread ID 0 is impossible). 7970 [Bodo Moeller] 7971 7972 *) Add "-rand" option also to s_client and s_server. 7973 [Lutz Jaenicke] 7974 7975 *) Fix CPU detection on Irix 6.x. 7976 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 7977 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 7978 7979 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 7980 was empty. 7981 [Steve Henson] 7982 [This change does not apply to 0.9.7.] 7983 7984 *) Use the cached encoding of an X509_NAME structure rather than 7985 copying it. This is apparently the reason for the libsafe "errors" 7986 but the code is actually correct. 7987 [Steve Henson] 7988 7989 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 7990 Bleichenbacher's DSA attack. 7991 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 7992 to be set and top=0 forces the highest bit to be set; top=-1 is new 7993 and leaves the highest bit random. 7994 [Ulf Moeller, Bodo Moeller] 7995 7996 *) In the NCONF_...-based implementations for CONF_... queries 7997 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 7998 a temporary CONF structure with the data component set to NULL 7999 (which gives segmentation faults in lh_retrieve). 8000 Instead, use NULL for the CONF pointer in CONF_get_string and 8001 CONF_get_number (which may use environment variables) and directly 8002 return NULL from CONF_get_section. 8003 [Bodo Moeller] 8004 8005 *) Fix potential buffer overrun for EBCDIC. 8006 [Ulf Moeller] 8007 8008 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 8009 keyUsage if basicConstraints absent for a CA. 8010 [Steve Henson] 8011 8012 *) Make SMIME_write_PKCS7() write mail header values with a format that 8013 is more generally accepted (no spaces before the semicolon), since 8014 some programs can't parse those values properly otherwise. Also make 8015 sure BIO's that break lines after each write do not create invalid 8016 headers. 8017 [Richard Levitte] 8018 8019 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 8020 macros previously used would not encode an empty SEQUENCE OF 8021 and break the signature. 8022 [Steve Henson] 8023 [This change does not apply to 0.9.7.] 8024 8025 *) Zero the premaster secret after deriving the master secret in 8026 DH ciphersuites. 8027 [Steve Henson] 8028 8029 *) Add some EVP_add_digest_alias registrations (as found in 8030 OpenSSL_add_all_digests()) to SSL_library_init() 8031 aka OpenSSL_add_ssl_algorithms(). This provides improved 8032 compatibility with peers using X.509 certificates 8033 with unconventional AlgorithmIdentifier OIDs. 8034 [Bodo Moeller] 8035 8036 *) Fix for Irix with NO_ASM. 8037 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 8038 8039 *) ./config script fixes. 8040 [Ulf Moeller, Richard Levitte] 8041 8042 *) Fix 'openssl passwd -1'. 8043 [Bodo Moeller] 8044 8045 *) Change PKCS12_key_gen_asc() so it can cope with non null 8046 terminated strings whose length is passed in the passlen 8047 parameter, for example from PEM callbacks. This was done 8048 by adding an extra length parameter to asc2uni(). 8049 [Steve Henson, reported by <oddissey@samsung.co.kr>] 8050 8051 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 8052 call failed, free the DSA structure. 8053 [Bodo Moeller] 8054 8055 *) Fix to uni2asc() to cope with zero length Unicode strings. 8056 These are present in some PKCS#12 files. 8057 [Steve Henson] 8058 8059 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 8060 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 8061 when writing a 32767 byte record. 8062 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 8063 8064 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 8065 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 8066 8067 (RSA objects have a reference count access to which is protected 8068 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 8069 so they are meant to be shared between threads.) 8070 [Bodo Moeller, Geoff Thorpe; original patch submitted by 8071 "Reddie, Steven" <Steven.Reddie@ca.com>] 8072 8073 *) Fix a deadlock in CRYPTO_mem_leaks(). 8074 [Bodo Moeller] 8075 8076 *) Use better test patterns in bntest. 8077 [Ulf M��ller] 8078 8079 *) rand_win.c fix for Borland C. 8080 [Ulf M��ller] 8081 8082 *) BN_rshift bugfix for n == 0. 8083 [Bodo Moeller] 8084 8085 *) Add a 'bctest' script that checks for some known 'bc' bugs 8086 so that 'make test' does not abort just because 'bc' is broken. 8087 [Bodo Moeller] 8088 8089 *) Store verify_result within SSL_SESSION also for client side to 8090 avoid potential security hole. (Re-used sessions on the client side 8091 always resulted in verify_result==X509_V_OK, not using the original 8092 result of the server certificate verification.) 8093 [Lutz Jaenicke] 8094 8095 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 8096 SSL3_RT_APPLICATION_DATA, return 0. 8097 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 8098 [Bodo Moeller] 8099 8100 *) Fix SSL_peek: 8101 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 8102 releases, have been re-implemented by renaming the previous 8103 implementations of ssl2_read and ssl3_read to ssl2_read_internal 8104 and ssl3_read_internal, respectively, and adding 'peek' parameters 8105 to them. The new ssl[23]_{read,peek} functions are calls to 8106 ssl[23]_read_internal with the 'peek' flag set appropriately. 8107 A 'peek' parameter has also been added to ssl3_read_bytes, which 8108 does the actual work for ssl3_read_internal. 8109 [Bodo Moeller] 8110 8111 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 8112 the method-specific "init()" handler. Also clean up ex_data after 8113 calling the method-specific "finish()" handler. Previously, this was 8114 happening the other way round. 8115 [Geoff Thorpe] 8116 8117 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 8118 The previous value, 12, was not always sufficient for BN_mod_exp(). 8119 [Bodo Moeller] 8120 8121 *) Make sure that shared libraries get the internal name engine with 8122 the full version number and not just 0. This should mark the 8123 shared libraries as not backward compatible. Of course, this should 8124 be changed again when we can guarantee backward binary compatibility. 8125 [Richard Levitte] 8126 8127 *) Fix typo in get_cert_by_subject() in by_dir.c 8128 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 8129 8130 *) Rework the system to generate shared libraries: 8131 8132 - Make note of the expected extension for the shared libraries and 8133 if there is a need for symbolic links from for example libcrypto.so.0 8134 to libcrypto.so.0.9.7. There is extended info in Configure for 8135 that. 8136 8137 - Make as few rebuilds of the shared libraries as possible. 8138 8139 - Still avoid linking the OpenSSL programs with the shared libraries. 8140 8141 - When installing, install the shared libraries separately from the 8142 static ones. 8143 [Richard Levitte] 8144 8145 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 8146 8147 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 8148 and not in SSL_clear because the latter is also used by the 8149 accept/connect functions; previously, the settings made by 8150 SSL_set_read_ahead would be lost during the handshake. 8151 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 8152 8153 *) Correct util/mkdef.pl to be selective about disabled algorithms. 8154 Previously, it would create entries for disableed algorithms no 8155 matter what. 8156 [Richard Levitte] 8157 8158 *) Added several new manual pages for SSL_* function. 8159 [Lutz Jaenicke] 8160 8161 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 8162 8163 *) In ssl23_get_client_hello, generate an error message when faced 8164 with an initial SSL 3.0/TLS record that is too small to contain the 8165 first two bytes of the ClientHello message, i.e. client_version. 8166 (Note that this is a pathologic case that probably has never happened 8167 in real life.) The previous approach was to use the version number 8168 from the record header as a substitute; but our protocol choice 8169 should not depend on that one because it is not authenticated 8170 by the Finished messages. 8171 [Bodo Moeller] 8172 8173 *) More robust randomness gathering functions for Windows. 8174 [Jeffrey Altman <jaltman@columbia.edu>] 8175 8176 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 8177 not set then we don't setup the error code for issuer check errors 8178 to avoid possibly overwriting other errors which the callback does 8179 handle. If an application does set the flag then we assume it knows 8180 what it is doing and can handle the new informational codes 8181 appropriately. 8182 [Steve Henson] 8183 8184 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 8185 a general "ANY" type, as such it should be able to decode anything 8186 including tagged types. However it didn't check the class so it would 8187 wrongly interpret tagged types in the same way as their universal 8188 counterpart and unknown types were just rejected. Changed so that the 8189 tagged and unknown types are handled in the same way as a SEQUENCE: 8190 that is the encoding is stored intact. There is also a new type 8191 "V_ASN1_OTHER" which is used when the class is not universal, in this 8192 case we have no idea what the actual type is so we just lump them all 8193 together. 8194 [Steve Henson] 8195 8196 *) On VMS, stdout may very well lead to a file that is written to 8197 in a record-oriented fashion. That means that every write() will 8198 write a separate record, which will be read separately by the 8199 programs trying to read from it. This can be very confusing. 8200 8201 The solution is to put a BIO filter in the way that will buffer 8202 text until a linefeed is reached, and then write everything a 8203 line at a time, so every record written will be an actual line, 8204 not chunks of lines and not (usually doesn't happen, but I've 8205 seen it once) several lines in one record. BIO_f_linebuffer() is 8206 the answer. 8207 8208 Currently, it's a VMS-only method, because that's where it has 8209 been tested well enough. 8210 [Richard Levitte] 8211 8212 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 8213 it can return incorrect results. 8214 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 8215 but it was in 0.9.6-beta[12].) 8216 [Bodo Moeller] 8217 8218 *) Disable the check for content being present when verifying detached 8219 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 8220 include zero length content when signing messages. 8221 [Steve Henson] 8222 8223 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 8224 BIO_ctrl (for BIO pairs). 8225 [Bodo M��ller] 8226 8227 *) Add DSO method for VMS. 8228 [Richard Levitte] 8229 8230 *) Bug fix: Montgomery multiplication could produce results with the 8231 wrong sign. 8232 [Ulf M��ller] 8233 8234 *) Add RPM specification openssl.spec and modify it to build three 8235 packages. The default package contains applications, application 8236 documentation and run-time libraries. The devel package contains 8237 include files, static libraries and function documentation. The 8238 doc package contains the contents of the doc directory. The original 8239 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 8240 [Richard Levitte] 8241 8242 *) Add a large number of documentation files for many SSL routines. 8243 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 8244 8245 *) Add a configuration entry for Sony News 4. 8246 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 8247 8248 *) Don't set the two most significant bits to one when generating a 8249 random number < q in the DSA library. 8250 [Ulf M��ller] 8251 8252 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 8253 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 8254 the underlying transport is blocking) if a handshake took place. 8255 (The default behaviour is needed by applications such as s_client 8256 and s_server that use select() to determine when to use SSL_read; 8257 but for applications that know in advance when to expect data, it 8258 just makes things more complicated.) 8259 [Bodo Moeller] 8260 8261 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 8262 from EGD. 8263 [Ben Laurie] 8264 8265 *) Add a few more EBCDIC conditionals that make `req' and `x509' 8266 work better on such systems. 8267 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 8268 8269 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 8270 Update PKCS12_parse() so it copies the friendlyName and the 8271 keyid to the certificates aux info. 8272 [Steve Henson] 8273 8274 *) Fix bug in PKCS7_verify() which caused an infinite loop 8275 if there was more than one signature. 8276 [Sven Uszpelkat <su@celocom.de>] 8277 8278 *) Major change in util/mkdef.pl to include extra information 8279 about each symbol, as well as presentig variables as well 8280 as functions. This change means that there's n more need 8281 to rebuild the .num files when some algorithms are excluded. 8282 [Richard Levitte] 8283 8284 *) Allow the verify time to be set by an application, 8285 rather than always using the current time. 8286 [Steve Henson] 8287 8288 *) Phase 2 verify code reorganisation. The certificate 8289 verify code now looks up an issuer certificate by a 8290 number of criteria: subject name, authority key id 8291 and key usage. It also verifies self signed certificates 8292 by the same criteria. The main comparison function is 8293 X509_check_issued() which performs these checks. 8294 8295 Lot of changes were necessary in order to support this 8296 without completely rewriting the lookup code. 8297 8298 Authority and subject key identifier are now cached. 8299 8300 The LHASH 'certs' is X509_STORE has now been replaced 8301 by a STACK_OF(X509_OBJECT). This is mainly because an 8302 LHASH can't store or retrieve multiple objects with 8303 the same hash value. 8304 8305 As a result various functions (which were all internal 8306 use only) have changed to handle the new X509_STORE 8307 structure. This will break anything that messed round 8308 with X509_STORE internally. 8309 8310 The functions X509_STORE_add_cert() now checks for an 8311 exact match, rather than just subject name. 8312 8313 The X509_STORE API doesn't directly support the retrieval 8314 of multiple certificates matching a given criteria, however 8315 this can be worked round by performing a lookup first 8316 (which will fill the cache with candidate certificates) 8317 and then examining the cache for matches. This is probably 8318 the best we can do without throwing out X509_LOOKUP 8319 entirely (maybe later...). 8320 8321 The X509_VERIFY_CTX structure has been enhanced considerably. 8322 8323 All certificate lookup operations now go via a get_issuer() 8324 callback. Although this currently uses an X509_STORE it 8325 can be replaced by custom lookups. This is a simple way 8326 to bypass the X509_STORE hackery necessary to make this 8327 work and makes it possible to use more efficient techniques 8328 in future. A very simple version which uses a simple 8329 STACK for its trusted certificate store is also provided 8330 using X509_STORE_CTX_trusted_stack(). 8331 8332 The verify_cb() and verify() callbacks now have equivalents 8333 in the X509_STORE_CTX structure. 8334 8335 X509_STORE_CTX also has a 'flags' field which can be used 8336 to customise the verify behaviour. 8337 [Steve Henson] 8338 8339 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 8340 excludes S/MIME capabilities. 8341 [Steve Henson] 8342 8343 *) When a certificate request is read in keep a copy of the 8344 original encoding of the signed data and use it when outputing 8345 again. Signatures then use the original encoding rather than 8346 a decoded, encoded version which may cause problems if the 8347 request is improperly encoded. 8348 [Steve Henson] 8349 8350 *) For consistency with other BIO_puts implementations, call 8351 buffer_write(b, ...) directly in buffer_puts instead of calling 8352 BIO_write(b, ...). 8353 8354 In BIO_puts, increment b->num_write as in BIO_write. 8355 [Peter.Sylvester@EdelWeb.fr] 8356 8357 *) Fix BN_mul_word for the case where the word is 0. (We have to use 8358 BN_zero, we may not return a BIGNUM with an array consisting of 8359 words set to zero.) 8360 [Bodo Moeller] 8361 8362 *) Avoid calling abort() from within the library when problems are 8363 detected, except if preprocessor symbols have been defined 8364 (such as REF_CHECK, BN_DEBUG etc.). 8365 [Bodo Moeller] 8366 8367 *) New openssl application 'rsautl'. This utility can be 8368 used for low level RSA operations. DER public key 8369 BIO/fp routines also added. 8370 [Steve Henson] 8371 8372 *) New Configure entry and patches for compiling on QNX 4. 8373 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 8374 8375 *) A demo state-machine implementation was sponsored by 8376 Nuron (http://www.nuron.com/) and is now available in 8377 demos/state_machine. 8378 [Ben Laurie] 8379 8380 *) New options added to the 'dgst' utility for signature 8381 generation and verification. 8382 [Steve Henson] 8383 8384 *) Unrecognized PKCS#7 content types are now handled via a 8385 catch all ASN1_TYPE structure. This allows unsupported 8386 types to be stored as a "blob" and an application can 8387 encode and decode it manually. 8388 [Steve Henson] 8389 8390 *) Fix various signed/unsigned issues to make a_strex.c 8391 compile under VC++. 8392 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 8393 8394 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 8395 length if passed a buffer. ASN1_INTEGER_to_BN failed 8396 if passed a NULL BN and its argument was negative. 8397 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 8398 8399 *) Modification to PKCS#7 encoding routines to output definite 8400 length encoding. Since currently the whole structures are in 8401 memory there's not real point in using indefinite length 8402 constructed encoding. However if OpenSSL is compiled with 8403 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 8404 [Steve Henson] 8405 8406 *) Added BIO_vprintf() and BIO_vsnprintf(). 8407 [Richard Levitte] 8408 8409 *) Added more prefixes to parse for in the the strings written 8410 through a logging bio, to cover all the levels that are available 8411 through syslog. The prefixes are now: 8412 8413 PANIC, EMERG, EMR => LOG_EMERG 8414 ALERT, ALR => LOG_ALERT 8415 CRIT, CRI => LOG_CRIT 8416 ERROR, ERR => LOG_ERR 8417 WARNING, WARN, WAR => LOG_WARNING 8418 NOTICE, NOTE, NOT => LOG_NOTICE 8419 INFO, INF => LOG_INFO 8420 DEBUG, DBG => LOG_DEBUG 8421 8422 and as before, if none of those prefixes are present at the 8423 beginning of the string, LOG_ERR is chosen. 8424 8425 On Win32, the LOG_* levels are mapped according to this: 8426 8427 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 8428 LOG_WARNING => EVENTLOG_WARNING_TYPE 8429 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 8430 8431 [Richard Levitte] 8432 8433 *) Made it possible to reconfigure with just the configuration 8434 argument "reconf" or "reconfigure". The command line arguments 8435 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 8436 and are retrieved from there when reconfiguring. 8437 [Richard Levitte] 8438 8439 *) MD4 implemented. 8440 [Assar Westerlund <assar@sics.se>, Richard Levitte] 8441 8442 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 8443 [Richard Levitte] 8444 8445 *) The obj_dat.pl script was messing up the sorting of object 8446 names. The reason was that it compared the quoted version 8447 of strings as a result "OCSP" > "OCSP Signing" because 8448 " > SPACE. Changed script to store unquoted versions of 8449 names and add quotes on output. It was also omitting some 8450 names from the lookup table if they were given a default 8451 value (that is if SN is missing it is given the same 8452 value as LN and vice versa), these are now added on the 8453 grounds that if an object has a name we should be able to 8454 look it up. Finally added warning output when duplicate 8455 short or long names are found. 