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 1.1.1j and 1.1.1k [25 Mar 2021] 11 12 *) Fixed a problem with verifying a certificate chain when using the 13 X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks 14 of the certificates present in a certificate chain. It is not set by 15 default. 16 17 Starting from OpenSSL version 1.1.1h a check to disallow certificates in 18 the chain that have explicitly encoded elliptic curve parameters was added 19 as an additional strict check. 20 21 An error in the implementation of this check meant that the result of a 22 previous check to confirm that certificates in the chain are valid CA 23 certificates was overwritten. This effectively bypasses the check 24 that non-CA certificates must not be able to issue other certificates. 25 26 If a "purpose" has been configured then there is a subsequent opportunity 27 for checks that the certificate is a valid CA. All of the named "purpose" 28 values implemented in libcrypto perform this check. Therefore, where 29 a purpose is set the certificate chain will still be rejected even when the 30 strict flag has been used. A purpose is set by default in libssl client and 31 server certificate verification routines, but it can be overridden or 32 removed by an application. 33 34 In order to be affected, an application must explicitly set the 35 X509_V_FLAG_X509_STRICT verification flag and either not set a purpose 36 for the certificate verification or, in the case of TLS client or server 37 applications, override the default purpose. 38 (CVE-2021-3450) 39 [Tomáš Mráz] 40 41 *) Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously 42 crafted renegotiation ClientHello message from a client. If a TLSv1.2 43 renegotiation ClientHello omits the signature_algorithms extension (where 44 it was present in the initial ClientHello), but includes a 45 signature_algorithms_cert extension then a NULL pointer dereference will 46 result, leading to a crash and a denial of service attack. 47 48 A server is only vulnerable if it has TLSv1.2 and renegotiation enabled 49 (which is the default configuration). OpenSSL TLS clients are not impacted 50 by this issue. 51 (CVE-2021-3449) 52 [Peter Kästle and Samuel Sapalski] 53 54 Changes between 1.1.1i and 1.1.1j [16 Feb 2021] 55 56 *) Fixed the X509_issuer_and_serial_hash() function. It attempts to 57 create a unique hash value based on the issuer and serial number data 58 contained within an X509 certificate. However it was failing to correctly 59 handle any errors that may occur while parsing the issuer field (which might 60 occur if the issuer field is maliciously constructed). This may subsequently 61 result in a NULL pointer deref and a crash leading to a potential denial of 62 service attack. 63 (CVE-2021-23841) 64 [Matt Caswell] 65 66 *) Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING 67 padding mode to correctly check for rollback attacks. This is considered a 68 bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is 69 CVE-2021-23839. 70 [Matt Caswell] 71 72 *) Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate 73 functions. Previously they could overflow the output length argument in some 74 cases where the input length is close to the maximum permissable length for 75 an integer on the platform. In such cases the return value from the function 76 call would be 1 (indicating success), but the output length value would be 77 negative. This could cause applications to behave incorrectly or crash. 78 (CVE-2021-23840) 79 [Matt Caswell] 80 81 *) Fixed SRP_Calc_client_key so that it runs in constant time. The previous 82 implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This 83 could be exploited in a side channel attack to recover the password. Since 84 the attack is local host only this is outside of the current OpenSSL 85 threat model and therefore no CVE is assigned. 86 87 Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this 88 issue. 89 [Matt Caswell] 90 91 Changes between 1.1.1h and 1.1.1i [8 Dec 2020] 92 93 *) Fixed NULL pointer deref in the GENERAL_NAME_cmp function 94 This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME. 95 If an attacker can control both items being compared then this could lead 96 to a possible denial of service attack. OpenSSL itself uses the 97 GENERAL_NAME_cmp function for two purposes: 98 1) Comparing CRL distribution point names between an available CRL and a 99 CRL distribution point embedded in an X509 certificate 100 2) When verifying that a timestamp response token signer matches the 101 timestamp authority name (exposed via the API functions 102 TS_RESP_verify_response and TS_RESP_verify_token) 103 (CVE-2020-1971) 104 [Matt Caswell] 105 106 *) Add support for Apple Silicon M1 Macs with the darwin64-arm64-cc target. 107 [Stuart Carnie] 108 109 *) The security callback, which can be customised by application code, supports 110 the security operation SSL_SECOP_TMP_DH. This is defined to take an EVP_PKEY 111 in the "other" parameter. In most places this is what is passed. All these 112 places occur server side. However there was one client side call of this 113 security operation and it passed a DH object instead. This is incorrect 114 according to the definition of SSL_SECOP_TMP_DH, and is inconsistent with all 115 of the other locations. Therefore this client side call has been changed to 116 pass an EVP_PKEY instead. 117 [Matt Caswell] 118 119 *) In 1.1.1h, an expired trusted (root) certificate was not anymore rejected 120 when validating a certificate path. This check is restored in 1.1.1i. 121 [David von Oheimb] 122 123 Changes between 1.1.1g and 1.1.1h [22 Sep 2020] 124 125 *) Certificates with explicit curve parameters are now disallowed in 126 verification chains if the X509_V_FLAG_X509_STRICT flag is used. 127 [Tomas Mraz] 128 129 *) The 'MinProtocol' and 'MaxProtocol' configuration commands now silently 130 ignore TLS protocol version bounds when configuring DTLS-based contexts, and 131 conversely, silently ignore DTLS protocol version bounds when configuring 132 TLS-based contexts. The commands can be repeated to set bounds of both 133 types. The same applies with the corresponding "min_protocol" and 134 "max_protocol" command-line switches, in case some application uses both TLS 135 and DTLS. 136 137 SSL_CTX instances that are created for a fixed protocol version (e.g. 138 TLSv1_server_method()) also silently ignore version bounds. Previously 139 attempts to apply bounds to these protocol versions would result in an 140 error. Now only the "version-flexible" SSL_CTX instances are subject to 141 limits in configuration files in command-line options. 142 [Viktor Dukhovni] 143 144 *) Handshake now fails if Extended Master Secret extension is dropped 145 on renegotiation. 146 [Tomas Mraz] 147 148 *) Accidentally, an expired trusted (root) certificate is not anymore rejected 149 when validating a certificate path. 150 [David von Oheimb] 151 152 *) The Oracle Developer Studio compiler will start reporting deprecated APIs 153 154 Changes between 1.1.1f and 1.1.1g [21 Apr 2020] 155 156 *) Fixed segmentation fault in SSL_check_chain() 157 Server or client applications that call the SSL_check_chain() function 158 during or after a TLS 1.3 handshake may crash due to a NULL pointer 159 dereference as a result of incorrect handling of the 160 "signature_algorithms_cert" TLS extension. The crash occurs if an invalid 161 or unrecognised signature algorithm is received from the peer. This could 162 be exploited by a malicious peer in a Denial of Service attack. 163 (CVE-2020-1967) 164 [Benjamin Kaduk] 165 166 *) Added AES consttime code for no-asm configurations 167 an optional constant time support for AES was added 168 when building openssl for no-asm. 169 Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME 170 Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME 171 At this time this feature is by default disabled. 172 It will be enabled by default in 3.0. 173 [Bernd Edlinger] 174 175 Changes between 1.1.1e and 1.1.1f [31 Mar 2020] 176 177 *) Revert the change of EOF detection while reading in libssl to avoid 178 regressions in applications depending on the current way of reporting 179 the EOF. As the existing method is not fully accurate the change to 180 reporting the EOF via SSL_ERROR_SSL is kept on the current development 181 branch and will be present in the 3.0 release. 182 [Tomas Mraz] 183 184 *) Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1 185 when primes for RSA keys are computed. 186 Since we previously always generated primes == 2 (mod 3) for RSA keys, 187 the 2-prime and 3-prime RSA modules were easy to distinguish, since 188 N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting 189 2-prime vs. 3-prime RSA keys was possible by computing N mod 3. 190 This avoids possible fingerprinting of newly generated RSA modules. 191 [Bernd Edlinger] 192 193 Changes between 1.1.1d and 1.1.1e [17 Mar 2020] 194 *) Properly detect EOF while reading in libssl. Previously if we hit an EOF 195 while reading in libssl then we would report an error back to the 196 application (SSL_ERROR_SYSCALL) but errno would be 0. We now add 197 an error to the stack (which means we instead return SSL_ERROR_SSL) and 198 therefore give a hint as to what went wrong. 199 [Matt Caswell] 200 201 *) Check that ed25519 and ed448 are allowed by the security level. Previously 202 signature algorithms not using an MD were not being checked that they were 203 allowed by the security level. 204 [Kurt Roeckx] 205 206 *) Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername() 207 was not quite right. The behaviour was not consistent between resumption 208 and normal handshakes, and also not quite consistent with historical 209 behaviour. The behaviour in various scenarios has been clarified and 210 it has been updated to make it match historical behaviour as closely as 211 possible. 212 [Matt Caswell] 213 214 *) [VMS only] The header files that the VMS compilers include automatically, 215 __DECC_INCLUDE_PROLOGUE.H and __DECC_INCLUDE_EPILOGUE.H, use pragmas that 216 the C++ compiler doesn't understand. This is a shortcoming in the 217 compiler, but can be worked around with __cplusplus guards. 218 219 C++ applications that use OpenSSL libraries must be compiled using the 220 qualifier '/NAMES=(AS_IS,SHORTENED)' to be able to use all the OpenSSL 221 functions. Otherwise, only functions with symbols of less than 31 222 characters can be used, as the linker will not be able to successfully 223 resolve symbols with longer names. 224 [Richard Levitte] 225 226 *) Corrected the documentation of the return values from the EVP_DigestSign* 227 set of functions. The documentation mentioned negative values for some 228 errors, but this was never the case, so the mention of negative values 229 was removed. 230 231 Code that followed the documentation and thereby check with something 232 like 'EVP_DigestSignInit(...) <= 0' will continue to work undisturbed. 233 [Richard Levitte] 234 235 *) Fixed an an overflow bug in the x64_64 Montgomery squaring procedure 236 used in exponentiation with 512-bit moduli. No EC algorithms are 237 affected. Analysis suggests that attacks against 2-prime RSA1024, 238 3-prime RSA1536, and DSA1024 as a result of this defect would be very 239 difficult to perform and are not believed likely. Attacks against DH512 240 are considered just feasible. However, for an attack the target would 241 have to re-use the DH512 private key, which is not recommended anyway. 242 Also applications directly using the low level API BN_mod_exp may be 243 affected if they use BN_FLG_CONSTTIME. 244 (CVE-2019-1551) 245 [Andy Polyakov] 246 247 *) Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY. 248 The presence of this system service is determined at run-time. 249 [Richard Levitte] 250 251 *) Added newline escaping functionality to a filename when using openssl dgst. 252 This output format is to replicate the output format found in the '*sum' 253 checksum programs. This aims to preserve backward compatibility. 254 [Matt Eaton, Richard Levitte, and Paul Dale] 255 256 *) Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just 257 the first value. 258 [Jon Spillett] 259 260 Changes between 1.1.1c and 1.1.1d [10 Sep 2019] 261 262 *) Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random 263 number generator (RNG). This was intended to include protection in the 264 event of a fork() system call in order to ensure that the parent and child 265 processes did not share the same RNG state. However this protection was not 266 being used in the default case. 267 268 A partial mitigation for this issue is that the output from a high 269 precision timer is mixed into the RNG state so the likelihood of a parent 270 and child process sharing state is significantly reduced. 271 272 If an application already calls OPENSSL_init_crypto() explicitly using 273 OPENSSL_INIT_ATFORK then this problem does not occur at all. 274 (CVE-2019-1549) 275 [Matthias St. Pierre] 276 277 *) For built-in EC curves, ensure an EC_GROUP built from the curve name is 278 used even when parsing explicit parameters, when loading a serialized key 279 or calling `EC_GROUP_new_from_ecpkparameters()`/ 280 `EC_GROUP_new_from_ecparameters()`. 281 This prevents bypass of security hardening and performance gains, 282 especially for curves with specialized EC_METHODs. 283 By default, if a key encoded with explicit parameters is loaded and later 284 serialized, the output is still encoded with explicit parameters, even if 285 internally a "named" EC_GROUP is used for computation. 286 [Nicola Tuveri] 287 288 *) Compute ECC cofactors if not provided during EC_GROUP construction. Before 289 this change, EC_GROUP_set_generator would accept order and/or cofactor as 290 NULL. After this change, only the cofactor parameter can be NULL. It also 291 does some minimal sanity checks on the passed order. 292 (CVE-2019-1547) 293 [Billy Bob Brumley] 294 295 *) Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey. 296 An attack is simple, if the first CMS_recipientInfo is valid but the 297 second CMS_recipientInfo is chosen ciphertext. If the second 298 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct 299 encryption key will be replaced by garbage, and the message cannot be 300 decoded, but if the RSA decryption fails, the correct encryption key is 301 used and the recipient will not notice the attack. 302 As a work around for this potential attack the length of the decrypted 303 key must be equal to the cipher default key length, in case the 304 certifiate is not given and all recipientInfo are tried out. 305 The old behaviour can be re-enabled in the CMS code by setting the 306 CMS_DEBUG_DECRYPT flag. 307 (CVE-2019-1563) 308 [Bernd Edlinger] 309 310 *) Early start up entropy quality from the DEVRANDOM seed source has been 311 improved for older Linux systems. The RAND subsystem will wait for 312 /dev/random to be producing output before seeding from /dev/urandom. 313 The seeded state is stored for future library initialisations using 314 a system global shared memory segment. The shared memory identifier 315 can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to 316 the desired value. The default identifier is 114. 317 [Paul Dale] 318 319 *) Correct the extended master secret constant on EBCDIC systems. Without this 320 fix TLS connections between an EBCDIC system and a non-EBCDIC system that 321 negotiate EMS will fail. Unfortunately this also means that TLS connections 322 between EBCDIC systems with this fix, and EBCDIC systems without this 323 fix will fail if they negotiate EMS. 324 [Matt Caswell] 325 326 *) Use Windows installation paths in the mingw builds 327 328 Mingw isn't a POSIX environment per se, which means that Windows 329 paths should be used for installation. 330 (CVE-2019-1552) 331 [Richard Levitte] 332 333 *) Changed DH_check to accept parameters with order q and 2q subgroups. 334 With order 2q subgroups the bit 0 of the private key is not secret 335 but DH_generate_key works around that by clearing bit 0 of the 336 private key for those. This avoids leaking bit 0 of the private key. 337 [Bernd Edlinger] 338 339 *) Significantly reduce secure memory usage by the randomness pools. 340 [Paul Dale] 341 342 *) Revert the DEVRANDOM_WAIT feature for Linux systems 343 344 The DEVRANDOM_WAIT feature added a select() call to wait for the 345 /dev/random device to become readable before reading from the 346 /dev/urandom device. 347 348 It turned out that this change had negative side effects on 349 performance which were not acceptable. After some discussion it 350 was decided to revert this feature and leave it up to the OS 351 resp. the platform maintainer to ensure a proper initialization 352 during early boot time. 353 [Matthias St. Pierre] 354 355 Changes between 1.1.1b and 1.1.1c [28 May 2019] 356 357 *) Add build tests for C++. These are generated files that only do one 358 thing, to include one public OpenSSL head file each. This tests that 359 the public header files can be usefully included in a C++ application. 360 361 This test isn't enabled by default. It can be enabled with the option 362 'enable-buildtest-c++'. 363 [Richard Levitte] 364 365 *) Enable SHA3 pre-hashing for ECDSA and DSA. 366 [Patrick Steuer] 367 368 *) Change the default RSA, DSA and DH size to 2048 bit instead of 1024. 369 This changes the size when using the genpkey app when no size is given. It 370 fixes an omission in earlier changes that changed all RSA, DSA and DH 371 generation apps to use 2048 bits by default. 372 [Kurt Roeckx] 373 374 *) Reorganize the manual pages to consistently have RETURN VALUES, 375 EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust 376 util/fix-doc-nits accordingly. 377 [Paul Yang, Joshua Lock] 378 379 *) Add the missing accessor EVP_PKEY_get0_engine() 380 [Matt Caswell] 381 382 *) Have apps like 's_client' and 's_server' output the signature scheme 383 along with other cipher suite parameters when debugging. 384 [Lorinczy Zsigmond] 385 386 *) Make OPENSSL_config() error agnostic again. 387 [Richard Levitte] 388 389 *) Do the error handling in RSA decryption constant time. 390 [Bernd Edlinger] 391 392 *) Prevent over long nonces in ChaCha20-Poly1305. 393 394 ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input 395 for every encryption operation. RFC 7539 specifies that the nonce value 396 (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length 397 and front pads the nonce with 0 bytes if it is less than 12 398 bytes. However it also incorrectly allows a nonce to be set of up to 16 399 bytes. In this case only the last 12 bytes are significant and any 400 additional leading bytes are ignored. 401 402 It is a requirement of using this cipher that nonce values are 403 unique. Messages encrypted using a reused nonce value are susceptible to 404 serious confidentiality and integrity attacks. If an application changes 405 the default nonce length to be longer than 12 bytes and then makes a 406 change to the leading bytes of the nonce expecting the new value to be a 407 new unique nonce then such an application could inadvertently encrypt 408 messages with a reused nonce. 409 410 Additionally the ignored bytes in a long nonce are not covered by the 411 integrity guarantee of this cipher. Any application that relies on the 412 integrity of these ignored leading bytes of a long nonce may be further 413 affected. Any OpenSSL internal use of this cipher, including in SSL/TLS, 414 is safe because no such use sets such a long nonce value. However user 415 applications that use this cipher directly and set a non-default nonce 416 length to be longer than 12 bytes may be vulnerable. 417 418 This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk 419 Greef of Ronomon. 420 (CVE-2019-1543) 421 [Matt Caswell] 422 423 *) Add DEVRANDOM_WAIT feature for Linux systems 424 425 On older Linux systems where the getrandom() system call is not available, 426 OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG. 427 Contrary to getrandom(), the /dev/urandom device will not block during 428 early boot when the kernel CSPRNG has not been seeded yet. 429 430 To mitigate this known weakness, use select() to wait for /dev/random to 431 become readable before reading from /dev/urandom. 432 433 *) Ensure that SM2 only uses SM3 as digest algorithm 434 [Paul Yang] 435 436 Changes between 1.1.1a and 1.1.1b [26 Feb 2019] 437 438 *) Added SCA hardening for modular field inversion in EC_GROUP through 439 a new dedicated field_inv() pointer in EC_METHOD. 440 This also addresses a leakage affecting conversions from projective 441 to affine coordinates. 442 [Billy Bob Brumley, Nicola Tuveri] 443 444 *) Change the info callback signals for the start and end of a post-handshake 445 message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START 446 and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get 447 confused by this and assume that a TLSv1.2 renegotiation has started. This 448 can break KeyUpdate handling. Instead we no longer signal the start and end 449 of a post handshake message exchange (although the messages themselves are 450 still signalled). This could break some applications that were expecting 451 the old signals. However without this KeyUpdate is not usable for many 452 applications. 453 [Matt Caswell] 454 455 *) Fix a bug in the computation of the endpoint-pair shared secret used 456 by DTLS over SCTP. This breaks interoperability with older versions 457 of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime 458 switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling 459 interoperability with such broken implementations. However, enabling 460 this switch breaks interoperability with correct implementations. 461 462 *) Fix a use after free bug in d2i_X509_PUBKEY when overwriting a 463 re-used X509_PUBKEY object if the second PUBKEY is malformed. 464 [Bernd Edlinger] 465 466 *) Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0(). 467 [Richard Levitte] 468 469 *) Remove the 'dist' target and add a tarball building script. The 470 'dist' target has fallen out of use, and it shouldn't be 471 necessary to configure just to create a source distribution. 472 [Richard Levitte] 473 474 Changes between 1.1.1 and 1.1.1a [20 Nov 2018] 475 476 *) Timing vulnerability in DSA signature generation 477 478 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a 479 timing side channel attack. An attacker could use variations in the signing 480 algorithm to recover the private key. 481 482 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser. 483 (CVE-2018-0734) 484 [Paul Dale] 485 486 *) Timing vulnerability in ECDSA signature generation 487 488 The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a 489 timing side channel attack. An attacker could use variations in the signing 490 algorithm to recover the private key. 491 492 This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser. 493 (CVE-2018-0735) 494 [Paul Dale] 495 496 *) Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for 497 the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names 498 are retained for backwards compatibility. 499 [Antoine Salon] 500 501 *) Fixed the issue that RAND_add()/RAND_seed() silently discards random input 502 if its length exceeds 4096 bytes. The limit has been raised to a buffer size 503 of two gigabytes and the error handling improved. 504 505 This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been 506 categorized as a normal bug, not a security issue, because the DRBG reseeds 507 automatically and is fully functional even without additional randomness 508 provided by the application. 509 510 Changes between 1.1.0i and 1.1.1 [11 Sep 2018] 511 512 *) Add a new ClientHello callback. Provides a callback interface that gives 513 the application the ability to adjust the nascent SSL object at the 514 earliest stage of ClientHello processing, immediately after extensions have 515 been collected but before they have been processed. In particular, this 516 callback can adjust the supported TLS versions in response to the contents 517 of the ClientHello 518 [Benjamin Kaduk] 519 520 *) Add SM2 base algorithm support. 521 [Jack Lloyd] 522 523 *) s390x assembly pack: add (improved) hardware-support for the following 524 cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb, 525 aes-cfb/cfb8, aes-ecb. 526 [Patrick Steuer] 527 528 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 529 parameter is no longer accepted, as it leads to a corrupt table. NULL 530 pem_str is reserved for alias entries only. 531 [Richard Levitte] 532 533 *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 534 step for prime curves. The new implementation is based on formulae from 535 differential addition-and-doubling in homogeneous projective coordinates 536 from Izu-Takagi "A fast parallel elliptic curve multiplication resistant 537 against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves 538 and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified 539 to work in projective coordinates. 540 [Billy Bob Brumley, Nicola Tuveri] 541 542 *) Change generating and checking of primes so that the error rate of not 543 being prime depends on the intended use based on the size of the input. 544 For larger primes this will result in more rounds of Miller-Rabin. 545 The maximal error rate for primes with more than 1080 bits is lowered 546 to 2^-128. 547 [Kurt Roeckx, Annie Yousar] 548 549 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 550 [Kurt Roeckx] 551 552 *) The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when 553 moving between systems, and to avoid confusion when a Windows build is 554 done with mingw vs with MSVC. For POSIX installs, there's still a 555 symlink or copy named 'tsget' to avoid that confusion as well. 556 [Richard Levitte] 557 558 *) Revert blinding in ECDSA sign and instead make problematic addition 559 length-invariant. Switch even to fixed-length Montgomery multiplication. 560 [Andy Polyakov] 561 562 *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 563 step for binary curves. The new implementation is based on formulae from 564 differential addition-and-doubling in mixed Lopez-Dahab projective 565 coordinates, modified to independently blind the operands. 566 [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri] 567 568 *) Add a scaffold to optionally enhance the Montgomery ladder implementation 569 for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing 570 EC_METHODs to implement their own specialized "ladder step", to take 571 advantage of more favorable coordinate systems or more efficient 572 differential addition-and-doubling algorithms. 573 [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri] 574 575 *) Modified the random device based seed sources to keep the relevant 576 file descriptors open rather than reopening them on each access. 577 This allows such sources to operate in a chroot() jail without 578 the associated device nodes being available. This behaviour can be 579 controlled using RAND_keep_random_devices_open(). 580 [Paul Dale] 581 582 *) Numerous side-channel attack mitigations have been applied. This may have 583 performance impacts for some algorithms for the benefit of improved 584 security. Specific changes are noted in this change log by their respective 585 authors. 586 [Matt Caswell] 587 588 *) AIX shared library support overhaul. Switch to AIX "natural" way of 589 handling shared libraries, which means collecting shared objects of 590 different versions and bitnesses in one common archive. This allows to 591 mitigate conflict between 1.0 and 1.1 side-by-side installations. It 592 doesn't affect the way 3rd party applications are linked, only how 593 multi-version installation is managed. 594 [Andy Polyakov] 595 596 *) Make ec_group_do_inverse_ord() more robust and available to other 597 EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA 598 mitigations are applied to the fallback BN_mod_inverse(). 599 When using this function rather than BN_mod_inverse() directly, new 600 EC cryptosystem implementations are then safer-by-default. 601 [Billy Bob Brumley] 602 603 *) Add coordinate blinding for EC_POINT and implement projective 604 coordinate blinding for generic prime curves as a countermeasure to 605 chosen point SCA attacks. 606 [Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley] 607 608 *) Add blinding to ECDSA and DSA signatures to protect against side channel 609 attacks discovered by Keegan Ryan (NCC Group). 610 [Matt Caswell] 611 612 *) Enforce checking in the pkeyutl command line app to ensure that the input 613 length does not exceed the maximum supported digest length when performing 614 a sign, verify or verifyrecover operation. 615 [Matt Caswell] 616 617 *) SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking 618 I/O in combination with something like select() or poll() will hang. This 619 can be turned off again using SSL_CTX_clear_mode(). 620 Many applications do not properly handle non-application data records, and 621 TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works 622 around the problems in those applications, but can also break some. 623 It's recommended to read the manpages about SSL_read(), SSL_write(), 624 SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and 625 SSL_CTX_set_read_ahead() again. 626 [Kurt Roeckx] 627 628 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 629 now allow empty (zero character) pass phrases. 630 [Richard Levitte] 631 632 *) Apply blinding to binary field modular inversion and remove patent 633 pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation. 634 [Billy Bob Brumley] 635 636 *) Deprecate ec2_mult.c and unify scalar multiplication code paths for 637 binary and prime elliptic curves. 638 [Billy Bob Brumley] 639 640 *) Remove ECDSA nonce padding: EC_POINT_mul is now responsible for 641 constant time fixed point multiplication. 642 [Billy Bob Brumley] 643 644 *) Revise elliptic curve scalar multiplication with timing attack 645 defenses: ec_wNAF_mul redirects to a constant time implementation 646 when computing fixed point and variable point multiplication (which 647 in OpenSSL are mostly used with secret scalars in keygen, sign, 648 ECDH derive operations). 649 [Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García, 650 Sohaib ul Hassan] 651 652 *) Updated CONTRIBUTING 653 [Rich Salz] 654 655 *) Updated DRBG / RAND to request nonce and additional low entropy 656 randomness from the system. 657 [Matthias St. Pierre] 658 659 *) Updated 'openssl rehash' to use OpenSSL consistent default. 660 [Richard Levitte] 661 662 *) Moved the load of the ssl_conf module to libcrypto, which helps 663 loading engines that libssl uses before libssl is initialised. 664 [Matt Caswell] 665 666 *) Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA 667 [Matt Caswell] 668 669 *) Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases. 670 [Ingo Schwarze, Rich Salz] 671 672 *) Added output of accepting IP address and port for 'openssl s_server' 673 [Richard Levitte] 674 675 *) Added a new API for TLSv1.3 ciphersuites: 676 SSL_CTX_set_ciphersuites() 677 SSL_set_ciphersuites() 678 [Matt Caswell] 679 680 *) Memory allocation failures consistently add an error to the error 681 stack. 682 [Rich Salz] 683 684 *) Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values 685 in libcrypto when run as setuid/setgid. 686 [Bernd Edlinger] 687 688 *) Load any config file by default when libssl is used. 689 [Matt Caswell] 690 691 *) Added new public header file <openssl/rand_drbg.h> and documentation 692 for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview. 693 [Matthias St. Pierre] 694 695 *) QNX support removed (cannot find contributors to get their approval 696 for the license change). 697 [Rich Salz] 698 699 *) TLSv1.3 replay protection for early data has been implemented. See the 700 SSL_read_early_data() man page for further details. 701 [Matt Caswell] 702 703 *) Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite 704 configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and 705 below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3. 706 In order to avoid issues where legacy TLSv1.2 ciphersuite configuration 707 would otherwise inadvertently disable all TLSv1.3 ciphersuites the 708 configuration has been separated out. See the ciphers man page or the 709 SSL_CTX_set_ciphersuites() man page for more information. 710 [Matt Caswell] 711 712 *) On POSIX (BSD, Linux, ...) systems the ocsp(1) command running 713 in responder mode now supports the new "-multi" option, which 714 spawns the specified number of child processes to handle OCSP 715 requests. The "-timeout" option now also limits the OCSP 716 responder's patience to wait to receive the full client request 717 on a newly accepted connection. Child processes are respawned 718 as needed, and the CA index file is automatically reloaded 719 when changed. This makes it possible to run the "ocsp" responder 720 as a long-running service, making the OpenSSL CA somewhat more 721 feature-complete. In this mode, most diagnostic messages logged 722 after entering the event loop are logged via syslog(3) rather than 723 written to stderr. 724 [Viktor Dukhovni] 725 726 *) Added support for X448 and Ed448. Heavily based on original work by 727 Mike Hamburg. 728 [Matt Caswell] 729 730 *) Extend OSSL_STORE with capabilities to search and to narrow the set of 731 objects loaded. This adds the functions OSSL_STORE_expect() and 732 OSSL_STORE_find() as well as needed tools to construct searches and 733 get the search data out of them. 734 [Richard Levitte] 735 736 *) Support for TLSv1.3 added. Note that users upgrading from an earlier 737 version of OpenSSL should review their configuration settings to ensure 738 that they are still appropriate for TLSv1.3. For further information see: 739 https://wiki.openssl.org/index.php/TLS1.3 740 [Matt Caswell] 741 742 *) Grand redesign of the OpenSSL random generator 743 744 The default RAND method now utilizes an AES-CTR DRBG according to 745 NIST standard SP 800-90Ar1. The new random generator is essentially 746 a port of the default random generator from the OpenSSL FIPS 2.0 747 object module. It is a hybrid deterministic random bit generator 748 using an AES-CTR bit stream and which seeds and reseeds itself 749 automatically using trusted system entropy sources. 750 751 Some of its new features are: 752 o Support for multiple DRBG instances with seed chaining. 753 o The default RAND method makes use of a DRBG. 754 o There is a public and private DRBG instance. 755 o The DRBG instances are fork-safe. 756 o Keep all global DRBG instances on the secure heap if it is enabled. 757 o The public and private DRBG instance are per thread for lock free 758 operation 759 [Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre] 760 761 *) Changed Configure so it only says what it does and doesn't dump 762 so much data. Instead, ./configdata.pm should be used as a script 763 to display all sorts of configuration data. 764 [Richard Levitte] 765 766 *) Added processing of "make variables" to Configure. 767 [Richard Levitte] 768 769 *) Added SHA512/224 and SHA512/256 algorithm support. 770 [Paul Dale] 771 772 *) The last traces of Netware support, first removed in 1.1.0, have 773 now been removed. 774 [Rich Salz] 775 776 *) Get rid of Makefile.shared, and in the process, make the processing 777 of certain files (rc.obj, or the .def/.map/.opt files produced from 778 the ordinal files) more visible and hopefully easier to trace and 779 debug (or make silent). 780 [Richard Levitte] 781 782 *) Make it possible to have environment variable assignments as 783 arguments to config / Configure. 784 [Richard Levitte] 785 786 *) Add multi-prime RSA (RFC 8017) support. 787 [Paul Yang] 788 789 *) Add SM3 implemented according to GB/T 32905-2016 790 [ Jack Lloyd <jack.lloyd@ribose.com>, 791 Ronald Tse <ronald.tse@ribose.com>, 792 Erick Borsboom <erick.borsboom@ribose.com> ] 793 794 *) Add 'Maximum Fragment Length' TLS extension negotiation and support 795 as documented in RFC6066. 796 Based on a patch from Tomasz Moń 797 [Filipe Raimundo da Silva] 798 799 *) Add SM4 implemented according to GB/T 32907-2016. 800 [ Jack Lloyd <jack.lloyd@ribose.com>, 801 Ronald Tse <ronald.tse@ribose.com>, 802 Erick Borsboom <erick.borsboom@ribose.com> ] 803 804 *) Reimplement -newreq-nodes and ERR_error_string_n; the 805 original author does not agree with the license change. 806 [Rich Salz] 807 808 *) Add ARIA AEAD TLS support. 809 [Jon Spillett] 810 811 *) Some macro definitions to support VS6 have been removed. Visual 812 Studio 6 has not worked since 1.1.0 813 [Rich Salz] 814 815 *) Add ERR_clear_last_mark(), to allow callers to clear the last mark 816 without clearing the errors. 817 [Richard Levitte] 818 819 *) Add "atfork" functions. If building on a system that without 820 pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application 821 requirements. The RAND facility now uses/requires this. 822 [Rich Salz] 823 824 *) Add SHA3. 825 [Andy Polyakov] 826 827 *) The UI API becomes a permanent and integral part of libcrypto, i.e. 828 not possible to disable entirely. However, it's still possible to 829 disable the console reading UI method, UI_OpenSSL() (use UI_null() 830 as a fallback). 831 832 To disable, configure with 'no-ui-console'. 'no-ui' is still 833 possible to use as an alias. Check at compile time with the 834 macro OPENSSL_NO_UI_CONSOLE. The macro OPENSSL_NO_UI is still 835 possible to check and is an alias for OPENSSL_NO_UI_CONSOLE. 836 [Richard Levitte] 837 838 *) Add a STORE module, which implements a uniform and URI based reader of 839 stores that can contain keys, certificates, CRLs and numerous other 840 objects. The main API is loosely based on a few stdio functions, 841 and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof, 842 OSSL_STORE_error and OSSL_STORE_close. 843 The implementation uses backends called "loaders" to implement arbitrary 844 URI schemes. There is one built in "loader" for the 'file' scheme. 845 [Richard Levitte] 846 847 *) Add devcrypto engine. This has been implemented against cryptodev-linux, 848 then adjusted to work on FreeBSD 8.4 as well. 849 Enable by configuring with 'enable-devcryptoeng'. This is done by default 850 on BSD implementations, as cryptodev.h is assumed to exist on all of them. 851 [Richard Levitte] 852 853 *) Module names can prefixed with OSSL_ or OPENSSL_. This affects 854 util/mkerr.pl, which is adapted to allow those prefixes, leading to 855 error code calls like this: 856 857 OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER); 858 859 With this change, we claim the namespaces OSSL and OPENSSL in a manner 860 that can be encoded in C. For the foreseeable future, this will only 861 affect new modules. 862 [Richard Levitte and Tim Hudson] 863 864 *) Removed BSD cryptodev engine. 865 [Rich Salz] 866 867 *) Add a build target 'build_all_generated', to build all generated files 868 and only that. This can be used to prepare everything that requires 869 things like perl for a system that lacks perl and then move everything 870 to that system and do the rest of the build there. 871 [Richard Levitte] 872 873 *) In the UI interface, make it possible to duplicate the user data. This 874 can be used by engines that need to retain the data for a longer time 875 than just the call where this user data is passed. 876 [Richard Levitte] 877 878 *) Ignore the '-named_curve auto' value for compatibility of applications 879 with OpenSSL 1.0.2. 