8456 [Steve Henson] 8457 8458 *) Changes needed for Tandem NSK. 8459 [Scott Uroff <scott@xypro.com>] 8460 8461 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 8462 RSA_padding_check_SSLv23(), special padding was never detected 8463 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 8464 version rollback attacks was not effective. 8465 8466 In s23_clnt.c, don't use special rollback-attack detection padding 8467 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 8468 client; similarly, in s23_srvr.c, don't do the rollback check if 8469 SSL 2.0 is the only protocol enabled in the server. 8470 [Bodo Moeller] 8471 8472 *) Make it possible to get hexdumps of unprintable data with 'openssl 8473 asn1parse'. By implication, the functions ASN1_parse_dump() and 8474 BIO_dump_indent() are added. 8475 [Richard Levitte] 8476 8477 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 8478 these print out strings and name structures based on various 8479 flags including RFC2253 support and proper handling of 8480 multibyte characters. Added options to the 'x509' utility 8481 to allow the various flags to be set. 8482 [Steve Henson] 8483 8484 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 8485 Also change the functions X509_cmp_current_time() and 8486 X509_gmtime_adj() work with an ASN1_TIME structure, 8487 this will enable certificates using GeneralizedTime in validity 8488 dates to be checked. 8489 [Steve Henson] 8490 8491 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 8492 negative public key encodings) on by default, 8493 NO_NEG_PUBKEY_BUG can be set to disable it. 8494 [Steve Henson] 8495 8496 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 8497 content octets. An i2c_ASN1_OBJECT is unnecessary because 8498 the encoding can be trivially obtained from the structure. 8499 [Steve Henson] 8500 8501 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 8502 not read locks (CRYPTO_r_[un]lock). 8503 [Bodo Moeller] 8504 8505 *) A first attempt at creating official support for shared 8506 libraries through configuration. I've kept it so the 8507 default is static libraries only, and the OpenSSL programs 8508 are always statically linked for now, but there are 8509 preparations for dynamic linking in place. 8510 This has been tested on Linux and Tru64. 8511 [Richard Levitte] 8512 8513 *) Randomness polling function for Win9x, as described in: 8514 Peter Gutmann, Software Generation of Practically Strong 8515 Random Numbers. 8516 [Ulf M��ller] 8517 8518 *) Fix so PRNG is seeded in req if using an already existing 8519 DSA key. 8520 [Steve Henson] 8521 8522 *) New options to smime application. -inform and -outform 8523 allow alternative formats for the S/MIME message including 8524 PEM and DER. The -content option allows the content to be 8525 specified separately. This should allow things like Netscape 8526 form signing output easier to verify. 8527 [Steve Henson] 8528 8529 *) Fix the ASN1 encoding of tags using the 'long form'. 8530 [Steve Henson] 8531 8532 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 8533 STRING types. These convert content octets to and from the 8534 underlying type. The actual tag and length octets are 8535 already assumed to have been read in and checked. These 8536 are needed because all other string types have virtually 8537 identical handling apart from the tag. By having versions 8538 of the ASN1 functions that just operate on content octets 8539 IMPLICIT tagging can be handled properly. It also allows 8540 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 8541 and ASN1_INTEGER are identical apart from the tag. 8542 [Steve Henson] 8543 8544 *) Change the handling of OID objects as follows: 8545 8546 - New object identifiers are inserted in objects.txt, following 8547 the syntax given in objects.README. 8548 - objects.pl is used to process obj_mac.num and create a new 8549 obj_mac.h. 8550 - obj_dat.pl is used to create a new obj_dat.h, using the data in 8551 obj_mac.h. 8552 8553 This is currently kind of a hack, and the perl code in objects.pl 8554 isn't very elegant, but it works as I intended. The simplest way 8555 to check that it worked correctly is to look in obj_dat.h and 8556 check the array nid_objs and make sure the objects haven't moved 8557 around (this is important!). Additions are OK, as well as 8558 consistent name changes. 8559 [Richard Levitte] 8560 8561 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 8562 [Bodo Moeller] 8563 8564 *) Addition of the command line parameter '-rand file' to 'openssl req'. 8565 The given file adds to whatever has already been seeded into the 8566 random pool through the RANDFILE configuration file option or 8567 environment variable, or the default random state file. 8568 [Richard Levitte] 8569 8570 *) mkstack.pl now sorts each macro group into lexical order. 8571 Previously the output order depended on the order the files 8572 appeared in the directory, resulting in needless rewriting 8573 of safestack.h . 8574 [Steve Henson] 8575 8576 *) Patches to make OpenSSL compile under Win32 again. Mostly 8577 work arounds for the VC++ problem that it treats func() as 8578 func(void). Also stripped out the parts of mkdef.pl that 8579 added extra typesafe functions: these no longer exist. 8580 [Steve Henson] 8581 8582 *) Reorganisation of the stack code. The macros are now all 8583 collected in safestack.h . Each macro is defined in terms of 8584 a "stack macro" of the form SKM_<name>(type, a, b). The 8585 DEBUG_SAFESTACK is now handled in terms of function casts, 8586 this has the advantage of retaining type safety without the 8587 use of additional functions. If DEBUG_SAFESTACK is not defined 8588 then the non typesafe macros are used instead. Also modified the 8589 mkstack.pl script to handle the new form. Needs testing to see 8590 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 8591 the default if no major problems. Similar behaviour for ASN1_SET_OF 8592 and PKCS12_STACK_OF. 8593 [Steve Henson] 8594 8595 *) When some versions of IIS use the 'NET' form of private key the 8596 key derivation algorithm is different. Normally MD5(password) is 8597 used as a 128 bit RC4 key. In the modified case 8598 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some 8599 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 8600 as the old Netscape_RSA functions except they have an additional 8601 'sgckey' parameter which uses the modified algorithm. Also added 8602 an -sgckey command line option to the rsa utility. Thanks to 8603 Adrian Peck <bertie@ncipher.com> for posting details of the modified 8604 algorithm to openssl-dev. 8605 [Steve Henson] 8606 8607 *) The evp_local.h macros were using 'c.##kname' which resulted in 8608 invalid expansion on some systems (SCO 5.0.5 for example). 8609 Corrected to 'c.kname'. 8610 [Phillip Porch <root@theporch.com>] 8611 8612 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 8613 a STACK of email addresses from a certificate or request, these look 8614 in the subject name and the subject alternative name extensions and 8615 omit any duplicate addresses. 8616 [Steve Henson] 8617 8618 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 8619 This makes DSA verification about 2 % faster. 8620 [Bodo Moeller] 8621 8622 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 8623 (meaning that now 2^5 values will be precomputed, which is only 4 KB 8624 plus overhead for 1024 bit moduli). 8625 This makes exponentiations about 0.5 % faster for 1024 bit 8626 exponents (as measured by "openssl speed rsa2048"). 8627 [Bodo Moeller] 8628 8629 *) Rename memory handling macros to avoid conflicts with other 8630 software: 8631 Malloc => OPENSSL_malloc 8632 Malloc_locked => OPENSSL_malloc_locked 8633 Realloc => OPENSSL_realloc 8634 Free => OPENSSL_free 8635 [Richard Levitte] 8636 8637 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 8638 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 8639 [Bodo Moeller] 8640 8641 *) CygWin32 support. 8642 [John Jarvie <jjarvie@newsguy.com>] 8643 8644 *) The type-safe stack code has been rejigged. It is now only compiled 8645 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 8646 by default all type-specific stack functions are "#define"d back to 8647 standard stack functions. This results in more streamlined output 8648 but retains the type-safety checking possibilities of the original 8649 approach. 8650 [Geoff Thorpe] 8651 8652 *) The STACK code has been cleaned up, and certain type declarations 8653 that didn't make a lot of sense have been brought in line. This has 8654 also involved a cleanup of sorts in safestack.h to more correctly 8655 map type-safe stack functions onto their plain stack counterparts. 8656 This work has also resulted in a variety of "const"ifications of 8657 lots of the code, especially "_cmp" operations which should normally 8658 be prototyped with "const" parameters anyway. 8659 [Geoff Thorpe] 8660 8661 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 8662 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 8663 (The PRNG state consists of two parts, the large pool 'state' and 'md', 8664 where all of 'md' is used each time the PRNG is used, but 'state' 8665 is used only indexed by a cyclic counter. As entropy may not be 8666 well distributed from the beginning, 'md' is important as a 8667 chaining variable. However, the output function chains only half 8668 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 8669 all of 'md', and seeding with STATE_SIZE dummy bytes will result 8670 in all of 'state' being rewritten, with the new values depending 8671 on virtually all of 'md'. This overcomes the 80 bit limitation.) 8672 [Bodo Moeller] 8673 8674 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 8675 the handshake is continued after ssl_verify_cert_chain(); 8676 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 8677 can lead to 'unexplainable' connection aborts later. 8678 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 8679 8680 *) Major EVP API cipher revision. 8681 Add hooks for extra EVP features. This allows various cipher 8682 parameters to be set in the EVP interface. Support added for variable 8683 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 8684 setting of RC2 and RC5 parameters. 8685 8686 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 8687 ciphers. 8688 8689 Remove lots of duplicated code from the EVP library. For example *every* 8690 cipher init() function handles the 'iv' in the same way according to the 8691 cipher mode. They also all do nothing if the 'key' parameter is NULL and 8692 for CFB and OFB modes they zero ctx->num. 8693 8694 New functionality allows removal of S/MIME code RC2 hack. 8695 8696 Most of the routines have the same form and so can be declared in terms 8697 of macros. 8698 8699 By shifting this to the top level EVP_CipherInit() it can be removed from 8700 all individual ciphers. If the cipher wants to handle IVs or keys 8701 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 8702 flags. 8703 8704 Change lots of functions like EVP_EncryptUpdate() to now return a 8705 value: although software versions of the algorithms cannot fail 8706 any installed hardware versions can. 8707 [Steve Henson] 8708 8709 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 8710 this option is set, tolerate broken clients that send the negotiated 8711 protocol version number instead of the requested protocol version 8712 number. 8713 [Bodo Moeller] 8714 8715 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 8716 i.e. non-zero for export ciphersuites, zero otherwise. 8717 Previous versions had this flag inverted, inconsistent with 8718 rsa_tmp_cb (..._TMP_RSA_CB). 8719 [Bodo Moeller; problem reported by Amit Chopra] 8720 8721 *) Add missing DSA library text string. Work around for some IIS 8722 key files with invalid SEQUENCE encoding. 8723 [Steve Henson] 8724 8725 *) Add a document (doc/standards.txt) that list all kinds of standards 8726 and so on that are implemented in OpenSSL. 8727 [Richard Levitte] 8728 8729 *) Enhance c_rehash script. Old version would mishandle certificates 8730 with the same subject name hash and wouldn't handle CRLs at all. 8731 Added -fingerprint option to crl utility, to support new c_rehash 8732 features. 8733 [Steve Henson] 8734 8735 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 8736 [Ulf M��ller] 8737 8738 *) Fix for SSL server purpose checking. Server checking was 8739 rejecting certificates which had extended key usage present 8740 but no ssl client purpose. 8741 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 8742 8743 *) Make PKCS#12 code work with no password. The PKCS#12 spec 8744 is a little unclear about how a blank password is handled. 8745 Since the password in encoded as a BMPString with terminating 8746 double NULL a zero length password would end up as just the 8747 double NULL. However no password at all is different and is 8748 handled differently in the PKCS#12 key generation code. NS 8749 treats a blank password as zero length. MSIE treats it as no 8750 password on export: but it will try both on import. We now do 8751 the same: PKCS12_parse() tries zero length and no password if 8752 the password is set to "" or NULL (NULL is now a valid password: 8753 it wasn't before) as does the pkcs12 application. 8754 [Steve Henson] 8755 8756 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 8757 perror when PEM_read_bio_X509_REQ fails, the error message must 8758 be obtained from the error queue. 8759 [Bodo Moeller] 8760 8761 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 8762 it in ERR_remove_state if appropriate, and change ERR_get_state 8763 accordingly to avoid race conditions (this is necessary because 8764 thread_hash is no longer constant once set). 8765 [Bodo Moeller] 8766 8767 *) Bugfix for linux-elf makefile.one. 8768 [Ulf M��ller] 8769 8770 *) RSA_get_default_method() will now cause a default 8771 RSA_METHOD to be chosen if one doesn't exist already. 8772 Previously this was only set during a call to RSA_new() 8773 or RSA_new_method(NULL) meaning it was possible for 8774 RSA_get_default_method() to return NULL. 8775 [Geoff Thorpe] 8776 8777 *) Added native name translation to the existing DSO code 8778 that will convert (if the flag to do so is set) filenames 8779 that are sufficiently small and have no path information 8780 into a canonical native form. Eg. "blah" converted to 8781 "libblah.so" or "blah.dll" etc. 8782 [Geoff Thorpe] 8783 8784 *) New function ERR_error_string_n(e, buf, len) which is like 8785 ERR_error_string(e, buf), but writes at most 'len' bytes 8786 including the 0 terminator. For ERR_error_string_n, 'buf' 8787 may not be NULL. 8788 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 8789 8790 *) CONF library reworked to become more general. A new CONF 8791 configuration file reader "class" is implemented as well as a 8792 new functions (NCONF_*, for "New CONF") to handle it. The now 8793 old CONF_* functions are still there, but are reimplemented to 8794 work in terms of the new functions. Also, a set of functions 8795 to handle the internal storage of the configuration data is 8796 provided to make it easier to write new configuration file 8797 reader "classes" (I can definitely see something reading a 8798 configuration file in XML format, for example), called _CONF_*, 8799 or "the configuration storage API"... 8800 8801 The new configuration file reading functions are: 8802 8803 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 8804 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 8805 8806 NCONF_default, NCONF_WIN32 8807 8808 NCONF_dump_fp, NCONF_dump_bio 8809 8810 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 8811 NCONF_new creates a new CONF object. This works in the same way 8812 as other interfaces in OpenSSL, like the BIO interface. 8813 NCONF_dump_* dump the internal storage of the configuration file, 8814 which is useful for debugging. All other functions take the same 8815 arguments as the old CONF_* functions wth the exception of the 8816 first that must be a `CONF *' instead of a `LHASH *'. 8817 8818 To make it easer to use the new classes with the old CONF_* functions, 8819 the function CONF_set_default_method is provided. 8820 [Richard Levitte] 8821 8822 *) Add '-tls1' option to 'openssl ciphers', which was already 8823 mentioned in the documentation but had not been implemented. 8824 (This option is not yet really useful because even the additional 8825 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 8826 [Bodo Moeller] 8827 8828 *) Initial DSO code added into libcrypto for letting OpenSSL (and 8829 OpenSSL-based applications) load shared libraries and bind to 8830 them in a portable way. 8831 [Geoff Thorpe, with contributions from Richard Levitte] 8832 8833 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 8834 8835 *) Make sure _lrotl and _lrotr are only used with MSVC. 8836 8837 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 8838 (the default implementation of RAND_status). 8839 8840 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 8841 to '-clrext' (= clear extensions), as intended and documented. 8842 [Bodo Moeller; inconsistency pointed out by Michael Attili 8843 <attili@amaxo.com>] 8844 8845 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 8846 was larger than the MD block size. 8847 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 8848 8849 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 8850 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 8851 using the passed key: if the passed key was a private key the result 8852 of X509_print(), for example, would be to print out all the private key 8853 components. 8854 [Steve Henson] 8855 8856 *) des_quad_cksum() byte order bug fix. 8857 [Ulf M��ller, using the problem description in krb4-0.9.7, where 8858 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 8859 8860 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 8861 discouraged. 8862 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 8863 8864 *) For easily testing in shell scripts whether some command 8865 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 8866 returns with exit code 0 iff no command of the given name is available. 8867 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 8868 the output goes to stdout and nothing is printed to stderr. 8869 Additional arguments are always ignored. 8870 8871 Since for each cipher there is a command of the same name, 8872 the 'no-cipher' compilation switches can be tested this way. 8873 8874 ('openssl no-XXX' is not able to detect pseudo-commands such 8875 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 8876 [Bodo Moeller] 8877 8878 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 8879 [Bodo Moeller] 8880 8881 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 8882 is set; it will be thrown away anyway because each handshake creates 8883 its own key. 8884 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 8885 to parameters -- in previous versions (since OpenSSL 0.9.3) the 8886 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining 8887 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro. 8888 [Bodo Moeller] 8889 8890 *) New s_client option -ign_eof: EOF at stdin is ignored, and 8891 'Q' and 'R' lose their special meanings (quit/renegotiate). 8892 This is part of what -quiet does; unlike -quiet, -ign_eof 8893 does not suppress any output. 8894 [Richard Levitte] 8895 8896 *) Add compatibility options to the purpose and trust code. The 8897 purpose X509_PURPOSE_ANY is "any purpose" which automatically 8898 accepts a certificate or CA, this was the previous behaviour, 8899 with all the associated security issues. 8900 8901 X509_TRUST_COMPAT is the old trust behaviour: only and 8902 automatically trust self signed roots in certificate store. A 8903 new trust setting X509_TRUST_DEFAULT is used to specify that 8904 a purpose has no associated trust setting and it should instead 8905 use the value in the default purpose. 8906 [Steve Henson] 8907 8908 *) Fix the PKCS#8 DSA private key code so it decodes keys again 8909 and fix a memory leak. 8910 [Steve Henson] 8911 8912 *) In util/mkerr.pl (which implements 'make errors'), preserve 8913 reason strings from the previous version of the .c file, as 8914 the default to have only downcase letters (and digits) in 8915 automatically generated reasons codes is not always appropriate. 8916 [Bodo Moeller] 8917 8918 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 8919 using strerror. Previously, ERR_reason_error_string() returned 8920 library names as reason strings for SYSerr; but SYSerr is a special 8921 case where small numbers are errno values, not library numbers. 8922 [Bodo Moeller] 8923 8924 *) Add '-dsaparam' option to 'openssl dhparam' application. This 8925 converts DSA parameters into DH parameters. (When creating parameters, 8926 DSA_generate_parameters is used.) 8927 [Bodo Moeller] 8928 8929 *) Include 'length' (recommended exponent length) in C code generated 8930 by 'openssl dhparam -C'. 8931 [Bodo Moeller] 8932 8933 *) The second argument to set_label in perlasm was already being used 8934 so couldn't be used as a "file scope" flag. Moved to third argument 8935 which was free. 8936 [Steve Henson] 8937 8938 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 8939 instead of RAND_bytes for encryption IVs and salts. 8940 [Bodo Moeller] 8941 8942 *) Include RAND_status() into RAND_METHOD instead of implementing 8943 it only for md_rand.c Otherwise replacing the PRNG by calling 8944 RAND_set_rand_method would be impossible. 8945 [Bodo Moeller] 8946 8947 *) Don't let DSA_generate_key() enter an infinite loop if the random 8948 number generation fails. 8949 [Bodo Moeller] 8950 8951 *) New 'rand' application for creating pseudo-random output. 8952 [Bodo Moeller] 8953 8954 *) Added configuration support for Linux/IA64 8955 [Rolf Haberrecker <rolf@suse.de>] 8956 8957 *) Assembler module support for Mingw32. 8958 [Ulf M��ller] 8959 8960 *) Shared library support for HPUX (in shlib/). 8961 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 8962 8963 *) Shared library support for Solaris gcc. 8964 [Lutz Behnke <behnke@trustcenter.de>] 8965 8966 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 8967 8968 *) PKCS7_encrypt() was adding text MIME headers twice because they 8969 were added manually and by SMIME_crlf_copy(). 8970 [Steve Henson] 8971 8972 *) In bntest.c don't call BN_rand with zero bits argument. 8973 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 8974 8975 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 8976 case was implemented. This caused BN_div_recp() to fail occasionally. 8977 [Ulf M��ller] 8978 8979 *) Add an optional second argument to the set_label() in the perl 8980 assembly language builder. If this argument exists and is set 8981 to 1 it signals that the assembler should use a symbol whose 8982 scope is the entire file, not just the current function. This 8983 is needed with MASM which uses the format label:: for this scope. 8984 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 8985 8986 *) Change the ASN1 types so they are typedefs by default. Before 8987 almost all types were #define'd to ASN1_STRING which was causing 8988 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 8989 for example. 8990 [Steve Henson] 8991 8992 *) Change names of new functions to the new get1/get0 naming 8993 convention: After 'get1', the caller owns a reference count 8994 and has to call ..._free; 'get0' returns a pointer to some 8995 data structure without incrementing reference counters. 8996 (Some of the existing 'get' functions increment a reference 8997 counter, some don't.) 8998 Similarly, 'set1' and 'add1' functions increase reference 8999 counters or duplicate objects. 9000 [Steve Henson] 9001 9002 *) Allow for the possibility of temp RSA key generation failure: 9003 the code used to assume it always worked and crashed on failure. 9004 [Steve Henson] 9005 9006 *) Fix potential buffer overrun problem in BIO_printf(). 9007 [Ulf M��ller, using public domain code by Patrick Powell; problem 9008 pointed out by David Sacerdote <das33@cornell.edu>] 9009 9010 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 9011 RAND_egd() and RAND_status(). In the command line application, 9012 the EGD socket can be specified like a seed file using RANDFILE 9013 or -rand. 9014 [Ulf M��ller] 9015 9016 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 9017 Some CAs (e.g. Verisign) distribute certificates in this form. 9018 [Steve Henson] 9019 9020 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 9021 list to exclude them. This means that no special compilation option 9022 is needed to use anonymous DH: it just needs to be included in the 9023 cipher list. 9024 [Steve Henson] 9025 9026 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 9027 EVP_MD_type. The old functionality is available in a new macro called 9028 EVP_MD_md(). Change code that uses it and update docs. 9029 [Steve Henson] 9030 9031 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 9032 where the 'void *' argument is replaced by a function pointer argument. 9033 Previously 'void *' was abused to point to functions, which works on 9034 many platforms, but is not correct. As these functions are usually 9035 called by macros defined in OpenSSL header files, most source code 9036 should work without changes. 9037 [Richard Levitte] 9038 9039 *) <openssl/opensslconf.h> (which is created by Configure) now contains 9040 sections with information on -D... compiler switches used for 9041 compiling the library so that applications can see them. To enable 9042 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 9043 must be defined. E.g., 9044 #define OPENSSL_ALGORITHM_DEFINES 9045 #include <openssl/opensslconf.h> 9046 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 9047 [Richard Levitte, Ulf and Bodo M��ller] 9048 9049 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 9050 record layer. 9051 [Bodo Moeller] 9052 9053 *) Change the 'other' type in certificate aux info to a STACK_OF 9054 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 9055 the required ASN1 format: arbitrary types determined by an OID. 9056 [Steve Henson] 9057 9058 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 9059 argument to 'req'. This is not because the function is newer or 9060 better than others it just uses the work 'NEW' in the certificate 9061 request header lines. Some software needs this. 9062 [Steve Henson] 9063 9064 *) Reorganise password command line arguments: now passwords can be 9065 obtained from various sources. Delete the PEM_cb function and make 9066 it the default behaviour: i.e. if the callback is NULL and the 9067 usrdata argument is not NULL interpret it as a null terminated pass 9068 phrase. If usrdata and the callback are NULL then the pass phrase 9069 is prompted for as usual. 9070 [Steve Henson] 9071 9072 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 9073 the support is automatically enabled. The resulting binaries will 9074 autodetect the card and use it if present. 9075 [Ben Laurie and Compaq Inc.] 9076 9077 *) Work around for Netscape hang bug. This sends certificate request 9078 and server done in one record. Since this is perfectly legal in the 9079 SSL/TLS protocol it isn't a "bug" option and is on by default. See 9080 the bugs/SSLv3 entry for more info. 9081 [Steve Henson] 9082 9083 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 9084 [Andy Polyakov] 9085 9086 *) Add -rand argument to smime and pkcs12 applications and read/write 9087 of seed file. 9088 [Steve Henson] 9089 9090 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 9091 [Bodo Moeller] 9092 9093 *) Add command line password options to the remaining applications. 9094 [Steve Henson] 9095 9096 *) Bug fix for BN_div_recp() for numerators with an even number of 9097 bits. 9098 [Ulf M��ller] 9099 9100 *) More tests in bntest.c, and changed test_bn output. 9101 [Ulf M��ller] 9102 9103 *) ./config recognizes MacOS X now. 9104 [Andy Polyakov] 9105 9106 *) Bug fix for BN_div() when the first words of num and divsor are 9107 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 9108 [Ulf M��ller] 9109 9110 *) Add support for various broken PKCS#8 formats, and command line 9111 options to produce them. 9112 [Steve Henson] 9113 9114 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 9115 get temporary BIGNUMs from a BN_CTX. 9116 [Ulf M��ller] 9117 9118 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 9119 for p == 0. 9120 [Ulf M��ller] 9121 9122 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 9123 include a #define from the old name to the new. The original intent 9124 was that statically linked binaries could for example just call 9125 SSLeay_add_all_ciphers() to just add ciphers to the table and not 9126 link with digests. This never worked becayse SSLeay_add_all_digests() 9127 and SSLeay_add_all_ciphers() were in the same source file so calling 9128 one would link with the other. They are now in separate source files. 9129 [Steve Henson] 9130 9131 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 9132 [Steve Henson] 9133 9134 *) Use a less unusual form of the Miller-Rabin primality test (it used 9135 a binary algorithm for exponentiation integrated into the Miller-Rabin 9136 loop, our standard modexp algorithms are faster). 9137 [Bodo Moeller] 9138 9139 *) Support for the EBCDIC character set completed. 9140 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 9141 9142 *) Source code cleanups: use const where appropriate, eliminate casts, 9143 use void * instead of char * in lhash. 9144 [Ulf M��ller] 9145 9146 *) Bugfix: ssl3_send_server_key_exchange was not restartable 9147 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 9148 this the server could overwrite ephemeral keys that the client 9149 has already seen). 9150 [Bodo Moeller] 9151 9152 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 9153 using 50 iterations of the Rabin-Miller test. 9154 9155 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 9156 iterations of the Rabin-Miller test as required by the appendix 9157 to FIPS PUB 186[-1]) instead of DSA_is_prime. 9158 As BN_is_prime_fasttest includes trial division, DSA parameter 9159 generation becomes much faster. 9160 9161 This implies a change for the callback functions in DSA_is_prime 9162 and DSA_generate_parameters: The callback function is called once 9163 for each positive witness in the Rabin-Miller test, not just 9164 occasionally in the inner loop; and the parameters to the 9165 callback function now provide an iteration count for the outer 9166 loop rather than for the current invocation of the inner loop. 9167 DSA_generate_parameters additionally can call the callback 9168 function with an 'iteration count' of -1, meaning that a 9169 candidate has passed the trial division test (when q is generated 9170 from an application-provided seed, trial division is skipped). 9171 [Bodo Moeller] 9172 9173 *) New function BN_is_prime_fasttest that optionally does trial 9174 division before starting the Rabin-Miller test and has 9175 an additional BN_CTX * argument (whereas BN_is_prime always 9176 has to allocate at least one BN_CTX). 9177 'callback(1, -1, cb_arg)' is called when a number has passed the 9178 trial division stage. 9179 [Bodo Moeller] 9180 9181 *) Fix for bug in CRL encoding. The validity dates weren't being handled 9182 as ASN1_TIME. 9183 [Steve Henson] 9184 9185 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 9186 [Steve Henson] 9187 9188 *) New function BN_pseudo_rand(). 9189 [Ulf M��ller] 9190 9191 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 9192 bignum version of BN_from_montgomery() with the working code from 9193 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 9194 the comments. 9195 [Ulf M��ller] 9196 9197 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 9198 made it impossible to use the same SSL_SESSION data structure in 9199 SSL2 clients in multiple threads. 9200 [Bodo Moeller] 9201 9202 *) The return value of RAND_load_file() no longer counts bytes obtained 9203 by stat(). RAND_load_file(..., -1) is new and uses the complete file 9204 to seed the PRNG (previously an explicit byte count was required). 9205 [Ulf M��ller, Bodo M��ller] 9206 9207 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 9208 used (char *) instead of (void *) and had casts all over the place. 9209 [Steve Henson] 9210 9211 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 9212 [Ulf M��ller] 9213 9214 *) Retain source code compatibility for BN_prime_checks macro: 9215 BN_is_prime(..., BN_prime_checks, ...) now uses 9216 BN_prime_checks_for_size to determine the appropriate number of 9217 Rabin-Miller iterations. 9218 [Ulf M��ller] 9219 9220 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 9221 DH_CHECK_P_NOT_SAFE_PRIME. 9222 (Check if this is true? OpenPGP calls them "strong".) 9223 [Ulf M��ller] 9224 9225 *) Merge the functionality of "dh" and "gendh" programs into a new program 9226 "dhparam". The old programs are retained for now but will handle DH keys 9227 (instead of parameters) in future. 9228 [Steve Henson] 9229 9230 *) Make the ciphers, s_server and s_client programs check the return values 9231 when a new cipher list is set. 9232 [Steve Henson] 9233 9234 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 9235 ciphers. Before when the 56bit ciphers were enabled the sorting was 9236 wrong. 9237 9238 The syntax for the cipher sorting has been extended to support sorting by 9239 cipher-strength (using the strength_bits hard coded in the tables). 9240 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 9241 9242 Fix a bug in the cipher-command parser: when supplying a cipher command 9243 string with an "undefined" symbol (neither command nor alphanumeric 9244 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 9245 an error is flagged. 9246 9247 Due to the strength-sorting extension, the code of the 9248 ssl_create_cipher_list() function was completely rearranged. I hope that 9249 the readability was also increased :-) 9250 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 9251 9252 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 9253 for the first serial number and places 2 in the serial number file. This 9254 avoids problems when the root CA is created with serial number zero and 9255 the first user certificate has the same issuer name and serial number 9256 as the root CA. 9257 [Steve Henson] 9258 9259 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 9260 the new code. Add documentation for this stuff. 9261 [Steve Henson] 9262 9263 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 9264 X509_*() to X509at_*() on the grounds that they don't handle X509 9265 structures and behave in an analagous way to the X509v3 functions: 9266 they shouldn't be called directly but wrapper functions should be used 9267 instead. 9268 9269 So we also now have some wrapper functions that call the X509at functions 9270 when passed certificate requests. (TO DO: similar things can be done with 9271 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 9272 things. Some of these need some d2i or i2d and print functionality 9273 because they handle more complex structures.) 9274 [Steve Henson] 9275 9276 *) Add missing #ifndefs that caused missing symbols when building libssl 9277 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 9278 NO_RSA in ssl/s2*.c. 9279 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf M��ller] 9280 9281 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 9282 has a return value which indicates the quality of the random data 9283 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 9284 error queue. New function RAND_pseudo_bytes() generates output that is 9285 guaranteed to be unique but not unpredictable. RAND_add is like 9286 RAND_seed, but takes an extra argument for an entropy estimate 9287 (RAND_seed always assumes full entropy). 9288 [Ulf M��ller] 9289 9290 *) Do more iterations of Rabin-Miller probable prime test (specifically, 9291 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 9292 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 9293 in crypto/bn/bn_prime.c for the complete table). This guarantees a 9294 false-positive rate of at most 2^-80 for random input. 9295 [Bodo Moeller] 9296 9297 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 9298 [Bodo Moeller] 9299 9300 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 9301 in the 0.9.5 release), this returns the chain 9302 from an X509_CTX structure with a dup of the stack and all 9303 the X509 reference counts upped: so the stack will exist 9304 after X509_CTX_cleanup() has been called. Modify pkcs12.c 9305 to use this. 9306 9307 Also make SSL_SESSION_print() print out the verify return 9308 code. 9309 [Steve Henson] 9310 9311 *) Add manpage for the pkcs12 command. Also change the default 9312 behaviour so MAC iteration counts are used unless the new 9313 -nomaciter option is used. This improves file security and 9314 only older versions of MSIE (4.0 for example) need it. 9315 [Steve Henson] 9316 9317 *) Honor the no-xxx Configure options when creating .DEF files. 9318 [Ulf M��ller] 9319 9320 *) Add PKCS#10 attributes to field table: challengePassword, 9321 unstructuredName and unstructuredAddress. These are taken from 9322 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 9323 international characters are used. 9324 9325 More changes to X509_ATTRIBUTE code: allow the setting of types 9326 based on strings. Remove the 'loc' parameter when adding 9327 attributes because these will be a SET OF encoding which is sorted 9328 in ASN1 order. 9329 [Steve Henson] 9330 9331 *) Initial changes to the 'req' utility to allow request generation 9332 automation. This will allow an application to just generate a template 9333 file containing all the field values and have req construct the 9334 request. 9335 9336 Initial support for X509_ATTRIBUTE handling. Stacks of these are 9337 used all over the place including certificate requests and PKCS#7 9338 structures. They are currently handled manually where necessary with 9339 some primitive wrappers for PKCS#7. The new functions behave in a 9340 manner analogous to the X509 extension functions: they allow 9341 attributes to be looked up by NID and added. 9342 9343 Later something similar to the X509V3 code would be desirable to 9344 automatically handle the encoding, decoding and printing of the 9345 more complex types. The string types like challengePassword can 9346 be handled by the string table functions. 9347 9348 Also modified the multi byte string table handling. Now there is 9349 a 'global mask' which masks out certain types. The table itself 9350 can use the flag STABLE_NO_MASK to ignore the mask setting: this 9351 is useful when for example there is only one permissible type 9352 (as in countryName) and using the mask might result in no valid 9353 types at all. 9354 [Steve Henson] 9355 9356 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 9357 SSL_get_peer_finished to allow applications to obtain the latest 9358 Finished messages sent to the peer or expected from the peer, 9359 respectively. (SSL_get_peer_finished is usually the Finished message 9360 actually received from the peer, otherwise the protocol will be aborted.) 9361 9362 As the Finished message are message digests of the complete handshake 9363 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 9364 be used for external authentication procedures when the authentication 9365 provided by SSL/TLS is not desired or is not enough. 9366 [Bodo Moeller] 9367 9368 *) Enhanced support for Alpha Linux is added. Now ./config checks if 9369 the host supports BWX extension and if Compaq C is present on the 9370 $PATH. Just exploiting of the BWX extension results in 20-30% 9371 performance kick for some algorithms, e.g. DES and RC4 to mention 9372 a couple. Compaq C in turn generates ~20% faster code for MD5 and 9373 SHA1. 