880 [Tomas Mraz <tmraz@fedoraproject.org>] 881 882 *) Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2 883 bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such 884 alerts across multiple records (some of which could be empty). In practice 885 it make no sense to send an empty alert record, or to fragment one. TLSv1.3 886 prohibits this altogether and other libraries (BoringSSL, NSS) do not 887 support this at all. Supporting it adds significant complexity to the 888 record layer, and its removal is unlikely to cause interoperability 889 issues. 890 [Matt Caswell] 891 892 *) Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed 893 with Z. These are meant to replace LONG and ZLONG and to be size safe. 894 The use of LONG and ZLONG is discouraged and scheduled for deprecation 895 in OpenSSL 1.2.0. 896 [Richard Levitte] 897 898 *) Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string, 899 'z' is to be used for [s]size_t, and 'j' - with [u]int64_t. 900 [Richard Levitte, Andy Polyakov] 901 902 *) Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine() 903 does for RSA, etc. 904 [Richard Levitte] 905 906 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 907 platform rather than 'mingw'. 908 [Richard Levitte] 909 910 *) The functions X509_STORE_add_cert and X509_STORE_add_crl return 911 success if they are asked to add an object which already exists 912 in the store. This change cascades to other functions which load 913 certificates and CRLs. 914 [Paul Dale] 915 916 *) x86_64 assembly pack: annotate code with DWARF CFI directives to 917 facilitate stack unwinding even from assembly subroutines. 918 [Andy Polyakov] 919 920 *) Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN. 921 Also remove OPENSSL_GLOBAL entirely, as it became a no-op. 922 [Richard Levitte] 923 924 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 925 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 926 which is the minimum version we support. 927 [Richard Levitte] 928 929 *) Certificate time validation (X509_cmp_time) enforces stricter 930 compliance with RFC 5280. Fractional seconds and timezone offsets 931 are no longer allowed. 932 [Emilia Käsper] 933 934 *) Add support for ARIA 935 [Paul Dale] 936 937 *) s_client will now send the Server Name Indication (SNI) extension by 938 default unless the new "-noservername" option is used. The server name is 939 based on the host provided to the "-connect" option unless overridden by 940 using "-servername". 941 [Matt Caswell] 942 943 *) Add support for SipHash 944 [Todd Short] 945 946 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0 947 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 948 prevent issues where no progress is being made and the peer continually 949 sends unrecognised record types, using up resources processing them. 950 [Matt Caswell] 951 952 *) 'openssl passwd' can now produce SHA256 and SHA512 based output, 953 using the algorithm defined in 954 https://www.akkadia.org/drepper/SHA-crypt.txt 955 [Richard Levitte] 956 957 *) Heartbeat support has been removed; the ABI is changed for now. 958 [Richard Levitte, Rich Salz] 959 960 *) Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd. 961 [Emilia Käsper] 962 963 *) The RSA "null" method, which was partially supported to avoid patent 964 issues, has been replaced to always returns NULL. 965 [Rich Salz] 966 967 968 Changes between 1.1.0h and 1.1.0i [xx XXX xxxx] 969 970 *) Client DoS due to large DH parameter 971 972 During key agreement in a TLS handshake using a DH(E) based ciphersuite a 973 malicious server can send a very large prime value to the client. This will 974 cause the client to spend an unreasonably long period of time generating a 975 key for this prime resulting in a hang until the client has finished. This 976 could be exploited in a Denial Of Service attack. 977 978 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken 979 (CVE-2018-0732) 980 [Guido Vranken] 981 982 *) Cache timing vulnerability in RSA Key Generation 983 984 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to 985 a cache timing side channel attack. An attacker with sufficient access to 986 mount cache timing attacks during the RSA key generation process could 987 recover the private key. 988 989 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera 990 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia. 991 (CVE-2018-0737) 992 [Billy Brumley] 993 994 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 995 parameter is no longer accepted, as it leads to a corrupt table. NULL 996 pem_str is reserved for alias entries only. 997 [Richard Levitte] 998 999 *) Revert blinding in ECDSA sign and instead make problematic addition 1000 length-invariant. Switch even to fixed-length Montgomery multiplication. 1001 [Andy Polyakov] 1002 1003 *) Change generating and checking of primes so that the error rate of not 1004 being prime depends on the intended use based on the size of the input. 1005 For larger primes this will result in more rounds of Miller-Rabin. 1006 The maximal error rate for primes with more than 1080 bits is lowered 1007 to 2^-128. 1008 [Kurt Roeckx, Annie Yousar] 1009 1010 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 1011 [Kurt Roeckx] 1012 1013 *) Add blinding to ECDSA and DSA signatures to protect against side channel 1014 attacks discovered by Keegan Ryan (NCC Group). 1015 [Matt Caswell] 1016 1017 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 1018 now allow empty (zero character) pass phrases. 1019 [Richard Levitte] 1020 1021 *) Certificate time validation (X509_cmp_time) enforces stricter 1022 compliance with RFC 5280. Fractional seconds and timezone offsets 1023 are no longer allowed. 1024 [Emilia Käsper] 1025 1026 *) Fixed a text canonicalisation bug in CMS 1027 1028 Where a CMS detached signature is used with text content the text goes 1029 through a canonicalisation process first prior to signing or verifying a 1030 signature. This process strips trailing space at the end of lines, converts 1031 line terminators to CRLF and removes additional trailing line terminators 1032 at the end of a file. A bug in the canonicalisation process meant that 1033 some characters, such as form-feed, were incorrectly treated as whitespace 1034 and removed. This is contrary to the specification (RFC5485). This fix 1035 could mean that detached text data signed with an earlier version of 1036 OpenSSL 1.1.0 may fail to verify using the fixed version, or text data 1037 signed with a fixed OpenSSL may fail to verify with an earlier version of 1038 OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data 1039 and use the "-binary" flag (for the "cms" command line application) or set 1040 the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()). 1041 [Matt Caswell] 1042 1043 Changes between 1.1.0g and 1.1.0h [27 Mar 2018] 1044 1045 *) Constructed ASN.1 types with a recursive definition could exceed the stack 1046 1047 Constructed ASN.1 types with a recursive definition (such as can be found 1048 in PKCS7) could eventually exceed the stack given malicious input with 1049 excessive recursion. This could result in a Denial Of Service attack. There 1050 are no such structures used within SSL/TLS that come from untrusted sources 1051 so this is considered safe. 1052 1053 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz 1054 project. 1055 (CVE-2018-0739) 1056 [Matt Caswell] 1057 1058 *) Incorrect CRYPTO_memcmp on HP-UX PA-RISC 1059 1060 Because of an implementation bug the PA-RISC CRYPTO_memcmp function is 1061 effectively reduced to only comparing the least significant bit of each 1062 byte. This allows an attacker to forge messages that would be considered as 1063 authenticated in an amount of tries lower than that guaranteed by the 1064 security claims of the scheme. The module can only be compiled by the 1065 HP-UX assembler, so that only HP-UX PA-RISC targets are affected. 1066 1067 This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg 1068 (IBM). 1069 (CVE-2018-0733) 1070 [Andy Polyakov] 1071 1072 *) Add a build target 'build_all_generated', to build all generated files 1073 and only that. This can be used to prepare everything that requires 1074 things like perl for a system that lacks perl and then move everything 1075 to that system and do the rest of the build there. 1076 [Richard Levitte] 1077 1078 *) Backport SSL_OP_NO_RENGOTIATION 1079 1080 OpenSSL 1.0.2 and below had the ability to disable renegotiation using the 1081 (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity 1082 changes this is no longer possible in 1.1.0. Therefore the new 1083 SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to 1084 1.1.0 to provide equivalent functionality. 1085 1086 Note that if an application built against 1.1.0h headers (or above) is run 1087 using an older version of 1.1.0 (prior to 1.1.0h) then the option will be 1088 accepted but nothing will happen, i.e. renegotiation will not be prevented. 1089 [Matt Caswell] 1090 1091 *) Removed the OS390-Unix config target. It relied on a script that doesn't 1092 exist. 1093 [Rich Salz] 1094 1095 *) rsaz_1024_mul_avx2 overflow bug on x86_64 1096 1097 There is an overflow bug in the AVX2 Montgomery multiplication procedure 1098 used in exponentiation with 1024-bit moduli. No EC algorithms are affected. 1099 Analysis suggests that attacks against RSA and DSA as a result of this 1100 defect would be very difficult to perform and are not believed likely. 1101 Attacks against DH1024 are considered just feasible, because most of the 1102 work necessary to deduce information about a private key may be performed 1103 offline. The amount of resources required for such an attack would be 1104 significant. However, for an attack on TLS to be meaningful, the server 1105 would have to share the DH1024 private key among multiple clients, which is 1106 no longer an option since CVE-2016-0701. 1107 1108 This only affects processors that support the AVX2 but not ADX extensions 1109 like Intel Haswell (4th generation). 1110 1111 This issue was reported to OpenSSL by David Benjamin (Google). The issue 1112 was originally found via the OSS-Fuzz project. 1113 (CVE-2017-3738) 1114 [Andy Polyakov] 1115 1116 Changes between 1.1.0f and 1.1.0g [2 Nov 2017] 1117 1118 *) bn_sqrx8x_internal carry bug on x86_64 1119 1120 There is a carry propagating bug in the x86_64 Montgomery squaring 1121 procedure. No EC algorithms are affected. Analysis suggests that attacks 1122 against RSA and DSA as a result of this defect would be very difficult to 1123 perform and are not believed likely. Attacks against DH are considered just 1124 feasible (although very difficult) because most of the work necessary to 1125 deduce information about a private key may be performed offline. The amount 1126 of resources required for such an attack would be very significant and 1127 likely only accessible to a limited number of attackers. An attacker would 1128 additionally need online access to an unpatched system using the target 1129 private key in a scenario with persistent DH parameters and a private 1130 key that is shared between multiple clients. 1131 1132 This only affects processors that support the BMI1, BMI2 and ADX extensions 1133 like Intel Broadwell (5th generation) and later or AMD Ryzen. 1134 1135 This issue was reported to OpenSSL by the OSS-Fuzz project. 1136 (CVE-2017-3736) 1137 [Andy Polyakov] 1138 1139 *) Malformed X.509 IPAddressFamily could cause OOB read 1140 1141 If an X.509 certificate has a malformed IPAddressFamily extension, 1142 OpenSSL could do a one-byte buffer overread. The most likely result 1143 would be an erroneous display of the certificate in text format. 1144 1145 This issue was reported to OpenSSL by the OSS-Fuzz project. 1146 (CVE-2017-3735) 1147 [Rich Salz] 1148 1149 Changes between 1.1.0e and 1.1.0f [25 May 2017] 1150 1151 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 1152 platform rather than 'mingw'. 1153 [Richard Levitte] 1154 1155 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 1156 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 1157 which is the minimum version we support. 1158 [Richard Levitte] 1159 1160 Changes between 1.1.0d and 1.1.0e [16 Feb 2017] 1161 1162 *) Encrypt-Then-Mac renegotiation crash 1163 1164 During a renegotiation handshake if the Encrypt-Then-Mac extension is 1165 negotiated where it was not in the original handshake (or vice-versa) then 1166 this can cause OpenSSL to crash (dependant on ciphersuite). Both clients 1167 and servers are affected. 1168 1169 This issue was reported to OpenSSL by Joe Orton (Red Hat). 1170 (CVE-2017-3733) 1171 [Matt Caswell] 1172 1173 Changes between 1.1.0c and 1.1.0d [26 Jan 2017] 1174 1175 *) Truncated packet could crash via OOB read 1176 1177 If one side of an SSL/TLS path is running on a 32-bit host and a specific 1178 cipher is being used, then a truncated packet can cause that host to 1179 perform an out-of-bounds read, usually resulting in a crash. 1180 1181 This issue was reported to OpenSSL by Robert Święcki of Google. 1182 (CVE-2017-3731) 1183 [Andy Polyakov] 1184 1185 *) Bad (EC)DHE parameters cause a client crash 1186 1187 If a malicious server supplies bad parameters for a DHE or ECDHE key 1188 exchange then this can result in the client attempting to dereference a 1189 NULL pointer leading to a client crash. This could be exploited in a Denial 1190 of Service attack. 1191 1192 This issue was reported to OpenSSL by Guido Vranken. 1193 (CVE-2017-3730) 1194 [Matt Caswell] 1195 1196 *) BN_mod_exp may produce incorrect results on x86_64 1197 1198 There is a carry propagating bug in the x86_64 Montgomery squaring 1199 procedure. No EC algorithms are affected. Analysis suggests that attacks 1200 against RSA and DSA as a result of this defect would be very difficult to 1201 perform and are not believed likely. Attacks against DH are considered just 1202 feasible (although very difficult) because most of the work necessary to 1203 deduce information about a private key may be performed offline. The amount 1204 of resources required for such an attack would be very significant and 1205 likely only accessible to a limited number of attackers. An attacker would 1206 additionally need online access to an unpatched system using the target 1207 private key in a scenario with persistent DH parameters and a private 1208 key that is shared between multiple clients. For example this can occur by 1209 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 1210 similar to CVE-2015-3193 but must be treated as a separate problem. 1211 1212 This issue was reported to OpenSSL by the OSS-Fuzz project. 1213 (CVE-2017-3732) 1214 [Andy Polyakov] 1215 1216 Changes between 1.1.0b and 1.1.0c [10 Nov 2016] 1217 1218 *) ChaCha20/Poly1305 heap-buffer-overflow 1219 1220 TLS connections using *-CHACHA20-POLY1305 ciphersuites are susceptible to 1221 a DoS attack by corrupting larger payloads. This can result in an OpenSSL 1222 crash. This issue is not considered to be exploitable beyond a DoS. 1223 1224 This issue was reported to OpenSSL by Robert Święcki (Google Security Team) 1225 (CVE-2016-7054) 1226 [Richard Levitte] 1227 1228 *) CMS Null dereference 1229 1230 Applications parsing invalid CMS structures can crash with a NULL pointer 1231 dereference. This is caused by a bug in the handling of the ASN.1 CHOICE 1232 type in OpenSSL 1.1.0 which can result in a NULL value being passed to the 1233 structure callback if an attempt is made to free certain invalid encodings. 1234 Only CHOICE structures using a callback which do not handle NULL value are 1235 affected. 1236 1237 This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure. 1238 (CVE-2016-7053) 1239 [Stephen Henson] 1240 1241 *) Montgomery multiplication may produce incorrect results 1242 1243 There is a carry propagating bug in the Broadwell-specific Montgomery 1244 multiplication procedure that handles input lengths divisible by, but 1245 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 1246 and DH private keys are impossible. This is because the subroutine in 1247 question is not used in operations with the private key itself and an input 1248 of the attacker's direct choice. Otherwise the bug can manifest itself as 1249 transient authentication and key negotiation failures or reproducible 1250 erroneous outcome of public-key operations with specially crafted input. 1251 Among EC algorithms only Brainpool P-512 curves are affected and one 1252 presumably can attack ECDH key negotiation. Impact was not analyzed in 1253 detail, because pre-requisites for attack are considered unlikely. Namely 1254 multiple clients have to choose the curve in question and the server has to 1255 share the private key among them, neither of which is default behaviour. 1256 Even then only clients that chose the curve will be affected. 1257 1258 This issue was publicly reported as transient failures and was not 1259 initially recognized as a security issue. Thanks to Richard Morgan for 1260 providing reproducible case. 1261 (CVE-2016-7055) 1262 [Andy Polyakov] 1263 1264 *) Removed automatic addition of RPATH in shared libraries and executables, 1265 as this was a remainder from OpenSSL 1.0.x and isn't needed any more. 1266 [Richard Levitte] 1267 1268 Changes between 1.1.0a and 1.1.0b [26 Sep 2016] 1269 1270 *) Fix Use After Free for large message sizes 1271 1272 The patch applied to address CVE-2016-6307 resulted in an issue where if a 1273 message larger than approx 16k is received then the underlying buffer to 1274 store the incoming message is reallocated and moved. Unfortunately a 1275 dangling pointer to the old location is left which results in an attempt to 1276 write to the previously freed location. This is likely to result in a 1277 crash, however it could potentially lead to execution of arbitrary code. 1278 1279 This issue only affects OpenSSL 1.1.0a. 1280 1281 This issue was reported to OpenSSL by Robert Święcki. 1282 (CVE-2016-6309) 1283 [Matt Caswell] 1284 1285 Changes between 1.1.0 and 1.1.0a [22 Sep 2016] 1286 1287 *) OCSP Status Request extension unbounded memory growth 1288 1289 A malicious client can send an excessively large OCSP Status Request 1290 extension. If that client continually requests renegotiation, sending a 1291 large OCSP Status Request extension each time, then there will be unbounded 1292 memory growth on the server. This will eventually lead to a Denial Of 1293 Service attack through memory exhaustion. Servers with a default 1294 configuration are vulnerable even if they do not support OCSP. Builds using 1295 the "no-ocsp" build time option are not affected. 1296 1297 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 1298 (CVE-2016-6304) 1299 [Matt Caswell] 1300 1301 *) SSL_peek() hang on empty record 1302 1303 OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer 1304 sends an empty record. This could be exploited by a malicious peer in a 1305 Denial Of Service attack. 1306 1307 This issue was reported to OpenSSL by Alex Gaynor. 1308 (CVE-2016-6305) 1309 [Matt Caswell] 1310 1311 *) Excessive allocation of memory in tls_get_message_header() and 1312 dtls1_preprocess_fragment() 1313 1314 A (D)TLS message includes 3 bytes for its length in the header for the 1315 message. This would allow for messages up to 16Mb in length. Messages of 1316 this length are excessive and OpenSSL includes a check to ensure that a 1317 peer is sending reasonably sized messages in order to avoid too much memory 1318 being consumed to service a connection. A flaw in the logic of version 1319 1.1.0 means that memory for the message is allocated too early, prior to 1320 the excessive message length check. Due to way memory is allocated in 1321 OpenSSL this could mean an attacker could force up to 21Mb to be allocated 1322 to service a connection. This could lead to a Denial of Service through 1323 memory exhaustion. However, the excessive message length check still takes 1324 place, and this would cause the connection to immediately fail. Assuming 1325 that the application calls SSL_free() on the failed connection in a timely 1326 manner then the 21Mb of allocated memory will then be immediately freed 1327 again. Therefore the excessive memory allocation will be transitory in 1328 nature. This then means that there is only a security impact if: 1329 1330 1) The application does not call SSL_free() in a timely manner in the event 1331 that the connection fails 1332 or 1333 2) The application is working in a constrained environment where there is 1334 very little free memory 1335 or 1336 3) The attacker initiates multiple connection attempts such that there are 1337 multiple connections in a state where memory has been allocated for the 1338 connection; SSL_free() has not yet been called; and there is insufficient 1339 memory to service the multiple requests. 1340 1341 Except in the instance of (1) above any Denial Of Service is likely to be 1342 transitory because as soon as the connection fails the memory is 1343 subsequently freed again in the SSL_free() call. However there is an 1344 increased risk during this period of application crashes due to the lack of 1345 memory - which would then mean a more serious Denial of Service. 1346 1347 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 1348 (CVE-2016-6307 and CVE-2016-6308) 1349 [Matt Caswell] 1350 1351 *) solaris-x86-cc, i.e. 32-bit configuration with vendor compiler, 1352 had to be removed. Primary reason is that vendor assembler can't 1353 assemble our modules with -KPIC flag. As result it, assembly 1354 support, was not even available as option. But its lack means 1355 lack of side-channel resistant code, which is incompatible with 1356 security by todays standards. Fortunately gcc is readily available 1357 prepackaged option, which we firmly point at... 1358 [Andy Polyakov] 1359 1360 Changes between 1.0.2h and 1.1.0 [25 Aug 2016] 1361 1362 *) Windows command-line tool supports UTF-8 opt-in option for arguments 1363 and console input. Setting OPENSSL_WIN32_UTF8 environment variable 1364 (to any value) allows Windows user to access PKCS#12 file generated 1365 with Windows CryptoAPI and protected with non-ASCII password, as well 1366 as files generated under UTF-8 locale on Linux also protected with 1367 non-ASCII password. 1368 [Andy Polyakov] 1369 1370 *) To mitigate the SWEET32 attack (CVE-2016-2183), 3DES cipher suites 1371 have been disabled by default and removed from DEFAULT, just like RC4. 1372 See the RC4 item below to re-enable both. 1373 [Rich Salz] 1374 1375 *) The method for finding the storage location for the Windows RAND seed file 1376 has changed. First we check %RANDFILE%. If that is not set then we check 1377 the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If 1378 all else fails we fall back to C:\. 1379 [Matt Caswell] 1380 1381 *) The EVP_EncryptUpdate() function has had its return type changed from void 1382 to int. A return of 0 indicates and error while a return of 1 indicates 1383 success. 1384 [Matt Caswell] 1385 1386 *) The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and 1387 DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch 1388 off the constant time implementation for RSA, DSA and DH have been made 1389 no-ops and deprecated. 1390 [Matt Caswell] 1391 1392 *) Windows RAND implementation was simplified to only get entropy by 1393 calling CryptGenRandom(). Various other RAND-related tickets 1394 were also closed. 1395 [Joseph Wylie Yandle, Rich Salz] 1396 1397 *) The stack and lhash API's were renamed to start with OPENSSL_SK_ 1398 and OPENSSL_LH_, respectively. The old names are available 1399 with API compatibility. They new names are now completely documented. 1400 [Rich Salz] 1401 1402 *) Unify TYPE_up_ref(obj) methods signature. 1403 SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(), 1404 X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an 1405 int (instead of void) like all others TYPE_up_ref() methods. 1406 So now these methods also check the return value of CRYPTO_atomic_add(), 1407 and the validity of object reference counter. 1408 [fdasilvayy@gmail.com] 1409 1410 *) With Windows Visual Studio builds, the .pdb files are installed 1411 alongside the installed libraries and executables. For a static 1412 library installation, ossl_static.pdb is the associate compiler 1413 generated .pdb file to be used when linking programs. 1414 [Richard Levitte] 1415 1416 *) Remove openssl.spec. Packaging files belong with the packagers. 1417 [Richard Levitte] 1418 1419 *) Automatic Darwin/OSX configuration has had a refresh, it will now 1420 recognise x86_64 architectures automatically. You can still decide 1421 to build for a different bitness with the environment variable 1422 KERNEL_BITS (can be 32 or 64), for example: 1423 1424 KERNEL_BITS=32 ./config 1425 1426 [Richard Levitte] 1427 1428 *) Change default algorithms in pkcs8 utility to use PKCS#5 v2.0, 1429 256 bit AES and HMAC with SHA256. 1430 [Steve Henson] 1431 1432 *) Remove support for MIPS o32 ABI on IRIX (and IRIX only). 1433 [Andy Polyakov] 1434 1435 *) Triple-DES ciphers have been moved from HIGH to MEDIUM. 1436 [Rich Salz] 1437 1438 *) To enable users to have their own config files and build file templates, 1439 Configure looks in the directory indicated by the environment variable 1440 OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/ 1441 directory. On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical 1442 name and is used as is. 1443 [Richard Levitte] 1444 1445 *) The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX, 1446 X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD. The unused type 1447 X509_CERT_FILE_CTX was removed. 1448 [Rich Salz] 1449 1450 *) "shared" builds are now the default. To create only static libraries use 1451 the "no-shared" Configure option. 1452 [Matt Caswell] 1453 1454 *) Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options. 1455 All of these option have not worked for some while and are fundamental 1456 algorithms. 1457 [Matt Caswell] 1458 1459 *) Make various cleanup routines no-ops and mark them as deprecated. Most 1460 global cleanup functions are no longer required because they are handled 1461 via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages). 1462 Explicitly de-initing can cause problems (e.g. where a library that uses 1463 OpenSSL de-inits, but an application is still using it). The affected 1464 functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(), 1465 EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(), 1466 RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and 1467 COMP_zlib_cleanup(). 1468 [Matt Caswell] 1469 1470 *) --strict-warnings no longer enables runtime debugging options 1471 such as REF_DEBUG. Instead, debug options are automatically 1472 enabled with '--debug' builds. 1473 [Andy Polyakov, Emilia Käsper] 1474 1475 *) Made DH and DH_METHOD opaque. The structures for managing DH objects 1476 have been moved out of the public header files. New functions for managing 1477 these have been added. 1478 [Matt Caswell] 1479 1480 *) Made RSA and RSA_METHOD opaque. The structures for managing RSA 1481 objects have been moved out of the public header files. New 1482 functions for managing these have been added. 1483 [Richard Levitte] 1484 1485 *) Made DSA and DSA_METHOD opaque. The structures for managing DSA objects 1486 have been moved out of the public header files. New functions for managing 1487 these have been added. 1488 [Matt Caswell] 1489 1490 *) Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been 1491 moved out of the public header files. New functions for managing these 1492 have been added. 1493 [Matt Caswell] 1494 1495 *) Removed no-rijndael as a config option. Rijndael is an old name for AES. 1496 [Matt Caswell] 1497 1498 *) Removed the mk1mf build scripts. 1499 [Richard Levitte] 1500 1501 *) Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so 1502 it is always safe to #include a header now. 1503 [Rich Salz] 1504 1505 *) Removed the aged BC-32 config and all its supporting scripts 1506 [Richard Levitte] 1507 1508 *) Removed support for Ultrix, Netware, and OS/2. 1509 [Rich Salz] 1510 1511 *) Add support for HKDF. 1512 [Alessandro Ghedini] 1513 1514 *) Add support for blake2b and blake2s 1515 [Bill Cox] 1516 1517 *) Added support for "pipelining". Ciphers that have the 1518 EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple 1519 encryptions/decryptions simultaneously. There are currently no built-in 1520 ciphers with this property but the expectation is that engines will be able 1521 to offer it to significantly improve throughput. Support has been extended 1522 into libssl so that multiple records for a single connection can be 1523 processed in one go (for >=TLS 1.1). 1524 [Matt Caswell] 1525 1526 *) Added the AFALG engine. This is an async capable engine which is able to 1527 offload work to the Linux kernel. In this initial version it only supports 1528 AES128-CBC. The kernel must be version 4.1.0 or greater. 1529 [Catriona Lucey] 1530 1531 *) OpenSSL now uses a new threading API. It is no longer necessary to 1532 set locking callbacks to use OpenSSL in a multi-threaded environment. There 1533 are two supported threading models: pthreads and windows threads. It is 1534 also possible to configure OpenSSL at compile time for "no-threads". The 1535 old threading API should no longer be used. The functions have been 1536 replaced with "no-op" compatibility macros. 1537 [Alessandro Ghedini, Matt Caswell] 1538 1539 *) Modify behavior of ALPN to invoke callback after SNI/servername 1540 callback, such that updates to the SSL_CTX affect ALPN. 1541 [Todd Short] 1542 1543 *) Add SSL_CIPHER queries for authentication and key-exchange. 1544 [Todd Short] 1545 1546 *) Changes to the DEFAULT cipherlist: 1547 - Prefer (EC)DHE handshakes over plain RSA. 1548 - Prefer AEAD ciphers over legacy ciphers. 1549 - Prefer ECDSA over RSA when both certificates are available. 1550 - Prefer TLSv1.2 ciphers/PRF. 1551 - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the 1552 default cipherlist. 1553 [Emilia Käsper] 1554 1555 *) Change the ECC default curve list to be this, in order: x25519, 1556 secp256r1, secp521r1, secp384r1. 1557 [Rich Salz] 1558 1559 *) RC4 based libssl ciphersuites are now classed as "weak" ciphers and are 1560 disabled by default. They can be re-enabled using the 1561 enable-weak-ssl-ciphers option to Configure. 1562 [Matt Caswell] 1563 1564 *) If the server has ALPN configured, but supports no protocols that the 1565 client advertises, send a fatal "no_application_protocol" alert. 1566 This behaviour is SHALL in RFC 7301, though it isn't universally 1567 implemented by other servers. 1568 [Emilia Käsper] 1569 1570 *) Add X25519 support. 1571 Add ASN.1 and EVP_PKEY methods for X25519. This includes support 1572 for public and private key encoding using the format documented in 1573 draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports 1574 key generation and key derivation. 1575 1576 TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses 1577 X25519(29). 1578 [Steve Henson] 1579 1580 *) Deprecate SRP_VBASE_get_by_user. 1581 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 1582 In order to fix an unavoidable memory leak (CVE-2016-0798), 1583 SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP 1584 seed, even if the seed is configured. 1585 1586 Users should use SRP_VBASE_get1_by_user instead. Note that in 1587 SRP_VBASE_get1_by_user, caller must free the returned value. Note 1588 also that even though configuring the SRP seed attempts to hide 1589 invalid usernames by continuing the handshake with fake 1590 credentials, this behaviour is not constant time and no strong 1591 guarantees are made that the handshake is indistinguishable from 1592 that of a valid user. 1593 [Emilia Käsper] 1594 1595 *) Configuration change; it's now possible to build dynamic engines 1596 without having to build shared libraries and vice versa. This 1597 only applies to the engines in engines/, those in crypto/engine/ 1598 will always be built into libcrypto (i.e. "static"). 1599 1600 Building dynamic engines is enabled by default; to disable, use 1601 the configuration option "disable-dynamic-engine". 1602 1603 The only requirements for building dynamic engines are the 1604 presence of the DSO module and building with position independent 1605 code, so they will also automatically be disabled if configuring 1606 with "disable-dso" or "disable-pic". 1607 1608 The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE 1609 are also taken away from openssl/opensslconf.h, as they are 1610 irrelevant. 1611 [Richard Levitte] 1612 1613 *) Configuration change; if there is a known flag to compile 1614 position independent code, it will always be applied on the 1615 libcrypto and libssl object files, and never on the application 1616 object files. This means other libraries that use routines from 1617 libcrypto / libssl can be made into shared libraries regardless 1618 of how OpenSSL was configured. 1619 1620 If this isn't desirable, the configuration options "disable-pic" 1621 or "no-pic" can be used to disable the use of PIC. This will 1622 also disable building shared libraries and dynamic engines. 1623 [Richard Levitte] 1624 1625 *) Removed JPAKE code. It was experimental and has no wide use. 1626 [Rich Salz] 1627 1628 *) The INSTALL_PREFIX Makefile variable has been renamed to 1629 DESTDIR. That makes for less confusion on what this variable 1630 is for. Also, the configuration option --install_prefix is 1631 removed. 1632 [Richard Levitte] 1633 1634 *) Heartbeat for TLS has been removed and is disabled by default 1635 for DTLS; configure with enable-heartbeats. Code that uses the 1636 old #define's might need to be updated. 1637 [Emilia Käsper, Rich Salz] 1638 1639 *) Rename REF_CHECK to REF_DEBUG. 1640 [Rich Salz] 1641 1642 *) New "unified" build system 1643 1644 The "unified" build system is aimed to be a common system for all 1645 platforms we support. With it comes new support for VMS. 1646 1647 This system builds supports building in a different directory tree 1648 than the source tree. It produces one Makefile (for unix family 1649 or lookalikes), or one descrip.mms (for VMS). 1650 1651 The source of information to make the Makefile / descrip.mms is 1652 small files called 'build.info', holding the necessary 1653 information for each directory with source to compile, and a 1654 template in Configurations, like unix-Makefile.tmpl or 1655 descrip.mms.tmpl. 1656 1657 With this change, the library names were also renamed on Windows 1658 and on VMS. They now have names that are closer to the standard 1659 on Unix, and include the major version number, and in certain 1660 cases, the architecture they are built for. See "Notes on shared 1661 libraries" in INSTALL. 1662 1663 We rely heavily on the perl module Text::Template. 1664 [Richard Levitte] 1665 1666 *) Added support for auto-initialisation and de-initialisation of the library. 1667 OpenSSL no longer requires explicit init or deinit routines to be called, 1668 except in certain circumstances. See the OPENSSL_init_crypto() and 1669 OPENSSL_init_ssl() man pages for further information. 1670 [Matt Caswell] 1671 1672 *) The arguments to the DTLSv1_listen function have changed. Specifically the 1673 "peer" argument is now expected to be a BIO_ADDR object. 1674 1675 *) Rewrite of BIO networking library. The BIO library lacked consistent 1676 support of IPv6, and adding it required some more extensive 1677 modifications. This introduces the BIO_ADDR and BIO_ADDRINFO types, 1678 which hold all types of addresses and chains of address information. 1679 It also introduces a new API, with functions like BIO_socket, 1680 BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept. 1681 The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram 1682 have been adapted accordingly. 1683 [Richard Levitte] 1684 1685 *) RSA_padding_check_PKCS1_type_1 now accepts inputs with and without 1686 the leading 0-byte. 1687 [Emilia Käsper] 1688 1689 *) CRIME protection: disable compression by default, even if OpenSSL is 1690 compiled with zlib enabled. Applications can still enable compression 1691 by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by 1692 using the SSL_CONF library to configure compression. 1693 [Emilia Käsper] 1694 1695 *) The signature of the session callback configured with 1696 SSL_CTX_sess_set_get_cb was changed. The read-only input buffer 1697 was explicitly marked as 'const unsigned char*' instead of 1698 'unsigned char*'. 1699 [Emilia Käsper] 1700 1701 *) Always DPURIFY. Remove the use of uninitialized memory in the 1702 RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op. 1703 [Emilia Käsper] 1704 1705 *) Removed many obsolete configuration items, including 1706 DES_PTR, DES_RISC1, DES_RISC2, DES_INT 1707 MD2_CHAR, MD2_INT, MD2_LONG 1708 BF_PTR, BF_PTR2 1709 IDEA_SHORT, IDEA_LONG 1710 RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX 1711 [Rich Salz, with advice from Andy Polyakov] 1712 1713 *) Many BN internals have been moved to an internal header file. 1714 [Rich Salz with help from Andy Polyakov] 1715 1716 *) Configuration and writing out the results from it has changed. 1717 Files such as Makefile include/openssl/opensslconf.h and are now 1718 produced through general templates, such as Makefile.in and 1719 crypto/opensslconf.h.in and some help from the perl module 1720 Text::Template. 1721 1722 Also, the center of configuration information is no longer 1723 Makefile. Instead, Configure produces a perl module in 1724 configdata.pm which holds most of the config data (in the hash 1725 table %config), the target data that comes from the target 1726 configuration in one of the Configurations/*.conf files (in 1727 %target). 1728 [Richard Levitte] 1729 1730 *) To clarify their intended purposes, the Configure options 1731 --prefix and --openssldir change their semantics, and become more 1732 straightforward and less interdependent. 1733 1734 --prefix shall be used exclusively to give the location INSTALLTOP 1735 where programs, scripts, libraries, include files and manuals are 1736 going to be installed. The default is now /usr/local. 1737 1738 --openssldir shall be used exclusively to give the default 1739 location OPENSSLDIR where certificates, private keys, CRLs are 1740 managed. This is also where the default openssl.cnf gets 1741 installed. 1742 If the directory given with this option is a relative path, the 1743 values of both the --prefix value and the --openssldir value will 1744 be combined to become OPENSSLDIR. 1745 The default for --openssldir is INSTALLTOP/ssl. 1746 1747 Anyone who uses --openssldir to specify where OpenSSL is to be 1748 installed MUST change to use --prefix instead. 1749 [Richard Levitte] 1750 1751 *) The GOST engine was out of date and therefore it has been removed. An up 1752 to date GOST engine is now being maintained in an external repository. 1753 See: https://wiki.openssl.org/index.php/Binaries. Libssl still retains 1754 support for GOST ciphersuites (these are only activated if a GOST engine 1755 is present). 1756 [Matt Caswell] 1757 1758 *) EGD is no longer supported by default; use enable-egd when 1759 configuring. 1760 [Ben Kaduk and Rich Salz] 1761 1762 *) The distribution now has Makefile.in files, which are used to 1763 create Makefile's when Configure is run. *Configure must be run 1764 before trying to build now.* 1765 [Rich Salz] 1766 1767 *) The return value for SSL_CIPHER_description() for error conditions 1768 has changed. 1769 [Rich Salz] 1770 1771 *) Support for RFC6698/RFC7671 DANE TLSA peer authentication. 