9374 [Andy Polyakov] 9375 9376 *) Add support for MS "fast SGC". This is arguably a violation of the 9377 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 9378 weak crypto and after checking the certificate is SGC a second one 9379 with strong crypto. MS SGC stops the first handshake after receiving 9380 the server certificate message and sends a second client hello. Since 9381 a server will typically do all the time consuming operations before 9382 expecting any further messages from the client (server key exchange 9383 is the most expensive) there is little difference between the two. 9384 9385 To get OpenSSL to support MS SGC we have to permit a second client 9386 hello message after we have sent server done. In addition we have to 9387 reset the MAC if we do get this second client hello. 9388 [Steve Henson] 9389 9390 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 9391 if a DER encoded private key is RSA or DSA traditional format. Changed 9392 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 9393 format DER encoded private key. Newer code should use PKCS#8 format which 9394 has the key type encoded in the ASN1 structure. Added DER private key 9395 support to pkcs8 application. 9396 [Steve Henson] 9397 9398 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 9399 ciphersuites has been selected (as required by the SSL 3/TLS 1 9400 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 9401 is set, we interpret this as a request to violate the specification 9402 (the worst that can happen is a handshake failure, and 'correct' 9403 behaviour would result in a handshake failure anyway). 9404 [Bodo Moeller] 9405 9406 *) In SSL_CTX_add_session, take into account that there might be multiple 9407 SSL_SESSION structures with the same session ID (e.g. when two threads 9408 concurrently obtain them from an external cache). 9409 The internal cache can handle only one SSL_SESSION with a given ID, 9410 so if there's a conflict, we now throw out the old one to achieve 9411 consistency. 9412 [Bodo Moeller] 9413 9414 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 9415 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 9416 some routines that use cipher OIDs: some ciphers do not have OIDs 9417 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 9418 example. 9419 [Steve Henson] 9420 9421 *) Simplify the trust setting structure and code. Now we just have 9422 two sequences of OIDs for trusted and rejected settings. These will 9423 typically have values the same as the extended key usage extension 9424 and any application specific purposes. 9425 9426 The trust checking code now has a default behaviour: it will just 9427 check for an object with the same NID as the passed id. Functions can 9428 be provided to override either the default behaviour or the behaviour 9429 for a given id. SSL client, server and email already have functions 9430 in place for compatibility: they check the NID and also return "trusted" 9431 if the certificate is self signed. 9432 [Steve Henson] 9433 9434 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 9435 traditional format into an EVP_PKEY structure. 9436 [Steve Henson] 9437 9438 *) Add a password callback function PEM_cb() which either prompts for 9439 a password if usr_data is NULL or otherwise assumes it is a null 9440 terminated password. Allow passwords to be passed on command line 9441 environment or config files in a few more utilities. 9442 [Steve Henson] 9443 9444 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 9445 keys. Add some short names for PKCS#8 PBE algorithms and allow them 9446 to be specified on the command line for the pkcs8 and pkcs12 utilities. 9447 Update documentation. 9448 [Steve Henson] 9449 9450 *) Support for ASN1 "NULL" type. This could be handled before by using 9451 ASN1_TYPE but there wasn't any function that would try to read a NULL 9452 and produce an error if it couldn't. For compatibility we also have 9453 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 9454 don't allocate anything because they don't need to. 9455 [Steve Henson] 9456 9457 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 9458 for details. 9459 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 9460 9461 *) Rebuild of the memory allocation routines used by OpenSSL code and 9462 possibly others as well. The purpose is to make an interface that 9463 provide hooks so anyone can build a separate set of allocation and 9464 deallocation routines to be used by OpenSSL, for example memory 9465 pool implementations, or something else, which was previously hard 9466 since Malloc(), Realloc() and Free() were defined as macros having 9467 the values malloc, realloc and free, respectively (except for Win32 9468 compilations). The same is provided for memory debugging code. 9469 OpenSSL already comes with functionality to find memory leaks, but 9470 this gives people a chance to debug other memory problems. 9471 9472 With these changes, a new set of functions and macros have appeared: 9473 9474 CRYPTO_set_mem_debug_functions() [F] 9475 CRYPTO_get_mem_debug_functions() [F] 9476 CRYPTO_dbg_set_options() [F] 9477 CRYPTO_dbg_get_options() [F] 9478 CRYPTO_malloc_debug_init() [M] 9479 9480 The memory debug functions are NULL by default, unless the library 9481 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 9482 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 9483 gives the standard debugging functions that come with OpenSSL) or 9484 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 9485 provided by the library user) must be used. When the standard 9486 debugging functions are used, CRYPTO_dbg_set_options can be used to 9487 request additional information: 9488 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 9489 the CRYPTO_MDEBUG_xxx macro when compiling the library. 9490 9491 Also, things like CRYPTO_set_mem_functions will always give the 9492 expected result (the new set of functions is used for allocation 9493 and deallocation) at all times, regardless of platform and compiler 9494 options. 9495 9496 To finish it up, some functions that were never use in any other 9497 way than through macros have a new API and new semantic: 9498 9499 CRYPTO_dbg_malloc() 9500 CRYPTO_dbg_realloc() 9501 CRYPTO_dbg_free() 9502 9503 All macros of value have retained their old syntax. 9504 [Richard Levitte and Bodo Moeller] 9505 9506 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 9507 ordering of SMIMECapabilities wasn't in "strength order" and there 9508 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 9509 algorithm. 9510 [Steve Henson] 9511 9512 *) Some ASN1 types with illegal zero length encoding (INTEGER, 9513 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 9514 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 9515 9516 *) Merge in my S/MIME library for OpenSSL. This provides a simple 9517 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 9518 functionality to handle multipart/signed properly) and a utility 9519 called 'smime' to call all this stuff. This is based on code I 9520 originally wrote for Celo who have kindly allowed it to be 9521 included in OpenSSL. 9522 [Steve Henson] 9523 9524 *) Add variants des_set_key_checked and des_set_key_unchecked of 9525 des_set_key (aka des_key_sched). Global variable des_check_key 9526 decides which of these is called by des_set_key; this way 9527 des_check_key behaves as it always did, but applications and 9528 the library itself, which was buggy for des_check_key == 1, 9529 have a cleaner way to pick the version they need. 9530 [Bodo Moeller] 9531 9532 *) New function PKCS12_newpass() which changes the password of a 9533 PKCS12 structure. 9534 [Steve Henson] 9535 9536 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 9537 dynamic mix. In both cases the ids can be used as an index into the 9538 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 9539 functions so they accept a list of the field values and the 9540 application doesn't need to directly manipulate the X509_TRUST 9541 structure. 9542 [Steve Henson] 9543 9544 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 9545 need initialising. 9546 [Steve Henson] 9547 9548 *) Modify the way the V3 extension code looks up extensions. This now 9549 works in a similar way to the object code: we have some "standard" 9550 extensions in a static table which is searched with OBJ_bsearch() 9551 and the application can add dynamic ones if needed. The file 9552 crypto/x509v3/ext_dat.h now has the info: this file needs to be 9553 updated whenever a new extension is added to the core code and kept 9554 in ext_nid order. There is a simple program 'tabtest.c' which checks 9555 this. New extensions are not added too often so this file can readily 9556 be maintained manually. 9557 9558 There are two big advantages in doing things this way. The extensions 9559 can be looked up immediately and no longer need to be "added" using 9560 X509V3_add_standard_extensions(): this function now does nothing. 9561 [Side note: I get *lots* of email saying the extension code doesn't 9562 work because people forget to call this function] 9563 Also no dynamic allocation is done unless new extensions are added: 9564 so if we don't add custom extensions there is no need to call 9565 X509V3_EXT_cleanup(). 9566 [Steve Henson] 9567 9568 *) Modify enc utility's salting as follows: make salting the default. Add a 9569 magic header, so unsalted files fail gracefully instead of just decrypting 9570 to garbage. This is because not salting is a big security hole, so people 9571 should be discouraged from doing it. 9572 [Ben Laurie] 9573 9574 *) Fixes and enhancements to the 'x509' utility. It allowed a message 9575 digest to be passed on the command line but it only used this 9576 parameter when signing a certificate. Modified so all relevant 9577 operations are affected by the digest parameter including the 9578 -fingerprint and -x509toreq options. Also -x509toreq choked if a 9579 DSA key was used because it didn't fix the digest. 9580 [Steve Henson] 9581 9582 *) Initial certificate chain verify code. Currently tests the untrusted 9583 certificates for consistency with the verify purpose (which is set 9584 when the X509_STORE_CTX structure is set up) and checks the pathlength. 9585 9586 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 9587 this is because it will reject chains with invalid extensions whereas 9588 every previous version of OpenSSL and SSLeay made no checks at all. 9589 9590 Trust code: checks the root CA for the relevant trust settings. Trust 9591 settings have an initial value consistent with the verify purpose: e.g. 9592 if the verify purpose is for SSL client use it expects the CA to be 9593 trusted for SSL client use. However the default value can be changed to 9594 permit custom trust settings: one example of this would be to only trust 9595 certificates from a specific "secure" set of CAs. 9596 9597 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 9598 which should be used for version portability: especially since the 9599 verify structure is likely to change more often now. 9600 9601 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 9602 to set them. If not set then assume SSL clients will verify SSL servers 9603 and vice versa. 9604 9605 Two new options to the verify program: -untrusted allows a set of 9606 untrusted certificates to be passed in and -purpose which sets the 9607 intended purpose of the certificate. If a purpose is set then the 9608 new chain verify code is used to check extension consistency. 9609 [Steve Henson] 9610 9611 *) Support for the authority information access extension. 9612 [Steve Henson] 9613 9614 *) Modify RSA and DSA PEM read routines to transparently handle 9615 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 9616 public keys in a format compatible with certificate 9617 SubjectPublicKeyInfo structures. Unfortunately there were already 9618 functions called *_PublicKey_* which used various odd formats so 9619 these are retained for compatibility: however the DSA variants were 9620 never in a public release so they have been deleted. Changed dsa/rsa 9621 utilities to handle the new format: note no releases ever handled public 9622 keys so we should be OK. 9623 9624 The primary motivation for this change is to avoid the same fiasco 9625 that dogs private keys: there are several incompatible private key 9626 formats some of which are standard and some OpenSSL specific and 9627 require various evil hacks to allow partial transparent handling and 9628 even then it doesn't work with DER formats. Given the option anything 9629 other than PKCS#8 should be dumped: but the other formats have to 9630 stay in the name of compatibility. 9631 9632 With public keys and the benefit of hindsight one standard format 9633 is used which works with EVP_PKEY, RSA or DSA structures: though 9634 it clearly returns an error if you try to read the wrong kind of key. 9635 9636 Added a -pubkey option to the 'x509' utility to output the public key. 9637 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 9638 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 9639 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 9640 that do the same as the EVP_PKEY_assign_*() except they up the 9641 reference count of the added key (they don't "swallow" the 9642 supplied key). 9643 [Steve Henson] 9644 9645 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 9646 CRLs would fail if the file contained no certificates or no CRLs: 9647 added a new function to read in both types and return the number 9648 read: this means that if none are read it will be an error. The 9649 DER versions of the certificate and CRL reader would always fail 9650 because it isn't possible to mix certificates and CRLs in DER format 9651 without choking one or the other routine. Changed this to just read 9652 a certificate: this is the best we can do. Also modified the code 9653 in apps/verify.c to take notice of return codes: it was previously 9654 attempting to read in certificates from NULL pointers and ignoring 9655 any errors: this is one reason why the cert and CRL reader seemed 9656 to work. It doesn't check return codes from the default certificate 9657 routines: these may well fail if the certificates aren't installed. 9658 [Steve Henson] 9659 9660 *) Code to support otherName option in GeneralName. 9661 [Steve Henson] 9662 9663 *) First update to verify code. Change the verify utility 9664 so it warns if it is passed a self signed certificate: 9665 for consistency with the normal behaviour. X509_verify 9666 has been modified to it will now verify a self signed 9667 certificate if *exactly* the same certificate appears 9668 in the store: it was previously impossible to trust a 9669 single self signed certificate. This means that: 9670 openssl verify ss.pem 9671 now gives a warning about a self signed certificate but 9672 openssl verify -CAfile ss.pem ss.pem 9673 is OK. 9674 [Steve Henson] 9675 9676 *) For servers, store verify_result in SSL_SESSION data structure 9677 (and add it to external session representation). 9678 This is needed when client certificate verifications fails, 9679 but an application-provided verification callback (set by 9680 SSL_CTX_set_cert_verify_callback) allows accepting the session 9681 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 9682 but returns 1): When the session is reused, we have to set 9683 ssl->verify_result to the appropriate error code to avoid 9684 security holes. 9685 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 9686 9687 *) Fix a bug in the new PKCS#7 code: it didn't consider the 9688 case in PKCS7_dataInit() where the signed PKCS7 structure 9689 didn't contain any existing data because it was being created. 9690 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 9691 9692 *) Add a salt to the key derivation routines in enc.c. This 9693 forms the first 8 bytes of the encrypted file. Also add a 9694 -S option to allow a salt to be input on the command line. 9695 [Steve Henson] 9696 9697 *) New function X509_cmp(). Oddly enough there wasn't a function 9698 to compare two certificates. We do this by working out the SHA1 9699 hash and comparing that. X509_cmp() will be needed by the trust 9700 code. 9701 [Steve Henson] 9702 9703 *) SSL_get1_session() is like SSL_get_session(), but increments 9704 the reference count in the SSL_SESSION returned. 9705 [Geoff Thorpe <geoff@eu.c2.net>] 9706 9707 *) Fix for 'req': it was adding a null to request attributes. 9708 Also change the X509_LOOKUP and X509_INFO code to handle 9709 certificate auxiliary information. 9710 [Steve Henson] 9711 9712 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 9713 the 'enc' command. 9714 [Steve Henson] 9715 9716 *) Add the possibility to add extra information to the memory leak 9717 detecting output, to form tracebacks, showing from where each 9718 allocation was originated: CRYPTO_push_info("constant string") adds 9719 the string plus current file name and line number to a per-thread 9720 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 9721 is like calling CYRPTO_pop_info() until the stack is empty. 9722 Also updated memory leak detection code to be multi-thread-safe. 9723 [Richard Levitte] 9724 9725 *) Add options -text and -noout to pkcs7 utility and delete the 9726 encryption options which never did anything. Update docs. 9727 [Steve Henson] 9728 9729 *) Add options to some of the utilities to allow the pass phrase 9730 to be included on either the command line (not recommended on 9731 OSes like Unix) or read from the environment. Update the 9732 manpages and fix a few bugs. 9733 [Steve Henson] 9734 9735 *) Add a few manpages for some of the openssl commands. 9736 [Steve Henson] 9737 9738 *) Fix the -revoke option in ca. It was freeing up memory twice, 9739 leaking and not finding already revoked certificates. 9740 [Steve Henson] 9741 9742 *) Extensive changes to support certificate auxiliary information. 9743 This involves the use of X509_CERT_AUX structure and X509_AUX 9744 functions. An X509_AUX function such as PEM_read_X509_AUX() 9745 can still read in a certificate file in the usual way but it 9746 will also read in any additional "auxiliary information". By 9747 doing things this way a fair degree of compatibility can be 9748 retained: existing certificates can have this information added 9749 using the new 'x509' options. 9750 9751 Current auxiliary information includes an "alias" and some trust 9752 settings. The trust settings will ultimately be used in enhanced 9753 certificate chain verification routines: currently a certificate 9754 can only be trusted if it is self signed and then it is trusted 9755 for all purposes. 9756 [Steve Henson] 9757 9758 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 9759 The problem was that one of the replacement routines had not been working 9760 since SSLeay releases. For now the offending routine has been replaced 9761 with non-optimised assembler. Even so, this now gives around 95% 9762 performance improvement for 1024 bit RSA signs. 9763 [Mark Cox] 9764 9765 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 9766 handling. Most clients have the effective key size in bits equal to 9767 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 9768 A few however don't do this and instead use the size of the decrypted key 9769 to determine the RC2 key length and the AlgorithmIdentifier to determine 9770 the effective key length. In this case the effective key length can still 9771 be 40 bits but the key length can be 168 bits for example. This is fixed 9772 by manually forcing an RC2 key into the EVP_PKEY structure because the 9773 EVP code can't currently handle unusual RC2 key sizes: it always assumes 9774 the key length and effective key length are equal. 