1772 1773 Obtaining and performing DNSSEC validation of TLSA records is 1774 the application's responsibility. The application provides 1775 the TLSA records of its choice to OpenSSL, and these are then 1776 used to authenticate the peer. 1777 1778 The TLSA records need not even come from DNS. They can, for 1779 example, be used to implement local end-entity certificate or 1780 trust-anchor "pinning", where the "pin" data takes the form 1781 of TLSA records, which can augment or replace verification 1782 based on the usual WebPKI public certification authorities. 1783 [Viktor Dukhovni] 1784 1785 *) Revert default OPENSSL_NO_DEPRECATED setting. Instead OpenSSL 1786 continues to support deprecated interfaces in default builds. 1787 However, applications are strongly advised to compile their 1788 source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides 1789 the declarations of all interfaces deprecated in 0.9.8, 1.0.0 1790 or the 1.1.0 releases. 1791 1792 In environments in which all applications have been ported to 1793 not use any deprecated interfaces OpenSSL's Configure script 1794 should be used with the --api=1.1.0 option to entirely remove 1795 support for the deprecated features from the library and 1796 unconditionally disable them in the installed headers. 1797 Essentially the same effect can be achieved with the "no-deprecated" 1798 argument to Configure, except that this will always restrict 1799 the build to just the latest API, rather than a fixed API 1800 version. 1801 1802 As applications are ported to future revisions of the API, 1803 they should update their compile-time OPENSSL_API_COMPAT define 1804 accordingly, but in most cases should be able to continue to 1805 compile with later releases. 1806 1807 The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are 1808 0x10000000L and 0x00908000L, respectively. However those 1809 versions did not support the OPENSSL_API_COMPAT feature, and 1810 so applications are not typically tested for explicit support 1811 of just the undeprecated features of either release. 1812 [Viktor Dukhovni] 1813 1814 *) Add support for setting the minimum and maximum supported protocol. 1815 It can bet set via the SSL_set_min_proto_version() and 1816 SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and 1817 MaxProtocol. It's recommended to use the new APIs to disable 1818 protocols instead of disabling individual protocols using 1819 SSL_set_options() or SSL_CONF's Protocol. This change also 1820 removes support for disabling TLS 1.2 in the OpenSSL TLS 1821 client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT. 1822 [Kurt Roeckx] 1823 1824 *) Support for ChaCha20 and Poly1305 added to libcrypto and libssl. 1825 [Andy Polyakov] 1826 1827 *) New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD 1828 and integrates ECDSA and ECDH functionality into EC. Implementations can 1829 now redirect key generation and no longer need to convert to or from 1830 ECDSA_SIG format. 1831 1832 Note: the ecdsa.h and ecdh.h headers are now no longer needed and just 1833 include the ec.h header file instead. 1834 [Steve Henson] 1835 1836 *) Remove support for all 40 and 56 bit ciphers. This includes all the export 1837 ciphers who are no longer supported and drops support the ephemeral RSA key 1838 exchange. The LOW ciphers currently doesn't have any ciphers in it. 1839 [Kurt Roeckx] 1840 1841 *) Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX 1842 opaque. For HMAC_CTX, the following constructors and destructors 1843 were added: 1844 1845 HMAC_CTX *HMAC_CTX_new(void); 1846 void HMAC_CTX_free(HMAC_CTX *ctx); 1847 1848 For EVP_MD and EVP_CIPHER, complete APIs to create, fill and 1849 destroy such methods has been added. See EVP_MD_meth_new(3) and 1850 EVP_CIPHER_meth_new(3) for documentation. 1851 1852 Additional changes: 1853 1) EVP_MD_CTX_cleanup(), EVP_CIPHER_CTX_cleanup() and 1854 HMAC_CTX_cleanup() were removed. HMAC_CTX_reset() and 1855 EVP_MD_CTX_reset() should be called instead to reinitialise 1856 an already created structure. 1857 2) For consistency with the majority of our object creators and 1858 destructors, EVP_MD_CTX_(create|destroy) were renamed to 1859 EVP_MD_CTX_(new|free). The old names are retained as macros 1860 for deprecated builds. 1861 [Richard Levitte] 1862 1863 *) Added ASYNC support. Libcrypto now includes the async sub-library to enable 1864 cryptographic operations to be performed asynchronously as long as an 1865 asynchronous capable engine is used. See the ASYNC_start_job() man page for 1866 further details. Libssl has also had this capability integrated with the 1867 introduction of the new mode SSL_MODE_ASYNC and associated error 1868 SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man 1869 pages. This work was developed in partnership with Intel Corp. 1870 [Matt Caswell] 1871 1872 *) SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is 1873 always enabled now. If you want to disable the support you should 1874 exclude it using the list of supported ciphers. This also means that the 1875 "-no_ecdhe" option has been removed from s_server. 1876 [Kurt Roeckx] 1877 1878 *) SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls 1879 SSL_{CTX_}set1_curves() which can set a list. 1880 [Kurt Roeckx] 1881 1882 *) Remove support for SSL_{CTX_}set_tmp_ecdh_callback(). You should set the 1883 curve you want to support using SSL_{CTX_}set1_curves(). 1884 [Kurt Roeckx] 1885 1886 *) State machine rewrite. The state machine code has been significantly 1887 refactored in order to remove much duplication of code and solve issues 1888 with the old code (see ssl/statem/README for further details). This change 1889 does have some associated API changes. Notably the SSL_state() function 1890 has been removed and replaced by SSL_get_state which now returns an 1891 "OSSL_HANDSHAKE_STATE" instead of an int. SSL_set_state() has been removed 1892 altogether. The previous handshake states defined in ssl.h and ssl3.h have 1893 also been removed. 1894 [Matt Caswell] 1895 1896 *) All instances of the string "ssleay" in the public API were replaced 1897 with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's) 1898 Some error codes related to internal RSA_eay API's were renamed. 1899 [Rich Salz] 1900 1901 *) The demo files in crypto/threads were moved to demo/threads. 1902 [Rich Salz] 1903 1904 *) Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp, 1905 sureware and ubsec. 1906 [Matt Caswell, Rich Salz] 1907 1908 *) New ASN.1 embed macro. 1909 1910 New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the 1911 structure is not allocated: it is part of the parent. That is instead of 1912 1913 FOO *x; 1914 1915 it must be: 1916 1917 FOO x; 1918 1919 This reduces memory fragmentation and make it impossible to accidentally 1920 set a mandatory field to NULL. 1921 1922 This currently only works for some fields specifically a SEQUENCE, CHOICE, 1923 or ASN1_STRING type which is part of a parent SEQUENCE. Since it is 1924 equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or 1925 SEQUENCE OF. 1926 [Steve Henson] 1927 1928 *) Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled. 1929 [Emilia Käsper] 1930 1931 *) Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although 1932 in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also 1933 an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add 1934 DES and RC4 ciphersuites. 1935 [Matt Caswell] 1936 1937 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 1938 This changes the decoding behaviour for some invalid messages, 1939 though the change is mostly in the more lenient direction, and 1940 legacy behaviour is preserved as much as possible. 1941 [Emilia Käsper] 1942 1943 *) Fix no-stdio build. 1944 [ David Woodhouse <David.Woodhouse@intel.com> and also 1945 Ivan Nestlerode <ivan.nestlerode@sonos.com> ] 1946 1947 *) New testing framework 1948 The testing framework has been largely rewritten and is now using 1949 perl and the perl modules Test::Harness and an extended variant of 1950 Test::More called OpenSSL::Test to do its work. All test scripts in 1951 test/ have been rewritten into test recipes, and all direct calls to 1952 executables in test/Makefile have become individual recipes using the 1953 simplified testing OpenSSL::Test::Simple. 1954 1955 For documentation on our testing modules, do: 1956 1957 perldoc test/testlib/OpenSSL/Test/Simple.pm 1958 perldoc test/testlib/OpenSSL/Test.pm 1959 1960 [Richard Levitte] 1961 1962 *) Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT 1963 are used; the latter aborts on memory leaks (usually checked on exit). 1964 Some undocumented "set malloc, etc., hooks" functions were removed 1965 and others were changed. All are now documented. 1966 [Rich Salz] 1967 1968 *) In DSA_generate_parameters_ex, if the provided seed is too short, 1969 return an error 1970 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 1971 1972 *) Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites 1973 from RFC4279, RFC4785, RFC5487, RFC5489. 1974 1975 Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the 1976 original RSA_PSK patch. 1977 [Steve Henson] 1978 1979 *) Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay 1980 era flag was never set throughout the codebase (only read). Also removed 1981 SSL3_FLAGS_POP_BUFFER which was only used if 1982 SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set. 1983 [Matt Caswell] 1984 1985 *) Changed the default name options in the "ca", "crl", "req" and "x509" 1986 to be "oneline" instead of "compat". 1987 [Richard Levitte] 1988 1989 *) Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're 1990 not aware of clients that still exhibit this bug, and the workaround 1991 hasn't been working properly for a while. 1992 [Emilia Käsper] 1993 1994 *) The return type of BIO_number_read() and BIO_number_written() as well as 1995 the corresponding num_read and num_write members in the BIO structure has 1996 changed from unsigned long to uint64_t. On platforms where an unsigned 1997 long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is 1998 transferred. 1999 [Matt Caswell] 2000 2001 *) Given the pervasive nature of TLS extensions it is inadvisable to run 2002 OpenSSL without support for them. It also means that maintaining 2003 the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably 2004 not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed. 2005 [Matt Caswell] 2006 2007 *) Removed support for the two export grade static DH ciphersuites 2008 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 2009 were newly added (along with a number of other static DH ciphersuites) to 2010 1.0.2. However the two export ones have *never* worked since they were 2011 introduced. It seems strange in any case to be adding new export 2012 ciphersuites, and given "logjam" it also does not seem correct to fix them. 2013 [Matt Caswell] 2014 2015 *) Version negotiation has been rewritten. In particular SSLv23_method(), 2016 SSLv23_client_method() and SSLv23_server_method() have been deprecated, 2017 and turned into macros which simply call the new preferred function names 2018 TLS_method(), TLS_client_method() and TLS_server_method(). All new code 2019 should use the new names instead. Also as part of this change the ssl23.h 2020 header file has been removed. 2021 [Matt Caswell] 2022 2023 *) Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This 2024 code and the associated standard is no longer considered fit-for-purpose. 2025 [Matt Caswell] 2026 2027 *) RT2547 was closed. When generating a private key, try to make the 2028 output file readable only by the owner. This behavior change might 2029 be noticeable when interacting with other software. 2030 2031 *) Documented all exdata functions. Added CRYPTO_free_ex_index. 2032 Added a test. 2033 [Rich Salz] 2034 2035 *) Added HTTP GET support to the ocsp command. 2036 [Rich Salz] 2037 2038 *) Changed default digest for the dgst and enc commands from MD5 to 2039 sha256 2040 [Rich Salz] 2041 2042 *) RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead. 2043 [Matt Caswell] 2044 2045 *) Added support for TLS extended master secret from 2046 draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an 2047 initial patch which was a great help during development. 2048 [Steve Henson] 2049 2050 *) All libssl internal structures have been removed from the public header 2051 files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is 2052 now redundant). Users should not attempt to access internal structures 2053 directly. Instead they should use the provided API functions. 2054 [Matt Caswell] 2055 2056 *) config has been changed so that by default OPENSSL_NO_DEPRECATED is used. 2057 Access to deprecated functions can be re-enabled by running config with 2058 "enable-deprecated". In addition applications wishing to use deprecated 2059 functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour 2060 will, by default, disable some transitive includes that previously existed 2061 in the header files (e.g. ec.h will no longer, by default, include bn.h) 2062 [Matt Caswell] 2063 2064 *) Added support for OCB mode. OpenSSL has been granted a patent license 2065 compatible with the OpenSSL license for use of OCB. Details are available 2066 at https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf. Support 2067 for OCB can be removed by calling config with no-ocb. 2068 [Matt Caswell] 2069 2070 *) SSLv2 support has been removed. It still supports receiving a SSLv2 2071 compatible client hello. 2072 [Kurt Roeckx] 2073 2074 *) Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz], 2075 done while fixing the error code for the key-too-small case. 2076 [Annie Yousar <a.yousar@informatik.hu-berlin.de>] 2077 2078 *) CA.sh has been removed; use CA.pl instead. 2079 [Rich Salz] 2080 2081 *) Removed old DES API. 2082 [Rich Salz] 2083 2084 *) Remove various unsupported platforms: 2085 Sony NEWS4 2086 BEOS and BEOS_R5 2087 NeXT 2088 SUNOS 2089 MPE/iX 2090 Sinix/ReliantUNIX RM400 2091 DGUX 2092 NCR 2093 Tandem 2094 Cray 2095 16-bit platforms such as WIN16 2096 [Rich Salz] 2097 2098 *) Clean up OPENSSL_NO_xxx #define's 2099 Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF 2100 Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx 2101 OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC 2102 OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160 2103 OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO 2104 Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY 2105 OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP 2106 OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK 2107 OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY 2108 Remove MS_STATIC; it's a relic from platforms <32 bits. 2109 [Rich Salz] 2110 2111 *) Cleaned up dead code 2112 Remove all but one '#ifdef undef' which is to be looked at. 2113 [Rich Salz] 2114 2115 *) Clean up calling of xxx_free routines. 2116 Just like free(), fix most of the xxx_free routines to accept 2117 NULL. Remove the non-null checks from callers. Save much code. 2118 [Rich Salz] 2119 2120 *) Add secure heap for storage of private keys (when possible). 2121 Add BIO_s_secmem(), CBIGNUM, etc. 2122 Contributed by Akamai Technologies under our Corporate CLA. 2123 [Rich Salz] 2124 2125 *) Experimental support for a new, fast, unbiased prime candidate generator, 2126 bn_probable_prime_dh_coprime(). Not currently used by any prime generator. 2127 [Felix Laurie von Massenbach <felix@erbridge.co.uk>] 2128 2129 *) New output format NSS in the sess_id command line tool. This allows 2130 exporting the session id and the master key in NSS keylog format. 2131 [Martin Kaiser <martin@kaiser.cx>] 2132 2133 *) Harmonize version and its documentation. -f flag is used to display 2134 compilation flags. 2135 [mancha <mancha1@zoho.com>] 2136 2137 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 2138 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 2139 [mancha <mancha1@zoho.com>] 2140 2141 *) Fix some double frees. These are not thought to be exploitable. 2142 [mancha <mancha1@zoho.com>] 2143 2144 *) A missing bounds check in the handling of the TLS heartbeat extension 2145 can be used to reveal up to 64k of memory to a connected client or 2146 server. 2147 2148 Thanks for Neel Mehta of Google Security for discovering this bug and to 2149 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 2150 preparing the fix (CVE-2014-0160) 2151 [Adam Langley, Bodo Moeller] 2152 2153 *) Fix for the attack described in the paper "Recovering OpenSSL 2154 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 2155 by Yuval Yarom and Naomi Benger. Details can be obtained from: 2156 http://eprint.iacr.org/2014/140 2157 2158 Thanks to Yuval Yarom and Naomi Benger for discovering this 2159 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 2160 [Yuval Yarom and Naomi Benger] 2161 2162 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 2163 this fixes a limitation in previous versions of OpenSSL. 2164 [Steve Henson] 2165 2166 *) Experimental encrypt-then-mac support. 2167 2168 Experimental support for encrypt then mac from 2169 draft-gutmann-tls-encrypt-then-mac-02.txt 2170 2171 To enable it set the appropriate extension number (0x42 for the test 2172 server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42 2173 2174 For non-compliant peers (i.e. just about everything) this should have no 2175 effect. 2176 2177 WARNING: EXPERIMENTAL, SUBJECT TO CHANGE. 2178 2179 [Steve Henson] 2180 2181 *) Add EVP support for key wrapping algorithms, to avoid problems with 2182 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 2183 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 2184 algorithms and include tests cases. 2185 [Steve Henson] 2186 2187 *) Extend CMS code to support RSA-PSS signatures and RSA-OAEP for 2188 enveloped data. 2189 [Steve Henson] 2190 2191 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 2192 MGF1 digest and OAEP label. 2193 [Steve Henson] 2194 2195 *) Make openssl verify return errors. 2196 [Chris Palmer <palmer@google.com> and Ben Laurie] 2197 2198 *) New function ASN1_TIME_diff to calculate the difference between two 2199 ASN1_TIME structures or one structure and the current time. 2200 [Steve Henson] 2201 2202 *) Update fips_test_suite to support multiple command line options. New 2203 test to induce all self test errors in sequence and check expected 2204 failures. 2205 [Steve Henson] 2206 2207 *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and 2208 sign or verify all in one operation. 2209 [Steve Henson] 2210 2211 *) Add fips_algvs: a multicall fips utility incorporating all the algorithm 2212 test programs and fips_test_suite. Includes functionality to parse 2213 the minimal script output of fipsalgest.pl directly. 2214 [Steve Henson] 2215 2216 *) Add authorisation parameter to FIPS_module_mode_set(). 2217 [Steve Henson] 2218 2219 *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves. 2220 [Steve Henson] 2221 2222 *) Use separate DRBG fields for internal and external flags. New function 2223 FIPS_drbg_health_check() to perform on demand health checking. Add 2224 generation tests to fips_test_suite with reduced health check interval to 2225 demonstrate periodic health checking. Add "nodh" option to 2226 fips_test_suite to skip very slow DH test. 2227 [Steve Henson] 2228 2229 *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers 2230 based on NID. 2231 [Steve Henson] 2232 2233 *) More extensive health check for DRBG checking many more failure modes. 2234 New function FIPS_selftest_drbg_all() to handle every possible DRBG 2235 combination: call this in fips_test_suite. 2236 [Steve Henson] 2237 2238 *) Add support for canonical generation of DSA parameter 'g'. See 2239 FIPS 186-3 A.2.3. 2240 2241 *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and 2242 POST to handle HMAC cases. 2243 [Steve Henson] 2244 2245 *) Add functions FIPS_module_version() and FIPS_module_version_text() 2246 to return numerical and string versions of the FIPS module number. 2247 [Steve Henson] 2248 2249 *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and 2250 FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented 2251 outside the validated module in the FIPS capable OpenSSL. 2252 [Steve Henson] 2253 2254 *) Minor change to DRBG entropy callback semantics. In some cases 2255 there is no multiple of the block length between min_len and 2256 max_len. Allow the callback to return more than max_len bytes 2257 of entropy but discard any extra: it is the callback's responsibility 2258 to ensure that the extra data discarded does not impact the 2259 requested amount of entropy. 2260 [Steve Henson] 2261 2262 *) Add PRNG security strength checks to RSA, DSA and ECDSA using 2263 information in FIPS186-3, SP800-57 and SP800-131A. 2264 [Steve Henson] 2265 2266 *) CCM support via EVP. Interface is very similar to GCM case except we 2267 must supply all data in one chunk (i.e. no update, final) and the 2268 message length must be supplied if AAD is used. Add algorithm test 2269 support. 2270 [Steve Henson] 2271 2272 *) Initial version of POST overhaul. Add POST callback to allow the status 2273 of POST to be monitored and/or failures induced. Modify fips_test_suite 2274 to use callback. Always run all selftests even if one fails. 2275 [Steve Henson] 2276 2277 *) XTS support including algorithm test driver in the fips_gcmtest program. 2278 Note: this does increase the maximum key length from 32 to 64 bytes but 2279 there should be no binary compatibility issues as existing applications 2280 will never use XTS mode. 2281 [Steve Henson] 2282 2283 *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies 2284 to OpenSSL RAND code and replace with a tiny FIPS RAND API which also 2285 performs algorithm blocking for unapproved PRNG types. Also do not 2286 set PRNG type in FIPS_mode_set(): leave this to the application. 2287 Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with 2288 the standard OpenSSL PRNG: set additional data to a date time vector. 2289 [Steve Henson] 2290 2291 *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*. 2292 This shouldn't present any incompatibility problems because applications 2293 shouldn't be using these directly and any that are will need to rethink 2294 anyway as the X9.31 PRNG is now deprecated by FIPS 140-2 2295 [Steve Henson] 2296 2297 *) Extensive self tests and health checking required by SP800-90 DRBG. 2298 Remove strength parameter from FIPS_drbg_instantiate and always 2299 instantiate at maximum supported strength. 2300 [Steve Henson] 2301 2302 *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing. 2303 [Steve Henson] 2304 2305 *) New algorithm test program fips_dhvs to handle DH primitives only testing. 2306 [Steve Henson] 2307 2308 *) New function DH_compute_key_padded() to compute a DH key and pad with 2309 leading zeroes if needed: this complies with SP800-56A et al. 2310 [Steve Henson] 2311 2312 *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by 2313 anything, incomplete, subject to change and largely untested at present. 2314 [Steve Henson] 2315 2316 *) Modify fipscanisteronly build option to only build the necessary object 2317 files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile. 2318 [Steve Henson] 2319 2320 *) Add experimental option FIPSSYMS to give all symbols in 2321 fipscanister.o and FIPS or fips prefix. This will avoid 2322 conflicts with future versions of OpenSSL. Add perl script 2323 util/fipsas.pl to preprocess assembly language source files 2324 and rename any affected symbols. 2325 [Steve Henson] 2326 2327 *) Add selftest checks and algorithm block of non-fips algorithms in 2328 FIPS mode. Remove DES2 from selftests. 2329 [Steve Henson] 2330 2331 *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just 2332 return internal method without any ENGINE dependencies. Add new 2333 tiny fips sign and verify functions. 2334 [Steve Henson] 2335 2336 *) New build option no-ec2m to disable characteristic 2 code. 2337 [Steve Henson] 2338 2339 *) New build option "fipscanisteronly". This only builds fipscanister.o 2340 and (currently) associated fips utilities. Uses the file Makefile.fips 2341 instead of Makefile.org as the prototype. 2342 [Steve Henson] 2343 2344 *) Add some FIPS mode restrictions to GCM. Add internal IV generator. 2345 Update fips_gcmtest to use IV generator. 2346 [Steve Henson] 2347 2348 *) Initial, experimental EVP support for AES-GCM. AAD can be input by 2349 setting output buffer to NULL. The *Final function must be 2350 called although it will not retrieve any additional data. The tag 2351 can be set or retrieved with a ctrl. The IV length is by default 12 2352 bytes (96 bits) but can be set to an alternative value. If the IV 2353 length exceeds the maximum IV length (currently 16 bytes) it cannot be 2354 set before the key. 2355 [Steve Henson] 2356 2357 *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the 2358 underlying do_cipher function handles all cipher semantics itself 2359 including padding and finalisation. This is useful if (for example) 2360 an ENGINE cipher handles block padding itself. The behaviour of 2361 do_cipher is subtly changed if this flag is set: the return value 2362 is the number of characters written to the output buffer (zero is 2363 no longer an error code) or a negative error code. Also if the 2364 input buffer is NULL and length 0 finalisation should be performed. 2365 [Steve Henson] 2366 2367 *) If a candidate issuer certificate is already part of the constructed 2368 path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case. 2369 [Steve Henson] 2370 2371 *) Improve forward-security support: add functions 2372 2373 void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure)) 2374 void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure)) 2375 2376 for use by SSL/TLS servers; the callback function will be called whenever a 2377 new session is created, and gets to decide whether the session may be 2378 cached to make it resumable (return 0) or not (return 1). (As by the 2379 SSL/TLS protocol specifications, the session_id sent by the server will be 2380 empty to indicate that the session is not resumable; also, the server will 2381 not generate RFC 4507 (RFC 5077) session tickets.) 2382 2383 A simple reasonable callback implementation is to return is_forward_secure. 2384 This parameter will be set to 1 or 0 depending on the ciphersuite selected 2385 by the SSL/TLS server library, indicating whether it can provide forward 2386 security. 2387 [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)] 2388 2389 *) New -verify_name option in command line utilities to set verification 2390 parameters by name. 2391 [Steve Henson] 2392 2393 *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE. 2394 Add CMAC pkey methods. 2395 [Steve Henson] 2396 2397 *) Experimental renegotiation in s_server -www mode. If the client 2398 browses /reneg connection is renegotiated. If /renegcert it is 2399 renegotiated requesting a certificate. 2400 [Steve Henson] 2401 2402 *) Add an "external" session cache for debugging purposes to s_server. This 2403 should help trace issues which normally are only apparent in deployed 2404 multi-process servers. 2405 [Steve Henson] 2406 2407 *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where 2408 return value is ignored. NB. The functions RAND_add(), RAND_seed(), 2409 BIO_set_cipher() and some obscure PEM functions were changed so they 2410 can now return an error. The RAND changes required a change to the 2411 RAND_METHOD structure. 2412 [Steve Henson] 2413 2414 *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of 2415 a gcc attribute to warn if the result of a function is ignored. This 2416 is enable if DEBUG_UNUSED is set. Add to several functions in evp.h 2417 whose return value is often ignored. 2418 [Steve Henson] 2419 2420 *) New -noct, -requestct, -requirect and -ctlogfile options for s_client. 2421 These allow SCTs (signed certificate timestamps) to be requested and 2422 validated when establishing a connection. 2423 [Rob Percival <robpercival@google.com>] 2424 2425 Changes between 1.0.2g and 1.0.2h [3 May 2016] 2426 2427 *) Prevent padding oracle in AES-NI CBC MAC check 2428 2429 A MITM attacker can use a padding oracle attack to decrypt traffic 2430 when the connection uses an AES CBC cipher and the server support 2431 AES-NI. 2432 2433 This issue was introduced as part of the fix for Lucky 13 padding 2434 attack (CVE-2013-0169). The padding check was rewritten to be in 2435 constant time by making sure that always the same bytes are read and 2436 compared against either the MAC or padding bytes. But it no longer 2437 checked that there was enough data to have both the MAC and padding 2438 bytes. 2439 2440 This issue was reported by Juraj Somorovsky using TLS-Attacker. 2441 (CVE-2016-2107) 2442 [Kurt Roeckx] 2443 2444 *) Fix EVP_EncodeUpdate overflow 2445 2446 An overflow can occur in the EVP_EncodeUpdate() function which is used for 2447 Base64 encoding of binary data. If an attacker is able to supply very large 2448 amounts of input data then a length check can overflow resulting in a heap 2449 corruption. 2450 2451 Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by 2452 the PEM_write_bio* family of functions. These are mainly used within the 2453 OpenSSL command line applications, so any application which processes data 2454 from an untrusted source and outputs it as a PEM file should be considered 2455 vulnerable to this issue. User applications that call these APIs directly 2456 with large amounts of untrusted data may also be vulnerable. 2457 2458 This issue was reported by Guido Vranken. 2459 (CVE-2016-2105) 2460 [Matt Caswell] 2461 2462 *) Fix EVP_EncryptUpdate overflow 2463 2464 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 2465 is able to supply very large amounts of input data after a previous call to 2466 EVP_EncryptUpdate() with a partial block then a length check can overflow 2467 resulting in a heap corruption. Following an analysis of all OpenSSL 2468 internal usage of the EVP_EncryptUpdate() function all usage is one of two 2469 forms. The first form is where the EVP_EncryptUpdate() call is known to be 2470 the first called function after an EVP_EncryptInit(), and therefore that 2471 specific call must be safe. The second form is where the length passed to 2472 EVP_EncryptUpdate() can be seen from the code to be some small value and 2473 therefore there is no possibility of an overflow. Since all instances are 2474 one of these two forms, it is believed that there can be no overflows in 2475 internal code due to this problem. It should be noted that 2476 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 2477 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 2478 of these calls have also been analysed too and it is believed there are no 2479 instances in internal usage where an overflow could occur. 2480 2481 This issue was reported by Guido Vranken. 2482 (CVE-2016-2106) 2483 [Matt Caswell] 2484 2485 *) Prevent ASN.1 BIO excessive memory allocation 2486 2487 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 2488 a short invalid encoding can cause allocation of large amounts of memory 2489 potentially consuming excessive resources or exhausting memory. 2490 2491 Any application parsing untrusted data through d2i BIO functions is 2492 affected. The memory based functions such as d2i_X509() are *not* affected. 2493 Since the memory based functions are used by the TLS library, TLS 2494 applications are not affected. 2495 2496 This issue was reported by Brian Carpenter. 2497 (CVE-2016-2109) 2498 [Stephen Henson] 2499 2500 *) EBCDIC overread 2501 2502 ASN1 Strings that are over 1024 bytes can cause an overread in applications 2503 using the X509_NAME_oneline() function on EBCDIC systems. This could result 2504 in arbitrary stack data being returned in the buffer. 2505 2506 This issue was reported by Guido Vranken. 2507 (CVE-2016-2176) 2508 [Matt Caswell] 2509 2510 *) Modify behavior of ALPN to invoke callback after SNI/servername 2511 callback, such that updates to the SSL_CTX affect ALPN. 2512 [Todd Short] 2513 2514 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the 2515 default. 2516 [Kurt Roeckx] 2517 2518 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the 2519 methods are enabled and ssl2 is disabled the methods return NULL. 2520 [Kurt Roeckx] 2521 2522 Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 2523 2524 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 2525 Builds that are not configured with "enable-weak-ssl-ciphers" will not 2526 provide any "EXPORT" or "LOW" strength ciphers. 2527 [Viktor Dukhovni] 2528 2529 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 2530 is by default disabled at build-time. Builds that are not configured with 2531 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 2532 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 2533 will need to explicitly call either of: 2534 2535 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 2536 or 2537 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 2538 2539 as appropriate. Even if either of those is used, or the application 2540 explicitly uses the version-specific SSLv2_method() or its client and 2541 server variants, SSLv2 ciphers vulnerable to exhaustive search key 2542 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 2543 ciphers, and SSLv2 56-bit DES are no longer available. 2544 (CVE-2016-0800) 2545 [Viktor Dukhovni] 2546 2547 *) Fix a double-free in DSA code 2548 2549 A double free bug was discovered when OpenSSL parses malformed DSA private 2550 keys and could lead to a DoS attack or memory corruption for applications 2551 that receive DSA private keys from untrusted sources. This scenario is 2552 considered rare. 2553 2554 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 2555 libFuzzer. 2556 (CVE-2016-0705) 2557 [Stephen Henson] 2558 2559 *) Disable SRP fake user seed to address a server memory leak. 2560 2561 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 2562 2563 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 2564 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 2565 was changed to ignore the "fake user" SRP seed, even if the seed 2566 is configured. 2567 2568 Users should use SRP_VBASE_get1_by_user instead. Note that in 2569 SRP_VBASE_get1_by_user, caller must free the returned value. Note 2570 also that even though configuring the SRP seed attempts to hide 2571 invalid usernames by continuing the handshake with fake 2572 credentials, this behaviour is not constant time and no strong 2573 guarantees are made that the handshake is indistinguishable from 2574 that of a valid user. 2575 (CVE-2016-0798) 2576 [Emilia Käsper] 2577 2578 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 2579 2580 In the BN_hex2bn function the number of hex digits is calculated using an 2581 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 2582 large values of |i| this can result in |bn_expand| not allocating any 2583 memory because |i * 4| is negative. This can leave the internal BIGNUM data 2584 field as NULL leading to a subsequent NULL ptr deref. For very large values 2585 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 2586 In this case memory is allocated to the internal BIGNUM data field, but it 2587 is insufficiently sized leading to heap corruption. A similar issue exists 2588 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 2589 is ever called by user applications with very large untrusted hex/dec data. 2590 This is anticipated to be a rare occurrence. 2591 2592 All OpenSSL internal usage of these functions use data that is not expected 2593 to be untrusted, e.g. config file data or application command line 2594 arguments. If user developed applications generate config file data based 2595 on untrusted data then it is possible that this could also lead to security 2596 consequences. This is also anticipated to be rare. 2597 2598 This issue was reported to OpenSSL by Guido Vranken. 2599 (CVE-2016-0797) 2600 [Matt Caswell] 2601 2602 *) Fix memory issues in BIO_*printf functions 2603 2604 The internal |fmtstr| function used in processing a "%s" format string in 2605 the BIO_*printf functions could overflow while calculating the length of a 2606 string and cause an OOB read when printing very long strings. 2607 2608 Additionally the internal |doapr_outch| function can attempt to write to an 2609 OOB memory location (at an offset from the NULL pointer) in the event of a 2610 memory allocation failure. In 1.0.2 and below this could be caused where 2611 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 2612 could be in processing a very long "%s" format string. Memory leaks can 2613 also occur. 2614 2615 The first issue may mask the second issue dependent on compiler behaviour. 2616 These problems could enable attacks where large amounts of untrusted data 2617 is passed to the BIO_*printf functions. If applications use these functions 2618 in this way then they could be vulnerable. OpenSSL itself uses these 2619 functions when printing out human-readable dumps of ASN.1 data. Therefore 2620 applications that print this data could be vulnerable if the data is from 2621 untrusted sources. OpenSSL command line applications could also be 2622 vulnerable where they print out ASN.1 data, or if untrusted data is passed 2623 as command line arguments. 2624 2625 Libssl is not considered directly vulnerable. Additionally certificates etc 2626 received via remote connections via libssl are also unlikely to be able to 2627 trigger these issues because of message size limits enforced within libssl. 2628 2629 This issue was reported to OpenSSL Guido Vranken. 2630 (CVE-2016-0799) 2631 [Matt Caswell] 2632 2633 *) Side channel attack on modular exponentiation 2634 2635 A side-channel attack was found which makes use of cache-bank conflicts on 2636 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 2637 of RSA keys. The ability to exploit this issue is limited as it relies on 2638 an attacker who has control of code in a thread running on the same 2639 hyper-threaded core as the victim thread which is performing decryptions. 2640 2641 This issue was reported to OpenSSL by Yuval Yarom, The University of 2642 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 2643 Nadia Heninger, University of Pennsylvania with more information at 2644 http://cachebleed.info. 2645 (CVE-2016-0702) 2646 [Andy Polyakov] 2647 2648 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 2649 if no keysize is specified with default_bits. This fixes an 2650 omission in an earlier change that changed all RSA/DSA key generation 2651 apps to use 2048 bits by default. 2652 [Emilia Käsper] 2653 2654 Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 2655 *) DH small subgroups 2656 2657 Historically OpenSSL only ever generated DH parameters based on "safe" 2658 primes. More recently (in version 1.0.2) support was provided for 2659 generating X9.42 style parameter files such as those required for RFC 5114 2660 support. The primes used in such files may not be "safe". Where an 2661 application is using DH configured with parameters based on primes that are 2662 not "safe" then an attacker could use this fact to find a peer's private 2663 DH exponent. This attack requires that the attacker complete multiple 2664 handshakes in which the peer uses the same private DH exponent. For example 2665 this could be used to discover a TLS server's private DH exponent if it's 2666 reusing the private DH exponent or it's using a static DH ciphersuite. 2667 2668 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 2669 TLS. It is not on by default. If the option is not set then the server 2670 reuses the same private DH exponent for the life of the server process and 2671 would be vulnerable to this attack. It is believed that many popular 2672 applications do set this option and would therefore not be at risk. 2673 2674 The fix for this issue adds an additional check where a "q" parameter is 2675 available (as is the case in X9.42 based parameters). This detects the 2676 only known attack, and is the only possible defense for static DH 2677 ciphersuites. This could have some performance impact. 2678 2679 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 2680 default and cannot be disabled. This could have some performance impact. 2681 2682 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 2683 (CVE-2016-0701) 2684 [Matt Caswell] 2685 2686 *) SSLv2 doesn't block disabled ciphers 2687 2688 A malicious client can negotiate SSLv2 ciphers that have been disabled on 2689 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 2690 been disabled, provided that the SSLv2 protocol was not also disabled via 2691 SSL_OP_NO_SSLv2. 2692 2693 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 2694 and Sebastian Schinzel. 2695 (CVE-2015-3197) 2696 [Viktor Dukhovni] 2697 2698 Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 2699 2700 *) BN_mod_exp may produce incorrect results on x86_64 2701 2702 There is a carry propagating bug in the x86_64 Montgomery squaring 2703 procedure. No EC algorithms are affected. Analysis suggests that attacks 2704 against RSA and DSA as a result of this defect would be very difficult to 2705 perform and are not believed likely. Attacks against DH are considered just 2706 feasible (although very difficult) because most of the work necessary to 2707 deduce information about a private key may be performed offline. The amount 2708 of resources required for such an attack would be very significant and 2709 likely only accessible to a limited number of attackers. An attacker would 2710 additionally need online access to an unpatched system using the target 2711 private key in a scenario with persistent DH parameters and a private 2712 key that is shared between multiple clients. For example this can occur by 2713 default in OpenSSL DHE based SSL/TLS ciphersuites. 2714 2715 This issue was reported to OpenSSL by Hanno Böck. 2716 (CVE-2015-3193) 2717 [Andy Polyakov] 2718 2719 *) Certificate verify crash with missing PSS parameter 2720 2721 The signature verification routines will crash with a NULL pointer 2722 dereference if presented with an ASN.1 signature using the RSA PSS 2723 algorithm and absent mask generation function parameter. Since these 2724 routines are used to verify certificate signature algorithms this can be 2725 used to crash any certificate verification operation and exploited in a 2726 DoS attack. Any application which performs certificate verification is 2727 vulnerable including OpenSSL clients and servers which enable client 2728 authentication. 2729 2730 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG). 2731 (CVE-2015-3194) 2732 [Stephen Henson] 2733 2734 *) X509_ATTRIBUTE memory leak 2735 2736 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 2737 memory. This structure is used by the PKCS#7 and CMS routines so any 2738 application which reads PKCS#7 or CMS data from untrusted sources is 2739 affected. SSL/TLS is not affected. 2740 2741 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 2742 libFuzzer. 2743 (CVE-2015-3195) 2744 [Stephen Henson] 2745 2746 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 2747 This changes the decoding behaviour for some invalid messages, 2748 though the change is mostly in the more lenient direction, and 2749 legacy behaviour is preserved as much as possible. 2750 [Emilia Käsper] 2751 2752 *) In DSA_generate_parameters_ex, if the provided seed is too short, 2753 return an error 2754 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 2755 2756 Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 2757 2758 *) Alternate chains certificate forgery 2759 2760 During certificate verification, OpenSSL will attempt to find an 2761 alternative certificate chain if the first attempt to build such a chain 2762 fails. An error in the implementation of this logic can mean that an 2763 attacker could cause certain checks on untrusted certificates to be 2764 bypassed, such as the CA flag, enabling them to use a valid leaf 2765 certificate to act as a CA and "issue" an invalid certificate. 2766 2767 This issue was reported to OpenSSL by Adam Langley/David Benjamin 2768 (Google/BoringSSL). 2769 [Matt Caswell] 2770 2771 Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 2772 2773 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 2774 incompatibility in the handling of HMAC. The previous ABI has now been 2775 restored. 2776 [Matt Caswell] 2777 2778 Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 2779 2780 *) Malformed ECParameters causes infinite loop 2781 2782 When processing an ECParameters structure OpenSSL enters an infinite loop 2783 if the curve specified is over a specially malformed binary polynomial 2784 field. 2785 2786 This can be used to perform denial of service against any 2787 system which processes public keys, certificate requests or 2788 certificates. This includes TLS clients and TLS servers with 2789 client authentication enabled. 2790 2791 This issue was reported to OpenSSL by Joseph Barr-Pixton. 2792 (CVE-2015-1788) 2793 [Andy Polyakov] 2794 2795 *) Exploitable out-of-bounds read in X509_cmp_time 2796 2797 X509_cmp_time does not properly check the length of the ASN1_TIME 2798 string and can read a few bytes out of bounds. In addition, 2799 X509_cmp_time accepts an arbitrary number of fractional seconds in the 2800 time string. 2801 2802 An attacker can use this to craft malformed certificates and CRLs of 2803 various sizes and potentially cause a segmentation fault, resulting in 2804 a DoS on applications that verify certificates or CRLs. TLS clients 2805 that verify CRLs are affected. TLS clients and servers with client 2806 authentication enabled may be affected if they use custom verification 2807 callbacks. 2808 2809 This issue was reported to OpenSSL by Robert Swiecki (Google), and 2810 independently by Hanno Böck. 2811 (CVE-2015-1789) 2812 [Emilia Käsper] 2813 2814 *) PKCS7 crash with missing EnvelopedContent 2815 2816 The PKCS#7 parsing code does not handle missing inner EncryptedContent 2817 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 2818 with missing content and trigger a NULL pointer dereference on parsing. 2819 2820 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 2821 structures from untrusted sources are affected. OpenSSL clients and 2822 servers are not affected. 2823 2824 This issue was reported to OpenSSL by Michal Zalewski (Google). 2825 (CVE-2015-1790) 2826 [Emilia Käsper] 2827 2828 *) CMS verify infinite loop with unknown hash function 2829 2830 When verifying a signedData message the CMS code can enter an infinite loop 2831 if presented with an unknown hash function OID. This can be used to perform 2832 denial of service against any system which verifies signedData messages using 2833 the CMS code. 2834 This issue was reported to OpenSSL by Johannes Bauer. 2835 (CVE-2015-1792) 2836 [Stephen Henson] 2837 2838 *) Race condition handling NewSessionTicket 2839 2840 If a NewSessionTicket is received by a multi-threaded client when attempting to 2841 reuse a previous ticket then a race condition can occur potentially leading to 2842 a double free of the ticket data. 2843 (CVE-2015-1791) 2844 [Matt Caswell] 2845 2846 *) Only support 256-bit or stronger elliptic curves with the 2847 'ecdh_auto' setting (server) or by default (client). Of supported 2848 curves, prefer P-256 (both). 2849 [Emilia Kasper] 2850 2851 Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 2852 2853 *) ClientHello sigalgs DoS fix 2854 2855 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 2856 invalid signature algorithms extension a NULL pointer dereference will 2857 occur. This can be exploited in a DoS attack against the server. 2858 2859 This issue was was reported to OpenSSL by David Ramos of Stanford 2860 University. 2861 (CVE-2015-0291) 2862 [Stephen Henson and Matt Caswell] 2863 2864 *) Multiblock corrupted pointer fix 2865 2866 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 2867 feature only applies on 64 bit x86 architecture platforms that support AES 2868 NI instructions. A defect in the implementation of "multiblock" can cause 2869 OpenSSL's internal write buffer to become incorrectly set to NULL when 2870 using non-blocking IO. Typically, when the user application is using a 2871 socket BIO for writing, this will only result in a failed connection. 2872 However if some other BIO is used then it is likely that a segmentation 2873 fault will be triggered, thus enabling a potential DoS attack. 2874 2875 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 2876 (CVE-2015-0290) 2877 [Matt Caswell] 2878 2879 *) Segmentation fault in DTLSv1_listen fix 2880 2881 The DTLSv1_listen function is intended to be stateless and processes the 2882 initial ClientHello from many peers. It is common for user code to loop 2883 over the call to DTLSv1_listen until a valid ClientHello is received with 2884 an associated cookie. A defect in the implementation of DTLSv1_listen means 2885 that state is preserved in the SSL object from one invocation to the next 2886 that can lead to a segmentation fault. Errors processing the initial 2887 ClientHello can trigger this scenario. An example of such an error could be 2888 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 2889 server. 2890 2891 This issue was reported to OpenSSL by Per Allansson. 2892 (CVE-2015-0207) 2893 [Matt Caswell] 2894 2895 *) Segmentation fault in ASN1_TYPE_cmp fix 2896 2897 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 2898 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 2899 certificate signature algorithm consistency this can be used to crash any 2900 certificate verification operation and exploited in a DoS attack. Any 2901 application which performs certificate verification is vulnerable including 2902 OpenSSL clients and servers which enable client authentication. 2903 (CVE-2015-0286) 2904 [Stephen Henson] 2905 2906 *) Segmentation fault for invalid PSS parameters fix 2907 2908 The signature verification routines will crash with a NULL pointer 2909 dereference if presented with an ASN.1 signature using the RSA PSS 2910 algorithm and invalid parameters. Since these routines are used to verify 2911 certificate signature algorithms this can be used to crash any 2912 certificate verification operation and exploited in a DoS attack. Any 2913 application which performs certificate verification is vulnerable including 2914 OpenSSL clients and servers which enable client authentication. 2915 2916 This issue was was reported to OpenSSL by Brian Carpenter. 2917 (CVE-2015-0208) 2918 [Stephen Henson] 2919 2920 *) ASN.1 structure reuse memory corruption fix 2921 2922 Reusing a structure in ASN.1 parsing may allow an attacker to cause 2923 memory corruption via an invalid write. Such reuse is and has been 2924 strongly discouraged and is believed to be rare. 2925 2926 Applications that parse structures containing CHOICE or ANY DEFINED BY 2927 components may be affected. Certificate parsing (d2i_X509 and related 2928 functions) are however not affected. OpenSSL clients and servers are 2929 not affected. 2930 (CVE-2015-0287) 2931 [Stephen Henson] 2932 2933 *) PKCS7 NULL pointer dereferences fix 2934 2935 The PKCS#7 parsing code does not handle missing outer ContentInfo 2936 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 2937 missing content and trigger a NULL pointer dereference on parsing. 2938 2939 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 2940 otherwise parse PKCS#7 structures from untrusted sources are 2941 affected. OpenSSL clients and servers are not affected. 2942 2943 This issue was reported to OpenSSL by Michal Zalewski (Google). 2944 (CVE-2015-0289) 2945 [Emilia Käsper] 2946 2947 *) DoS via reachable assert in SSLv2 servers fix 2948 2949 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 2950 servers that both support SSLv2 and enable export cipher suites by sending 2951 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 2952 2953 This issue was discovered by Sean Burford (Google) and Emilia Käsper 2954 (OpenSSL development team). 2955 (CVE-2015-0293) 2956 [Emilia Käsper] 2957 2958 *) Empty CKE with client auth and DHE fix 2959 2960 If client auth is used then a server can seg fault in the event of a DHE 2961 ciphersuite being selected and a zero length ClientKeyExchange message 2962 being sent by the client. This could be exploited in a DoS attack. 2963 (CVE-2015-1787) 2964 [Matt Caswell] 2965 2966 *) Handshake with unseeded PRNG fix 2967 2968 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 2969 with an unseeded PRNG. The conditions are: 2970 - The client is on a platform where the PRNG has not been seeded 2971 automatically, and the user has not seeded manually 2972 - A protocol specific client method version has been used (i.e. not 2973 SSL_client_methodv23) 2974 - A ciphersuite is used that does not require additional random data from 2975 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 2976 2977 If the handshake succeeds then the client random that has been used will 2978 have been generated from a PRNG with insufficient entropy and therefore the 2979 output may be predictable. 2980 2981 For example using the following command with an unseeded openssl will 2982 succeed on an unpatched platform: 2983 2984 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 2985 (CVE-2015-0285) 2986 [Matt Caswell] 2987 2988 *) Use After Free following d2i_ECPrivatekey error fix 2989 2990 A malformed EC private key file consumed via the d2i_ECPrivateKey function 2991 could cause a use after free condition. This, in turn, could cause a double 2992 free in several private key parsing functions (such as d2i_PrivateKey 2993 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 2994 for applications that receive EC private keys from untrusted 2995 sources. This scenario is considered rare. 2996 2997 This issue was discovered by the BoringSSL project and fixed in their 2998 commit 517073cd4b. 2999 (CVE-2015-0209) 3000 [Matt Caswell] 3001 3002 *) X509_to_X509_REQ NULL pointer deref fix 3003 3004 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 3005 the certificate key is invalid. This function is rarely used in practice. 3006 3007 This issue was discovered by Brian Carpenter. 3008 (CVE-2015-0288) 3009 [Stephen Henson] 3010 3011 *) Removed the export ciphers from the DEFAULT ciphers 3012 [Kurt Roeckx] 3013 3014 Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 3015 3016 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 3017 ARMv5 through ARMv8, as opposite to "locking" it to single one. 3018 So far those who have to target multiple platforms would compromise 3019 and argue that binary targeting say ARMv5 would still execute on 3020 ARMv8. "Universal" build resolves this compromise by providing 3021 near-optimal performance even on newer platforms. 3022 [Andy Polyakov] 3023 3024 *) Accelerated NIST P-256 elliptic curve implementation for x86_64 3025 (other platforms pending). 3026 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov] 3027 3028 *) Add support for the SignedCertificateTimestampList certificate and 3029 OCSP response extensions from RFC6962. 3030 [Rob Stradling] 3031 3032 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 3033 for corner cases. (Certain input points at infinity could lead to 3034 bogus results, with non-infinity inputs mapped to infinity too.) 3035 [Bodo Moeller] 3036 3037 *) Initial support for PowerISA 2.0.7, first implemented in POWER8. 3038 This covers AES, SHA256/512 and GHASH. "Initial" means that most 3039 common cases are optimized and there still is room for further 3040 improvements. Vector Permutation AES for Altivec is also added. 3041 [Andy Polyakov] 3042 3043 *) Add support for little-endian ppc64 Linux target. 3044 [Marcelo Cerri (IBM)] 3045 3046 *) Initial support for AMRv8 ISA crypto extensions. This covers AES, 3047 SHA1, SHA256 and GHASH. "Initial" means that most common cases 3048 are optimized and there still is room for further improvements. 3049 Both 32- and 64-bit modes are supported. 3050 [Andy Polyakov, Ard Biesheuvel (Linaro)] 3051 3052 *) Improved ARMv7 NEON support. 3053 [Andy Polyakov] 3054 3055 *) Support for SPARC Architecture 2011 crypto extensions, first 3056 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 3057 SHA256/512, MD5, GHASH and modular exponentiation. 3058 [Andy Polyakov, David Miller] 3059 3060 *) Accelerated modular exponentiation for Intel processors, a.k.a. 3061 RSAZ. 3062 [Shay Gueron & Vlad Krasnov (Intel Corp)] 3063 3064 *) Support for new and upcoming Intel processors, including AVX2, 3065 BMI and SHA ISA extensions. This includes additional "stitched" 3066 implementations, AESNI-SHA256 and GCM, and multi-buffer support 3067 for TLS encrypt. 3068 3069 This work was sponsored by Intel Corp. 3070 [Andy Polyakov] 3071 3072 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 3073 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 3074 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 3075 [Steve Henson] 3076 3077 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 3078 this fixes a limitation in previous versions of OpenSSL. 3079 [Steve Henson] 3080 3081 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 3082 MGF1 digest and OAEP label. 3083 [Steve Henson] 3084 3085 *) Add EVP support for key wrapping algorithms, to avoid problems with 3086 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 3087 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 3088 algorithms and include tests cases. 3089 [Steve Henson] 3090 3091 *) Add functions to allocate and set the fields of an ECDSA_METHOD 3092 structure. 3093 [Douglas E. Engert, Steve Henson] 3094 3095 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 3096 difference in days and seconds between two tm or ASN1_TIME structures. 3097 [Steve Henson] 3098 3099 *) Add -rev test option to s_server to just reverse order of characters 3100 received by client and send back to server. Also prints an abbreviated 3101 summary of the connection parameters. 3102 [Steve Henson] 3103 3104 *) New option -brief for s_client and s_server to print out a brief summary 3105 of connection parameters. 3106 [Steve Henson] 3107 3108 *) Add callbacks for arbitrary TLS extensions. 3109 [Trevor Perrin <trevp@trevp.net> and Ben Laurie] 3110 3111 *) New option -crl_download in several openssl utilities to download CRLs 3112 from CRLDP extension in certificates. 3113 [Steve Henson] 3114 3115 *) New options -CRL and -CRLform for s_client and s_server for CRLs. 3116 [Steve Henson] 3117 3118 *) New function X509_CRL_diff to generate a delta CRL from the difference 3119 of two full CRLs. Add support to "crl" utility. 3120 [Steve Henson] 3121 3122 *) New functions to set lookup_crls function and to retrieve 3123 X509_STORE from X509_STORE_CTX. 3124 [Steve Henson] 3125 3126 *) Print out deprecated issuer and subject unique ID fields in 3127 certificates. 3128 [Steve Henson] 3129 3130 *) Extend OCSP I/O functions so they can be used for simple general purpose 3131 HTTP as well as OCSP. New wrapper function which can be used to download 3132 CRLs using the OCSP API. 3133 [Steve Henson] 3134 3135 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs. 3136 [Steve Henson] 3137 3138 *) SSL_CONF* functions. These provide a common framework for application 3139 configuration using configuration files or command lines. 3140 [Steve Henson] 3141 3142 *) SSL/TLS tracing code. This parses out SSL/TLS records using the 3143 message callback and prints the results. Needs compile time option 3144 "enable-ssl-trace". New options to s_client and s_server to enable 3145 tracing. 3146 [Steve Henson] 3147 3148 *) New ctrl and macro to retrieve supported points extensions. 3149 Print out extension in s_server and s_client. 3150 [Steve Henson] 3151 3152 *) New functions to retrieve certificate signature and signature 3153 OID NID. 3154 [Steve Henson] 3155 3156 *) Add functions to retrieve and manipulate the raw cipherlist sent by a 3157 client to OpenSSL. 3158 [Steve Henson] 3159 3160 *) New Suite B modes for TLS code. These use and enforce the requirements 3161 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 3162 only use Suite B curves. The Suite B modes can be set by using the 3163 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 3164 [Steve Henson] 3165 3166 *) New chain verification flags for Suite B levels of security. Check 3167 algorithms are acceptable when flags are set in X509_verify_cert. 3168 [Steve Henson] 3169 3170 *) Make tls1_check_chain return a set of flags indicating checks passed 3171 by a certificate chain. Add additional tests to handle client 3172 certificates: checks for matching certificate type and issuer name 3173 comparison. 3174 [Steve Henson] 3175 3176 *) If an attempt is made to use a signature algorithm not in the peer 3177 preference list abort the handshake. If client has no suitable 3178 signature algorithms in response to a certificate request do not 3179 use the certificate. 3180 [Steve Henson] 3181 3182 *) If server EC tmp key is not in client preference list abort handshake. 3183 [Steve Henson] 3184 3185 *) Add support for certificate stores in CERT structure. This makes it 3186 possible to have different stores per SSL structure or one store in 3187 the parent SSL_CTX. Include distinct stores for certificate chain 3188 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 3189 to build and store a certificate chain in CERT structure: returning 3190 an error if the chain cannot be built: this will allow applications 3191 to test if a chain is correctly configured. 3192 3193 Note: if the CERT based stores are not set then the parent SSL_CTX 3194 store is used to retain compatibility with existing behaviour. 3195 3196 [Steve Henson] 3197 3198 *) New function ssl_set_client_disabled to set a ciphersuite disabled 3199 mask based on the current session, check mask when sending client 3200 hello and checking the requested ciphersuite. 3201 [Steve Henson] 3202 3203 *) New ctrls to retrieve and set certificate types in a certificate 3204 request message. Print out received values in s_client. If certificate 3205 types is not set with custom values set sensible values based on 3206 supported signature algorithms. 3207 [Steve Henson] 3208 3209 *) Support for distinct client and server supported signature algorithms. 3210 [Steve Henson] 3211 3212 *) Add certificate callback. If set this is called whenever a certificate 3213 is required by client or server. An application can decide which 3214 certificate chain to present based on arbitrary criteria: for example 3215 supported signature algorithms. Add very simple example to s_server. 3216 This fixes many of the problems and restrictions of the existing client 3217 certificate callback: for example you can now clear an existing 3218 certificate and specify the whole chain. 3219 [Steve Henson] 3220 3221 *) Add new "valid_flags" field to CERT_PKEY structure which determines what 3222 the certificate can be used for (if anything). Set valid_flags field 3223 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 3224 to have similar checks in it. 3225 3226 Add new "cert_flags" field to CERT structure and include a "strict mode". 3227 This enforces some TLS certificate requirements (such as only permitting 3228 certificate signature algorithms contained in the supported algorithms 3229 extension) which some implementations ignore: this option should be used 3230 with caution as it could cause interoperability issues. 3231 [Steve Henson] 3232 3233 *) Update and tidy signature algorithm extension processing. Work out 3234 shared signature algorithms based on preferences and peer algorithms 3235 and print them out in s_client and s_server. Abort handshake if no 3236 shared signature algorithms. 3237 [Steve Henson] 3238 3239 *) Add new functions to allow customised supported signature algorithms 3240 for SSL and SSL_CTX structures. Add options to s_client and s_server 3241 to support them. 3242 [Steve Henson] 3243 3244 *) New function SSL_certs_clear() to delete all references to certificates 3245 from an SSL structure. Before this once a certificate had been added 3246 it couldn't be removed. 3247 [Steve Henson] 3248 3249 *) Integrate hostname, email address and IP address checking with certificate 3250 verification. New verify options supporting checking in openssl utility. 3251 [Steve Henson] 3252 3253 *) Fixes and wildcard matching support to hostname and email checking 3254 functions. Add manual page. 3255 [Florian Weimer (Red Hat Product Security Team)] 3256 3257 *) New functions to check a hostname email or IP address against a 3258 certificate. Add options x509 utility to print results of checks against 3259 a certificate. 3260 [Steve Henson] 3261 3262 *) Fix OCSP checking. 3263 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie] 3264 3265 *) Initial experimental support for explicitly trusted non-root CAs. 3266 OpenSSL still tries to build a complete chain to a root but if an 3267 intermediate CA has a trust setting included that is used. The first 3268 setting is used: whether to trust (e.g., -addtrust option to the x509 3269 utility) or reject. 3270 [Steve Henson] 3271 3272 *) Add -trusted_first option which attempts to find certificates in the 3273 trusted store even if an untrusted chain is also supplied. 3274 [Steve Henson] 3275 3276 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 3277 platform support for Linux and Android. 3278 [Andy Polyakov] 3279 3280 *) Support for linux-x32, ILP32 environment in x86_64 framework. 3281 [Andy Polyakov] 3282 3283 *) Experimental multi-implementation support for FIPS capable OpenSSL. 3284 When in FIPS mode the approved implementations are used as normal, 3285 when not in FIPS mode the internal unapproved versions are used instead. 3286 This means that the FIPS capable OpenSSL isn't forced to use the 3287 (often lower performance) FIPS implementations outside FIPS mode. 3288 [Steve Henson] 3289 3290 *) Transparently support X9.42 DH parameters when calling 3291 PEM_read_bio_DHparameters. This means existing applications can handle 3292 the new parameter format automatically. 3293 [Steve Henson] 3294 3295 *) Initial experimental support for X9.42 DH parameter format: mainly 3296 to support use of 'q' parameter for RFC5114 parameters. 3297 [Steve Henson] 3298 3299 *) Add DH parameters from RFC5114 including test data to dhtest. 3300 [Steve Henson] 3301 3302 *) Support for automatic EC temporary key parameter selection. If enabled 3303 the most preferred EC parameters are automatically used instead of 3304 hardcoded fixed parameters. Now a server just has to call: 3305 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 3306 support ECDH and use the most appropriate parameters. 3307 [Steve Henson] 3308 3309 *) Enhance and tidy EC curve and point format TLS extension code. Use 3310 static structures instead of allocation if default values are used. 3311 New ctrls to set curves we wish to support and to retrieve shared curves. 3312 Print out shared curves in s_server. New options to s_server and s_client 3313 to set list of supported curves. 3314 [Steve Henson] 3315 3316 *) New ctrls to retrieve supported signature algorithms and 3317 supported curve values as an array of NIDs. Extend openssl utility 3318 to print out received values. 3319 [Steve Henson] 3320 3321 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 3322 between NIDs and the more common NIST names such as "P-256". Enhance 3323 ecparam utility and ECC method to recognise the NIST names for curves. 3324 [Steve Henson] 3325 3326 *) Enhance SSL/TLS certificate chain handling to support different 3327 chains for each certificate instead of one chain in the parent SSL_CTX. 3328 [Steve Henson] 3329 3330 *) Support for fixed DH ciphersuite client authentication: where both 3331 server and client use DH certificates with common parameters. 3332 [Steve Henson] 3333 3334 *) Support for fixed DH ciphersuites: those requiring DH server 3335 certificates. 3336 [Steve Henson] 3337 3338 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of 3339 the certificate. 3340 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 3341 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 3342 X509_CINF_get_signature were reverted post internal team review. 3343 3344 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 3345 3346 *) Build fixes for the Windows and OpenVMS platforms 3347 [Matt Caswell and Richard Levitte] 3348 3349 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 3350 3351 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 3352 message can cause a segmentation fault in OpenSSL due to a NULL pointer 3353 dereference. This could lead to a Denial Of Service attack. Thanks to 3354 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 3355 (CVE-2014-3571) 3356 [Steve Henson] 3357 3358 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 3359 dtls1_buffer_record function under certain conditions. In particular this 3360 could occur if an attacker sent repeated DTLS records with the same 3361 sequence number but for the next epoch. The memory leak could be exploited 3362 by an attacker in a Denial of Service attack through memory exhaustion. 3363 Thanks to Chris Mueller for reporting this issue. 3364 (CVE-2015-0206) 3365 [Matt Caswell] 3366 3367 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 3368 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 3369 method would be set to NULL which could later result in a NULL pointer 3370 dereference. Thanks to Frank Schmirler for reporting this issue. 3371 (CVE-2014-3569) 3372 [Kurt Roeckx] 3373 3374 *) Abort handshake if server key exchange message is omitted for ephemeral 3375 ECDH ciphersuites. 3376 3377 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 3378 reporting this issue. 3379 (CVE-2014-3572) 3380 [Steve Henson] 3381 3382 *) Remove non-export ephemeral RSA code on client and server. This code 3383 violated the TLS standard by allowing the use of temporary RSA keys in 3384 non-export ciphersuites and could be used by a server to effectively 3385 downgrade the RSA key length used to a value smaller than the server 3386 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 3387 INRIA or reporting this issue. 3388 (CVE-2015-0204) 3389 [Steve Henson] 3390 3391 *) Fixed issue where DH client certificates are accepted without verification. 3392 An OpenSSL server will accept a DH certificate for client authentication 3393 without the certificate verify message. This effectively allows a client to 3394 authenticate without the use of a private key. This only affects servers 3395 which trust a client certificate authority which issues certificates 3396 containing DH keys: these are extremely rare and hardly ever encountered. 3397 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 3398 this issue. 3399 (CVE-2015-0205) 3400 [Steve Henson] 3401 3402 *) Ensure that the session ID context of an SSL is updated when its 3403 SSL_CTX is updated via SSL_set_SSL_CTX. 3404 3405 The session ID context is typically set from the parent SSL_CTX, 3406 and can vary with the CTX. 3407 [Adam Langley] 3408 3409 *) Fix various certificate fingerprint issues. 3410 3411 By using non-DER or invalid encodings outside the signed portion of a 3412 certificate the fingerprint can be changed without breaking the signature. 3413 Although no details of the signed portion of the certificate can be changed 3414 this can cause problems with some applications: e.g. those using the 3415 certificate fingerprint for blacklists. 3416 3417 1. Reject signatures with non zero unused bits. 3418 3419 If the BIT STRING containing the signature has non zero unused bits reject 3420 the signature. All current signature algorithms require zero unused bits. 3421 3422 2. Check certificate algorithm consistency. 3423 3424 Check the AlgorithmIdentifier inside TBS matches the one in the 3425 certificate signature. NB: this will result in signature failure 3426 errors for some broken certificates. 3427 3428 Thanks to Konrad Kraszewski from Google for reporting this issue. 3429 3430 3. Check DSA/ECDSA signatures use DER. 3431 3432 Re-encode DSA/ECDSA signatures and compare with the original received 3433 signature. Return an error if there is a mismatch. 3434 3435 This will reject various cases including garbage after signature 3436 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 3437 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 3438 (negative or with leading zeroes). 3439 3440 Further analysis was conducted and fixes were developed by Stephen Henson 3441 of the OpenSSL core team. 3442 3443 (CVE-2014-8275) 3444 [Steve Henson] 3445 3446 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 3447 results on some platforms, including x86_64. This bug occurs at random 3448 with a very low probability, and is not known to be exploitable in any 3449 way, though its exact impact is difficult to determine. Thanks to Pieter 3450 Wuille (Blockstream) who reported this issue and also suggested an initial 3451 fix. Further analysis was conducted by the OpenSSL development team and 3452 Adam Langley of Google. The final fix was developed by Andy Polyakov of 3453 the OpenSSL core team. 3454 (CVE-2014-3570) 3455 [Andy Polyakov] 3456 3457 *) Do not resume sessions on the server if the negotiated protocol 3458 version does not match the session's version. Resuming with a different 3459 version, while not strictly forbidden by the RFC, is of questionable 3460 sanity and breaks all known clients. 3461 [David Benjamin, Emilia Käsper] 3462 3463 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 3464 early CCS messages during renegotiation. (Note that because 3465 renegotiation is encrypted, this early CCS was not exploitable.) 3466 [Emilia Käsper] 3467 3468 *) Tighten client-side session ticket handling during renegotiation: 3469 ensure that the client only accepts a session ticket if the server sends 3470 the extension anew in the ServerHello. Previously, a TLS client would 3471 reuse the old extension state and thus accept a session ticket if one was 3472 announced in the initial ServerHello. 3473 3474 Similarly, ensure that the client requires a session ticket if one 3475 was advertised in the ServerHello. Previously, a TLS client would 3476 ignore a missing NewSessionTicket message. 3477 [Emilia Käsper] 3478 3479 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 3480 3481 *) SRTP Memory Leak. 3482 3483 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 3484 sends a carefully crafted handshake message, to cause OpenSSL to fail 3485 to free up to 64k of memory causing a memory leak. This could be 3486 exploited in a Denial Of Service attack. This issue affects OpenSSL 3487 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 3488 whether SRTP is used or configured. Implementations of OpenSSL that 3489 have been compiled with OPENSSL_NO_SRTP defined are not affected. 3490 3491 The fix was developed by the OpenSSL team. 3492 (CVE-2014-3513) 3493 [OpenSSL team] 3494 3495 *) Session Ticket Memory Leak. 3496 3497 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 3498 integrity of that ticket is first verified. In the event of a session 3499 ticket integrity check failing, OpenSSL will fail to free memory 3500 causing a memory leak. By sending a large number of invalid session 3501 tickets an attacker could exploit this issue in a Denial Of Service 3502 attack. 3503 (CVE-2014-3567) 3504 [Steve Henson] 3505 3506 *) Build option no-ssl3 is incomplete. 3507 3508 When OpenSSL is configured with "no-ssl3" as a build option, servers 3509 could accept and complete a SSL 3.0 handshake, and clients could be 3510 configured to send them. 3511 (CVE-2014-3568) 3512 [Akamai and the OpenSSL team] 3513 3514 *) Add support for TLS_FALLBACK_SCSV. 3515 Client applications doing fallback retries should call 3516 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 3517 (CVE-2014-3566) 3518 [Adam Langley, Bodo Moeller] 3519 3520 *) Add additional DigestInfo checks. 3521 3522 Re-encode DigestInto in DER and check against the original when 3523 verifying RSA signature: this will reject any improperly encoded 3524 DigestInfo structures. 3525 3526 Note: this is a precautionary measure and no attacks are currently known. 3527 3528 [Steve Henson] 3529 3530 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 3531 3532 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 3533 SRP code can be overrun an internal buffer. Add sanity check that 3534 g, A, B < N to SRP code. 3535 3536 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 3537 Group for discovering this issue. 3538 (CVE-2014-3512) 3539 [Steve Henson] 3540 3541 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 3542 TLS 1.0 instead of higher protocol versions when the ClientHello message 3543 is badly fragmented. This allows a man-in-the-middle attacker to force a 3544 downgrade to TLS 1.0 even if both the server and the client support a 3545 higher protocol version, by modifying the client's TLS records. 3546 3547 Thanks to David Benjamin and Adam Langley (Google) for discovering and 3548 researching this issue. 3549 (CVE-2014-3511) 3550 [David Benjamin] 3551 3552 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 3553 to a denial of service attack. A malicious server can crash the client 3554 with a null pointer dereference (read) by specifying an anonymous (EC)DH 3555 ciphersuite and sending carefully crafted handshake messages. 3556 3557 Thanks to Felix Gröbert (Google) for discovering and researching this 3558 issue. 3559 (CVE-2014-3510) 3560 [Emilia Käsper] 3561 3562 *) By sending carefully crafted DTLS packets an attacker could cause openssl 3563 to leak memory. This can be exploited through a Denial of Service attack. 3564 Thanks to Adam Langley for discovering and researching this issue. 3565 (CVE-2014-3507) 3566 [Adam Langley] 3567 3568 *) An attacker can force openssl to consume large amounts of memory whilst 3569 processing DTLS handshake messages. This can be exploited through a 3570 Denial of Service attack. 3571 Thanks to Adam Langley for discovering and researching this issue. 3572 (CVE-2014-3506) 3573 [Adam Langley] 3574 3575 *) An attacker can force an error condition which causes openssl to crash 3576 whilst processing DTLS packets due to memory being freed twice. This 3577 can be exploited through a Denial of Service attack. 3578 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 3579 this issue. 3580 (CVE-2014-3505) 3581 [Adam Langley] 3582 3583 *) If a multithreaded client connects to a malicious server using a resumed 3584 session and the server sends an ec point format extension it could write 3585 up to 255 bytes to freed memory. 3586 3587 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 3588 issue. 3589 (CVE-2014-3509) 3590 [Gabor Tyukasz] 3591 3592 *) A malicious server can crash an OpenSSL client with a null pointer 3593 dereference (read) by specifying an SRP ciphersuite even though it was not 3594 properly negotiated with the client. This can be exploited through a 3595 Denial of Service attack. 3596 3597 Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for 3598 discovering and researching this issue. 3599 (CVE-2014-5139) 3600 [Steve Henson] 3601 3602 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 3603 X509_name_oneline, X509_name_print_ex et al. to leak some information 3604 from the stack. Applications may be affected if they echo pretty printing 3605 output to the attacker. 3606 3607 Thanks to Ivan Fratric (Google) for discovering this issue. 