9775 [Steve Henson] 9776 9777 *) Add a bunch of functions that should simplify the creation of 9778 X509_NAME structures. Now you should be able to do: 9779 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 9780 and have it automatically work out the correct field type and fill in 9781 the structures. The more adventurous can try: 9782 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 9783 and it will (hopefully) work out the correct multibyte encoding. 9784 [Steve Henson] 9785 9786 *) Change the 'req' utility to use the new field handling and multibyte 9787 copy routines. Before the DN field creation was handled in an ad hoc 9788 way in req, ca, and x509 which was rather broken and didn't support 9789 BMPStrings or UTF8Strings. Since some software doesn't implement 9790 BMPStrings or UTF8Strings yet, they can be enabled using the config file 9791 using the dirstring_type option. See the new comment in the default 9792 openssl.cnf for more info. 9793 [Steve Henson] 9794 9795 *) Make crypto/rand/md_rand.c more robust: 9796 - Assure unique random numbers after fork(). 9797 - Make sure that concurrent threads access the global counter and 9798 md serializably so that we never lose entropy in them 9799 or use exactly the same state in multiple threads. 9800 Access to the large state is not always serializable because 9801 the additional locking could be a performance killer, and 9802 md should be large enough anyway. 9803 [Bodo Moeller] 9804 9805 *) New file apps/app_rand.c with commonly needed functionality 9806 for handling the random seed file. 9807 9808 Use the random seed file in some applications that previously did not: 9809 ca, 9810 dsaparam -genkey (which also ignored its '-rand' option), 9811 s_client, 9812 s_server, 9813 x509 (when signing). 9814 Except on systems with /dev/urandom, it is crucial to have a random 9815 seed file at least for key creation, DSA signing, and for DH exchanges; 9816 for RSA signatures we could do without one. 9817 9818 gendh and gendsa (unlike genrsa) used to read only the first byte 9819 of each file listed in the '-rand' option. The function as previously 9820 found in genrsa is now in app_rand.c and is used by all programs 9821 that support '-rand'. 9822 [Bodo Moeller] 9823 9824 *) In RAND_write_file, use mode 0600 for creating files; 9825 don't just chmod when it may be too late. 9826 [Bodo Moeller] 9827 9828 *) Report an error from X509_STORE_load_locations 9829 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 9830 [Bill Perry] 9831 9832 *) New function ASN1_mbstring_copy() this copies a string in either 9833 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 9834 into an ASN1_STRING type. A mask of permissible types is passed 9835 and it chooses the "minimal" type to use or an error if not type 9836 is suitable. 9837 [Steve Henson] 9838 9839 *) Add function equivalents to the various macros in asn1.h. The old 9840 macros are retained with an M_ prefix. Code inside the library can 9841 use the M_ macros. External code (including the openssl utility) 9842 should *NOT* in order to be "shared library friendly". 9843 [Steve Henson] 9844 9845 *) Add various functions that can check a certificate's extensions 9846 to see if it usable for various purposes such as SSL client, 9847 server or S/MIME and CAs of these types. This is currently 9848 VERY EXPERIMENTAL but will ultimately be used for certificate chain 9849 verification. Also added a -purpose flag to x509 utility to 9850 print out all the purposes. 9851 [Steve Henson] 9852 9853 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 9854 functions. 9855 [Steve Henson] 9856 9857 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 9858 for, obtain and decode and extension and obtain its critical flag. 9859 This allows all the necessary extension code to be handled in a 9860 single function call. 9861 [Steve Henson] 9862 9863 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 9864 platforms. See crypto/rc4/rc4_enc.c for further details. 9865 [Andy Polyakov] 9866 9867 *) New -noout option to asn1parse. This causes no output to be produced 9868 its main use is when combined with -strparse and -out to extract data 9869 from a file (which may not be in ASN.1 format). 9870 [Steve Henson] 9871 9872 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 9873 when producing the local key id. 9874 [Richard Levitte <levitte@stacken.kth.se>] 9875 9876 *) New option -dhparam in s_server. This allows a DH parameter file to be 9877 stated explicitly. If it is not stated then it tries the first server 9878 certificate file. The previous behaviour hard coded the filename 9879 "server.pem". 9880 [Steve Henson] 9881 9882 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 9883 a public key to be input or output. For example: 9884 openssl rsa -in key.pem -pubout -out pubkey.pem 9885 Also added necessary DSA public key functions to handle this. 9886 [Steve Henson] 9887 9888 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 9889 in the message. This was handled by allowing 9890 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 9891 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 9892 9893 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 9894 to the end of the strings whereas this didn't. This would cause problems 9895 if strings read with d2i_ASN1_bytes() were later modified. 9896 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 9897 9898 *) Fix for base64 decode bug. When a base64 bio reads only one line of 9899 data and it contains EOF it will end up returning an error. This is 9900 caused by input 46 bytes long. The cause is due to the way base64 9901 BIOs find the start of base64 encoded data. They do this by trying a 9902 trial decode on each line until they find one that works. When they 9903 do a flag is set and it starts again knowing it can pass all the 9904 data directly through the decoder. Unfortunately it doesn't reset 9905 the context it uses. This means that if EOF is reached an attempt 9906 is made to pass two EOFs through the context and this causes the 9907 resulting error. This can also cause other problems as well. As is 9908 usual with these problems it takes *ages* to find and the fix is 9909 trivial: move one line. 9910 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 9911 9912 *) Ugly workaround to get s_client and s_server working under Windows. The 9913 old code wouldn't work because it needed to select() on sockets and the 9914 tty (for keypresses and to see if data could be written). Win32 only 9915 supports select() on sockets so we select() with a 1s timeout on the 9916 sockets and then see if any characters are waiting to be read, if none 9917 are present then we retry, we also assume we can always write data to 9918 the tty. This isn't nice because the code then blocks until we've 9919 received a complete line of data and it is effectively polling the 9920 keyboard at 1s intervals: however it's quite a bit better than not 9921 working at all :-) A dedicated Windows application might handle this 9922 with an event loop for example. 9923 [Steve Henson] 9924 9925 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 9926 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 9927 will be called when RSA_sign() and RSA_verify() are used. This is useful 9928 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 9929 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 9930 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 9931 This necessitated the support of an extra signature type NID_md5_sha1 9932 for SSL signatures and modifications to the SSL library to use it instead 9933 of calling RSA_public_decrypt() and RSA_private_encrypt(). 9934 [Steve Henson] 9935 9936 *) Add new -verify -CAfile and -CApath options to the crl program, these 9937 will lookup a CRL issuers certificate and verify the signature in a 9938 similar way to the verify program. Tidy up the crl program so it 9939 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 9940 less strict. It will now permit CRL extensions even if it is not 9941 a V2 CRL: this will allow it to tolerate some broken CRLs. 9942 [Steve Henson] 9943 9944 *) Initialize all non-automatic variables each time one of the openssl 9945 sub-programs is started (this is necessary as they may be started 9946 multiple times from the "OpenSSL>" prompt). 9947 [Lennart Bang, Bodo Moeller] 9948 9949 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 9950 removing all other RSA functionality (this is what NO_RSA does). This 9951 is so (for example) those in the US can disable those operations covered 9952 by the RSA patent while allowing storage and parsing of RSA keys and RSA 9953 key generation. 9954 [Steve Henson] 9955 9956 *) Non-copying interface to BIO pairs. 9957 (still largely untested) 9958 [Bodo Moeller] 9959 9960 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 9961 ASCII string. This was handled independently in various places before. 9962 [Steve Henson] 9963 9964 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 9965 UTF8 strings a character at a time. 9966 [Steve Henson] 9967 9968 *) Use client_version from client hello to select the protocol 9969 (s23_srvr.c) and for RSA client key exchange verification 9970 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 9971 [Bodo Moeller] 9972 9973 *) Add various utility functions to handle SPKACs, these were previously 9974 handled by poking round in the structure internals. Added new function 9975 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 9976 print, verify and generate SPKACs. Based on an original idea from 9977 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 9978 [Steve Henson] 9979 9980 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 9981 [Andy Polyakov] 9982 9983 *) Allow the config file extension section to be overwritten on the 9984 command line. Based on an original idea from Massimiliano Pala 9985 <madwolf@comune.modena.it>. The new option is called -extensions 9986 and can be applied to ca, req and x509. Also -reqexts to override 9987 the request extensions in req and -crlexts to override the crl extensions 9988 in ca. 9989 [Steve Henson] 9990 9991 *) Add new feature to the SPKAC handling in ca. Now you can include 9992 the same field multiple times by preceding it by "XXXX." for example: 9993 1.OU="Unit name 1" 9994 2.OU="Unit name 2" 9995 this is the same syntax as used in the req config file. 9996 [Steve Henson] 9997 9998 *) Allow certificate extensions to be added to certificate requests. These 9999 are specified in a 'req_extensions' option of the req section of the 10000 config file. They can be printed out with the -text option to req but 10001 are otherwise ignored at present. 10002 [Steve Henson] 10003 10004 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 10005 data read consists of only the final block it would not decrypted because 10006 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 10007 A misplaced 'break' also meant the decrypted final block might not be 10008 copied until the next read. 10009 [Steve Henson] 10010 10011 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 10012 a few extra parameters to the DH structure: these will be useful if 10013 for example we want the value of 'q' or implement X9.42 DH. 10014 [Steve Henson] 10015 10016 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 10017 provides hooks that allow the default DSA functions or functions on a 10018 "per key" basis to be replaced. This allows hardware acceleration and 10019 hardware key storage to be handled without major modification to the 10020 library. Also added low level modexp hooks and CRYPTO_EX structure and 10021 associated functions. 10022 [Steve Henson] 10023 10024 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 10025 as "read only": it can't be written to and the buffer it points to will 10026 not be freed. Reading from a read only BIO is much more efficient than 10027 a normal memory BIO. This was added because there are several times when 10028 an area of memory needs to be read from a BIO. The previous method was 10029 to create a memory BIO and write the data to it, this results in two 10030 copies of the data and an O(n^2) reading algorithm. There is a new 10031 function BIO_new_mem_buf() which creates a read only memory BIO from 10032 an area of memory. Also modified the PKCS#7 routines to use read only 10033 memory BIOs. 10034 [Steve Henson] 10035 10036 *) Bugfix: ssl23_get_client_hello did not work properly when called in 10037 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 10038 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 10039 but a retry condition occured while trying to read the rest. 10040 [Bodo Moeller] 10041 10042 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 10043 NID_pkcs7_encrypted by default: this was wrong since this should almost 10044 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 10045 the encrypted data type: this is a more sensible place to put it and it 10046 allows the PKCS#12 code to be tidied up that duplicated this 10047 functionality. 10048 [Steve Henson] 10049 10050 *) Changed obj_dat.pl script so it takes its input and output files on 10051 the command line. This should avoid shell escape redirection problems 10052 under Win32. 10053 [Steve Henson] 10054 10055 *) Initial support for certificate extension requests, these are included 10056 in things like Xenroll certificate requests. Included functions to allow 10057 extensions to be obtained and added. 10058 [Steve Henson] 10059 10060 *) -crlf option to s_client and s_server for sending newlines as 10061 CRLF (as required by many protocols). 10062 [Bodo Moeller] 10063 10064 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 10065 10066 *) Install libRSAglue.a when OpenSSL is built with RSAref. 10067 [Ralf S. Engelschall] 10068 10069 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 10070 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 10071 10072 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 10073 program. 10074 [Steve Henson] 10075 10076 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 10077 DH parameters/keys (q is lost during that conversion, but the resulting 10078 DH parameters contain its length). 10079 10080 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 10081 much faster than DH_generate_parameters (which creates parameters 10082 where p = 2*q + 1), and also the smaller q makes DH computations 10083 much more efficient (160-bit exponentiation instead of 1024-bit 10084 exponentiation); so this provides a convenient way to support DHE 10085 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 10086 utter importance to use 10087 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 10088 or 10089 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 10090 when such DH parameters are used, because otherwise small subgroup 10091 attacks may become possible! 10092 [Bodo Moeller] 10093 10094 *) Avoid memory leak in i2d_DHparams. 10095 [Bodo Moeller] 10096 10097 *) Allow the -k option to be used more than once in the enc program: 10098 this allows the same encrypted message to be read by multiple recipients. 10099 [Steve Henson] 10100 10101 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 10102 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 10103 it will always use the numerical form of the OID, even if it has a short 10104 or long name. 10105 [Steve Henson] 10106 10107 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 10108 method only got called if p,q,dmp1,dmq1,iqmp components were present, 10109 otherwise bn_mod_exp was called. In the case of hardware keys for example 10110 no private key components need be present and it might store extra data 10111 in the RSA structure, which cannot be accessed from bn_mod_exp. 10112 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 10113 private key operations. 10114 [Steve Henson] 10115 10116 *) Added support for SPARC Linux. 10117 [Andy Polyakov] 10118 10119 *) pem_password_cb function type incompatibly changed from 10120 typedef int pem_password_cb(char *buf, int size, int rwflag); 10121 to 10122 ....(char *buf, int size, int rwflag, void *userdata); 10123 so that applications can pass data to their callbacks: 10124 The PEM[_ASN1]_{read,write}... functions and macros now take an 10125 additional void * argument, which is just handed through whenever 10126 the password callback is called. 10127 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 10128 10129 New function SSL_CTX_set_default_passwd_cb_userdata. 10130 10131 Compatibility note: As many C implementations push function arguments 10132 onto the stack in reverse order, the new library version is likely to 10133 interoperate with programs that have been compiled with the old 10134 pem_password_cb definition (PEM_whatever takes some data that 10135 happens to be on the stack as its last argument, and the callback 10136 just ignores this garbage); but there is no guarantee whatsoever that 10137 this will work. 10138 10139 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 10140 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 10141 problems not only on Windows, but also on some Unix platforms. 10142 To avoid problematic command lines, these definitions are now in an 10143 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 10144 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 10145 [Bodo Moeller] 10146 10147 *) MIPS III/IV assembler module is reimplemented. 10148 [Andy Polyakov] 10149 10150 *) More DES library cleanups: remove references to srand/rand and 10151 delete an unused file. 10152 [Ulf M��ller] 10153 10154 *) Add support for the the free Netwide assembler (NASM) under Win32, 10155 since not many people have MASM (ml) and it can be hard to obtain. 10156 This is currently experimental but it seems to work OK and pass all 10157 the tests. Check out INSTALL.W32 for info. 10158 [Steve Henson] 10159 10160 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 10161 without temporary keys kept an extra copy of the server key, 10162 and connections with temporary keys did not free everything in case 10163 of an error. 10164 [Bodo Moeller] 10165 10166 *) New function RSA_check_key and new openssl rsa option -check 10167 for verifying the consistency of RSA keys. 10168 [Ulf Moeller, Bodo Moeller] 10169 10170 *) Various changes to make Win32 compile work: 10171 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 10172 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 10173 comparison" warnings. 10174 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 10175 [Steve Henson] 10176 10177 *) Add a debugging option to PKCS#5 v2 key generation function: when 10178 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 10179 derived keys are printed to stderr. 10180 [Steve Henson] 10181 10182 *) Copy the flags in ASN1_STRING_dup(). 10183 [Roman E. Pavlov <pre@mo.msk.ru>] 10184 10185 *) The x509 application mishandled signing requests containing DSA 10186 keys when the signing key was also DSA and the parameters didn't match. 10187 10188 It was supposed to omit the parameters when they matched the signing key: 10189 the verifying software was then supposed to automatically use the CA's 10190 parameters if they were absent from the end user certificate. 10191 10192 Omitting parameters is no longer recommended. The test was also 10193 the wrong way round! This was probably due to unusual behaviour in 10194 EVP_cmp_parameters() which returns 1 if the parameters match. 10195 This meant that parameters were omitted when they *didn't* match and 10196 the certificate was useless. Certificates signed with 'ca' didn't have 10197 this bug. 10198 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 10199 10200 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 10201 The interface is as follows: 10202 Applications can use 10203 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 10204 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 10205 "off" is now the default. 10206 The library internally uses 10207 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 10208 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 10209 to disable memory-checking temporarily. 