3608 (CVE-2014-3508) 3609 [Emilia Käsper, and Steve Henson] 3610 3611 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 3612 for corner cases. (Certain input points at infinity could lead to 3613 bogus results, with non-infinity inputs mapped to infinity too.) 3614 [Bodo Moeller] 3615 3616 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 3617 3618 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 3619 handshake can force the use of weak keying material in OpenSSL 3620 SSL/TLS clients and servers. 3621 3622 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 3623 researching this issue. (CVE-2014-0224) 3624 [KIKUCHI Masashi, Steve Henson] 3625 3626 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 3627 OpenSSL DTLS client the code can be made to recurse eventually crashing 3628 in a DoS attack. 3629 3630 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 3631 (CVE-2014-0221) 3632 [Imre Rad, Steve Henson] 3633 3634 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 3635 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 3636 client or server. This is potentially exploitable to run arbitrary 3637 code on a vulnerable client or server. 3638 3639 Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195) 3640 [Jüri Aedla, Steve Henson] 3641 3642 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 3643 are subject to a denial of service attack. 3644 3645 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering 3646 this issue. (CVE-2014-3470) 3647 [Felix Gröbert, Ivan Fratric, Steve Henson] 3648 3649 *) Harmonize version and its documentation. -f flag is used to display 3650 compilation flags. 3651 [mancha <mancha1@zoho.com>] 3652 3653 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 3654 in i2d_ECPrivateKey. 3655 [mancha <mancha1@zoho.com>] 3656 3657 *) Fix some double frees. These are not thought to be exploitable. 3658 [mancha <mancha1@zoho.com>] 3659 3660 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 3661 3662 *) A missing bounds check in the handling of the TLS heartbeat extension 3663 can be used to reveal up to 64k of memory to a connected client or 3664 server. 3665 3666 Thanks for Neel Mehta of Google Security for discovering this bug and to 3667 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 3668 preparing the fix (CVE-2014-0160) 3669 [Adam Langley, Bodo Moeller] 3670 3671 *) Fix for the attack described in the paper "Recovering OpenSSL 3672 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 3673 by Yuval Yarom and Naomi Benger. Details can be obtained from: 3674 http://eprint.iacr.org/2014/140 3675 3676 Thanks to Yuval Yarom and Naomi Benger for discovering this 3677 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 3678 [Yuval Yarom and Naomi Benger] 3679 3680 *) TLS pad extension: draft-agl-tls-padding-03 3681 3682 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 3683 TLS client Hello record length value would otherwise be > 255 and 3684 less that 512 pad with a dummy extension containing zeroes so it 3685 is at least 512 bytes long. 3686 3687 [Adam Langley, Steve Henson] 3688 3689 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 3690 3691 *) Fix for TLS record tampering bug. A carefully crafted invalid 3692 handshake could crash OpenSSL with a NULL pointer exception. 3693 Thanks to Anton Johansson for reporting this issues. 3694 (CVE-2013-4353) 3695 3696 *) Keep original DTLS digest and encryption contexts in retransmission 3697 structures so we can use the previous session parameters if they need 3698 to be resent. (CVE-2013-6450) 3699 [Steve Henson] 3700 3701 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 3702 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 3703 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 3704 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 3705 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 3706 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 3707 [Rob Stradling, Adam Langley] 3708 3709 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 3710 3711 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 3712 supporting platforms or when small records were transferred. 3713 [Andy Polyakov, Steve Henson] 3714 3715 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 3716 3717 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 3718 3719 This addresses the flaw in CBC record processing discovered by 3720 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 3721 at: http://www.isg.rhul.ac.uk/tls/ 3722 3723 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 3724 Security Group at Royal Holloway, University of London 3725 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 3726 Emilia Käsper for the initial patch. 3727 (CVE-2013-0169) 3728 [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 3729 3730 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 3731 ciphersuites which can be exploited in a denial of service attack. 3732 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 3733 and detecting this bug and to Wolfgang Ettlinger 3734 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 3735 (CVE-2012-2686) 3736 [Adam Langley] 3737 3738 *) Return an error when checking OCSP signatures when key is NULL. 3739 This fixes a DoS attack. (CVE-2013-0166) 3740 [Steve Henson] 3741 3742 *) Make openssl verify return errors. 3743 [Chris Palmer <palmer@google.com> and Ben Laurie] 3744 3745 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 3746 the right response is stapled. Also change SSL_get_certificate() 3747 so it returns the certificate actually sent. 3748 See http://rt.openssl.org/Ticket/Display.html?id=2836. 3749 [Rob Stradling <rob.stradling@comodo.com>] 3750 3751 *) Fix possible deadlock when decoding public keys. 3752 [Steve Henson] 3753 3754 *) Don't use TLS 1.0 record version number in initial client hello 3755 if renegotiating. 3756 [Steve Henson] 3757 3758 Changes between 1.0.1b and 1.0.1c [10 May 2012] 3759 3760 *) Sanity check record length before skipping explicit IV in TLS 3761 1.2, 1.1 and DTLS to fix DoS attack. 3762 3763 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 3764 fuzzing as a service testing platform. 3765 (CVE-2012-2333) 3766 [Steve Henson] 3767 3768 *) Initialise tkeylen properly when encrypting CMS messages. 3769 Thanks to Solar Designer of Openwall for reporting this issue. 3770 [Steve Henson] 3771 3772 *) In FIPS mode don't try to use composite ciphers as they are not 3773 approved. 3774 [Steve Henson] 3775 3776 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 3777 3778 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 3779 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 3780 mean any application compiled against OpenSSL 1.0.0 headers setting 3781 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling 3782 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 3783 0x10000000L Any application which was previously compiled against 3784 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 3785 will need to be recompiled as a result. Letting be results in 3786 inability to disable specifically TLS 1.1 and in client context, 3787 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 3788 [Steve Henson] 3789 3790 *) In order to ensure interoperability SSL_OP_NO_protocolX does not 3791 disable just protocol X, but all protocols above X *if* there are 3792 protocols *below* X still enabled. In more practical terms it means 3793 that if application wants to disable TLS1.0 in favor of TLS1.1 and 3794 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 3795 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 3796 client side. 3797 [Andy Polyakov] 3798 3799 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 3800 3801 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 3802 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 3803 in CRYPTO_realloc_clean. 3804 3805 Thanks to Tavis Ormandy, Google Security Team, for discovering this 3806 issue and to Adam Langley <agl@chromium.org> for fixing it. 3807 (CVE-2012-2110) 3808 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 3809 3810 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 3811 [Adam Langley] 3812 3813 *) Workarounds for some broken servers that "hang" if a client hello 3814 record length exceeds 255 bytes. 3815 3816 1. Do not use record version number > TLS 1.0 in initial client 3817 hello: some (but not all) hanging servers will now work. 3818 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 3819 the number of ciphers sent in the client hello. This should be 3820 set to an even number, such as 50, for example by passing: 3821 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 3822 Most broken servers should now work. 3823 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 3824 TLS 1.2 client support entirely. 3825 [Steve Henson] 3826 3827 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 3828 [Andy Polyakov] 3829 3830 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 3831 3832 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 3833 STRING form instead of a DigestInfo. 3834 [Steve Henson] 3835 3836 *) The format used for MDC2 RSA signatures is inconsistent between EVP 3837 and the RSA_sign/RSA_verify functions. This was made more apparent when 3838 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 3839 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 3840 the correct format in RSA_verify so both forms transparently work. 3841 [Steve Henson] 3842 3843 *) Some servers which support TLS 1.0 can choke if we initially indicate 3844 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 3845 encrypted premaster secret. As a workaround use the maximum permitted 3846 client version in client hello, this should keep such servers happy 3847 and still work with previous versions of OpenSSL. 3848 [Steve Henson] 3849 3850 *) Add support for TLS/DTLS heartbeats. 3851 [Robin Seggelmann <seggelmann@fh-muenster.de>] 3852 3853 *) Add support for SCTP. 3854 [Robin Seggelmann <seggelmann@fh-muenster.de>] 3855 3856 *) Improved PRNG seeding for VOS. 3857 [Paul Green <Paul.Green@stratus.com>] 3858 3859 *) Extensive assembler packs updates, most notably: 3860 3861 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 3862 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 3863 - x86_64: bit-sliced AES implementation; 3864 - ARM: NEON support, contemporary platforms optimizations; 3865 - s390x: z196 support; 3866 - *: GHASH and GF(2^m) multiplication implementations; 3867 3868 [Andy Polyakov] 3869 3870 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 3871 (removal of unnecessary code) 3872 [Peter Sylvester <peter.sylvester@edelweb.fr>] 3873 3874 *) Add TLS key material exporter from RFC 5705. 3875 [Eric Rescorla] 3876 3877 *) Add DTLS-SRTP negotiation from RFC 5764. 3878 [Eric Rescorla] 3879 3880 *) Add Next Protocol Negotiation, 3881 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 3882 disabled with a no-npn flag to config or Configure. Code donated 3883 by Google. 3884 [Adam Langley <agl@google.com> and Ben Laurie] 3885 3886 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 3887 NIST-P256, NIST-P521, with constant-time single point multiplication on 3888 typical inputs. Compiler support for the nonstandard type __uint128_t is 3889 required to use this (present in gcc 4.4 and later, for 64-bit builds). 3890 Code made available under Apache License version 2.0. 3891 3892 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 3893 line to include this in your build of OpenSSL, and run "make depend" (or 3894 "make update"). This enables the following EC_METHODs: 3895 3896 EC_GFp_nistp224_method() 3897 EC_GFp_nistp256_method() 3898 EC_GFp_nistp521_method() 3899 3900 EC_GROUP_new_by_curve_name() will automatically use these (while 3901 EC_GROUP_new_curve_GFp() currently prefers the more flexible 3902 implementations). 3903 [Emilia Käsper, Adam Langley, Bodo Moeller (Google)] 3904 3905 *) Use type ossl_ssize_t instead of ssize_t which isn't available on 3906 all platforms. Move ssize_t definition from e_os.h to the public 3907 header file e_os2.h as it now appears in public header file cms.h 3908 [Steve Henson] 3909 3910 *) New -sigopt option to the ca, req and x509 utilities. Additional 3911 signature parameters can be passed using this option and in 3912 particular PSS. 3913 [Steve Henson] 3914 3915 *) Add RSA PSS signing function. This will generate and set the 3916 appropriate AlgorithmIdentifiers for PSS based on those in the 3917 corresponding EVP_MD_CTX structure. No application support yet. 3918 [Steve Henson] 3919 3920 *) Support for companion algorithm specific ASN1 signing routines. 3921 New function ASN1_item_sign_ctx() signs a pre-initialised 3922 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 3923 the appropriate parameters. 3924 [Steve Henson] 3925 3926 *) Add new algorithm specific ASN1 verification initialisation function 3927 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 3928 handling will be the same no matter what EVP_PKEY_METHOD is used. 3929 Add a PSS handler to support verification of PSS signatures: checked 3930 against a number of sample certificates. 3931 [Steve Henson] 3932 3933 *) Add signature printing for PSS. Add PSS OIDs. 3934 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 3935 3936 *) Add algorithm specific signature printing. An individual ASN1 method 3937 can now print out signatures instead of the standard hex dump. 3938 3939 More complex signatures (e.g. PSS) can print out more meaningful 3940 information. Include DSA version that prints out the signature 3941 parameters r, s. 3942 [Steve Henson] 3943 3944 *) Password based recipient info support for CMS library: implementing 3945 RFC3211. 3946 [Steve Henson] 3947 3948 *) Split password based encryption into PBES2 and PBKDF2 functions. This 3949 neatly separates the code into cipher and PBE sections and is required 3950 for some algorithms that split PBES2 into separate pieces (such as 3951 password based CMS). 3952 [Steve Henson] 3953 3954 *) Session-handling fixes: 3955 - Fix handling of connections that are resuming with a session ID, 3956 but also support Session Tickets. 3957 - Fix a bug that suppressed issuing of a new ticket if the client 3958 presented a ticket with an expired session. 3959 - Try to set the ticket lifetime hint to something reasonable. 3960 - Make tickets shorter by excluding irrelevant information. 3961 - On the client side, don't ignore renewed tickets. 3962 [Adam Langley, Bodo Moeller (Google)] 3963 3964 *) Fix PSK session representation. 3965 [Bodo Moeller] 3966 3967 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 3968 3969 This work was sponsored by Intel. 3970 [Andy Polyakov] 3971 3972 *) Add GCM support to TLS library. Some custom code is needed to split 3973 the IV between the fixed (from PRF) and explicit (from TLS record) 3974 portions. This adds all GCM ciphersuites supported by RFC5288 and 3975 RFC5289. Generalise some AES* cipherstrings to include GCM and 3976 add a special AESGCM string for GCM only. 3977 [Steve Henson] 3978 3979 *) Expand range of ctrls for AES GCM. Permit setting invocation 3980 field on decrypt and retrieval of invocation field only on encrypt. 3981 [Steve Henson] 3982 3983 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 3984 As required by RFC5289 these ciphersuites cannot be used if for 3985 versions of TLS earlier than 1.2. 3986 [Steve Henson] 3987 3988 *) For FIPS capable OpenSSL interpret a NULL default public key method 3989 as unset and return the appropriate default but do *not* set the default. 3990 This means we can return the appropriate method in applications that 3991 switch between FIPS and non-FIPS modes. 3992 [Steve Henson] 3993 3994 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 3995 ENGINE is used then we cannot handle that in the FIPS module so we 3996 keep original code iff non-FIPS operations are allowed. 3997 [Steve Henson] 3998 3999 *) Add -attime option to openssl utilities. 4000 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 4001 4002 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 4003 [Steve Henson] 4004 4005 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 4006 FIPS EC methods unconditionally for now. 4007 [Steve Henson] 4008 4009 *) New build option no-ec2m to disable characteristic 2 code. 4010 [Steve Henson] 4011 4012 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 4013 all cases can be covered as some introduce binary incompatibilities. 4014 [Steve Henson] 4015 4016 *) Redirect RSA operations to FIPS module including keygen, 4017 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 4018 [Steve Henson] 4019 4020 *) Add similar low level API blocking to ciphers. 4021 [Steve Henson] 4022 4023 *) Low level digest APIs are not approved in FIPS mode: any attempt 4024 to use these will cause a fatal error. Applications that *really* want 4025 to use them can use the private_* version instead. 4026 [Steve Henson] 4027 4028 *) Redirect cipher operations to FIPS module for FIPS builds. 4029 [Steve Henson] 4030 4031 *) Redirect digest operations to FIPS module for FIPS builds. 4032 [Steve Henson] 4033 4034 *) Update build system to add "fips" flag which will link in fipscanister.o 4035 for static and shared library builds embedding a signature if needed. 4036 [Steve Henson] 4037 4038 *) Output TLS supported curves in preference order instead of numerical 4039 order. This is currently hardcoded for the highest order curves first. 4040 This should be configurable so applications can judge speed vs strength. 4041 [Steve Henson] 4042 4043 *) Add TLS v1.2 server support for client authentication. 4044 [Steve Henson] 4045 4046 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 4047 and enable MD5. 4048 [Steve Henson] 4049 4050 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 4051 FIPS modules versions. 4052 [Steve Henson] 4053 4054 *) Add TLS v1.2 client side support for client authentication. Keep cache 4055 of handshake records longer as we don't know the hash algorithm to use 4056 until after the certificate request message is received. 4057 [Steve Henson] 4058 4059 *) Initial TLS v1.2 client support. Add a default signature algorithms 4060 extension including all the algorithms we support. Parse new signature 4061 format in client key exchange. Relax some ECC signing restrictions for 4062 TLS v1.2 as indicated in RFC5246. 4063 [Steve Henson] 4064 4065 *) Add server support for TLS v1.2 signature algorithms extension. Switch 4066 to new signature format when needed using client digest preference. 4067 All server ciphersuites should now work correctly in TLS v1.2. No client 4068 support yet and no support for client certificates. 4069 [Steve Henson] 4070 4071 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 4072 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 4073 ciphersuites. At present only RSA key exchange ciphersuites work with 4074 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 4075 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 4076 and version checking. 4077 [Steve Henson] 4078 4079 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 4080 with this defined it will not be affected by any changes to ssl internal 4081 structures. Add several utility functions to allow openssl application 4082 to work with OPENSSL_NO_SSL_INTERN defined. 4083 [Steve Henson] 4084 4085 *) A long standing patch to add support for SRP from EdelWeb (Peter 4086 Sylvester and Christophe Renou) was integrated. 4087 [Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester 4088 <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and 4089 Ben Laurie] 4090 4091 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 4092 [Steve Henson] 4093 4094 *) Permit abbreviated handshakes when renegotiating using the function 4095 SSL_renegotiate_abbreviated(). 4096 [Robin Seggelmann <seggelmann@fh-muenster.de>] 4097 4098 *) Add call to ENGINE_register_all_complete() to 4099 ENGINE_load_builtin_engines(), so some implementations get used 4100 automatically instead of needing explicit application support. 4101 [Steve Henson] 4102 4103 *) Add support for TLS key exporter as described in RFC5705. 4104 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 4105 4106 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 4107 a few changes are required: 4108 4109 Add SSL_OP_NO_TLSv1_1 flag. 4110 Add TLSv1_1 methods. 4111 Update version checking logic to handle version 1.1. 4112 Add explicit IV handling (ported from DTLS code). 4113 Add command line options to s_client/s_server. 4114 [Steve Henson] 4115 4116 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 4117 4118 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 4119 in CMS and PKCS7 code. When RSA decryption fails use a random key for 4120 content decryption and always return the same error. Note: this attack 4121 needs on average 2^20 messages so it only affects automated senders. The 4122 old behaviour can be re-enabled in the CMS code by setting the 4123 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 4124 an MMA defence is not necessary. 4125 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 4126 this issue. (CVE-2012-0884) 4127 [Steve Henson] 4128 4129 *) Fix CVE-2011-4619: make sure we really are receiving a 4130 client hello before rejecting multiple SGC restarts. Thanks to 4131 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 4132 [Steve Henson] 4133 4134 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 4135 4136 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 4137 Thanks to Antonio Martin, Enterprise Secure Access Research and 4138 Development, Cisco Systems, Inc. for discovering this bug and 4139 preparing a fix. (CVE-2012-0050) 4140 [Antonio Martin] 4141 4142 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 4143 4144 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 4145 of the Vaudenay padding oracle attack on CBC mode encryption 4146 which enables an efficient plaintext recovery attack against 4147 the OpenSSL implementation of DTLS. Their attack exploits timing 4148 differences arising during decryption processing. A research 4149 paper describing this attack can be found at: 4150 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 4151 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 4152 Security Group at Royal Holloway, University of London 4153 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 4154 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 4155 for preparing the fix. (CVE-2011-4108) 4156 [Robin Seggelmann, Michael Tuexen] 4157 4158 *) Clear bytes used for block padding of SSL 3.0 records. 4159 (CVE-2011-4576) 4160 [Adam Langley (Google)] 4161 4162 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 4163 Kadianakis <desnacked@gmail.com> for discovering this issue and 4164 Adam Langley for preparing the fix. (CVE-2011-4619) 4165 [Adam Langley (Google)] 4166 4167 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 4168 [Andrey Kulikov <amdeich@gmail.com>] 4169 4170 *) Prevent malformed RFC3779 data triggering an assertion failure. 4171 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 4172 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 4173 [Rob Austein <sra@hactrn.net>] 4174 4175 *) Improved PRNG seeding for VOS. 4176 [Paul Green <Paul.Green@stratus.com>] 4177 4178 *) Fix ssl_ciph.c set-up race. 4179 [Adam Langley (Google)] 4180 4181 *) Fix spurious failures in ecdsatest.c. 4182 [Emilia Käsper (Google)] 4183 4184 *) Fix the BIO_f_buffer() implementation (which was mixing different 4185 interpretations of the '..._len' fields). 4186 [Adam Langley (Google)] 4187 4188 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 4189 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 4190 threads won't reuse the same blinding coefficients. 4191 4192 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 4193 lock to call BN_BLINDING_invert_ex, and avoids one use of 4194 BN_BLINDING_update for each BN_BLINDING structure (previously, 4195 the last update always remained unused). 4196 [Emilia Käsper (Google)] 4197 4198 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 4199 [Bob Buckholz (Google)] 4200 4201 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 4202 4203 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 4204 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 4205 [Kaspar Brand <ossl@velox.ch>] 4206 4207 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 4208 for multi-threaded use of ECDH. (CVE-2011-3210) 4209 [Adam Langley (Google)] 4210 4211 *) Fix x509_name_ex_d2i memory leak on bad inputs. 4212 [Bodo Moeller] 4213 4214 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 4215 signature public key algorithm by using OID xref utilities instead. 4216 Before this you could only use some ECC ciphersuites with SHA1 only. 4217 [Steve Henson] 4218 4219 *) Add protection against ECDSA timing attacks as mentioned in the paper 4220 by Billy Bob Brumley and Nicola Tuveri, see: 4221 4222 http://eprint.iacr.org/2011/232.pdf 4223 4224 [Billy Bob Brumley and Nicola Tuveri] 4225 4226 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 4227 4228 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 4229 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 4230 4231 *) Fix bug in string printing code: if *any* escaping is enabled we must 4232 escape the escape character (backslash) or the resulting string is 4233 ambiguous. 4234 [Steve Henson] 4235 4236 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 4237 4238 *) Disable code workaround for ancient and obsolete Netscape browsers 4239 and servers: an attacker can use it in a ciphersuite downgrade attack. 4240 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 4241 [Steve Henson] 4242 4243 *) Fixed J-PAKE implementation error, originally discovered by 4244 Sebastien Martini, further info and confirmation from Stefan 4245 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 4246 [Ben Laurie] 4247 4248 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 4249 4250 *) Fix extension code to avoid race conditions which can result in a buffer 4251 overrun vulnerability: resumed sessions must not be modified as they can 4252 be shared by multiple threads. CVE-2010-3864 4253 [Steve Henson] 4254 4255 *) Fix WIN32 build system to correctly link an ENGINE directory into 4256 a DLL. 4257 [Steve Henson] 4258 4259 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 4260 4261 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 4262 (CVE-2010-1633) 4263 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 4264 4265 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 4266 4267 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 4268 context. The operation can be customised via the ctrl mechanism in 4269 case ENGINEs want to include additional functionality. 4270 [Steve Henson] 4271 4272 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 4273 [Steve Henson] 4274 4275 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 4276 output hashes compatible with older versions of OpenSSL. 4277 [Willy Weisz <weisz@vcpc.univie.ac.at>] 4278 4279 *) Fix compression algorithm handling: if resuming a session use the 4280 compression algorithm of the resumed session instead of determining 4281 it from client hello again. Don't allow server to change algorithm. 4282 [Steve Henson] 4283 4284 *) Add load_crls() function to apps tidying load_certs() too. Add option 4285 to verify utility to allow additional CRLs to be included. 4286 [Steve Henson] 4287 4288 *) Update OCSP request code to permit adding custom headers to the request: 4289 some responders need this. 4290 [Steve Henson] 4291 4292 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 4293 correctly. 4294 [Julia Lawall <julia@diku.dk>] 4295 4296 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 4297 needlessly dereferenced structures, used obsolete functions and 4298 didn't handle all updated verify codes correctly. 4299 [Steve Henson] 4300 4301 *) Disable MD2 in the default configuration. 4302 [Steve Henson] 4303 4304 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 4305 indicate the initial BIO being pushed or popped. This makes it possible 4306 to determine whether the BIO is the one explicitly called or as a result 4307 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 4308 it handles reference counts correctly and doesn't zero out the I/O bio 4309 when it is not being explicitly popped. WARNING: applications which 4310 included workarounds for the old buggy behaviour will need to be modified 4311 or they could free up already freed BIOs. 4312 [Steve Henson] 4313 4314 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 4315 renaming to all platforms (within the 0.9.8 branch, this was 4316 done conditionally on Netware platforms to avoid a name clash). 4317 [Guenter <lists@gknw.net>] 4318 4319 *) Add ECDHE and PSK support to DTLS. 4320 [Michael Tuexen <tuexen@fh-muenster.de>] 4321 4322 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 4323 be used on C++. 4324 [Steve Henson] 4325 4326 *) Add "missing" function EVP_MD_flags() (without this the only way to 4327 retrieve a digest flags is by accessing the structure directly. Update 4328 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 4329 or cipher is registered as in the "from" argument. Print out all 4330 registered digests in the dgst usage message instead of manually 4331 attempting to work them out. 4332 [Steve Henson] 4333 4334 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 4335 this allows the use of compression and extensions. Change default cipher 4336 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 4337 by default unless an application cipher string requests it. 4338 [Steve Henson] 4339 4340 *) Alter match criteria in PKCS12_parse(). It used to try to use local 4341 key ids to find matching certificates and keys but some PKCS#12 files 4342 don't follow the (somewhat unwritten) rules and this strategy fails. 4343 Now just gather all certificates together and the first private key 4344 then look for the first certificate that matches the key. 4345 [Steve Henson] 4346 4347 *) Support use of registered digest and cipher names for dgst and cipher 4348 commands instead of having to add each one as a special case. So now 4349 you can do: 4350 4351 openssl sha256 foo 4352 4353 as well as: 4354 4355 openssl dgst -sha256 foo 4356 4357 and this works for ENGINE based algorithms too. 4358 4359 [Steve Henson] 4360 4361 *) Update Gost ENGINE to support parameter files. 4362 [Victor B. Wagner <vitus@cryptocom.ru>] 4363 4364 *) Support GeneralizedTime in ca utility. 4365 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 4366 4367 *) Enhance the hash format used for certificate directory links. The new 4368 form uses the canonical encoding (meaning equivalent names will work 4369 even if they aren't identical) and uses SHA1 instead of MD5. This form 4370 is incompatible with the older format and as a result c_rehash should 4371 be used to rebuild symbolic links. 4372 [Steve Henson] 4373 4374 *) Make PKCS#8 the default write format for private keys, replacing the 4375 traditional format. This form is standardised, more secure and doesn't 4376 include an implicit MD5 dependency. 4377 [Steve Henson] 4378 4379 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 4380 committed to OpenSSL should pass this lot as a minimum. 4381 [Steve Henson] 4382 4383 *) Add session ticket override functionality for use by EAP-FAST. 4384 [Jouni Malinen <j@w1.fi>] 4385 4386 *) Modify HMAC functions to return a value. Since these can be implemented 4387 in an ENGINE errors can occur. 4388 [Steve Henson] 4389 4390 *) Type-checked OBJ_bsearch_ex. 4391 [Ben Laurie] 4392 4393 *) Type-checked OBJ_bsearch. Also some constification necessitated 4394 by type-checking. Still to come: TXT_DB, bsearch(?), 4395 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 4396 CONF_VALUE. 4397 [Ben Laurie] 4398 4399 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 4400 seconds to a tm structure directly, instead of going through OS 4401 specific date routines. This avoids any issues with OS routines such 4402 as the year 2038 bug. New *_adj() functions for ASN1 time structures 4403 and X509_time_adj_ex() to cover the extended range. The existing 4404 X509_time_adj() is still usable and will no longer have any date issues. 4405 [Steve Henson] 4406 4407 *) Delta CRL support. New use deltas option which will attempt to locate 4408 and search any appropriate delta CRLs available. 4409 4410 This work was sponsored by Google. 4411 [Steve Henson] 4412 4413 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 4414 code and add additional score elements. Validate alternate CRL paths 4415 as part of the CRL checking and indicate a new error "CRL path validation 4416 error" in this case. Applications wanting additional details can use 4417 the verify callback and check the new "parent" field. If this is not 4418 NULL CRL path validation is taking place. Existing applications won't 4419 see this because it requires extended CRL support which is off by 4420 default. 4421 4422 This work was sponsored by Google. 4423 [Steve Henson] 4424 4425 *) Support for freshest CRL extension. 4426 4427 This work was sponsored by Google. 4428 [Steve Henson] 4429 4430 *) Initial indirect CRL support. Currently only supported in the CRLs 4431 passed directly and not via lookup. Process certificate issuer 4432 CRL entry extension and lookup CRL entries by bother issuer name 4433 and serial number. Check and process CRL issuer entry in IDP extension. 4434 4435 This work was sponsored by Google. 4436 [Steve Henson] 4437 4438 *) Add support for distinct certificate and CRL paths. The CRL issuer 4439 certificate is validated separately in this case. Only enabled if 4440 an extended CRL support flag is set: this flag will enable additional 4441 CRL functionality in future. 4442 4443 This work was sponsored by Google. 4444 [Steve Henson] 4445 4446 *) Add support for policy mappings extension. 4447 4448 This work was sponsored by Google. 4449 [Steve Henson] 4450 4451 *) Fixes to pathlength constraint, self issued certificate handling, 4452 policy processing to align with RFC3280 and PKITS tests. 4453 4454 This work was sponsored by Google. 4455 [Steve Henson] 4456 4457 *) Support for name constraints certificate extension. DN, email, DNS 4458 and URI types are currently supported. 4459 4460 This work was sponsored by Google. 4461 [Steve Henson] 4462 4463 *) To cater for systems that provide a pointer-based thread ID rather 4464 than numeric, deprecate the current numeric thread ID mechanism and 4465 replace it with a structure and associated callback type. This 4466 mechanism allows a numeric "hash" to be extracted from a thread ID in 4467 either case, and on platforms where pointers are larger than 'long', 4468 mixing is done to help ensure the numeric 'hash' is usable even if it 4469 can't be guaranteed unique. The default mechanism is to use "&errno" 4470 as a pointer-based thread ID to distinguish between threads. 4471 4472 Applications that want to provide their own thread IDs should now use 4473 CRYPTO_THREADID_set_callback() to register a callback that will call 4474 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 4475 4476 Note that ERR_remove_state() is now deprecated, because it is tied 4477 to the assumption that thread IDs are numeric. ERR_remove_state(0) 4478 to free the current thread's error state should be replaced by 4479 ERR_remove_thread_state(NULL). 4480 4481 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 4482 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 4483 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 4484 application was previously providing a numeric thread callback that 4485 was inappropriate for distinguishing threads, then uniqueness might 4486 have been obtained with &errno that happened immediately in the 4487 intermediate development versions of OpenSSL; this is no longer the 4488 case, the numeric thread callback will now override the automatic use 4489 of &errno.) 4490 [Geoff Thorpe, with help from Bodo Moeller] 4491 4492 *) Initial support for different CRL issuing certificates. This covers a 4493 simple case where the self issued certificates in the chain exist and 4494 the real CRL issuer is higher in the existing chain. 4495 4496 This work was sponsored by Google. 4497 [Steve Henson] 4498 4499 *) Removed effectively defunct crypto/store from the build. 4500 [Ben Laurie] 4501 4502 *) Revamp of STACK to provide stronger type-checking. Still to come: 4503 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 4504 ASN1_STRING, CONF_VALUE. 4505 [Ben Laurie] 4506 4507 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 4508 RAM on SSL connections. This option can save about 34k per idle SSL. 4509 [Nick Mathewson] 4510 4511 *) Revamp of LHASH to provide stronger type-checking. Still to come: 4512 STACK, TXT_DB, bsearch, qsort. 4513 [Ben Laurie] 4514 4515 *) Initial support for Cryptographic Message Syntax (aka CMS) based 4516 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 4517 support for data, signedData, compressedData, digestedData and 4518 encryptedData, envelopedData types included. Scripts to check against 4519 RFC4134 examples draft and interop and consistency checks of many 4520 content types and variants. 4521 [Steve Henson] 4522 4523 *) Add options to enc utility to support use of zlib compression BIO. 4524 [Steve Henson] 4525 4526 *) Extend mk1mf to support importing of options and assembly language 4527 files from Configure script, currently only included in VC-WIN32. 4528 The assembly language rules can now optionally generate the source 4529 files from the associated perl scripts. 4530 [Steve Henson] 4531 4532 *) Implement remaining functionality needed to support GOST ciphersuites. 4533 Interop testing has been performed using CryptoPro implementations. 4534 [Victor B. Wagner <vitus@cryptocom.ru>] 4535 4536 *) s390x assembler pack. 4537 [Andy Polyakov] 4538 4539 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 4540 "family." 4541 [Andy Polyakov] 4542 4543 *) Implement Opaque PRF Input TLS extension as specified in 4544 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 4545 official specification yet and no extension type assignment by 4546 IANA exists, this extension (for now) will have to be explicitly 4547 enabled when building OpenSSL by providing the extension number 4548 to use. For example, specify an option 4549 4550 -DTLSEXT_TYPE_opaque_prf_input=0x9527 4551 4552 to the "config" or "Configure" script to enable the extension, 4553 assuming extension number 0x9527 (which is a completely arbitrary 4554 and unofficial assignment based on the MD5 hash of the Internet 4555 Draft). Note that by doing so, you potentially lose 4556 interoperability with other TLS implementations since these might 4557 be using the same extension number for other purposes. 4558 4559 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 4560 opaque PRF input value to use in the handshake. This will create 4561 an internal copy of the length-'len' string at 'src', and will 4562 return non-zero for success. 4563 4564 To get more control and flexibility, provide a callback function 4565 by using 4566 4567 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 4568 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 4569 4570 where 4571 4572 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 4573 void *arg; 4574 4575 Callback function 'cb' will be called in handshakes, and is 4576 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 4577 Argument 'arg' is for application purposes (the value as given to 4578 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 4579 be provided to the callback function). The callback function 4580 has to return non-zero to report success: usually 1 to use opaque 4581 PRF input just if possible, or 2 to enforce use of the opaque PRF 4582 input. In the latter case, the library will abort the handshake 4583 if opaque PRF input is not successfully negotiated. 