10210 10211 Some inconsistent states that previously were possible (and were 10212 even the default) are now avoided. 10213 10214 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 10215 with each memory chunk allocated; this is occasionally more helpful 10216 than just having a counter. 10217 10218 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 10219 10220 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 10221 extensions. 10222 [Bodo Moeller] 10223 10224 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 10225 which largely parallels "options", but is for changing API behaviour, 10226 whereas "options" are about protocol behaviour. 10227 Initial "mode" flags are: 10228 10229 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 10230 a single record has been written. 10231 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 10232 retries use the same buffer location. 10233 (But all of the contents must be 10234 copied!) 10235 [Bodo Moeller] 10236 10237 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 10238 worked. 10239 10240 *) Fix problems with no-hmac etc. 10241 [Ulf M��ller, pointed out by Brian Wellington <bwelling@tislabs.com>] 10242 10243 *) New functions RSA_get_default_method(), RSA_set_method() and 10244 RSA_get_method(). These allows replacement of RSA_METHODs without having 10245 to mess around with the internals of an RSA structure. 10246 [Steve Henson] 10247 10248 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 10249 Also really enable memory leak checks in openssl.c and in some 10250 test programs. 10251 [Chad C. Mulligan, Bodo Moeller] 10252 10253 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 10254 up the length of negative integers. This has now been simplified to just 10255 store the length when it is first determined and use it later, rather 10256 than trying to keep track of where data is copied and updating it to 10257 point to the end. 10258 [Steve Henson, reported by Brien Wheeler 10259 <bwheeler@authentica-security.com>] 10260 10261 *) Add a new function PKCS7_signatureVerify. This allows the verification 10262 of a PKCS#7 signature but with the signing certificate passed to the 10263 function itself. This contrasts with PKCS7_dataVerify which assumes the 10264 certificate is present in the PKCS#7 structure. This isn't always the 10265 case: certificates can be omitted from a PKCS#7 structure and be 10266 distributed by "out of band" means (such as a certificate database). 10267 [Steve Henson] 10268 10269 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 10270 function prototypes in pem.h, also change util/mkdef.pl to add the 10271 necessary function names. 10272 [Steve Henson] 10273 10274 *) mk1mf.pl (used by Windows builds) did not properly read the 10275 options set by Configure in the top level Makefile, and Configure 10276 was not even able to write more than one option correctly. 10277 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 10278 [Bodo Moeller] 10279 10280 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 10281 file to be loaded from a BIO or FILE pointer. The BIO version will 10282 for example allow memory BIOs to contain config info. 10283 [Steve Henson] 10284 10285 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 10286 Whoever hopes to achieve shared-library compatibility across versions 10287 must use this, not the compile-time macro. 10288 (Exercise 0.9.4: Which is the minimum library version required by 10289 such programs?) 10290 Note: All this applies only to multi-threaded programs, others don't 10291 need locks. 10292 [Bodo Moeller] 10293 10294 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 10295 through a BIO pair triggered the default case, i.e. 10296 SSLerr(...,SSL_R_UNKNOWN_STATE). 10297 [Bodo Moeller] 10298 10299 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 10300 can use the SSL library even if none of the specific BIOs is 10301 appropriate. 10302 [Bodo Moeller] 10303 10304 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 10305 for the encoded length. 10306 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 10307 10308 *) Add initial documentation of the X509V3 functions. 10309 [Steve Henson] 10310 10311 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 10312 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 10313 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 10314 secure PKCS#8 private key format with a high iteration count. 10315 [Steve Henson] 10316 10317 *) Fix determination of Perl interpreter: A perl or perl5 10318 _directory_ in $PATH was also accepted as the interpreter. 10319 [Ralf S. Engelschall] 10320 10321 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 10322 wrong with it but it was very old and did things like calling 10323 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 10324 unusual formatting. 10325 [Steve Henson] 10326 10327 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 10328 to use the new extension code. 10329 [Steve Henson] 10330 10331 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 10332 with macros. This should make it easier to change their form, add extra 10333 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 10334 constant. 10335 [Steve Henson] 10336 10337 *) Add to configuration table a new entry that can specify an alternative 10338 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 10339 according to Mark Crispin <MRC@Panda.COM>. 10340 [Bodo Moeller] 10341 10342#if 0 10343 *) DES CBC did not update the IV. Weird. 10344 [Ben Laurie] 10345#else 10346 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 10347 Changing the behaviour of the former might break existing programs -- 10348 where IV updating is needed, des_ncbc_encrypt can be used. 10349#endif 10350 10351 *) When bntest is run from "make test" it drives bc to check its 10352 calculations, as well as internally checking them. If an internal check 10353 fails, it needs to cause bc to give a non-zero result or make test carries 10354 on without noticing the failure. Fixed. 10355 [Ben Laurie] 10356 10357 *) DES library cleanups. 10358 [Ulf M��ller] 10359 10360 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 10361 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 10362 ciphers. NOTE: although the key derivation function has been verified 10363 against some published test vectors it has not been extensively tested 10364 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 10365 of v2.0. 10366 [Steve Henson] 10367 10368 *) Instead of "mkdir -p", which is not fully portable, use new 10369 Perl script "util/mkdir-p.pl". 10370 [Bodo Moeller] 10371 10372 *) Rewrite the way password based encryption (PBE) is handled. It used to 10373 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 10374 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 10375 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 10376 the 'parameter' field of the AlgorithmIdentifier is passed to the 10377 underlying key generation function so it must do its own ASN1 parsing. 10378 This has also changed the EVP_PBE_CipherInit() function which now has a 10379 'parameter' argument instead of literal salt and iteration count values 10380 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 10381 [Steve Henson] 10382 10383 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 10384 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 10385 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 10386 KEY" because this clashed with PKCS#8 unencrypted string. Since this 10387 value was just used as a "magic string" and not used directly its 10388 value doesn't matter. 10389 [Steve Henson] 10390 10391 *) Introduce some semblance of const correctness to BN. Shame C doesn't 10392 support mutable. 10393 [Ben Laurie] 10394 10395 *) "linux-sparc64" configuration (ultrapenguin). 10396 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 10397 "linux-sparc" configuration. 10398 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 10399 10400 *) config now generates no-xxx options for missing ciphers. 10401 [Ulf M��ller] 10402 10403 *) Support the EBCDIC character set (work in progress). 10404 File ebcdic.c not yet included because it has a different license. 10405 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10406 10407 *) Support BS2000/OSD-POSIX. 10408 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10409 10410 *) Make callbacks for key generation use void * instead of char *. 10411 [Ben Laurie] 10412 10413 *) Make S/MIME samples compile (not yet tested). 10414 [Ben Laurie] 10415 10416 *) Additional typesafe stacks. 10417 [Ben Laurie] 10418 10419 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 10420 [Bodo Moeller] 10421 10422 10423 Changes between 0.9.3 and 0.9.3a [29 May 1999] 10424 10425 *) New configuration variant "sco5-gcc". 10426 10427 *) Updated some demos. 10428 [Sean O Riordain, Wade Scholine] 10429 10430 *) Add missing BIO_free at exit of pkcs12 application. 10431 [Wu Zhigang] 10432 10433 *) Fix memory leak in conf.c. 10434 [Steve Henson] 10435 10436 *) Updates for Win32 to assembler version of MD5. 10437 [Steve Henson] 10438 10439 *) Set #! path to perl in apps/der_chop to where we found it 10440 instead of using a fixed path. 10441 [Bodo Moeller] 10442 10443 *) SHA library changes for irix64-mips4-cc. 10444 [Andy Polyakov] 10445 10446 *) Improvements for VMS support. 10447 [Richard Levitte] 10448 10449 10450 Changes between 0.9.2b and 0.9.3 [24 May 1999] 10451 10452 *) Bignum library bug fix. IRIX 6 passes "make test" now! 10453 This also avoids the problems with SC4.2 and unpatched SC5. 10454 [Andy Polyakov <appro@fy.chalmers.se>] 10455 10456 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 10457 These are required because of the typesafe stack would otherwise break 10458 existing code. If old code used a structure member which used to be STACK 10459 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 10460 sk_num or sk_value it would produce an error because the num, data members 10461 are not present in STACK_OF. Now it just produces a warning. sk_set 10462 replaces the old method of assigning a value to sk_value 10463 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 10464 that does this will no longer work (and should use sk_set instead) but 10465 this could be regarded as a "questionable" behaviour anyway. 10466 [Steve Henson] 10467 10468 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 10469 correctly handle encrypted S/MIME data. 10470 [Steve Henson] 10471 10472 *) Change type of various DES function arguments from des_cblock 10473 (which means, in function argument declarations, pointer to char) 10474 to des_cblock * (meaning pointer to array with 8 char elements), 10475 which allows the compiler to do more typechecking; it was like 10476 that back in SSLeay, but with lots of ugly casts. 10477 10478 Introduce new type const_des_cblock. 10479 [Bodo Moeller] 10480 10481 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 10482 problems: find RecipientInfo structure that matches recipient certificate 10483 and initialise the ASN1 structures properly based on passed cipher. 10484 [Steve Henson] 10485 10486 *) Belatedly make the BN tests actually check the results. 10487 [Ben Laurie] 10488 10489 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 10490 to and from BNs: it was completely broken. New compilation option 10491 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 10492 key elements as negative integers. 10493 [Steve Henson] 10494 10495 *) Reorganize and speed up MD5. 10496 [Andy Polyakov <appro@fy.chalmers.se>] 10497 10498 *) VMS support. 10499 [Richard Levitte <richard@levitte.org>] 10500 10501 *) New option -out to asn1parse to allow the parsed structure to be 10502 output to a file. This is most useful when combined with the -strparse 10503 option to examine the output of things like OCTET STRINGS. 10504 [Steve Henson] 10505 10506 *) Make SSL library a little more fool-proof by not requiring any longer 10507 that SSL_set_{accept,connect}_state be called before 10508 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 10509 in many applications because usually everything *appeared* to work as 10510 intended anyway -- now it really works as intended). 10511 [Bodo Moeller] 10512 10513 *) Move openssl.cnf out of lib/. 10514 [Ulf M��ller] 10515 10516 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 10517 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 10518 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 10519 [Ralf S. Engelschall] 10520 10521 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 10522 handle PKCS#7 enveloped data properly. 10523 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 10524 10525 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 10526 copying pointers. The cert_st handling is changed by this in 10527 various ways (and thus what used to be known as ctx->default_cert 10528 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 10529 any longer when s->cert does not give us what we need). 10530 ssl_cert_instantiate becomes obsolete by this change. 10531 As soon as we've got the new code right (possibly it already is?), 10532 we have solved a couple of bugs of the earlier code where s->cert 10533 was used as if it could not have been shared with other SSL structures. 10534 10535 Note that using the SSL API in certain dirty ways now will result 10536 in different behaviour than observed with earlier library versions: 10537 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 10538 does not influence s as it used to. 10539 10540 In order to clean up things more thoroughly, inside SSL_SESSION 10541 we don't use CERT any longer, but a new structure SESS_CERT 10542 that holds per-session data (if available); currently, this is 10543 the peer's certificate chain and, for clients, the server's certificate 10544 and temporary key. CERT holds only those values that can have 10545 meaningful defaults in an SSL_CTX. 10546 [Bodo Moeller] 10547 10548 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 10549 from the internal representation. Various PKCS#7 fixes: remove some 10550 evil casts and set the enc_dig_alg field properly based on the signing 10551 key type. 10552 [Steve Henson] 10553 10554 *) Allow PKCS#12 password to be set from the command line or the 10555 environment. Let 'ca' get its config file name from the environment 10556 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 10557 and 'x509'). 10558 [Steve Henson] 10559 10560 *) Allow certificate policies extension to use an IA5STRING for the 10561 organization field. This is contrary to the PKIX definition but 10562 VeriSign uses it and IE5 only recognises this form. Document 'x509' 10563 extension option. 10564 [Steve Henson] 10565 10566 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 10567 without disallowing inline assembler and the like for non-pedantic builds. 10568 [Ben Laurie] 10569 10570 *) Support Borland C++ builder. 10571 [Janez Jere <jj@void.si>, modified by Ulf M��ller] 10572 10573 *) Support Mingw32. 10574 [Ulf M��ller] 10575 10576 *) SHA-1 cleanups and performance enhancements. 10577 [Andy Polyakov <appro@fy.chalmers.se>] 10578 10579 *) Sparc v8plus assembler for the bignum library. 10580 [Andy Polyakov <appro@fy.chalmers.se>] 10581 10582 *) Accept any -xxx and +xxx compiler options in Configure. 10583 [Ulf M��ller] 10584 10585 *) Update HPUX configuration. 10586 [Anonymous] 10587 10588 *) Add missing sk_<type>_unshift() function to safestack.h 10589 [Ralf S. Engelschall] 10590 10591 *) New function SSL_CTX_use_certificate_chain_file that sets the 10592 "extra_cert"s in addition to the certificate. (This makes sense 10593 only for "PEM" format files, as chains as a whole are not 10594 DER-encoded.) 10595 [Bodo Moeller] 10596 10597 *) Support verify_depth from the SSL API. 10598 x509_vfy.c had what can be considered an off-by-one-error: 10599 Its depth (which was not part of the external interface) 10600 was actually counting the number of certificates in a chain; 10601 now it really counts the depth. 10602 [Bodo Moeller] 10603 10604 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 10605 instead of X509err, which often resulted in confusing error 10606 messages since the error codes are not globally unique 10607 (e.g. an alleged error in ssl3_accept when a certificate 10608 didn't match the private key). 10609 10610 *) New function SSL_CTX_set_session_id_context that allows to set a default 10611 value (so that you don't need SSL_set_session_id_context for each 10612 connection using the SSL_CTX). 10613 [Bodo Moeller] 10614 10615 *) OAEP decoding bug fix. 10616 [Ulf M��ller] 10617 10618 *) Support INSTALL_PREFIX for package builders, as proposed by 10619 David Harris. 10620 [Bodo Moeller] 10621 10622 *) New Configure options "threads" and "no-threads". For systems 10623 where the proper compiler options are known (currently Solaris 10624 and Linux), "threads" is the default. 10625 [Bodo Moeller] 10626 10627 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 10628 [Bodo Moeller] 10629 10630 *) Install various scripts to $(OPENSSLDIR)/misc, not to 10631 $(INSTALLTOP)/bin -- they shouldn't clutter directories 10632 such as /usr/local/bin. 10633 [Bodo Moeller] 10634 10635 *) "make linux-shared" to build shared libraries. 10636 [Niels Poppe <niels@netbox.org>] 10637 10638 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 10639 [Ulf M��ller] 10640 10641 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 10642 extension adding in x509 utility. 10643 [Steve Henson] 10644 10645 *) Remove NOPROTO sections and error code comments. 10646 [Ulf M��ller] 10647 10648 *) Partial rewrite of the DEF file generator to now parse the ANSI 10649 prototypes. 10650 [Steve Henson] 10651 10652 *) New Configure options --prefix=DIR and --openssldir=DIR. 10653 [Ulf M��ller] 10654 10655 *) Complete rewrite of the error code script(s). It is all now handled 10656 by one script at the top level which handles error code gathering, 10657 header rewriting and C source file generation. It should be much better 10658 than the old method: it now uses a modified version of Ulf's parser to 10659 read the ANSI prototypes in all header files (thus the old K&R definitions 10660 aren't needed for error creation any more) and do a better job of 10661 translating function codes into names. The old 'ASN1 error code imbedded 10662 in a comment' is no longer necessary and it doesn't use .err files which 10663 have now been deleted. Also the error code call doesn't have to appear all 10664 on one line (which resulted in some large lines...). 10665 [Steve Henson] 10666 10667 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 10668 [Bodo Moeller] 10669 10670 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 10671 0 (which usually indicates a closed connection), but continue reading. 10672 [Bodo Moeller] 10673 10674 *) Fix some race conditions. 10675 [Bodo Moeller] 10676 10677 *) Add support for CRL distribution points extension. Add Certificate 10678 Policies and CRL distribution points documentation. 10679 [Steve Henson] 10680 10681 *) Move the autogenerated header file parts to crypto/opensslconf.h. 10682 [Ulf M��ller] 10683 10684 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 10685 8 of keying material. Merlin has also confirmed interop with this fix 10686 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 10687 [Merlin Hughes <merlin@baltimore.ie>] 10688 10689 *) Fix lots of warnings. 10690 [Richard Levitte <levitte@stacken.kth.se>] 10691 10692 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 10693 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 10694 [Richard Levitte <levitte@stacken.kth.se>] 10695 10696 *) Fix problems with sizeof(long) == 8. 10697 [Andy Polyakov <appro@fy.chalmers.se>] 10698 10699 *) Change functions to ANSI C. 10700 [Ulf M��ller] 10701 10702 *) Fix typos in error codes. 10703 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf M��ller] 10704 10705 *) Remove defunct assembler files from Configure. 10706 [Ulf M��ller] 10707 10708 *) SPARC v8 assembler BIGNUM implementation. 10709 [Andy Polyakov <appro@fy.chalmers.se>] 10710 10711 *) Support for Certificate Policies extension: both print and set. 10712 Various additions to support the r2i method this uses. 10713 [Steve Henson] 10714 10715 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 10716 return a const string when you are expecting an allocated buffer. 10717 [Ben Laurie] 10718 10719 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 10720 types DirectoryString and DisplayText. 10721 [Steve Henson] 10722 10723 *) Add code to allow r2i extensions to access the configuration database, 10724 add an LHASH database driver and add several ctx helper functions. 10725 [Steve Henson] 10726 10727 *) Fix an evil bug in bn_expand2() which caused various BN functions to 10728 fail when they extended the size of a BIGNUM. 10729 [Steve Henson] 10730 10731 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 10732 support typesafe stack. 10733 [Steve Henson] 10734 10735 *) Fix typo in SSL_[gs]et_options(). 10736 [Nils Frostberg <nils@medcom.se>] 10737 10738 *) Delete various functions and files that belonged to the (now obsolete) 10739 old X509V3 handling code. 10740 [Steve Henson] 10741 10742 *) New Configure option "rsaref". 10743 [Ulf M��ller] 10744 10745 *) Don't auto-generate pem.h. 10746 [Bodo Moeller] 10747 10748 *) Introduce type-safe ASN.1 SETs. 10749 [Ben Laurie] 10750 10751 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 10752 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 10753 10754 *) Introduce type-safe STACKs. This will almost certainly break lots of code 10755 that links with OpenSSL (well at least cause lots of warnings), but fear 10756 not: the conversion is trivial, and it eliminates loads of evil casts. A 10757 few STACKed things have been converted already. Feel free to convert more. 10758 In the fullness of time, I'll do away with the STACK type altogether. 10759 [Ben Laurie] 10760 10761 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 10762 specified in <certfile> by updating the entry in the index.txt file. 10763 This way one no longer has to edit the index.txt file manually for 10764 revoking a certificate. The -revoke option does the gory details now. 10765 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 10766 10767 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 10768 `-text' option at all and this way the `-noout -text' combination was 10769 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 10770 [Ralf S. Engelschall] 10771 10772 *) Make sure a corresponding plain text error message exists for the 10773 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 10774 verify callback function determined that a certificate was revoked. 10775 [Ralf S. Engelschall] 10776 10777 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 10778 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 10779 all available cipers including rc5, which was forgotten until now. 10780 In order to let the testing shell script know which algorithms 10781 are available, a new (up to now undocumented) command 10782 "openssl list-cipher-commands" is used. 10783 [Bodo Moeller] 10784 10785 *) Bugfix: s_client occasionally would sleep in select() when 10786 it should have checked SSL_pending() first. 10787 [Bodo Moeller] 10788 10789 *) New functions DSA_do_sign and DSA_do_verify to provide access to 10790 the raw DSA values prior to ASN.1 encoding. 10791 [Ulf M��ller] 10792 10793 *) Tweaks to Configure 10794 [Niels Poppe <niels@netbox.org>] 10795 10796 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 10797 yet... 10798 [Steve Henson] 10799 10800 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 10801 [Ulf M��ller] 10802 10803 *) New config option to avoid instructions that are illegal on the 80386. 10804 The default code is faster, but requires at least a 486. 10805 [Ulf M��ller] 10806 10807 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 10808 SSL2_SERVER_VERSION (not used at all) macros, which are now the 10809 same as SSL2_VERSION anyway. 10810 [Bodo Moeller] 10811 10812 *) New "-showcerts" option for s_client. 10813 [Bodo Moeller] 10814 10815 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 10816 application. Various cleanups and fixes. 10817 [Steve Henson] 10818 10819 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 10820 modify error routines to work internally. Add error codes and PBE init 10821 to library startup routines. 10822 [Steve Henson] 10823 10824 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 10825 packing functions to asn1 and evp. Changed function names and error 10826 codes along the way. 10827 [Steve Henson] 10828 10829 *) PKCS12 integration: and so it begins... First of several patches to 10830 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 10831 objects to objects.h 10832 [Steve Henson] 10833 10834 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 10835 and display support for Thawte strong extranet extension. 10836 [Steve Henson] 10837 10838 *) Add LinuxPPC support. 10839 [Jeff Dubrule <igor@pobox.org>] 10840 10841 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 10842 bn_div_words in alpha.s. 10843 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 10844 10845 *) Make sure the RSA OAEP test is skipped under -DRSAref because 10846 OAEP isn't supported when OpenSSL is built with RSAref. 10847 [Ulf Moeller <ulf@fitug.de>] 10848 10849 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 10850 so they no longer are missing under -DNOPROTO. 10851 [Soren S. Jorvang <soren@t.dk>] 10852 10853 10854 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 10855 10856 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 10857 doesn't work when the session is reused. Coming soon! 10858 [Ben Laurie] 10859 10860 *) Fix a security hole, that allows sessions to be reused in the wrong 10861 context thus bypassing client cert protection! All software that uses 10862 client certs and session caches in multiple contexts NEEDS PATCHING to 10863 allow session reuse! A fuller solution is in the works. 10864 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 10865 10866 *) Some more source tree cleanups (removed obsolete files 10867 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 10868 permission on "config" script to be executable) and a fix for the INSTALL 10869 document. 10870 [Ulf Moeller <ulf@fitug.de>] 10871 10872 *) Remove some legacy and erroneous uses of malloc, free instead of 10873 Malloc, Free. 10874 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 10875 10876 *) Make rsa_oaep_test return non-zero on error. 10877 [Ulf Moeller <ulf@fitug.de>] 10878 10879 *) Add support for native Solaris shared libraries. Configure 10880 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 10881 if someone would make that last step automatic. 10882 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 10883 10884 *) ctx_size was not built with the right compiler during "make links". Fixed. 10885 [Ben Laurie] 10886 10887 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 10888 except NULL ciphers". This means the default cipher list will no longer 10889 enable NULL ciphers. They need to be specifically enabled e.g. with 10890 the string "DEFAULT:eNULL". 10891 [Steve Henson] 10892 10893 *) Fix to RSA private encryption routines: if p < q then it would 10894 occasionally produce an invalid result. This will only happen with 10895 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 10896 [Steve Henson] 10897 10898 *) Be less restrictive and allow also `perl util/perlpath.pl 10899 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 10900 because this way one can also use an interpreter named `perl5' (which is 10901 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 10902 installed as `perl'). 10903 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10904 10905 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 10906 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 10907 10908 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 10909 advapi32.lib to Win32 build and change the pem test comparision 10910 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 10911 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 10912 and crypto/des/ede_cbcm_enc.c. 10913 [Steve Henson] 10914 10915 *) DES quad checksum was broken on big-endian architectures. Fixed. 10916 [Ben Laurie] 10917 10918 *) Comment out two functions in bio.h that aren't implemented. Fix up the 10919 Win32 test batch file so it (might) work again. The Win32 test batch file 10920 is horrible: I feel ill.... 10921 [Steve Henson] 10922 10923 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 10924 in e_os.h. Audit of header files to check ANSI and non ANSI 10925 sections: 10 functions were absent from non ANSI section and not exported 10926 from Windows DLLs. Fixed up libeay.num for new functions. 10927 [Steve Henson] 10928 10929 *) Make `openssl version' output lines consistent. 10930 [Ralf S. Engelschall] 10931 10932 *) Fix Win32 symbol export lists for BIO functions: Added 10933 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 10934 to ms/libeay{16,32}.def. 10935 [Ralf S. Engelschall] 10936 10937 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 10938 fine under Unix and passes some trivial tests I've now added. But the 10939 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 10940 added to make sure no one expects that this stuff really works in the 10941 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 10942 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 10943 openssl_bio.xs. 10944 [Ralf S. Engelschall] 10945 10946 *) Fix the generation of two part addresses in perl. 10947 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 10948 10949 *) Add config entry for Linux on MIPS. 10950 [John Tobey <jtobey@channel1.com>] 10951 10952 *) Make links whenever Configure is run, unless we are on Windoze. 10953 [Ben Laurie] 10954 10955 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 10956 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 10957 in CRLs. 10958 [Steve Henson] 10959 10960 *) Add a useful kludge to allow package maintainers to specify compiler and 10961 other platforms details on the command line without having to patch the 10962 Configure script everytime: One now can use ``perl Configure 10963 <id>:<details>'', i.e. platform ids are allowed to have details appended 10964 to them (seperated by colons). This is treated as there would be a static 10965 pre-configured entry in Configure's %table under key <id> with value 10966 <details> and ``perl Configure <id>'' is called. So, when you want to 10967 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 10968 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 10969 now, which overrides the FreeBSD-elf entry on-the-fly. 10970 [Ralf S. Engelschall] 10971 10972 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 10973 [Ben Laurie] 10974 10975 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 10976 on the `perl Configure ...' command line. This way one can compile 10977 OpenSSL libraries with Position Independent Code (PIC) which is needed 10978 for linking it into DSOs. 10979 [Ralf S. Engelschall] 10980 10981 *) Remarkably, export ciphers were totally broken and no-one had noticed! 10982 Fixed. 10983 [Ben Laurie] 10984 10985 *) Cleaned up the LICENSE document: The official contact for any license 10986 questions now is the OpenSSL core team under openssl-core@openssl.org. 10987 And add a paragraph about the dual-license situation to make sure people 10988 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 10989 to the OpenSSL toolkit. 10990 [Ralf S. Engelschall] 10991 10992 *) General source tree makefile cleanups: Made `making xxx in yyy...' 10993 display consistent in the source tree and replaced `/bin/rm' by `rm'. 10994 Additonally cleaned up the `make links' target: Remove unnecessary 10995 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 10996 to speed processing and no longer clutter the display with confusing 10997 stuff. Instead only the actually done links are displayed. 10998 [Ralf S. Engelschall] 10999 11000 *) Permit null encryption ciphersuites, used for authentication only. It used 11001 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 11002 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 11003 encryption. 11004 [Ben Laurie] 11005 11006 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 11007 signed attributes when verifying signatures (this would break them), 11008 the detached data encoding was wrong and public keys obtained using 11009 X509_get_pubkey() weren't freed. 11010 [Steve Henson] 11011 11012 *) Add text documentation for the BUFFER functions. Also added a work around 11013 to a Win95 console bug. This was triggered by the password read stuff: the 11014 last character typed gets carried over to the next fread(). If you were 11015 generating a new cert request using 'req' for example then the last 11016 character of the passphrase would be CR which would then enter the first 11017 field as blank. 11018 [Steve Henson] 11019 11020 *) Added the new `Includes OpenSSL Cryptography Software' button as 11021 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 11022 button and can be used by applications based on OpenSSL to show the 11023 relationship to the OpenSSL project. 11024 [Ralf S. Engelschall] 11025 11026 *) Remove confusing variables in function signatures in files 11027 ssl/ssl_lib.c and ssl/ssl.h. 11028 [Lennart Bong <lob@kulthea.stacken.kth.se>] 11029 11030 *) Don't install bss_file.c under PREFIX/include/ 11031 [Lennart Bong <lob@kulthea.stacken.kth.se>] 11032 11033 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 11034 functions that return function pointers and has support for NT specific 11035 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 11036 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 11037 unsigned to signed types: this was killing the Win32 compile. 11038 [Steve Henson] 11039 11040 *) Add new certificate file to stack functions, 11041 SSL_add_dir_cert_subjects_to_stack() and 11042 SSL_add_file_cert_subjects_to_stack(). These largely supplant 11043 SSL_load_client_CA_file(), and can be used to add multiple certs easily 11044 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 11045 This means that Apache-SSL and similar packages don't have to mess around 11046 to add as many CAs as they want to the preferred list. 11047 [Ben Laurie] 11048 11049 *) Experiment with doxygen documentation. Currently only partially applied to 11050 ssl/ssl_lib.c. 11051 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 11052 openssl.doxy as the configuration file. 11053 [Ben Laurie] 11054 11055 *) Get rid of remaining C++-style comments which strict C compilers hate. 11056 [Ralf S. Engelschall, pointed out by Carlos Amengual] 11057 11058 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 11059 compiled in by default: it has problems with large keys. 11060 [Steve Henson] 11061 11062 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 11063 DH private keys and/or callback functions which directly correspond to 11064 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 11065 is needed for applications which have to configure certificates on a 11066 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 11067 (e.g. s_server). 11068 For the RSA certificate situation is makes no difference, but 11069 for the DSA certificate situation this fixes the "no shared cipher" 11070 problem where the OpenSSL cipher selection procedure failed because the 11071 temporary keys were not overtaken from the context and the API provided 11072 no way to reconfigure them. 11073 The new functions now let applications reconfigure the stuff and they 11074 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 11075 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 11076 non-public-API function ssl_cert_instantiate() is used as a helper 11077 function and also to reduce code redundancy inside ssl_rsa.c. 11078 [Ralf S. Engelschall] 11079 11080 *) Move s_server -dcert and -dkey options out of the undocumented feature 11081 area because they are useful for the DSA situation and should be 11082 recognized by the users. 11083 [Ralf S. Engelschall] 11084 11085 *) Fix the cipher decision scheme for export ciphers: the export bits are 11086 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 11087 SSL_EXP_MASK. So, the original variable has to be used instead of the 11088 already masked variable. 11089 [Richard Levitte <levitte@stacken.kth.se>] 11090 11091 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 11092 [Richard Levitte <levitte@stacken.kth.se>] 11093 11094 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 11095 from `int' to `unsigned int' because it's a length and initialized by 11096 EVP_DigestFinal() which expects an `unsigned int *'. 11097 [Richard Levitte <levitte@stacken.kth.se>] 11098 11099 *) Don't hard-code path to Perl interpreter on shebang line of Configure 11100 script. Instead use the usual Shell->Perl transition trick. 11101 [Ralf S. Engelschall] 11102 11103 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 11104 (in addition to RSA certificates) to match the behaviour of `openssl dsa 11105 -noout -modulus' as it's already the case for `openssl rsa -noout 11106 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 11107 currently the public key is printed (a decision which was already done by 11108 `openssl dsa -modulus' in the past) which serves a similar purpose. 11109 Additionally the NO_RSA no longer completely removes the whole -modulus 11110 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 11111 now, too. 11112 [Ralf S. Engelschall] 11113 11114 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 11115 BIO. See the source (crypto/evp/bio_ok.c) for more info. 11116 [Arne Ansper <arne@ats.cyber.ee>] 11117 11118 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 11119 to be added. Now both 'req' and 'ca' can use new objects defined in the 11120 config file. 11121 [Steve Henson] 11122 11123 *) Add cool BIO that does syslog (or event log on NT). 11124 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 11125 11126 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 11127 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 11128 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 11129 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 11130 [Ben Laurie] 11131 11132 *) Add preliminary config info for new extension code. 11133 [Steve Henson] 11134 11135 *) Make RSA_NO_PADDING really use no padding. 11136 [Ulf Moeller <ulf@fitug.de>] 11137 11138 *) Generate errors when private/public key check is done. 11139 [Ben Laurie] 11140 11141 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 11142 for some CRL extensions and new objects added. 11143 [Steve Henson] 11144 11145 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 11146 key usage extension and fuller support for authority key id. 11147 [Steve Henson] 11148 11149 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 11150 padding method for RSA, which is recommended for new applications in PKCS 11151 #1 v2.0 (RFC 2437, October 1998). 11152 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 11153 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 11154 against Bleichbacher's attack on RSA. 11155 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 11156 Ben Laurie] 11157 11158 *) Updates to the new SSL compression code 11159 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11160 11161 *) Fix so that the version number in the master secret, when passed 11162 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 11163 (because the server will not accept higher), that the version number 11164 is 0x03,0x01, not 0x03,0x00 11165 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11166 11167 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 11168 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 11169 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 11170 [Steve Henson] 11171 11172 *) Support for RAW extensions where an arbitrary extension can be 11173 created by including its DER encoding. See apps/openssl.cnf for 11174 an example. 