4584 4585 Arguments 'peerinput' and 'len' given to the callback function 4586 will always be NULL and 0 in the case of a client. A server will 4587 see the client's opaque PRF input through these variables if 4588 available (NULL and 0 otherwise). Note that if the server 4589 provides an opaque PRF input, the length must be the same as the 4590 length of the client's opaque PRF input. 4591 4592 Note that the callback function will only be called when creating 4593 a new session (session resumption can resume whatever was 4594 previously negotiated), and will not be called in SSL 2.0 4595 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 4596 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 4597 for applications that need to enforce opaque PRF input. 4598 4599 [Bodo Moeller] 4600 4601 *) Update ssl code to support digests other than SHA1+MD5 for handshake 4602 MAC. 4603 4604 [Victor B. Wagner <vitus@cryptocom.ru>] 4605 4606 *) Add RFC4507 support to OpenSSL. This includes the corrections in 4607 RFC4507bis. The encrypted ticket format is an encrypted encoded 4608 SSL_SESSION structure, that way new session features are automatically 4609 supported. 4610 4611 If a client application caches session in an SSL_SESSION structure 4612 support is transparent because tickets are now stored in the encoded 4613 SSL_SESSION. 4614 4615 The SSL_CTX structure automatically generates keys for ticket 4616 protection in servers so again support should be possible 4617 with no application modification. 4618 4619 If a client or server wishes to disable RFC4507 support then the option 4620 SSL_OP_NO_TICKET can be set. 4621 4622 Add a TLS extension debugging callback to allow the contents of any client 4623 or server extensions to be examined. 4624 4625 This work was sponsored by Google. 4626 [Steve Henson] 4627 4628 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 4629 OpenSSL should now compile cleanly on gcc 4.2 4630 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 4631 4632 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 4633 support including streaming MAC support: this is required for GOST 4634 ciphersuite support. 4635 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 4636 4637 *) Add option -stream to use PKCS#7 streaming in smime utility. New 4638 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 4639 to output in BER and PEM format. 4640 [Steve Henson] 4641 4642 *) Experimental support for use of HMAC via EVP_PKEY interface. This 4643 allows HMAC to be handled via the EVP_DigestSign*() interface. The 4644 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 4645 ENGINE support for HMAC keys which are unextractable. New -mac and 4646 -macopt options to dgst utility. 4647 [Steve Henson] 4648 4649 *) New option -sigopt to dgst utility. Update dgst to use 4650 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 4651 alternative signing parameters such as X9.31 or PSS in the dgst 4652 utility. 4653 [Steve Henson] 4654 4655 *) Change ssl_cipher_apply_rule(), the internal function that does 4656 the work each time a ciphersuite string requests enabling 4657 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 4658 removing ("!foo+bar") a class of ciphersuites: Now it maintains 4659 the order of disabled ciphersuites such that those ciphersuites 4660 that most recently went from enabled to disabled not only stay 4661 in order with respect to each other, but also have higher priority 4662 than other disabled ciphersuites the next time ciphersuites are 4663 enabled again. 4664 4665 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 4666 the same ciphersuites as with "HIGH" alone, but in a specific 4667 order where the PSK ciphersuites come first (since they are the 4668 most recently disabled ciphersuites when "HIGH" is parsed). 4669 4670 Also, change ssl_create_cipher_list() (using this new 4671 functionality) such that between otherwise identical 4672 ciphersuites, ephemeral ECDH is preferred over ephemeral DH in 4673 the default order. 4674 [Bodo Moeller] 4675 4676 *) Change ssl_create_cipher_list() so that it automatically 4677 arranges the ciphersuites in reasonable order before starting 4678 to process the rule string. Thus, the definition for "DEFAULT" 4679 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 4680 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 4681 This makes it much easier to arrive at a reasonable default order 4682 in applications for which anonymous ciphers are OK (meaning 4683 that you can't actually use DEFAULT). 4684 [Bodo Moeller; suggested by Victor Duchovni] 4685 4686 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 4687 processing) into multiple integers instead of setting 4688 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 4689 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 4690 (These masks as well as the individual bit definitions are hidden 4691 away into the non-exported interface ssl/ssl_locl.h, so this 4692 change to the definition of the SSL_CIPHER structure shouldn't 4693 affect applications.) This give us more bits for each of these 4694 categories, so there is no longer a need to coagulate AES128 and 4695 AES256 into a single algorithm bit, and to coagulate Camellia128 4696 and Camellia256 into a single algorithm bit, which has led to all 4697 kinds of kludges. 4698 4699 Thus, among other things, the kludge introduced in 0.9.7m and 4700 0.9.8e for masking out AES256 independently of AES128 or masking 4701 out Camellia256 independently of AES256 is not needed here in 0.9.9. 4702 4703 With the change, we also introduce new ciphersuite aliases that 4704 so far were missing: "AES128", "AES256", "CAMELLIA128", and 4705 "CAMELLIA256". 4706 [Bodo Moeller] 4707 4708 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 4709 Use the leftmost N bytes of the signature input if the input is 4710 larger than the prime q (with N being the size in bytes of q). 4711 [Nils Larsch] 4712 4713 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 4714 it yet and it is largely untested. 4715 [Steve Henson] 4716 4717 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 4718 [Nils Larsch] 4719 4720 *) Initial incomplete changes to avoid need for function casts in OpenSSL 4721 some compilers (gcc 4.2 and later) reject their use. Safestack is 4722 reimplemented. Update ASN1 to avoid use of legacy functions. 4723 [Steve Henson] 4724 4725 *) Win32/64 targets are linked with Winsock2. 4726 [Andy Polyakov] 4727 4728 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 4729 to external functions. This can be used to increase CRL handling 4730 efficiency especially when CRLs are very large by (for example) storing 4731 the CRL revoked certificates in a database. 4732 [Steve Henson] 4733 4734 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 4735 new CRLs added to a directory can be used. New command line option 4736 -verify_return_error to s_client and s_server. This causes real errors 4737 to be returned by the verify callback instead of carrying on no matter 4738 what. This reflects the way a "real world" verify callback would behave. 4739 [Steve Henson] 4740 4741 *) GOST engine, supporting several GOST algorithms and public key formats. 4742 Kindly donated by Cryptocom. 4743 [Cryptocom] 4744 4745 *) Partial support for Issuing Distribution Point CRL extension. CRLs 4746 partitioned by DP are handled but no indirect CRL or reason partitioning 4747 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 4748 selected via a scoring technique which handles IDP and AKID in CRLs. 4749 [Steve Henson] 4750 4751 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 4752 will ultimately be used for all verify operations: this will remove the 4753 X509_STORE dependency on certificate verification and allow alternative 4754 lookup methods. X509_STORE based implementations of these two callbacks. 4755 [Steve Henson] 4756 4757 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 4758 Modify get_crl() to find a valid (unexpired) CRL if possible. 4759 [Steve Henson] 4760 4761 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 4762 this would be called X509_CRL_cmp() but that name is already used by 4763 a function that just compares CRL issuer names. Cache several CRL 4764 extensions in X509_CRL structure and cache CRLDP in X509. 4765 [Steve Henson] 4766 4767 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 4768 this maps equivalent X509_NAME structures into a consistent structure. 4769 Name comparison can then be performed rapidly using memcmp(). 4770 [Steve Henson] 4771 4772 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 4773 utility. 4774 [Steve Henson] 4775 4776 *) Allow digests to supply their own micalg string for S/MIME type using 4777 the ctrl EVP_MD_CTRL_MICALG. 4778 [Steve Henson] 4779 4780 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 4781 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 4782 ctrl. It can then customise the structure before and/or after signing 4783 if necessary. 4784 [Steve Henson] 4785 4786 *) New function OBJ_add_sigid() to allow application defined signature OIDs 4787 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 4788 to free up any added signature OIDs. 4789 [Steve Henson] 4790 4791 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 4792 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 4793 digest and cipher tables. New options added to openssl utility: 4794 list-message-digest-algorithms and list-cipher-algorithms. 4795 [Steve Henson] 4796 4797 *) Change the array representation of binary polynomials: the list 4798 of degrees of non-zero coefficients is now terminated with -1. 4799 Previously it was terminated with 0, which was also part of the 4800 value; thus, the array representation was not applicable to 4801 polynomials where t^0 has coefficient zero. This change makes 4802 the array representation useful in a more general context. 4803 [Douglas Stebila] 4804 4805 *) Various modifications and fixes to SSL/TLS cipher string 4806 handling. For ECC, the code now distinguishes between fixed ECDH 4807 with RSA certificates on the one hand and with ECDSA certificates 4808 on the other hand, since these are separate ciphersuites. The 4809 unused code for Fortezza ciphersuites has been removed. 4810 4811 For consistency with EDH, ephemeral ECDH is now called "EECDH" 4812 (not "ECDHE"). For consistency with the code for DH 4813 certificates, use of ECDH certificates is now considered ECDH 4814 authentication, not RSA or ECDSA authentication (the latter is 4815 merely the CA's signing algorithm and not actively used in the 4816 protocol). 4817 4818 The temporary ciphersuite alias "ECCdraft" is no longer 4819 available, and ECC ciphersuites are no longer excluded from "ALL" 4820 and "DEFAULT". The following aliases now exist for RFC 4492 4821 ciphersuites, most of these by analogy with the DH case: 4822 4823 kECDHr - ECDH cert, signed with RSA 4824 kECDHe - ECDH cert, signed with ECDSA 4825 kECDH - ECDH cert (signed with either RSA or ECDSA) 4826 kEECDH - ephemeral ECDH 4827 ECDH - ECDH cert or ephemeral ECDH 4828 4829 aECDH - ECDH cert 4830 aECDSA - ECDSA cert 4831 ECDSA - ECDSA cert 4832 4833 AECDH - anonymous ECDH 4834 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 4835 4836 [Bodo Moeller] 4837 4838 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 4839 Use correct micalg parameters depending on digest(s) in signed message. 4840 [Steve Henson] 4841 4842 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 4843 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 4844 [Steve Henson] 4845 4846 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 4847 an engine to register a method. Add ENGINE lookups for methods and 4848 functional reference processing. 4849 [Steve Henson] 4850 4851 *) New functions EVP_Digest{Sign,Verify)*. These are enhanced versions of 4852 EVP_{Sign,Verify}* which allow an application to customise the signature 4853 process. 4854 [Steve Henson] 4855 4856 *) New -resign option to smime utility. This adds one or more signers 4857 to an existing PKCS#7 signedData structure. Also -md option to use an 4858 alternative message digest algorithm for signing. 4859 [Steve Henson] 4860 4861 *) Tidy up PKCS#7 routines and add new functions to make it easier to 4862 create PKCS7 structures containing multiple signers. Update smime 4863 application to support multiple signers. 4864 [Steve Henson] 4865 4866 *) New -macalg option to pkcs12 utility to allow setting of an alternative 4867 digest MAC. 4868 [Steve Henson] 4869 4870 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 4871 Reorganize PBE internals to lookup from a static table using NIDs, 4872 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 4873 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 4874 PRF which will be automatically used with PBES2. 4875 [Steve Henson] 4876 4877 *) Replace the algorithm specific calls to generate keys in "req" with the 4878 new API. 4879 [Steve Henson] 4880 4881 *) Update PKCS#7 enveloped data routines to use new API. This is now 4882 supported by any public key method supporting the encrypt operation. A 4883 ctrl is added to allow the public key algorithm to examine or modify 4884 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 4885 a no op. 4886 [Steve Henson] 4887 4888 *) Add a ctrl to asn1 method to allow a public key algorithm to express 4889 a default digest type to use. In most cases this will be SHA1 but some 4890 algorithms (such as GOST) need to specify an alternative digest. The 4891 return value indicates how strong the preference is 1 means optional and 4892 2 is mandatory (that is it is the only supported type). Modify 4893 ASN1_item_sign() to accept a NULL digest argument to indicate it should 4894 use the default md. Update openssl utilities to use the default digest 4895 type for signing if it is not explicitly indicated. 4896 [Steve Henson] 4897 4898 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 4899 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 4900 signing method from the key type. This effectively removes the link 4901 between digests and public key types. 4902 [Steve Henson] 4903 4904 *) Add an OID cross reference table and utility functions. Its purpose is to 4905 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 4906 rsaEncryption. This will allow some of the algorithm specific hackery 4907 needed to use the correct OID to be removed. 4908 [Steve Henson] 4909 4910 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 4911 structures for PKCS7_sign(). They are now set up by the relevant public 4912 key ASN1 method. 4913 [Steve Henson] 4914 4915 *) Add provisional EC pkey method with support for ECDSA and ECDH. 4916 [Steve Henson] 4917 4918 *) Add support for key derivation (agreement) in the API, DH method and 4919 pkeyutl. 4920 [Steve Henson] 4921 4922 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 4923 public and private key formats. As a side effect these add additional 4924 command line functionality not previously available: DSA signatures can be 4925 generated and verified using pkeyutl and DH key support and generation in 4926 pkey, genpkey. 4927 [Steve Henson] 4928 4929 *) BeOS support. 4930 [Oliver Tappe <zooey@hirschkaefer.de>] 4931 4932 *) New make target "install_html_docs" installs HTML renditions of the 4933 manual pages. 4934 [Oliver Tappe <zooey@hirschkaefer.de>] 4935 4936 *) New utility "genpkey" this is analogous to "genrsa" etc except it can 4937 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 4938 support key and parameter generation and add initial key generation 4939 functionality for RSA. 4940 [Steve Henson] 4941 4942 *) Add functions for main EVP_PKEY_method operations. The undocumented 4943 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 4944 EVP_PKEY_{encrypt,decrypt}_old. 4945 [Steve Henson] 4946 4947 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 4948 key API, doesn't do much yet. 4949 [Steve Henson] 4950 4951 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 4952 public key algorithms. New option to openssl utility: 4953 "list-public-key-algorithms" to print out info. 4954 [Steve Henson] 4955 4956 *) Implement the Supported Elliptic Curves Extension for 4957 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 4958 [Douglas Stebila] 4959 4960 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 4961 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 4962 [Steve Henson] 4963 4964 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 4965 utilities such as rsa, dsa, dsaparam etc except they process any key 4966 type. 4967 [Steve Henson] 4968 4969 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 4970 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 4971 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 4972 structure. 4973 [Steve Henson] 4974 4975 *) Initial support for pluggable public key ASN1. 4976 De-spaghettify the public key ASN1 handling. Move public and private 4977 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 4978 algorithm specific handling to a single module within the relevant 4979 algorithm directory. Add functions to allow (near) opaque processing 4980 of public and private key structures. 4981 [Steve Henson] 4982 4983 *) Implement the Supported Point Formats Extension for 4984 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 4985 [Douglas Stebila] 4986 4987 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 4988 for the psk identity [hint] and the psk callback functions to the 4989 SSL_SESSION, SSL and SSL_CTX structure. 4990 4991 New ciphersuites: 4992 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 4993 PSK-AES256-CBC-SHA 4994 4995 New functions: 4996 SSL_CTX_use_psk_identity_hint 4997 SSL_get_psk_identity_hint 4998 SSL_get_psk_identity 4999 SSL_use_psk_identity_hint 5000 5001 [Mika Kousa and Pasi Eronen of Nokia Corporation] 5002 5003 *) Add RFC 3161 compliant time stamp request creation, response generation 5004 and response verification functionality. 5005 [Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project] 5006 5007 *) Add initial support for TLS extensions, specifically for the server_name 5008 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 5009 have new members for a host name. The SSL data structure has an 5010 additional member SSL_CTX *initial_ctx so that new sessions can be 5011 stored in that context to allow for session resumption, even after the 5012 SSL has been switched to a new SSL_CTX in reaction to a client's 5013 server_name extension. 5014 5015 New functions (subject to change): 5016 5017 SSL_get_servername() 5018 SSL_get_servername_type() 5019 SSL_set_SSL_CTX() 5020 5021 New CTRL codes and macros (subject to change): 5022 5023 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 5024 - SSL_CTX_set_tlsext_servername_callback() 5025 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 5026 - SSL_CTX_set_tlsext_servername_arg() 5027 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 5028 5029 openssl s_client has a new '-servername ...' option. 5030 5031 openssl s_server has new options '-servername_host ...', '-cert2 ...', 5032 '-key2 ...', '-servername_fatal' (subject to change). This allows 5033 testing the HostName extension for a specific single host name ('-cert' 5034 and '-key' remain fallbacks for handshakes without HostName 5035 negotiation). If the unrecognized_name alert has to be sent, this by 5036 default is a warning; it becomes fatal with the '-servername_fatal' 5037 option. 5038 5039 [Peter Sylvester, Remy Allais, Christophe Renou] 5040 5041 *) Whirlpool hash implementation is added. 5042 [Andy Polyakov] 5043 5044 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 5045 bn(64,32). Because of instruction set limitations it doesn't have 5046 any negative impact on performance. This was done mostly in order 5047 to make it possible to share assembler modules, such as bn_mul_mont 5048 implementations, between 32- and 64-bit builds without hassle. 5049 [Andy Polyakov] 5050 5051 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 5052 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 5053 macro. 5054 [Bodo Moeller] 5055 5056 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 5057 dedicated Montgomery multiplication procedure, is introduced. 5058 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 5059 "64-bit" performance on certain 32-bit targets. 5060 [Andy Polyakov] 5061 5062 *) New option SSL_OP_NO_COMP to disable use of compression selectively 5063 in SSL structures. New SSL ctrl to set maximum send fragment size. 5064 Save memory by setting the I/O buffer sizes dynamically instead of 5065 using the maximum available value. 5066 [Steve Henson] 5067 5068 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 5069 in addition to the text details. 5070 [Bodo Moeller] 5071 5072 *) Very, very preliminary EXPERIMENTAL support for printing of general 5073 ASN1 structures. This currently produces rather ugly output and doesn't 5074 handle several customised structures at all. 5075 [Steve Henson] 5076 5077 *) Integrated support for PVK file format and some related formats such 5078 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 5079 these in the 'rsa' and 'dsa' utilities. 5080 [Steve Henson] 5081 5082 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 5083 [Steve Henson] 5084 5085 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 5086 place for the (very old) "NETSCAPE" format certificates which are now 5087 handled using new ASN1 code equivalents. 5088 [Steve Henson] 5089 5090 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 5091 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 5092 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 5093 [Nils Larsch] 5094 5095 *) Modify CRL distribution points extension code to print out previously 5096 unsupported fields. Enhance extension setting code to allow setting of 5097 all fields. 5098 [Steve Henson] 5099 5100 *) Add print and set support for Issuing Distribution Point CRL extension. 5101 [Steve Henson] 5102 5103 *) Change 'Configure' script to enable Camellia by default. 5104 [NTT] 5105 5106 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 5107 5108 *) When rejecting SSL/TLS records due to an incorrect version number, never 5109 update s->server with a new major version number. As of 5110 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 5111 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 5112 the previous behavior could result in a read attempt at NULL when 5113 receiving specific incorrect SSL/TLS records once record payload 5114 protection is active. (CVE-2010-0740) 5115 [Bodo Moeller, Adam Langley <agl@chromium.org>] 5116 5117 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 5118 could be crashed if the relevant tables were not present (e.g. chrooted). 5119 [Tomas Hoger <thoger@redhat.com>] 5120 5121 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 5122 5123 *) Always check bn_wexpand() return values for failure. (CVE-2009-3245) 5124 [Martin Olsson, Neel Mehta] 5125 5126 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 5127 accommodate for stack sorting, always a write lock!). 5128 [Bodo Moeller] 5129 5130 *) On some versions of WIN32 Heap32Next is very slow. This can cause 5131 excessive delays in the RAND_poll(): over a minute. As a workaround 5132 include a time check in the inner Heap32Next loop too. 5133 [Steve Henson] 5134 5135 *) The code that handled flushing of data in SSL/TLS originally used the 5136 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 5137 the problem outlined in PR#1949. The fix suggested there however can 5138 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 5139 of Apache). So instead simplify the code to flush unconditionally. 5140 This should be fine since flushing with no data to flush is a no op. 5141 [Steve Henson] 5142 5143 *) Handle TLS versions 2.0 and later properly and correctly use the 5144 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 5145 off ancient servers have a habit of sticking around for a while... 5146 [Steve Henson] 5147 5148 *) Modify compression code so it frees up structures without using the 5149 ex_data callbacks. This works around a problem where some applications 5150 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 5151 restarting) then use compression (e.g. SSL with compression) later. 5152 This results in significant per-connection memory leaks and 5153 has caused some security issues including CVE-2008-1678 and 5154 CVE-2009-4355. 5155 [Steve Henson] 5156 5157 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 5158 change when encrypting or decrypting. 5159 [Bodo Moeller] 5160 5161 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 5162 connect and renegotiate with servers which do not support RI. 5163 Until RI is more widely deployed this option is enabled by default. 5164 [Steve Henson] 5165 5166 *) Add "missing" ssl ctrls to clear options and mode. 5167 [Steve Henson] 5168 5169 *) If client attempts to renegotiate and doesn't support RI respond with 5170 a no_renegotiation alert as required by RFC5746. Some renegotiating 5171 TLS clients will continue a connection gracefully when they receive 5172 the alert. Unfortunately OpenSSL mishandled this alert and would hang 5173 waiting for a server hello which it will never receive. Now we treat a 5174 received no_renegotiation alert as a fatal error. This is because 5175 applications requesting a renegotiation might well expect it to succeed 5176 and would have no code in place to handle the server denying it so the 5177 only safe thing to do is to terminate the connection. 5178 [Steve Henson] 5179 5180 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 5181 peer supports secure renegotiation and 0 otherwise. Print out peer 5182 renegotiation support in s_client/s_server. 5183 [Steve Henson] 5184 5185 *) Replace the highly broken and deprecated SPKAC certification method with 5186 the updated NID creation version. This should correctly handle UTF8. 5187 [Steve Henson] 5188 5189 *) Implement RFC5746. Re-enable renegotiation but require the extension 5190 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 5191 turns out to be a bad idea. It has been replaced by 5192 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 5193 SSL_CTX_set_options(). This is really not recommended unless you 5194 know what you are doing. 5195 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 5196 5197 *) Fixes to stateless session resumption handling. Use initial_ctx when 5198 issuing and attempting to decrypt tickets in case it has changed during 5199 servername handling. Use a non-zero length session ID when attempting 5200 stateless session resumption: this makes it possible to determine if 5201 a resumption has occurred immediately after receiving server hello 5202 (several places in OpenSSL subtly assume this) instead of later in 5203 the handshake. 5204 [Steve Henson] 5205 5206 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 5207 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 5208 fixes for a few places where the return code is not checked 5209 correctly. 5210 [Julia Lawall <julia@diku.dk>] 5211 5212 *) Add --strict-warnings option to Configure script to include devteam 5213 warnings in other configurations. 5214 [Steve Henson] 5215 5216 *) Add support for --libdir option and LIBDIR variable in makefiles. This 5217 makes it possible to install openssl libraries in locations which 5218 have names other than "lib", for example "/usr/lib64" which some 5219 systems need. 5220 [Steve Henson, based on patch from Jeremy Utley] 5221 5222 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 5223 X690 8.9.12 and can produce some misleading textual output of OIDs. 5224 [Steve Henson, reported by Dan Kaminsky] 5225 5226 *) Delete MD2 from algorithm tables. This follows the recommendation in 5227 several standards that it is not used in new applications due to 5228 several cryptographic weaknesses. For binary compatibility reasons 5229 the MD2 API is still compiled in by default. 5230 [Steve Henson] 5231 5232 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 5233 and restored. 5234 [Steve Henson] 5235 5236 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 5237 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 5238 clash. 5239 [Guenter <lists@gknw.net>] 5240 5241 *) Fix the server certificate chain building code to use X509_verify_cert(), 5242 it used to have an ad-hoc builder which was unable to cope with anything 5243 other than a simple chain. 5244 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 5245 5246 *) Don't check self signed certificate signatures in X509_verify_cert() 5247 by default (a flag can override this): it just wastes time without 5248 adding any security. As a useful side effect self signed root CAs 5249 with non-FIPS digests are now usable in FIPS mode. 5250 [Steve Henson] 5251 5252 *) In dtls1_process_out_of_seq_message() the check if the current message 5253 is already buffered was missing. For every new message was memory 5254 allocated, allowing an attacker to perform an denial of service attack 5255 with sending out of seq handshake messages until there is no memory 5256 left. Additionally every future message was buffered, even if the 5257 sequence number made no sense and would be part of another handshake. 5258 So only messages with sequence numbers less than 10 in advance will be 5259 buffered. (CVE-2009-1378) 5260 [Robin Seggelmann, discovered by Daniel Mentz] 5261 5262 *) Records are buffered if they arrive with a future epoch to be 5263 processed after finishing the corresponding handshake. There is 5264 currently no limitation to this buffer allowing an attacker to perform 5265 a DOS attack with sending records with future epochs until there is no 5266 memory left. This patch adds the pqueue_size() function to determine 5267 the size of a buffer and limits the record buffer to 100 entries. 5268 (CVE-2009-1377) 5269 [Robin Seggelmann, discovered by Daniel Mentz] 5270 5271 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 5272 parent structure is freed. (CVE-2009-1379) 5273 [Daniel Mentz] 5274 5275 *) Handle non-blocking I/O properly in SSL_shutdown() call. 5276 [Darryl Miles <darryl-mailinglists@netbauds.net>] 5277 5278 *) Add 2.5.4.* OIDs 5279 [Ilya O. <vrghost@gmail.com>] 5280 5281 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 5282 5283 *) Disable renegotiation completely - this fixes a severe security 5284 problem (CVE-2009-3555) at the cost of breaking all 5285 renegotiation. Renegotiation can be re-enabled by setting 5286 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 5287 run-time. This is really not recommended unless you know what 5288 you're doing. 5289 [Ben Laurie] 5290 5291 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 5292 5293 *) Don't set val to NULL when freeing up structures, it is freed up by 5294 underlying code. If sizeof(void *) > sizeof(long) this can result in 5295 zeroing past the valid field. (CVE-2009-0789) 5296 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 5297 5298 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 5299 checked correctly. This would allow some invalid signed attributes to 5300 appear to verify correctly. (CVE-2009-0591) 5301 [Ivan Nestlerode <inestlerode@us.ibm.com>] 5302 5303 *) Reject UniversalString and BMPString types with invalid lengths. This 5304 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 5305 a legal length. (CVE-2009-0590) 5306 [Steve Henson] 5307 5308 *) Set S/MIME signing as the default purpose rather than setting it 5309 unconditionally. This allows applications to override it at the store 5310 level. 5311 [Steve Henson] 5312 5313 *) Permit restricted recursion of ASN1 strings. This is needed in practice 5314 to handle some structures. 5315 [Steve Henson] 5316 5317 *) Improve efficiency of mem_gets: don't search whole buffer each time 5318 for a '\n' 5319 [Jeremy Shapiro <jnshapir@us.ibm.com>] 5320 5321 *) New -hex option for openssl rand. 5322 [Matthieu Herrb] 5323 5324 *) Print out UTF8String and NumericString when parsing ASN1. 5325 [Steve Henson] 5326 5327 *) Support NumericString type for name components. 5328 [Steve Henson] 5329 5330 *) Allow CC in the environment to override the automatically chosen 5331 compiler. Note that nothing is done to ensure flags work with the 5332 chosen compiler. 5333 [Ben Laurie] 5334 5335 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 5336 5337 *) Properly check EVP_VerifyFinal() and similar return values 5338 (CVE-2008-5077). 5339 [Ben Laurie, Bodo Moeller, Google Security Team] 5340 5341 *) Enable TLS extensions by default. 5342 [Ben Laurie] 5343 5344 *) Allow the CHIL engine to be loaded, whether the application is 5345 multithreaded or not. (This does not release the developer from the 5346 obligation to set up the dynamic locking callbacks.) 5347 [Sander Temme <sander@temme.net>] 5348 5349 *) Use correct exit code if there is an error in dgst command. 5350 [Steve Henson; problem pointed out by Roland Dirlewanger] 5351 5352 *) Tweak Configure so that you need to say "experimental-jpake" to enable 5353 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 5354 [Bodo Moeller] 5355 5356 *) Add experimental JPAKE support, including demo authentication in 5357 s_client and s_server. 5358 [Ben Laurie] 5359 5360 *) Set the comparison function in v3_addr_canonize(). 5361 [Rob Austein <sra@hactrn.net>] 5362 5363 *) Add support for XMPP STARTTLS in s_client. 5364 [Philip Paeps <philip@freebsd.org>] 5365 5366 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 5367 to ensure that even with this option, only ciphersuites in the 5368 server's preference list will be accepted. (Note that the option 5369 applies only when resuming a session, so the earlier behavior was 5370 just about the algorithm choice for symmetric cryptography.) 5371 [Bodo Moeller] 5372 5373 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 5374 5375 *) Fix NULL pointer dereference if a DTLS server received 5376 ChangeCipherSpec as first record (CVE-2009-1386). 5377 [PR #1679] 5378 5379 *) Fix a state transition in s3_srvr.c and d1_srvr.c 5380 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 5381 [Nagendra Modadugu] 5382 5383 *) The fix in 0.9.8c that supposedly got rid of unsafe 5384 double-checked locking was incomplete for RSA blinding, 5385 addressing just one layer of what turns out to have been 5386 doubly unsafe triple-checked locking. 5387 5388 So now fix this for real by retiring the MONT_HELPER macro 5389 in crypto/rsa/rsa_eay.c. 5390 5391 [Bodo Moeller; problem pointed out by Marius Schilder] 5392 5393 *) Various precautionary measures: 5394 5395 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 5396 5397 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 5398 (NB: This would require knowledge of the secret session ticket key 5399 to exploit, in which case you'd be SOL either way.) 5400 5401 - Change bn_nist.c so that it will properly handle input BIGNUMs 5402 outside the expected range. 5403 5404 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 5405 builds. 5406 5407 [Neel Mehta, Bodo Moeller] 5408 5409 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 5410 the load fails. Useful for distros. 5411 [Ben Laurie and the FreeBSD team] 5412 5413 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 5414 [Steve Henson] 5415 5416 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 5417 [Huang Ying] 5418 5419 *) Expand ENGINE to support engine supplied SSL client certificate functions. 5420 5421 This work was sponsored by Logica. 5422 [Steve Henson] 5423 5424 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 5425 keystores. Support for SSL/TLS client authentication too. 5426 Not compiled unless enable-capieng specified to Configure. 5427 5428 This work was sponsored by Logica. 5429 [Steve Henson] 5430 5431 *) Fix bug in X509_ATTRIBUTE creation: don't set attribute using 5432 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 5433 attribute creation routines such as certificate requests and PKCS#12 5434 files. 5435 [Steve Henson] 5436 5437 Changes between 0.9.8g and 0.9.8h [28 May 2008] 5438 5439 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 5440 handshake which could lead to a client crash as found using the 5441 Codenomicon TLS test suite (CVE-2008-1672) 5442 [Steve Henson, Mark Cox] 5443 5444 *) Fix double free in TLS server name extensions which could lead to 5445 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 5446 [Joe Orton] 5447 5448 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 5449 5450 Clear the error queue to ensure that error entries left from 5451 older function calls do not interfere with the correct operation. 5452 [Lutz Jaenicke, Erik de Castro Lopo] 5453 5454 *) Remove root CA certificates of commercial CAs: 5455 5456 The OpenSSL project does not recommend any specific CA and does not 5457 have any policy with respect to including or excluding any CA. 5458 Therefore it does not make any sense to ship an arbitrary selection 5459 of root CA certificates with the OpenSSL software. 5460 [Lutz Jaenicke] 5461 5462 *) RSA OAEP patches to fix two separate invalid memory reads. 5463 The first one involves inputs when 'lzero' is greater than 5464 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 5465 before the beginning of from). The second one involves inputs where 5466 the 'db' section contains nothing but zeroes (there is a one-byte 5467 invalid read after the end of 'db'). 5468 [Ivan Nestlerode <inestlerode@us.ibm.com>] 5469 5470 *) Partial backport from 0.9.9-dev: 5471 5472 Introduce bn_mul_mont (dedicated Montgomery multiplication 5473 procedure) as a candidate for BIGNUM assembler implementation. 5474 While 0.9.9-dev uses assembler for various architectures, only 5475 x86_64 is available by default here in the 0.9.8 branch, and 5476 32-bit x86 is available through a compile-time setting. 5477 5478 To try the 32-bit x86 assembler implementation, use Configure 5479 option "enable-montasm" (which exists only for this backport). 5480 5481 As "enable-montasm" for 32-bit x86 disclaims code stability 5482 anyway, in this constellation we activate additional code 5483 backported from 0.9.9-dev for further performance improvements, 5484 namely BN_from_montgomery_word. (To enable this otherwise, 5485 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 5486 5487 [Andy Polyakov (backport partially by Bodo Moeller)] 5488 5489 *) Add TLS session ticket callback. This allows an application to set 5490 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 5491 values. This is useful for key rollover for example where several key 5492 sets may exist with different names. 5493 [Steve Henson] 5494 5495 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 5496 This was broken until now in 0.9.8 releases, such that the only way 5497 a registered ENGINE could be used (assuming it initialises 5498 successfully on the host) was to explicitly set it as the default 5499 for the relevant algorithms. This is in contradiction with 0.9.7 5500 behaviour and the documentation. With this fix, when an ENGINE is 5501 registered into a given algorithm's table of implementations, the 5502 'uptodate' flag is reset so that auto-discovery will be used next 5503 time a new context for that algorithm attempts to select an 5504 implementation. 5505 [Ian Lister (tweaked by Geoff Thorpe)] 5506 5507 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 5508 implementation in the following ways: 5509 5510 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 5511 hard coded. 5512 5513 Lack of BER streaming support means one pass streaming processing is 5514 only supported if data is detached: setting the streaming flag is 5515 ignored for embedded content. 5516 5517 CMS support is disabled by default and must be explicitly enabled 5518 with the enable-cms configuration option. 5519 [Steve Henson] 5520 5521 *) Update the GMP engine glue to do direct copies between BIGNUM and 5522 mpz_t when openssl and GMP use the same limb size. Otherwise the 5523 existing "conversion via a text string export" trick is still used. 5524 [Paul Sheer <paulsheer@gmail.com>] 5525 5526 *) Zlib compression BIO. This is a filter BIO which compressed and 5527 uncompresses any data passed through it. 5528 [Steve Henson] 5529 5530 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 5531 RFC3394 compatible AES key wrapping. 5532 [Steve Henson] 5533 5534 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 5535 sets string data without copying. X509_ALGOR_set0() and 5536 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 5537 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 5538 from an X509_ATTRIBUTE structure optionally checking it occurs only 5539 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 5540 data. 5541 [Steve Henson] 5542 5543 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 5544 to get the expected BN_FLG_CONSTTIME behavior. 5545 [Bodo Moeller (Google)] 5546 5547 *) Netware support: 5548 5549 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 5550 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 5551 - added some more tests to do_tests.pl 5552 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 5553 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 5554 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 5555 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 5556 - various changes to netware.pl to enable gcc-cross builds on Win32 5557 platform 5558 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 5559 - various changes to fix missing prototype warnings 5560 - fixed x86nasm.pl to create correct asm files for NASM COFF output 5561 - added AES, WHIRLPOOL and CPUID assembler code to build files 5562 - added missing AES assembler make rules to mk1mf.pl 5563 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 5564 [Guenter Knauf <eflash@gmx.net>] 5565 5566 *) Implement certificate status request TLS extension defined in RFC3546. 5567 A client can set the appropriate parameters and receive the encoded 5568 OCSP response via a callback. A server can query the supplied parameters 5569 and set the encoded OCSP response in the callback. Add simplified examples 5570 to s_client and s_server. 5571 [Steve Henson] 5572 5573 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 5574 5575 *) Fix various bugs: 5576 + Binary incompatibility of ssl_ctx_st structure 5577 + DTLS interoperation with non-compliant servers 5578 + Don't call get_session_cb() without proposed session 5579 + Fix ia64 assembler code 5580 [Andy Polyakov, Steve Henson] 5581 5582 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 5583 5584 *) DTLS Handshake overhaul. There were longstanding issues with 5585 OpenSSL DTLS implementation, which were making it impossible for 5586 RFC 4347 compliant client to communicate with OpenSSL server. 5587 Unfortunately just fixing these incompatibilities would "cut off" 5588 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 5589 server keeps tolerating non RFC compliant syntax. The opposite is 5590 not true, 0.9.8f client can not communicate with earlier server. 5591 This update even addresses CVE-2007-4995. 5592 [Andy Polyakov] 5593 5594 *) Changes to avoid need for function casts in OpenSSL: some compilers 5595 (gcc 4.2 and later) reject their use. 5596 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 5597 Steve Henson] 5598 5599 *) Add RFC4507 support to OpenSSL. This includes the corrections in 5600 RFC4507bis. The encrypted ticket format is an encrypted encoded 5601 SSL_SESSION structure, that way new session features are automatically 5602 supported. 5603 5604 If a client application caches session in an SSL_SESSION structure 5605 support is transparent because tickets are now stored in the encoded 5606 SSL_SESSION. 5607 5608 The SSL_CTX structure automatically generates keys for ticket 5609 protection in servers so again support should be possible 5610 with no application modification. 5611 5612 If a client or server wishes to disable RFC4507 support then the option 5613 SSL_OP_NO_TICKET can be set. 5614 5615 Add a TLS extension debugging callback to allow the contents of any client 5616 or server extensions to be examined. 5617 5618 This work was sponsored by Google. 5619 [Steve Henson] 5620 5621 *) Add initial support for TLS extensions, specifically for the server_name 5622 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 5623 have new members for a host name. The SSL data structure has an 5624 additional member SSL_CTX *initial_ctx so that new sessions can be 5625 stored in that context to allow for session resumption, even after the 5626 SSL has been switched to a new SSL_CTX in reaction to a client's 5627 server_name extension. 5628 5629 New functions (subject to change): 5630 5631 SSL_get_servername() 5632 SSL_get_servername_type() 5633 SSL_set_SSL_CTX() 5634 5635 New CTRL codes and macros (subject to change): 5636 5637 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 5638 - SSL_CTX_set_tlsext_servername_callback() 5639 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 5640 - SSL_CTX_set_tlsext_servername_arg() 5641 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 5642 5643 openssl s_client has a new '-servername ...' option. 5644 5645 openssl s_server has new options '-servername_host ...', '-cert2 ...', 5646 '-key2 ...', '-servername_fatal' (subject to change). This allows 5647 testing the HostName extension for a specific single host name ('-cert' 5648 and '-key' remain fallbacks for handshakes without HostName 5649 negotiation). If the unrecognized_name alert has to be sent, this by 5650 default is a warning; it becomes fatal with the '-servername_fatal' 5651 option. 5652 5653 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 5654 5655 *) Add AES and SSE2 assembly language support to VC++ build. 5656 [Steve Henson] 5657 5658 *) Mitigate attack on final subtraction in Montgomery reduction. 5659 [Andy Polyakov] 5660 5661 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 5662 (which previously caused an internal error). 5663 [Bodo Moeller] 5664 5665 *) Squeeze another 10% out of IGE mode when in != out. 5666 [Ben Laurie] 5667 5668 *) AES IGE mode speedup. 5669 [Dean Gaudet (Google)] 5670 5671 *) Add the Korean symmetric 128-bit cipher SEED (see 5672 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 5673 add SEED ciphersuites from RFC 4162: 5674 5675 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 5676 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 5677 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 5678 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 5679 5680 To minimize changes between patchlevels in the OpenSSL 0.9.8 5681 series, SEED remains excluded from compilation unless OpenSSL 5682 is configured with 'enable-seed'. 5683 [KISA, Bodo Moeller] 5684 5685 *) Mitigate branch prediction attacks, which can be practical if a 5686 single processor is shared, allowing a spy process to extract 5687 information. For detailed background information, see 5688 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 5689 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 5690 and Necessary Software Countermeasures"). The core of the change 5691 are new versions BN_div_no_branch() and 5692 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 5693 respectively, which are slower, but avoid the security-relevant 5694 conditional branches. These are automatically called by BN_div() 5695 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 5696 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 5697 remove a conditional branch. 5698 5699 BN_FLG_CONSTTIME is the new name for the previous 5700 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 5701 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 5702 in the exponent causes BN_mod_exp_mont() to use the alternative 5703 implementation in BN_mod_exp_mont_consttime().) The old name 5704 remains as a deprecated alias. 5705 5706 Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 5707 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 5708 constant-time implementations for more than just exponentiation. 5709 Here too the old name is kept as a deprecated alias. 5710 5711 BN_BLINDING_new() will now use BN_dup() for the modulus so that 5712 the BN_BLINDING structure gets an independent copy of the 5713 modulus. This means that the previous "BIGNUM *m" argument to 5714 BN_BLINDING_new() and to BN_BLINDING_create_param() now 5715 essentially becomes "const BIGNUM *m", although we can't actually 5716 change this in the header file before 0.9.9. It allows 5717 RSA_setup_blinding() to use BN_with_flags() on the modulus to 5718 enable BN_FLG_CONSTTIME. 5719 5720 [Matthew D Wood (Intel Corp)] 5721 5722 *) In the SSL/TLS server implementation, be strict about session ID 5723 context matching (which matters if an application uses a single 5724 external cache for different purposes). Previously, 5725 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 5726 set. This did ensure strict client verification, but meant that, 5727 with applications using a single external cache for quite 5728 different requirements, clients could circumvent ciphersuite 5729 restrictions for a given session ID context by starting a session 5730 in a different context. 5731 [Bodo Moeller] 5732 5733 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 5734 a ciphersuite string such as "DEFAULT:RSA" cannot enable 5735 authentication-only ciphersuites. 5736 [Bodo Moeller] 5737 5738 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 5739 not complete and could lead to a possible single byte overflow 5740 (CVE-2007-5135) [Ben Laurie] 5741 5742 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 5743 5744 *) Since AES128 and AES256 (and similarly Camellia128 and 5745 Camellia256) share a single mask bit in the logic of 5746 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 5747 kludge to work properly if AES128 is available and AES256 isn't 5748 (or if Camellia128 is available and Camellia256 isn't). 5749 [Victor Duchovni] 5750 5751 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 5752 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 5753 When a point or a seed is encoded in a BIT STRING, we need to 5754 prevent the removal of trailing zero bits to get the proper DER 5755 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 5756 of a NamedBitList, for which trailing 0 bits need to be removed.) 5757 [Bodo Moeller] 5758 5759 *) Have SSL/TLS server implementation tolerate "mismatched" record 5760 protocol version while receiving ClientHello even if the 5761 ClientHello is fragmented. (The server can't insist on the 5762 particular protocol version it has chosen before the ServerHello 5763 message has informed the client about his choice.) 5764 [Bodo Moeller] 5765 5766 *) Add RFC 3779 support. 5767 [Rob Austein for ARIN, Ben Laurie] 5768 5769 *) Load error codes if they are not already present instead of using a 5770 static variable. This allows them to be cleanly unloaded and reloaded. 5771 Improve header file function name parsing. 5772 [Steve Henson] 5773 5774 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 5775 or CAPABILITY handshake as required by RFCs. 5776 [Goetz Babin-Ebell] 5777 5778 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 5779 5780 *) Introduce limits to prevent malicious keys being able to 5781 cause a denial of service. (CVE-2006-2940) 5782 [Steve Henson, Bodo Moeller] 5783 5784 *) Fix ASN.1 parsing of certain invalid structures that can result 5785 in a denial of service. (CVE-2006-2937) [Steve Henson] 5786 5787 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 5788 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 5789 5790 *) Fix SSL client code which could crash if connecting to a 5791 malicious SSLv2 server. (CVE-2006-4343) 5792 [Tavis Ormandy and Will Drewry, Google Security Team] 5793 5794 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 5795 match only those. Before that, "AES256-SHA" would be interpreted 5796 as a pattern and match "AES128-SHA" too (since AES128-SHA got 5797 the same strength classification in 0.9.7h) as we currently only 5798 have a single AES bit in the ciphersuite description bitmap. 5799 That change, however, also applied to ciphersuite strings such as 5800 "RC4-MD5" that intentionally matched multiple ciphersuites -- 5801 namely, SSL 2.0 ciphersuites in addition to the more common ones 5802 from SSL 3.0/TLS 1.0. 5803 5804 So we change the selection algorithm again: Naming an explicit 5805 ciphersuite selects this one ciphersuite, and any other similar 5806 ciphersuite (same bitmap) from *other* protocol versions. 5807 Thus, "RC4-MD5" again will properly select both the SSL 2.0 5808 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 5809 5810 Since SSL 2.0 does not have any ciphersuites for which the 5811 128/256 bit distinction would be relevant, this works for now. 5812 The proper fix will be to use different bits for AES128 and 5813 AES256, which would have avoided the problems from the beginning; 5814 however, bits are scarce, so we can only do this in a new release 5815 (not just a patchlevel) when we can change the SSL_CIPHER 5816 definition to split the single 'unsigned long mask' bitmap into 5817 multiple values to extend the available space. 5818 5819 [Bodo Moeller] 5820 5821 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 5822 5823 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 5824 (CVE-2006-4339) [Ben Laurie and Google Security Team] 5825 5826 *) Add AES IGE and biIGE modes. 5827 [Ben Laurie] 5828 5829 *) Change the Unix randomness entropy gathering to use poll() when 5830 possible instead of select(), since the latter has some 5831 undesirable limitations. 5832 [Darryl Miles via Richard Levitte and Bodo Moeller] 5833 5834 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 5835 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 5836 cannot be implicitly activated as part of, e.g., the "AES" alias. 5837 However, please upgrade to OpenSSL 0.9.9[-dev] for 5838 non-experimental use of the ECC ciphersuites to get TLS extension 5839 support, which is required for curve and point format negotiation 5840 to avoid potential handshake problems. 5841 [Bodo Moeller] 5842 5843 *) Disable rogue ciphersuites: 5844 5845 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 5846 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 5847 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 5848 5849 The latter two were purportedly from 5850 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 5851 appear there. 5852 5853 Also deactivate the remaining ciphersuites from 5854 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 5855 unofficial, and the ID has long expired. 5856 [Bodo Moeller] 5857 5858 *) Fix RSA blinding Heisenbug (problems sometimes occurred on 5859 dual-core machines) and other potential thread-safety issues. 5860 [Bodo Moeller] 5861 5862 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 5863 versions), which is now available for royalty-free use 5864 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 5865 Also, add Camellia TLS ciphersuites from RFC 4132. 5866 5867 To minimize changes between patchlevels in the OpenSSL 0.9.8 5868 series, Camellia remains excluded from compilation unless OpenSSL 5869 is configured with 'enable-camellia'. 5870 [NTT] 5871 5872 *) Disable the padding bug check when compression is in use. The padding 5873 bug check assumes the first packet is of even length, this is not 5874 necessarily true if compression is enabled and can result in false 5875 positives causing handshake failure. The actual bug test is ancient 5876 code so it is hoped that implementations will either have fixed it by 5877 now or any which still have the bug do not support compression. 5878 [Steve Henson] 5879 5880 Changes between 0.9.8a and 0.9.8b [04 May 2006] 5881 5882 *) When applying a cipher rule check to see if string match is an explicit 5883 cipher suite and only match that one cipher suite if it is. 5884 [Steve Henson] 5885 5886 *) Link in manifests for VC++ if needed. 5887 [Austin Ziegler <halostatue@gmail.com>] 5888 5889 *) Update support for ECC-based TLS ciphersuites according to 5890 draft-ietf-tls-ecc-12.txt with proposed changes (but without 5891 TLS extensions, which are supported starting with the 0.9.9 5892 branch, not in the OpenSSL 0.9.8 branch). 5893 [Douglas Stebila] 5894 5895 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 5896 opaque EVP_CIPHER_CTX handling. 5897 [Steve Henson] 5898 5899 *) Fixes and enhancements to zlib compression code. We now only use 5900 "zlib1.dll" and use the default __cdecl calling convention on Win32 5901 to conform with the standards mentioned here: 5902 http://www.zlib.net/DLL_FAQ.txt 5903 Static zlib linking now works on Windows and the new --with-zlib-include 5904 --with-zlib-lib options to Configure can be used to supply the location 5905 of the headers and library. Gracefully handle case where zlib library 5906 can't be loaded. 5907 [Steve Henson] 5908 5909 *) Several fixes and enhancements to the OID generation code. The old code 5910 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 5911 handle numbers larger than ULONG_MAX, truncated printing and had a 5912 non standard OBJ_obj2txt() behaviour. 5913 [Steve Henson] 5914 5915 *) Add support for building of engines under engine/ as shared libraries 5916 under VC++ build system. 5917 [Steve Henson] 5918 5919 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 5920 Hopefully, we will not see any false combination of paths any more. 5921 [Richard Levitte] 5922 5923 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 5924 5925 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 5926 (part of SSL_OP_ALL). This option used to disable the 5927 countermeasure against man-in-the-middle protocol-version 5928 rollback in the SSL 2.0 server implementation, which is a bad 5929 idea. (CVE-2005-2969) 5930 5931 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 5932 for Information Security, National Institute of Advanced Industrial 5933 Science and Technology [AIST], Japan)] 5934 5935 *) Add two function to clear and return the verify parameter flags. 5936 [Steve Henson] 5937 5938 *) Keep cipherlists sorted in the source instead of sorting them at 5939 runtime, thus removing the need for a lock. 5940 [Nils Larsch] 5941 5942 *) Avoid some small subgroup attacks in Diffie-Hellman. 5943 [Nick Mathewson and Ben Laurie] 5944 5945 *) Add functions for well-known primes. 5946 [Nick Mathewson] 5947 5948 *) Extended Windows CE support. 5949 [Satoshi Nakamura and Andy Polyakov] 5950 5951 *) Initialize SSL_METHOD structures at compile time instead of during 5952 runtime, thus removing the need for a lock. 5953 [Steve Henson] 5954 5955 *) Make PKCS7_decrypt() work even if no certificate is supplied by 5956 attempting to decrypt each encrypted key in turn. Add support to 5957 smime utility. 5958 [Steve Henson] 5959 5960 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 5961 5962 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 5963 OpenSSL 0.9.8.] 5964 5965 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 5966 [Richard Levitte] 5967 5968 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 5969 key into the same file any more. 5970 [Richard Levitte] 5971 5972 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 5973 [Andy Polyakov] 5974 5975 *) Add -utf8 command line and config file option to 'ca'. 5976 [Stefan <stf@udoma.org] 5977 5978 *) Removed the macro des_crypt(), as it seems to conflict with some 5979 libraries. Use DES_crypt(). 5980 [Richard Levitte] 5981 5982 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 5983 involves renaming the source and generated shared-libs for 5984 both. The engines will accept the corrected or legacy ids 5985 ('ncipher' and '4758_cca' respectively) when binding. NB, 5986 this only applies when building 'shared'. 5987 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 5988 5989 *) Add attribute functions to EVP_PKEY structure. Modify 5990 PKCS12_create() to recognize a CSP name attribute and 5991 use it. Make -CSP option work again in pkcs12 utility. 5992 [Steve Henson] 5993 5994 *) Add new functionality to the bn blinding code: 5995 - automatic re-creation of the BN_BLINDING parameters after 5996 a fixed number of uses (currently 32) 5997 - add new function for parameter creation 5998 - introduce flags to control the update behaviour of the 5999 BN_BLINDING parameters 6000 - hide BN_BLINDING structure 6001 Add a second BN_BLINDING slot to the RSA structure to improve 6002 performance when a single RSA object is shared among several 6003 threads. 6004 [Nils Larsch] 6005 6006 *) Add support for DTLS. 6007 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 6008 6009 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 6010 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 6011 [Walter Goulet] 6012 6013 *) Remove buggy and incomplete DH cert support from 6014 ssl/ssl_rsa.c and ssl/s3_both.c 6015 [Nils Larsch] 6016 6017 *) Use SHA-1 instead of MD5 as the default digest algorithm for 6018 the apps/openssl applications. 6019 [Nils Larsch] 6020 6021 *) Compile clean with "-Wall -Wmissing-prototypes 6022 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 6023 DEBUG_SAFESTACK must also be set. 6024 [Ben Laurie] 6025 6026 *) Change ./Configure so that certain algorithms can be disabled by default. 6027 The new counterpiece to "no-xxx" is "enable-xxx". 6028 6029 The patented RC5 and MDC2 algorithms will now be disabled unless 6030 "enable-rc5" and "enable-mdc2", respectively, are specified. 6031 6032 (IDEA remains enabled despite being patented. This is because IDEA 6033 is frequently required for interoperability, and there is no license 6034 fee for non-commercial use. As before, "no-idea" can be used to 6035 avoid this algorithm.) 6036 6037 [Bodo Moeller] 6038 6039 *) Add processing of proxy certificates (see RFC 3820). This work was 6040 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 6041 EGEE (Enabling Grids for E-science in Europe). 6042 [Richard Levitte] 6043 6044 *) RC4 performance overhaul on modern architectures/implementations, such 6045 as Intel P4, IA-64 and AMD64. 6046 [Andy Polyakov] 6047 6048 *) New utility extract-section.pl. This can be used specify an alternative 6049 section number in a pod file instead of having to treat each file as 6050 a separate case in Makefile. This can be done by adding two lines to the 6051 pod file: 6052 6053 =for comment openssl_section:XXX 6054 6055 The blank line is mandatory. 6056 6057 [Steve Henson] 6058 6059 *) New arguments -certform, -keyform and -pass for s_client and s_server 6060 to allow alternative format key and certificate files and passphrase 6061 sources. 6062 [Steve Henson] 6063 6064 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 6065 update associated structures and add various utility functions. 6066 6067 Add new policy related verify parameters, include policy checking in 6068 standard verify code. Enhance 'smime' application with extra parameters 6069 to support policy checking and print out. 6070 [Steve Henson] 6071 6072 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 6073 Nehemiah processors. These extensions support AES encryption in hardware 6074 as well as RNG (though RNG support is currently disabled). 6075 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 6076 6077 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 6078 [Geoff Thorpe] 6079 6080 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 6081 [Andy Polyakov and a number of other people] 6082 6083 *) Improved PowerPC platform support. Most notably BIGNUM assembler 6084 implementation contributed by IBM. 6085 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 6086 6087 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 6088 exponent rather than 'unsigned long'. There is a corresponding change to 6089 the new 'rsa_keygen' element of the RSA_METHOD structure. 6090 [Jelte Jansen, Geoff Thorpe] 6091 6092 *) Functionality for creating the initial serial number file is now 6093 moved from CA.pl to the 'ca' utility with a new option -create_serial. 6094 6095 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 6096 number file to 1, which is bound to cause problems. To avoid 6097 the problems while respecting compatibility between different 0.9.7 6098 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 6099 CA.pl for serial number initialization. With the new release 0.9.8, 6100 we can fix the problem directly in the 'ca' utility.) 6101 [Steve Henson] 6102 6103 *) Reduced header interdependencies by declaring more opaque objects in 6104 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 6105 give fewer recursive includes, which could break lazy source code - so 6106 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 6107 developers should define this symbol when building and using openssl to 6108 ensure they track the recommended behaviour, interfaces, [etc], but 6109 backwards-compatible behaviour prevails when this isn't defined. 6110 [Geoff Thorpe] 6111 6112 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 6113 [Steve Henson] 6114 6115 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 6116 This will generate a random key of the appropriate length based on the 6117 cipher context. The EVP_CIPHER can provide its own random key generation 6118 routine to support keys of a specific form. This is used in the des and 6119 3des routines to generate a key of the correct parity. Update S/MIME 6120 code to use new functions and hence generate correct parity DES keys. 6121 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 6122 valid (weak or incorrect parity). 6123 [Steve Henson] 6124 6125 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 6126 as looking them up. This is useful when the verified structure may contain 6127 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 6128 present unless the new PKCS7_NO_CRL flag is asserted. 6129 [Steve Henson] 6130 6131 *) Extend ASN1 oid configuration module. It now additionally accepts the 6132 syntax: 6133 6134 shortName = some long name, 1.2.3.4 6135 [Steve Henson] 6136 6137 *) Reimplemented the BN_CTX implementation. There is now no more static 6138 limitation on the number of variables it can handle nor the depth of the 6139 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 6140 information can now expand as required, and rather than having a single 6141 static array of bignums, BN_CTX now uses a linked-list of such arrays 6142 allowing it to expand on demand whilst maintaining the usefulness of 6143 BN_CTX's "bundling". 6144 [Geoff Thorpe] 6145 6146 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 6147 to allow all RSA operations to function using a single BN_CTX. 6148 [Geoff Thorpe] 6149 6150 *) Preliminary support for certificate policy evaluation and checking. This 6151 is initially intended to pass the tests outlined in "Conformance Testing 6152 of Relying Party Client Certificate Path Processing Logic" v1.07. 6153 [Steve Henson] 6154 6155 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 6156 remained unused and not that useful. A variety of other little bignum 6157 tweaks and fixes have also been made continuing on from the audit (see 6158 below). 6159 [Geoff Thorpe] 6160 6161 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 6162 associated ASN1, EVP and SSL functions and old ASN1 macros. 6163 [Richard Levitte] 6164 6165 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 6166 and this should never fail. So the return value from the use of 6167 BN_set_word() (which can fail due to needless expansion) is now deprecated; 6168 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 6169 [Geoff Thorpe] 6170 6171 *) BN_CTX_get() should return zero-valued bignums, providing the same 6172 initialised value as BN_new(). 6173 [Geoff Thorpe, suggested by Ulf Möller] 6174 6175 *) Support for inhibitAnyPolicy certificate extension. 6176 [Steve Henson] 6177 6178 *) An audit of the BIGNUM code is underway, for which debugging code is 6179 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 6180 is considered valid when processing BIGNUMs, and causes execution to 6181 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 6182 further steps are taken to deliberately pollute unused data in BIGNUM 6183 structures to try and expose faulty code further on. For now, openssl will 6184 (in its default mode of operation) continue to tolerate the inconsistent 6185 forms that it has tolerated in the past, but authors and packagers should 6186 consider trying openssl and their own applications when compiled with 6187 these debugging symbols defined. It will help highlight potential bugs in 6188 their own code, and will improve the test coverage for OpenSSL itself. At 6189 some point, these tighter rules will become openssl's default to improve 6190 maintainability, though the assert()s and other overheads will remain only 6191 in debugging configurations. See bn.h for more details. 6192 [Geoff Thorpe, Nils Larsch, Ulf Möller] 6193 6194 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 6195 that can only be obtained through BN_CTX_new() (which implicitly 6196 initialises it). The presence of this function only made it possible 6197 to overwrite an existing structure (and cause memory leaks). 6198 [Geoff Thorpe] 6199 6200 *) Because of the callback-based approach for implementing LHASH as a 6201 template type, lh_insert() adds opaque objects to hash-tables and 6202 lh_doall() or lh_doall_arg() are typically used with a destructor callback 6203 to clean up those corresponding objects before destroying the hash table 6204 (and losing the object pointers). So some over-zealous constifications in 6205 LHASH have been relaxed so that lh_insert() does not take (nor store) the 6206 objects as "const" and the lh_doall[_arg] callback wrappers are not 6207 prototyped to have "const" restrictions on the object pointers they are 6208 given (and so aren't required to cast them away any more). 6209 [Geoff Thorpe] 6210 6211 *) The tmdiff.h API was so ugly and minimal that our own timing utility 6212 (speed) prefers to use its own implementation. The two implementations 6213 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 6214 its object type properly exposed (MS_TM) instead of casting to/from "char 6215 *". This may still change yet if someone realises MS_TM and "ms_time_***" 6216 aren't necessarily the greatest nomenclatures - but this is what was used 6217 internally to the implementation so I've used that for now. 6218 [Geoff Thorpe] 6219 6220 *) Ensure that deprecated functions do not get compiled when 6221 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 6222 the self-tests were still using deprecated key-generation functions so 6223 these have been updated also. 6224 [Geoff Thorpe] 6225 6226 *) Reorganise PKCS#7 code to separate the digest location functionality 6227 into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest(). 6228 New function PKCS7_set_digest() to set the digest type for PKCS#7 6229 digestedData type. Add additional code to correctly generate the 6230 digestedData type and add support for this type in PKCS7 initialization 6231 functions. 6232 [Steve Henson] 6233 6234 *) New function PKCS7_set0_type_other() this initializes a PKCS7 6235 structure of type "other". 6236 [Steve Henson] 6237 6238 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 6239 sure the loop does correctly stop and breaking ("division by zero") 6240 modulus operations are not performed. The (pre-generated) prime 6241 table crypto/bn/bn_prime.h was already correct, but it could not be 6242 re-generated on some platforms because of the "division by zero" 6243 situation in the script. 6244 [Ralf S. Engelschall] 6245 6246 *) Update support for ECC-based TLS ciphersuites according to 6247 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 6248 SHA-1 now is only used for "small" curves (where the 6249 representation of a field element takes up to 24 bytes); for 6250 larger curves, the field element resulting from ECDH is directly 6251 used as premaster secret. 6252 [Douglas Stebila (Sun Microsystems Laboratories)] 6253 6254 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 6255 curve secp160r1 to the tests. 6256 [Douglas Stebila (Sun Microsystems Laboratories)] 6257 6258 *) Add the possibility to load symbols globally with DSO. 6259 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 6260 6261 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 6262 control of the error stack. 6263 [Richard Levitte] 6264 6265 *) Add support for STORE in ENGINE. 6266 [Richard Levitte] 6267 6268 *) Add the STORE type. The intention is to provide a common interface 6269 to certificate and key stores, be they simple file-based stores, or 6270 HSM-type store, or LDAP stores, or... 6271 NOTE: The code is currently UNTESTED and isn't really used anywhere. 6272 [Richard Levitte] 6273 6274 *) Add a generic structure called OPENSSL_ITEM. This can be used to 6275 pass a list of arguments to any function as well as provide a way 6276 for a function to pass data back to the caller. 6277 [Richard Levitte] 6278 6279 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 6280 works like BUF_strdup() but can be used to duplicate a portion of 6281 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 6282 a memory area. 6283 [Richard Levitte] 6284 6285 *) Add the function sk_find_ex() which works like sk_find(), but will 6286 return an index to an element even if an exact match couldn't be 6287 found. The index is guaranteed to point at the element where the 6288 searched-for key would be inserted to preserve sorting order. 6289 [Richard Levitte] 6290 6291 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 6292 takes an extra flags argument for optional functionality. Currently, 6293 the following flags are defined: 6294 6295 OBJ_BSEARCH_VALUE_ON_NOMATCH 6296 This one gets OBJ_bsearch_ex() to return a pointer to the first 6297 element where the comparing function returns a negative or zero 6298 number. 6299 6300 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 6301 This one gets OBJ_bsearch_ex() to return a pointer to the first 6302 element where the comparing function returns zero. This is useful 6303 if there are more than one element where the comparing function 6304 returns zero. 6305 [Richard Levitte] 6306 6307 *) Make it possible to create self-signed certificates with 'openssl ca' 6308 in such a way that the self-signed certificate becomes part of the 6309 CA database and uses the same mechanisms for serial number generation 6310 as all other certificate signing. The new flag '-selfsign' enables 6311 this functionality. Adapt CA.sh and CA.pl.in. 6312 [Richard Levitte] 6313 6314 *) Add functionality to check the public key of a certificate request 6315 against a given private. This is useful to check that a certificate 6316 request can be signed by that key (self-signing). 6317 [Richard Levitte] 6318 6319 *) Make it possible to have multiple active certificates with the same 6320 subject in the CA index file. This is done only if the keyword 6321 'unique_subject' is set to 'no' in the main CA section (default 6322 if 'CA_default') of the configuration file. The value is saved 6323 with the database itself in a separate index attribute file, 6324 named like the index file with '.attr' appended to the name. 6325 [Richard Levitte] 6326 6327 *) Generate multi-valued AVAs using '+' notation in config files for 6328 req and dirName. 6329 [Steve Henson] 6330 6331 *) Support for nameConstraints certificate extension. 6332 [Steve Henson] 6333 6334 *) Support for policyConstraints certificate extension. 6335 [Steve Henson] 6336 6337 *) Support for policyMappings certificate extension. 6338 [Steve Henson] 6339 6340 *) Make sure the default DSA_METHOD implementation only uses its 6341 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 6342 and change its own handlers to be NULL so as to remove unnecessary 6343 indirection. This lets alternative implementations fallback to the 6344 default implementation more easily. 6345 [Geoff Thorpe] 6346 6347 *) Support for directoryName in GeneralName related extensions 6348 in config files. 6349 [Steve Henson] 6350 6351 *) Make it possible to link applications using Makefile.shared. 6352 Make that possible even when linking against static libraries! 6353 [Richard Levitte] 6354 6355 *) Support for single pass processing for S/MIME signing. This now 6356 means that S/MIME signing can be done from a pipe, in addition 6357 cleartext signing (multipart/signed type) is effectively streaming 6358 and the signed data does not need to be all held in memory. 6359 6360 This is done with a new flag PKCS7_STREAM. When this flag is set 6361 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 6362 is done after the data is output (and digests calculated) in 6363 SMIME_write_PKCS7(). 6364 [Steve Henson] 6365 6366 *) Add full support for -rpath/-R, both in shared libraries and 6367 applications, at least on the platforms where it's known how 6368 to do it. 6369 [Richard Levitte] 6370 6371 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 6372 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 6373 will now compute a table of multiples of the generator that 6374 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 6375 faster (notably in the case of a single point multiplication, 6376 scalar * generator). 6377 [Nils Larsch, Bodo Moeller] 6378 6379 *) IPv6 support for certificate extensions. The various extensions 6380 which use the IP:a.b.c.d can now take IPv6 addresses using the 6381 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 6382 correctly. 6383 [Steve Henson] 6384 6385 *) Added an ENGINE that implements RSA by performing private key 6386 exponentiations with the GMP library. The conversions to and from 6387 GMP's mpz_t format aren't optimised nor are any montgomery forms 6388 cached, and on x86 it appears OpenSSL's own performance has caught up. 6389 However there are likely to be other architectures where GMP could 6390 provide a boost. This ENGINE is not built in by default, but it can be 6391 specified at Configure time and should be accompanied by the necessary 6392 linker additions, eg; 6393 ./config -DOPENSSL_USE_GMP -lgmp 6394 [Geoff Thorpe] 6395 6396 *) "openssl engine" will not display ENGINE/DSO load failure errors when 6397 testing availability of engines with "-t" - the old behaviour is 6398 produced by increasing the feature's verbosity with "-tt". 6399 [Geoff Thorpe] 6400 6401 *) ECDSA routines: under certain error conditions uninitialized BN objects 6402 could be freed. Solution: make sure initialization is performed early 6403 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 6404 via PR#459) 6405 [Lutz Jaenicke] 6406 6407 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 6408 and DH_METHOD (eg. by ENGINE implementations) to override the normal 6409 software implementations. For DSA and DH, parameter generation can 6410 also be overridden by providing the appropriate method callbacks. 