11175 [Steve Henson] 11176 11177 *) Make sure latest Perl versions don't interpret some generated C array 11178 code as Perl array code in the crypto/err/err_genc.pl script. 11179 [Lars Weber <3weber@informatik.uni-hamburg.de>] 11180 11181 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 11182 not many people have the assembler. Various Win32 compilation fixes and 11183 update to the INSTALL.W32 file with (hopefully) more accurate Win32 11184 build instructions. 11185 [Steve Henson] 11186 11187 *) Modify configure script 'Configure' to automatically create crypto/date.h 11188 file under Win32 and also build pem.h from pem.org. New script 11189 util/mkfiles.pl to create the MINFO file on environments that can't do a 11190 'make files': perl util/mkfiles.pl >MINFO should work. 11191 [Steve Henson] 11192 11193 *) Major rework of DES function declarations, in the pursuit of correctness 11194 and purity. As a result, many evil casts evaporated, and some weirdness, 11195 too. You may find this causes warnings in your code. Zapping your evil 11196 casts will probably fix them. Mostly. 11197 [Ben Laurie] 11198 11199 *) Fix for a typo in asn1.h. Bug fix to object creation script 11200 obj_dat.pl. It considered a zero in an object definition to mean 11201 "end of object": none of the objects in objects.h have any zeros 11202 so it wasn't spotted. 11203 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 11204 11205 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 11206 Masking (CBCM). In the absence of test vectors, the best I have been able 11207 to do is check that the decrypt undoes the encrypt, so far. Send me test 11208 vectors if you have them. 11209 [Ben Laurie] 11210 11211 *) Correct calculation of key length for export ciphers (too much space was 11212 allocated for null ciphers). This has not been tested! 11213 [Ben Laurie] 11214 11215 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 11216 message is now correct (it understands "crypto" and "ssl" on its 11217 command line). There is also now an "update" option. This will update 11218 the util/ssleay.num and util/libeay.num files with any new functions. 11219 If you do a: 11220 perl util/mkdef.pl crypto ssl update 11221 it will update them. 11222 [Steve Henson] 11223 11224 *) Overhauled the Perl interface (perl/*): 11225 - ported BN stuff to OpenSSL's different BN library 11226 - made the perl/ source tree CVS-aware 11227 - renamed the package from SSLeay to OpenSSL (the files still contain 11228 their history because I've copied them in the repository) 11229 - removed obsolete files (the test scripts will be replaced 11230 by better Test::Harness variants in the future) 11231 [Ralf S. Engelschall] 11232 11233 *) First cut for a very conservative source tree cleanup: 11234 1. merge various obsolete readme texts into doc/ssleay.txt 11235 where we collect the old documents and readme texts. 11236 2. remove the first part of files where I'm already sure that we no 11237 longer need them because of three reasons: either they are just temporary 11238 files which were left by Eric or they are preserved original files where 11239 I've verified that the diff is also available in the CVS via "cvs diff 11240 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 11241 the crypto/md/ stuff). 11242 [Ralf S. Engelschall] 11243 11244 *) More extension code. Incomplete support for subject and issuer alt 11245 name, issuer and authority key id. Change the i2v function parameters 11246 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 11247 what that's for :-) Fix to ASN1 macro which messed up 11248 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 11249 [Steve Henson] 11250 11251 *) Preliminary support for ENUMERATED type. This is largely copied from the 11252 INTEGER code. 11253 [Steve Henson] 11254 11255 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 11256 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11257 11258 *) Make sure `make rehash' target really finds the `openssl' program. 11259 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 11260 11261 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 11262 like to hear about it if this slows down other processors. 11263 [Ben Laurie] 11264 11265 *) Add CygWin32 platform information to Configure script. 11266 [Alan Batie <batie@aahz.jf.intel.com>] 11267 11268 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 11269 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 11270 11271 *) New program nseq to manipulate netscape certificate sequences 11272 [Steve Henson] 11273 11274 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 11275 few typos. 11276 [Steve Henson] 11277 11278 *) Fixes to BN code. Previously the default was to define BN_RECURSION 11279 but the BN code had some problems that would cause failures when 11280 doing certificate verification and some other functions. 11281 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 11282 11283 *) Add ASN1 and PEM code to support netscape certificate sequences. 11284 [Steve Henson] 11285 11286 *) Add ASN1 and PEM code to support netscape certificate sequences. 11287 [Steve Henson] 11288 11289 *) Add several PKIX and private extended key usage OIDs. 11290 [Steve Henson] 11291 11292 *) Modify the 'ca' program to handle the new extension code. Modify 11293 openssl.cnf for new extension format, add comments. 11294 [Steve Henson] 11295 11296 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 11297 and add a sample to openssl.cnf so req -x509 now adds appropriate 11298 CA extensions. 11299 [Steve Henson] 11300 11301 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 11302 error code, add initial support to X509_print() and x509 application. 11303 [Steve Henson] 11304 11305 *) Takes a deep breath and start addding X509 V3 extension support code. Add 11306 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 11307 stuff is currently isolated and isn't even compiled yet. 11308 [Steve Henson] 11309 11310 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 11311 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 11312 Removed the versions check from X509 routines when loading extensions: 11313 this allows certain broken certificates that don't set the version 11314 properly to be processed. 11315 [Steve Henson] 11316 11317 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 11318 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 11319 can still be regenerated with "make depend". 11320 [Ben Laurie] 11321 11322 *) Spelling mistake in C version of CAST-128. 11323 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 11324 11325 *) Changes to the error generation code. The perl script err-code.pl 11326 now reads in the old error codes and retains the old numbers, only 11327 adding new ones if necessary. It also only changes the .err files if new 11328 codes are added. The makefiles have been modified to only insert errors 11329 when needed (to avoid needlessly modifying header files). This is done 11330 by only inserting errors if the .err file is newer than the auto generated 11331 C file. To rebuild all the error codes from scratch (the old behaviour) 11332 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 11333 or delete all the .err files. 11334 [Steve Henson] 11335 11336 *) CAST-128 was incorrectly implemented for short keys. The C version has 11337 been fixed, but is untested. The assembler versions are also fixed, but 11338 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 11339 to regenerate it if needed. 11340 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 11341 Hagino <itojun@kame.net>] 11342 11343 *) File was opened incorrectly in randfile.c. 11344 [Ulf M��ller <ulf@fitug.de>] 11345 11346 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 11347 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 11348 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 11349 al: it's just almost always a UTCTime. Note this patch adds new error 11350 codes so do a "make errors" if there are problems. 11351 [Steve Henson] 11352 11353 *) Correct Linux 1 recognition in config. 11354 [Ulf M��ller <ulf@fitug.de>] 11355 11356 *) Remove pointless MD5 hash when using DSA keys in ca. 11357 [Anonymous <nobody@replay.com>] 11358 11359 *) Generate an error if given an empty string as a cert directory. Also 11360 generate an error if handed NULL (previously returned 0 to indicate an 11361 error, but didn't set one). 11362 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 11363 11364 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 11365 [Ben Laurie] 11366 11367 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 11368 parameters. This was causing a warning which killed off the Win32 compile. 11369 [Steve Henson] 11370 11371 *) Remove C++ style comments from crypto/bn/bn_local.h. 11372 [Neil Costigan <neil.costigan@celocom.com>] 11373 11374 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 11375 based on a text string, looking up short and long names and finally 11376 "dot" format. The "dot" format stuff didn't work. Added new function 11377 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 11378 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 11379 OID is not part of the table. 11380 [Steve Henson] 11381 11382 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 11383 X509_LOOKUP_by_alias(). 11384 [Ben Laurie] 11385 11386 *) Sort openssl functions by name. 11387 [Ben Laurie] 11388 11389 *) Get the gendsa program working (hopefully) and add it to app list. Remove 11390 encryption from sample DSA keys (in case anyone is interested the password 11391 was "1234"). 11392 [Steve Henson] 11393 11394 *) Make _all_ *_free functions accept a NULL pointer. 11395 [Frans Heymans <fheymans@isaserver.be>] 11396 11397 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 11398 NULL pointers. 11399 [Anonymous <nobody@replay.com>] 11400 11401 *) s_server should send the CAfile as acceptable CAs, not its own cert. 11402 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11403 11404 *) Don't blow it for numeric -newkey arguments to apps/req. 11405 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 11406 11407 *) Temp key "for export" tests were wrong in s3_srvr.c. 11408 [Anonymous <nobody@replay.com>] 11409 11410 *) Add prototype for temp key callback functions 11411 SSL_CTX_set_tmp_{rsa,dh}_callback(). 11412 [Ben Laurie] 11413 11414 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 11415 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 11416 [Steve Henson] 11417 11418 *) X509_name_add_entry() freed the wrong thing after an error. 11419 [Arne Ansper <arne@ats.cyber.ee>] 11420 11421 *) rsa_eay.c would attempt to free a NULL context. 11422 [Arne Ansper <arne@ats.cyber.ee>] 11423 11424 *) BIO_s_socket() had a broken should_retry() on Windoze. 11425 [Arne Ansper <arne@ats.cyber.ee>] 11426 11427 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 11428 [Arne Ansper <arne@ats.cyber.ee>] 11429 11430 *) Make sure the already existing X509_STORE->depth variable is initialized 11431 in X509_STORE_new(), but document the fact that this variable is still 11432 unused in the certificate verification process. 11433 [Ralf S. Engelschall] 11434 11435 *) Fix the various library and apps files to free up pkeys obtained from 11436 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 11437 [Steve Henson] 11438 11439 *) Fix reference counting in X509_PUBKEY_get(). This makes 11440 demos/maurice/example2.c work, amongst others, probably. 11441 [Steve Henson and Ben Laurie] 11442 11443 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 11444 `openssl' and second, the shortcut symlinks for the `openssl <command>' 11445 are no longer created. This way we have a single and consistent command 11446 line interface `openssl <command>', similar to `cvs <command>'. 11447 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 11448 11449 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 11450 BIT STRING wrapper always have zero unused bits. 11451 [Steve Henson] 11452 11453 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 11454 [Steve Henson] 11455 11456 *) Make the top-level INSTALL documentation easier to understand. 11457 [Paul Sutton] 11458 11459 *) Makefiles updated to exit if an error occurs in a sub-directory 11460 make (including if user presses ^C) [Paul Sutton] 11461 11462 *) Make Montgomery context stuff explicit in RSA data structure. 11463 [Ben Laurie] 11464 11465 *) Fix build order of pem and err to allow for generated pem.h. 11466 [Ben Laurie] 11467 11468 *) Fix renumbering bug in X509_NAME_delete_entry(). 11469 [Ben Laurie] 11470 11471 *) Enhanced the err-ins.pl script so it makes the error library number 11472 global and can add a library name. This is needed for external ASN1 and 11473 other error libraries. 11474 [Steve Henson] 11475 11476 *) Fixed sk_insert which never worked properly. 11477 [Steve Henson] 11478 11479 *) Fix ASN1 macros so they can handle indefinite length construted 11480 EXPLICIT tags. Some non standard certificates use these: they can now 11481 be read in. 11482 [Steve Henson] 11483 11484 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 11485 into a single doc/ssleay.txt bundle. This way the information is still 11486 preserved but no longer messes up this directory. Now it's new room for 11487 the new set of documenation files. 11488 [Ralf S. Engelschall] 11489 11490 *) SETs were incorrectly DER encoded. This was a major pain, because they 11491 shared code with SEQUENCEs, which aren't coded the same. This means that 11492 almost everything to do with SETs or SEQUENCEs has either changed name or 11493 number of arguments. 11494 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 11495 11496 *) Fix test data to work with the above. 11497 [Ben Laurie] 11498 11499 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 11500 was already fixed by Eric for 0.9.1 it seems. 11501 [Ben Laurie - pointed out by Ulf M��ller <ulf@fitug.de>] 11502 11503 *) Autodetect FreeBSD3. 11504 [Ben Laurie] 11505 11506 *) Fix various bugs in Configure. This affects the following platforms: 11507 nextstep 11508 ncr-scde 11509 unixware-2.0 11510 unixware-2.0-pentium 11511 sco5-cc. 11512 [Ben Laurie] 11513 11514 *) Eliminate generated files from CVS. Reorder tests to regenerate files 11515 before they are needed. 11516 [Ben Laurie] 11517 11518 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 11519 [Ben Laurie] 11520 11521 11522 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 11523 11524 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 11525 changed SSLeay to OpenSSL in version strings. 11526 [Ralf S. Engelschall] 11527 11528 *) Some fixups to the top-level documents. 11529 [Paul Sutton] 11530 11531 *) Fixed the nasty bug where rsaref.h was not found under compile-time 11532 because the symlink to include/ was missing. 11533 [Ralf S. Engelschall] 11534 11535 *) Incorporated the popular no-RSA/DSA-only patches 11536 which allow to compile a RSA-free SSLeay. 11537 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 11538 11539 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 11540 when "ssleay" is still not found. 11541 [Ralf S. Engelschall] 11542 11543 *) Added more platforms to Configure: Cray T3E, HPUX 11, 11544 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 11545 11546 *) Updated the README file. 11547 [Ralf S. Engelschall] 11548 11549 *) Added various .cvsignore files in the CVS repository subdirs 11550 to make a "cvs update" really silent. 11551 [Ralf S. Engelschall] 11552 11553 *) Recompiled the error-definition header files and added 11554 missing symbols to the Win32 linker tables. 11555 [Ralf S. Engelschall] 11556 11557 *) Cleaned up the top-level documents; 11558 o new files: CHANGES and LICENSE 11559 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 11560 o merged COPYRIGHT into LICENSE 11561 o removed obsolete TODO file 11562 o renamed MICROSOFT to INSTALL.W32 11563 [Ralf S. Engelschall] 11564 11565 *) Removed dummy files from the 0.9.1b source tree: 11566 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 11567 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 11568 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 11569 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 11570 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 11571 [Ralf S. Engelschall] 11572 11573 *) Added various platform portability fixes. 11574 [Mark J. Cox] 11575 11576 *) The Genesis of the OpenSSL rpject: 11577 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 11578 Young and Tim J. Hudson created while they were working for C2Net until 11579 summer 1998. 11580 [The OpenSSL Project] 11581 11582 11583 Changes between 0.9.0b and 0.9.1b [not released] 11584 11585 *) Updated a few CA certificates under certs/ 11586 [Eric A. Young] 11587 11588 *) Changed some BIGNUM api stuff. 11589 [Eric A. Young] 11590 11591 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 11592 DGUX x86, Linux Alpha, etc. 11593 [Eric A. Young] 11594 11595 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 11596 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 11597 available). 11598 [Eric A. Young] 11599 11600 *) Add -strparse option to asn1pars program which parses nested 11601 binary structures 11602 [Dr Stephen Henson <shenson@bigfoot.com>] 11603 11604 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 11605 [Eric A. Young] 11606 11607 *) DSA fix for "ca" program. 11608 [Eric A. Young] 11609 11610 *) Added "-genkey" option to "dsaparam" program. 11611 [Eric A. Young] 11612 11613 *) Added RIPE MD160 (rmd160) message digest. 11614 [Eric A. Young] 11615 11616 *) Added -a (all) option to "ssleay version" command. 11617 [Eric A. Young] 11618 11619 *) Added PLATFORM define which is the id given to Configure. 11620 [Eric A. Young] 11621 11622 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 11623 [Eric A. Young] 11624 11625 *) Extended the ASN.1 parser routines. 11626 [Eric A. Young] 11627 11628 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 11629 [Eric A. Young] 11630 11631 *) Added a BN_CTX to the BN library. 11632 [Eric A. Young] 11633 11634 *) Fixed the weak key values in DES library 11635 [Eric A. Young] 11636 11637 *) Changed API in EVP library for cipher aliases. 11638 [Eric A. Young] 11639 11640 *) Added support for RC2/64bit cipher. 11641 [Eric A. Young] 11642 11643 *) Converted the lhash library to the crypto/mem.c functions. 11644 [Eric A. Young] 11645 11646 *) Added more recognized ASN.1 object ids. 11647 [Eric A. Young] 11648 11649 *) Added more RSA padding checks for SSL/TLS. 11650 [Eric A. Young] 11651 11652 *) Added BIO proxy/filter functionality. 11653 [Eric A. Young] 11654 11655 *) Added extra_certs to SSL_CTX which can be used 11656 send extra CA certificates to the client in the CA cert chain sending 11657 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 11658 [Eric A. Young] 11659 11660 *) Now Fortezza is denied in the authentication phase because 11661 this is key exchange mechanism is not supported by SSLeay at all. 11662 [Eric A. Young] 11663 11664 *) Additional PKCS1 checks. 11665 [Eric A. Young] 11666 11667 *) Support the string "TLSv1" for all TLS v1 ciphers. 11668 [Eric A. Young] 11669 11670 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 11671 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 11672 [Eric A. Young] 11673 11674 *) Fixed a few memory leaks. 11675 [Eric A. Young] 11676 11677 *) Fixed various code and comment typos. 11678 [Eric A. Young] 11679 11680 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 11681 bytes sent in the client random. 11682 [Edward Bishop <ebishop@spyglass.com>] 11683 11684