6411 [Geoff Thorpe] 6412 6413 *) Change the "progress" mechanism used in key-generation and 6414 primality testing to functions that take a new BN_GENCB pointer in 6415 place of callback/argument pairs. The new API functions have "_ex" 6416 postfixes and the older functions are reimplemented as wrappers for 6417 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 6418 declarations of the old functions to help (graceful) attempts to 6419 migrate to the new functions. Also, the new key-generation API 6420 functions operate on a caller-supplied key-structure and return 6421 success/failure rather than returning a key or NULL - this is to 6422 help make "keygen" another member function of RSA_METHOD etc. 6423 6424 Example for using the new callback interface: 6425 6426 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 6427 void *my_arg = ...; 6428 BN_GENCB my_cb; 6429 6430 BN_GENCB_set(&my_cb, my_callback, my_arg); 6431 6432 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 6433 /* For the meaning of a, b in calls to my_callback(), see the 6434 * documentation of the function that calls the callback. 6435 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 6436 * my_callback should return 1 if it wants BN_is_prime_ex() 6437 * to continue, or 0 to stop. 6438 */ 6439 6440 [Geoff Thorpe] 6441 6442 *) Change the ZLIB compression method to be stateful, and make it 6443 available to TLS with the number defined in 6444 draft-ietf-tls-compression-04.txt. 6445 [Richard Levitte] 6446 6447 *) Add the ASN.1 structures and functions for CertificatePair, which 6448 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 6449 6450 CertificatePair ::= SEQUENCE { 6451 forward [0] Certificate OPTIONAL, 6452 reverse [1] Certificate OPTIONAL, 6453 -- at least one of the pair shall be present -- } 6454 6455 Also implement the PEM functions to read and write certificate 6456 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 6457 6458 This needed to be defined, mostly for the sake of the LDAP 6459 attribute crossCertificatePair, but may prove useful elsewhere as 6460 well. 6461 [Richard Levitte] 6462 6463 *) Make it possible to inhibit symlinking of shared libraries in 6464 Makefile.shared, for Cygwin's sake. 6465 [Richard Levitte] 6466 6467 *) Extend the BIGNUM API by creating a function 6468 void BN_set_negative(BIGNUM *a, int neg); 6469 and a macro that behave like 6470 int BN_is_negative(const BIGNUM *a); 6471 6472 to avoid the need to access 'a->neg' directly in applications. 6473 [Nils Larsch] 6474 6475 *) Implement fast modular reduction for pseudo-Mersenne primes 6476 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 6477 EC_GROUP_new_curve_GFp() will now automatically use this 6478 if applicable. 6479 [Nils Larsch <nla@trustcenter.de>] 6480 6481 *) Add new lock type (CRYPTO_LOCK_BN). 6482 [Bodo Moeller] 6483 6484 *) Change the ENGINE framework to automatically load engines 6485 dynamically from specific directories unless they could be 6486 found to already be built in or loaded. Move all the 6487 current engines except for the cryptodev one to a new 6488 directory engines/. 6489 The engines in engines/ are built as shared libraries if 6490 the "shared" options was given to ./Configure or ./config. 6491 Otherwise, they are inserted in libcrypto.a. 6492 /usr/local/ssl/engines is the default directory for dynamic 6493 engines, but that can be overridden at configure time through 6494 the usual use of --prefix and/or --openssldir, and at run 6495 time with the environment variable OPENSSL_ENGINES. 6496 [Geoff Thorpe and Richard Levitte] 6497 6498 *) Add Makefile.shared, a helper makefile to build shared 6499 libraries. Adapt Makefile.org. 6500 [Richard Levitte] 6501 6502 *) Add version info to Win32 DLLs. 6503 [Peter 'Luna' Runestig" <peter@runestig.com>] 6504 6505 *) Add new 'medium level' PKCS#12 API. Certificates and keys 6506 can be added using this API to created arbitrary PKCS#12 6507 files while avoiding the low level API. 6508 6509 New options to PKCS12_create(), key or cert can be NULL and 6510 will then be omitted from the output file. The encryption 6511 algorithm NIDs can be set to -1 for no encryption, the mac 6512 iteration count can be set to 0 to omit the mac. 6513 6514 Enhance pkcs12 utility by making the -nokeys and -nocerts 6515 options work when creating a PKCS#12 file. New option -nomac 6516 to omit the mac, NONE can be set for an encryption algorithm. 6517 New code is modified to use the enhanced PKCS12_create() 6518 instead of the low level API. 6519 [Steve Henson] 6520 6521 *) Extend ASN1 encoder to support indefinite length constructed 6522 encoding. This can output sequences tags and octet strings in 6523 this form. Modify pk7_asn1.c to support indefinite length 6524 encoding. This is experimental and needs additional code to 6525 be useful, such as an ASN1 bio and some enhanced streaming 6526 PKCS#7 code. 6527 6528 Extend template encode functionality so that tagging is passed 6529 down to the template encoder. 6530 [Steve Henson] 6531 6532 *) Let 'openssl req' fail if an argument to '-newkey' is not 6533 recognized instead of using RSA as a default. 6534 [Bodo Moeller] 6535 6536 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 6537 As these are not official, they are not included in "ALL"; 6538 the "ECCdraft" ciphersuite group alias can be used to select them. 6539 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 6540 6541 *) Add ECDH engine support. 6542 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 6543 6544 *) Add ECDH in new directory crypto/ecdh/. 6545 [Douglas Stebila (Sun Microsystems Laboratories)] 6546 6547 *) Let BN_rand_range() abort with an error after 100 iterations 6548 without success (which indicates a broken PRNG). 6549 [Bodo Moeller] 6550 6551 *) Change BN_mod_sqrt() so that it verifies that the input value 6552 is really the square of the return value. (Previously, 6553 BN_mod_sqrt would show GIGO behaviour.) 6554 [Bodo Moeller] 6555 6556 *) Add named elliptic curves over binary fields from X9.62, SECG, 6557 and WAP/WTLS; add OIDs that were still missing. 6558 6559 [Sheueling Chang Shantz and Douglas Stebila 6560 (Sun Microsystems Laboratories)] 6561 6562 *) Extend the EC library for elliptic curves over binary fields 6563 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 6564 New EC_METHOD: 6565 6566 EC_GF2m_simple_method 6567 6568 New API functions: 6569 6570 EC_GROUP_new_curve_GF2m 6571 EC_GROUP_set_curve_GF2m 6572 EC_GROUP_get_curve_GF2m 6573 EC_POINT_set_affine_coordinates_GF2m 6574 EC_POINT_get_affine_coordinates_GF2m 6575 EC_POINT_set_compressed_coordinates_GF2m 6576 6577 Point compression for binary fields is disabled by default for 6578 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 6579 enable it). 6580 6581 As binary polynomials are represented as BIGNUMs, various members 6582 of the EC_GROUP and EC_POINT data structures can be shared 6583 between the implementations for prime fields and binary fields; 6584 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 6585 are essentially identical to their ..._GFp counterparts. 6586 (For simplicity, the '..._GFp' prefix has been dropped from 6587 various internal method names.) 6588 6589 An internal 'field_div' method (similar to 'field_mul' and 6590 'field_sqr') has been added; this is used only for binary fields. 6591 6592 [Sheueling Chang Shantz and Douglas Stebila 6593 (Sun Microsystems Laboratories)] 6594 6595 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 6596 through methods ('mul', 'precompute_mult'). 6597 6598 The generic implementations (now internally called 'ec_wNAF_mul' 6599 and 'ec_wNAF_precomputed_mult') remain the default if these 6600 methods are undefined. 6601 6602 [Sheueling Chang Shantz and Douglas Stebila 6603 (Sun Microsystems Laboratories)] 6604 6605 *) New function EC_GROUP_get_degree, which is defined through 6606 EC_METHOD. For curves over prime fields, this returns the bit 6607 length of the modulus. 6608 6609 [Sheueling Chang Shantz and Douglas Stebila 6610 (Sun Microsystems Laboratories)] 6611 6612 *) New functions EC_GROUP_dup, EC_POINT_dup. 6613 (These simply call ..._new and ..._copy). 6614 6615 [Sheueling Chang Shantz and Douglas Stebila 6616 (Sun Microsystems Laboratories)] 6617 6618 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 6619 Polynomials are represented as BIGNUMs (where the sign bit is not 6620 used) in the following functions [macros]: 6621 6622 BN_GF2m_add 6623 BN_GF2m_sub [= BN_GF2m_add] 6624 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 6625 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 6626 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 6627 BN_GF2m_mod_inv 6628 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 6629 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 6630 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 6631 BN_GF2m_cmp [= BN_ucmp] 6632 6633 (Note that only the 'mod' functions are actually for fields GF(2^m). 6634 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 6635 6636 For some functions, an the irreducible polynomial defining a 6637 field can be given as an 'unsigned int[]' with strictly 6638 decreasing elements giving the indices of those bits that are set; 6639 i.e., p[] represents the polynomial 6640 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 6641 where 6642 p[0] > p[1] > ... > p[k] = 0. 6643 This applies to the following functions: 6644 6645 BN_GF2m_mod_arr 6646 BN_GF2m_mod_mul_arr 6647 BN_GF2m_mod_sqr_arr 6648 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 6649 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 6650 BN_GF2m_mod_exp_arr 6651 BN_GF2m_mod_sqrt_arr 6652 BN_GF2m_mod_solve_quad_arr 6653 BN_GF2m_poly2arr 6654 BN_GF2m_arr2poly 6655 6656 Conversion can be performed by the following functions: 6657 6658 BN_GF2m_poly2arr 6659 BN_GF2m_arr2poly 6660 6661 bntest.c has additional tests for binary polynomial arithmetic. 6662 6663 Two implementations for BN_GF2m_mod_div() are available. 6664 The default algorithm simply uses BN_GF2m_mod_inv() and 6665 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 6666 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 6667 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 6668 6669 [Sheueling Chang Shantz and Douglas Stebila 6670 (Sun Microsystems Laboratories)] 6671 6672 *) Add new error code 'ERR_R_DISABLED' that can be used when some 6673 functionality is disabled at compile-time. 6674 [Douglas Stebila <douglas.stebila@sun.com>] 6675 6676 *) Change default behaviour of 'openssl asn1parse' so that more 6677 information is visible when viewing, e.g., a certificate: 6678 6679 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 6680 mode the content of non-printable OCTET STRINGs is output in a 6681 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 6682 avoid the appearance of a printable string. 6683 [Nils Larsch <nla@trustcenter.de>] 6684 6685 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 6686 functions 6687 EC_GROUP_set_asn1_flag() 6688 EC_GROUP_get_asn1_flag() 6689 EC_GROUP_set_point_conversion_form() 6690 EC_GROUP_get_point_conversion_form() 6691 These control ASN1 encoding details: 6692 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 6693 has been set to OPENSSL_EC_NAMED_CURVE. 6694 - Points are encoded in uncompressed form by default; options for 6695 asn1_for are as for point2oct, namely 6696 POINT_CONVERSION_COMPRESSED 6697 POINT_CONVERSION_UNCOMPRESSED 6698 POINT_CONVERSION_HYBRID 6699 6700 Also add 'seed' and 'seed_len' members to EC_GROUP with access 6701 functions 6702 EC_GROUP_set_seed() 6703 EC_GROUP_get0_seed() 6704 EC_GROUP_get_seed_len() 6705 This is used only for ASN1 purposes (so far). 6706 [Nils Larsch <nla@trustcenter.de>] 6707 6708 *) Add 'field_type' member to EC_METHOD, which holds the NID 6709 of the appropriate field type OID. The new function 6710 EC_METHOD_get_field_type() returns this value. 6711 [Nils Larsch <nla@trustcenter.de>] 6712 6713 *) Add functions 6714 EC_POINT_point2bn() 6715 EC_POINT_bn2point() 6716 EC_POINT_point2hex() 6717 EC_POINT_hex2point() 6718 providing useful interfaces to EC_POINT_point2oct() and 6719 EC_POINT_oct2point(). 6720 [Nils Larsch <nla@trustcenter.de>] 6721 6722 *) Change internals of the EC library so that the functions 6723 EC_GROUP_set_generator() 6724 EC_GROUP_get_generator() 6725 EC_GROUP_get_order() 6726 EC_GROUP_get_cofactor() 6727 are implemented directly in crypto/ec/ec_lib.c and not dispatched 6728 to methods, which would lead to unnecessary code duplication when 6729 adding different types of curves. 6730 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 6731 6732 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 6733 arithmetic, and such that modified wNAFs are generated 6734 (which avoid length expansion in many cases). 6735 [Bodo Moeller] 6736 6737 *) Add a function EC_GROUP_check_discriminant() (defined via 6738 EC_METHOD) that verifies that the curve discriminant is non-zero. 6739 6740 Add a function EC_GROUP_check() that makes some sanity tests 6741 on a EC_GROUP, its generator and order. This includes 6742 EC_GROUP_check_discriminant(). 6743 [Nils Larsch <nla@trustcenter.de>] 6744 6745 *) Add ECDSA in new directory crypto/ecdsa/. 6746 6747 Add applications 'openssl ecparam' and 'openssl ecdsa' 6748 (these are based on 'openssl dsaparam' and 'openssl dsa'). 6749 6750 ECDSA support is also included in various other files across the 6751 library. Most notably, 6752 - 'openssl req' now has a '-newkey ecdsa:file' option; 6753 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 6754 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 6755 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 6756 them suitable for ECDSA where domain parameters must be 6757 extracted before the specific public key; 6758 - ECDSA engine support has been added. 6759 [Nils Larsch <nla@trustcenter.de>] 6760 6761 *) Include some named elliptic curves, and add OIDs from X9.62, 6762 SECG, and WAP/WTLS. Each curve can be obtained from the new 6763 function 6764 EC_GROUP_new_by_curve_name(), 6765 and the list of available named curves can be obtained with 6766 EC_get_builtin_curves(). 6767 Also add a 'curve_name' member to EC_GROUP objects, which can be 6768 accessed via 6769 EC_GROUP_set_curve_name() 6770 EC_GROUP_get_curve_name() 6771 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 6772 6773 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 6774 was actually never needed) and in BN_mul(). The removal in BN_mul() 6775 required a small change in bn_mul_part_recursive() and the addition 6776 of the functions bn_cmp_part_words(), bn_sub_part_words() and 6777 bn_add_part_words(), which do the same thing as bn_cmp_words(), 6778 bn_sub_words() and bn_add_words() except they take arrays with 6779 differing sizes. 6780 [Richard Levitte] 6781 6782 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 6783 6784 *) Cleanse PEM buffers before freeing them since they may contain 6785 sensitive data. 6786 [Benjamin Bennett <ben@psc.edu>] 6787 6788 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 6789 a ciphersuite string such as "DEFAULT:RSA" cannot enable 6790 authentication-only ciphersuites. 6791 [Bodo Moeller] 6792 6793 *) Since AES128 and AES256 share a single mask bit in the logic of 6794 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 6795 kludge to work properly if AES128 is available and AES256 isn't. 6796 [Victor Duchovni] 6797 6798 *) Expand security boundary to match 1.1.1 module. 6799 [Steve Henson] 6800 6801 *) Remove redundant features: hash file source, editing of test vectors 6802 modify fipsld to use external fips_premain.c signature. 6803 [Steve Henson] 6804 6805 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 6806 run algorithm test programs. 6807 [Steve Henson] 6808 6809 *) Make algorithm test programs more tolerant of whitespace. 6810 [Steve Henson] 6811 6812 *) Have SSL/TLS server implementation tolerate "mismatched" record 6813 protocol version while receiving ClientHello even if the 6814 ClientHello is fragmented. (The server can't insist on the 6815 particular protocol version it has chosen before the ServerHello 6816 message has informed the client about his choice.) 6817 [Bodo Moeller] 6818 6819 *) Load error codes if they are not already present instead of using a 6820 static variable. This allows them to be cleanly unloaded and reloaded. 6821 [Steve Henson] 6822 6823 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 6824 6825 *) Introduce limits to prevent malicious keys being able to 6826 cause a denial of service. (CVE-2006-2940) 6827 [Steve Henson, Bodo Moeller] 6828 6829 *) Fix ASN.1 parsing of certain invalid structures that can result 6830 in a denial of service. (CVE-2006-2937) [Steve Henson] 6831 6832 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 6833 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 6834 6835 *) Fix SSL client code which could crash if connecting to a 6836 malicious SSLv2 server. (CVE-2006-4343) 6837 [Tavis Ormandy and Will Drewry, Google Security Team] 6838 6839 *) Change ciphersuite string processing so that an explicit 6840 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 6841 will no longer include "AES128-SHA"), and any other similar 6842 ciphersuite (same bitmap) from *other* protocol versions (so that 6843 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 6844 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 6845 changes from 0.9.8b and 0.9.8d. 6846 [Bodo Moeller] 6847 6848 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 6849 6850 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 6851 (CVE-2006-4339) [Ben Laurie and Google Security Team] 6852 6853 *) Change the Unix randomness entropy gathering to use poll() when 6854 possible instead of select(), since the latter has some 6855 undesirable limitations. 6856 [Darryl Miles via Richard Levitte and Bodo Moeller] 6857 6858 *) Disable rogue ciphersuites: 6859 6860 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 6861 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 6862 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 6863 6864 The latter two were purportedly from 6865 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 6866 appear there. 6867 6868 Also deactivate the remaining ciphersuites from 6869 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 6870 unofficial, and the ID has long expired. 6871 [Bodo Moeller] 6872 6873 *) Fix RSA blinding Heisenbug (problems sometimes occurred on 6874 dual-core machines) and other potential thread-safety issues. 6875 [Bodo Moeller] 6876 6877 Changes between 0.9.7i and 0.9.7j [04 May 2006] 6878 6879 *) Adapt fipsld and the build system to link against the validated FIPS 6880 module in FIPS mode. 6881 [Steve Henson] 6882 6883 *) Fixes for VC++ 2005 build under Windows. 6884 [Steve Henson] 6885 6886 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 6887 from a Windows bash shell such as MSYS. It is autodetected from the 6888 "config" script when run from a VC++ environment. Modify standard VC++ 6889 build to use fipscanister.o from the GNU make build. 6890 [Steve Henson] 6891 6892 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 6893 6894 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 6895 The value now differs depending on if you build for FIPS or not. 6896 BEWARE! A program linked with a shared FIPSed libcrypto can't be 6897 safely run with a non-FIPSed libcrypto, as it may crash because of 6898 the difference induced by this change. 6899 [Andy Polyakov] 6900 6901 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 6902 6903 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 6904 (part of SSL_OP_ALL). This option used to disable the 6905 countermeasure against man-in-the-middle protocol-version 6906 rollback in the SSL 2.0 server implementation, which is a bad 6907 idea. (CVE-2005-2969) 6908 6909 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 6910 for Information Security, National Institute of Advanced Industrial 6911 Science and Technology [AIST], Japan)] 6912 6913 *) Minimal support for X9.31 signatures and PSS padding modes. This is 6914 mainly for FIPS compliance and not fully integrated at this stage. 6915 [Steve Henson] 6916 6917 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 6918 the exponentiation using a fixed-length exponent. (Otherwise, 6919 the information leaked through timing could expose the secret key 6920 after many signatures; cf. Bleichenbacher's attack on DSA with 6921 biased k.) 6922 [Bodo Moeller] 6923 6924 *) Make a new fixed-window mod_exp implementation the default for 6925 RSA, DSA, and DH private-key operations so that the sequence of 6926 squares and multiplies and the memory access pattern are 6927 independent of the particular secret key. This will mitigate 6928 cache-timing and potential related attacks. 6929 6930 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 6931 and this is automatically used by BN_mod_exp_mont() if the new flag 6932 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 6933 will use this BN flag for private exponents unless the flag 6934 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 6935 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 6936 6937 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 6938 6939 *) Change the client implementation for SSLv23_method() and 6940 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 6941 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 6942 (Previously, the SSL 2.0 backwards compatible Client Hello 6943 message format would be used even with SSL_OP_NO_SSLv2.) 6944 [Bodo Moeller] 6945 6946 *) Add support for smime-type MIME parameter in S/MIME messages which some 6947 clients need. 6948 [Steve Henson] 6949 6950 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 6951 a threadsafe manner. Modify rsa code to use new function and add calls 6952 to dsa and dh code (which had race conditions before). 6953 [Steve Henson] 6954 6955 *) Include the fixed error library code in the C error file definitions 6956 instead of fixing them up at runtime. This keeps the error code 6957 structures constant. 6958 [Steve Henson] 6959 6960 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 6961 6962 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 6963 OpenSSL 0.9.8.] 6964 6965 *) Fixes for newer kerberos headers. NB: the casts are needed because 6966 the 'length' field is signed on one version and unsigned on another 6967 with no (?) obvious way to tell the difference, without these VC++ 6968 complains. Also the "definition" of FAR (blank) is no longer included 6969 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 6970 some needed definitions. 6971 [Steve Henson] 6972 6973 *) Undo Cygwin change. 6974 [Ulf Möller] 6975 6976 *) Added support for proxy certificates according to RFC 3820. 6977 Because they may be a security thread to unaware applications, 6978 they must be explicitly allowed in run-time. See 6979 docs/HOWTO/proxy_certificates.txt for further information. 6980 [Richard Levitte] 6981 6982 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 6983 6984 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 6985 server and client random values. Previously 6986 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 6987 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 6988 6989 This change has negligible security impact because: 6990 6991 1. Server and client random values still have 24 bytes of pseudo random 6992 data. 6993 6994 2. Server and client random values are sent in the clear in the initial 6995 handshake. 6996 6997 3. The master secret is derived using the premaster secret (48 bytes in 6998 size for static RSA ciphersuites) as well as client server and random 6999 values. 7000 7001 The OpenSSL team would like to thank the UK NISCC for bringing this issue 7002 to our attention. 7003 7004 [Stephen Henson, reported by UK NISCC] 7005 7006 *) Use Windows randomness collection on Cygwin. 7007 [Ulf Möller] 7008 7009 *) Fix hang in EGD/PRNGD query when communication socket is closed 7010 prematurely by EGD/PRNGD. 7011 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014] 7012 7013 *) Prompt for pass phrases when appropriate for PKCS12 input format. 7014 [Steve Henson] 7015 7016 *) Back-port of selected performance improvements from development 7017 branch, as well as improved support for PowerPC platforms. 7018 [Andy Polyakov] 7019 7020 *) Add lots of checks for memory allocation failure, error codes to indicate 7021 failure and freeing up memory if a failure occurs. 7022 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 7023 7024 *) Add new -passin argument to dgst. 7025 [Steve Henson] 7026 7027 *) Perform some character comparisons of different types in X509_NAME_cmp: 7028 this is needed for some certificates that re-encode DNs into UTF8Strings 7029 (in violation of RFC3280) and can't or won't issue name rollover 7030 certificates. 7031 [Steve Henson] 7032 7033 *) Make an explicit check during certificate validation to see that 7034 the CA setting in each certificate on the chain is correct. As a 7035 side effect always do the following basic checks on extensions, 7036 not just when there's an associated purpose to the check: 7037 7038 - if there is an unhandled critical extension (unless the user 7039 has chosen to ignore this fault) 7040 - if the path length has been exceeded (if one is set at all) 7041 - that certain extensions fit the associated purpose (if one has 7042 been given) 7043 [Richard Levitte] 7044 7045 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 7046 7047 *) Avoid a race condition when CRLs are checked in a multi threaded 7048 environment. This would happen due to the reordering of the revoked 7049 entries during signature checking and serial number lookup. Now the 7050 encoding is cached and the serial number sort performed under a lock. 7051 Add new STACK function sk_is_sorted(). 7052 [Steve Henson] 7053 7054 *) Add Delta CRL to the extension code. 7055 [Steve Henson] 7056 7057 *) Various fixes to s3_pkt.c so alerts are sent properly. 7058 [David Holmes <d.holmes@f5.com>] 7059 7060 *) Reduce the chances of duplicate issuer name and serial numbers (in 7061 violation of RFC3280) using the OpenSSL certificate creation utilities. 7062 This is done by creating a random 64 bit value for the initial serial 7063 number when a serial number file is created or when a self signed 7064 certificate is created using 'openssl req -x509'. The initial serial 7065 number file is created using 'openssl x509 -next_serial' in CA.pl 7066 rather than being initialized to 1. 7067 [Steve Henson] 7068 7069 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 7070 7071 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 7072 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 7073 [Joe Orton, Steve Henson] 7074 7075 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 7076 (CVE-2004-0112) 7077 [Joe Orton, Steve Henson] 7078 7079 *) Make it possible to have multiple active certificates with the same 7080 subject in the CA index file. This is done only if the keyword 7081 'unique_subject' is set to 'no' in the main CA section (default 7082 if 'CA_default') of the configuration file. The value is saved 7083 with the database itself in a separate index attribute file, 7084 named like the index file with '.attr' appended to the name. 7085 [Richard Levitte] 7086 7087 *) X509 verify fixes. Disable broken certificate workarounds when 7088 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 7089 keyUsage extension present. Don't accept CRLs with unhandled critical 7090 extensions: since verify currently doesn't process CRL extensions this 7091 rejects a CRL with *any* critical extensions. Add new verify error codes 7092 for these cases. 7093 [Steve Henson] 7094 7095 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 7096 A clarification of RFC2560 will require the use of OCTET STRINGs and 7097 some implementations cannot handle the current raw format. Since OpenSSL 7098 copies and compares OCSP nonces as opaque blobs without any attempt at 7099 parsing them this should not create any compatibility issues. 7100 [Steve Henson] 7101 7102 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 7103 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 7104 this HMAC (and other) operations are several times slower than OpenSSL 7105 < 0.9.7. 7106 [Steve Henson] 7107 7108 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 7109 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 7110 7111 *) Use the correct content when signing type "other". 7112 [Steve Henson] 7113 7114 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 7115 7116 *) Fix various bugs revealed by running the NISCC test suite: 7117 7118 Stop out of bounds reads in the ASN1 code when presented with 7119 invalid tags (CVE-2003-0543 and CVE-2003-0544). 7120 7121 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 7122 7123 If verify callback ignores invalid public key errors don't try to check 7124 certificate signature with the NULL public key. 7125 7126 [Steve Henson] 7127 7128 *) New -ignore_err option in ocsp application to stop the server 7129 exiting on the first error in a request. 7130 [Steve Henson] 7131 7132 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 7133 if the server requested one: as stated in TLS 1.0 and SSL 3.0 7134 specifications. 7135 [Steve Henson] 7136 7137 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 7138 extra data after the compression methods not only for TLS 1.0 7139 but also for SSL 3.0 (as required by the specification). 7140 [Bodo Moeller; problem pointed out by Matthias Loepfe] 7141 7142 *) Change X509_certificate_type() to mark the key as exported/exportable 7143 when it's 512 *bits* long, not 512 bytes. 7144 [Richard Levitte] 7145 7146 *) Change AES_cbc_encrypt() so it outputs exact multiple of 7147 blocks during encryption. 7148 [Richard Levitte] 7149 7150 *) Various fixes to base64 BIO and non blocking I/O. On write 7151 flushes were not handled properly if the BIO retried. On read 7152 data was not being buffered properly and had various logic bugs. 7153 This also affects blocking I/O when the data being decoded is a 7154 certain size. 7155 [Steve Henson] 7156 7157 *) Various S/MIME bugfixes and compatibility changes: 7158 output correct application/pkcs7 MIME type if 7159 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 7160 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 7161 of files as .eml work). Correctly handle very long lines in MIME 7162 parser. 7163 [Steve Henson] 7164 7165 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 7166 7167 *) Countermeasure against the Klima-Pokorny-Rosa extension of 7168 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 7169 a protocol version number mismatch like a decryption error 7170 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 7171 [Bodo Moeller] 7172 7173 *) Turn on RSA blinding by default in the default implementation 7174 to avoid a timing attack. Applications that don't want it can call 7175 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 7176 They would be ill-advised to do so in most cases. 7177 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 7178 7179 *) Change RSA blinding code so that it works when the PRNG is not 7180 seeded (in this case, the secret RSA exponent is abused as 7181 an unpredictable seed -- if it is not unpredictable, there 7182 is no point in blinding anyway). Make RSA blinding thread-safe 7183 by remembering the creator's thread ID in rsa->blinding and 7184 having all other threads use local one-time blinding factors 7185 (this requires more computation than sharing rsa->blinding, but 7186 avoids excessive locking; and if an RSA object is not shared 7187 between threads, blinding will still be very fast). 7188 [Bodo Moeller] 7189 7190 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 7191 ENGINE as defaults for all supported algorithms irrespective of 7192 the 'flags' parameter. 'flags' is now honoured, so applications 7193 should make sure they are passing it correctly. 7194 [Geoff Thorpe] 7195 7196 *) Target "mingw" now allows native Windows code to be generated in 7197 the Cygwin environment as well as with the MinGW compiler. 7198 [Ulf Moeller] 7199 7200 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 7201 7202 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 7203 via timing by performing a MAC computation even if incorrect 7204 block cipher padding has been found. This is a countermeasure 7205 against active attacks where the attacker has to distinguish 7206 between bad padding and a MAC verification error. (CVE-2003-0078) 7207 7208 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 7209 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 7210 Martin Vuagnoux (EPFL, Ilion)] 7211 7212 *) Make the no-err option work as intended. The intention with no-err 7213 is not to have the whole error stack handling routines removed from 7214 libcrypto, it's only intended to remove all the function name and 7215 reason texts, thereby removing some of the footprint that may not 7216 be interesting if those errors aren't displayed anyway. 7217 7218 NOTE: it's still possible for any application or module to have its 7219 own set of error texts inserted. The routines are there, just not 7220 used by default when no-err is given. 7221 [Richard Levitte] 7222 7223 *) Add support for FreeBSD on IA64. 7224 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 7225 7226 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 7227 Kerberos function mit_des_cbc_cksum(). Before this change, 7228 the value returned by DES_cbc_cksum() was like the one from 7229 mit_des_cbc_cksum(), except the bytes were swapped. 7230 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 7231 7232 *) Allow an application to disable the automatic SSL chain building. 7233 Before this a rather primitive chain build was always performed in 7234 ssl3_output_cert_chain(): an application had no way to send the 7235 correct chain if the automatic operation produced an incorrect result. 7236 7237 Now the chain builder is disabled if either: 7238 7239 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 7240 7241 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 7242 7243 The reasoning behind this is that an application would not want the 7244 auto chain building to take place if extra chain certificates are 7245 present and it might also want a means of sending no additional 7246 certificates (for example the chain has two certificates and the 7247 root is omitted). 7248 [Steve Henson] 7249 7250 *) Add the possibility to build without the ENGINE framework. 7251 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 7252 7253 *) Under Win32 gmtime() can return NULL: check return value in 7254 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 7255 [Steve Henson] 7256 7257 *) DSA routines: under certain error conditions uninitialized BN objects 7258 could be freed. Solution: make sure initialization is performed early 7259 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 7260 Nils Larsch <nla@trustcenter.de> via PR#459) 7261 [Lutz Jaenicke] 7262 7263 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 7264 checked on reconnect on the client side, therefore session resumption 7265 could still fail with a "ssl session id is different" error. This 7266 behaviour is masked when SSL_OP_ALL is used due to 7267 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 7268 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 7269 followup to PR #377. 7270 [Lutz Jaenicke] 7271 7272 *) IA-32 assembler support enhancements: unified ELF targets, support 7273 for SCO/Caldera platforms, fix for Cygwin shared build. 7274 [Andy Polyakov] 7275 7276 *) Add support for FreeBSD on sparc64. As a consequence, support for 7277 FreeBSD on non-x86 processors is separate from x86 processors on 7278 the config script, much like the NetBSD support. 7279 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 7280 7281 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 7282 7283 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 7284 OpenSSL 0.9.7.] 7285 7286 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 7287 code (06) was taken as the first octet of the session ID and the last 7288 octet was ignored consequently. As a result SSLv2 client side session 7289 caching could not have worked due to the session ID mismatch between 7290 client and server. 7291 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 7292 PR #377. 7293 [Lutz Jaenicke] 7294 7295 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 7296 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 7297 removed entirely. 7298 [Richard Levitte] 7299 7300 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 7301 seems that in spite of existing for more than a year, many application 7302 author have done nothing to provide the necessary callbacks, which 7303 means that this particular engine will not work properly anywhere. 7304 This is a very unfortunate situation which forces us, in the name 7305 of usability, to give the hw_ncipher.c a static lock, which is part 7306 of libcrypto. 7307 NOTE: This is for the 0.9.7 series ONLY. This hack will never 7308 appear in 0.9.8 or later. We EXPECT application authors to have 7309 dealt properly with this when 0.9.8 is released (unless we actually 7310 make such changes in the libcrypto locking code that changes will 7311 have to be made anyway). 7312 [Richard Levitte] 7313 7314 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 7315 octets have been read, EOF or an error occurs. Without this change 7316 some truncated ASN1 structures will not produce an error. 7317 [Steve Henson] 7318 7319 *) Disable Heimdal support, since it hasn't been fully implemented. 7320 Still give the possibility to force the use of Heimdal, but with 7321 warnings and a request that patches get sent to openssl-dev. 7322 [Richard Levitte] 7323 7324 *) Add the VC-CE target, introduce the WINCE sysname, and add 7325 INSTALL.WCE and appropriate conditionals to make it build. 7326 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 7327 7328 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 7329 cygssl-x.y.z.dll, where x, y and z are the major, minor and 7330 edit numbers of the version. 7331 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 7332 7333 *) Introduce safe string copy and catenation functions 7334 (BUF_strlcpy() and BUF_strlcat()). 7335 [Ben Laurie (CHATS) and Richard Levitte] 7336 7337 *) Avoid using fixed-size buffers for one-line DNs. 7338 [Ben Laurie (CHATS)] 7339 7340 *) Add BUF_MEM_grow_clean() to avoid information leakage when 7341 resizing buffers containing secrets, and use where appropriate. 7342 [Ben Laurie (CHATS)] 7343 7344 *) Avoid using fixed size buffers for configuration file location. 7345 [Ben Laurie (CHATS)] 7346 7347 *) Avoid filename truncation for various CA files. 7348 [Ben Laurie (CHATS)] 7349 7350 *) Use sizeof in preference to magic numbers. 7351 [Ben Laurie (CHATS)] 7352 7353 *) Avoid filename truncation in cert requests. 7354 [Ben Laurie (CHATS)] 7355 7356 *) Add assertions to check for (supposedly impossible) buffer 7357 overflows. 7358 [Ben Laurie (CHATS)] 7359 7360 *) Don't cache truncated DNS entries in the local cache (this could 7361 potentially lead to a spoofing attack). 7362 [Ben Laurie (CHATS)] 7363 7364 *) Fix various buffers to be large enough for hex/decimal 7365 representations in a platform independent manner. 7366 [Ben Laurie (CHATS)] 7367 7368 *) Add CRYPTO_realloc_clean() to avoid information leakage when 7369 resizing buffers containing secrets, and use where appropriate. 7370 [Ben Laurie (CHATS)] 7371 7372 *) Add BIO_indent() to avoid much slightly worrying code to do 7373 indents. 7374 [Ben Laurie (CHATS)] 7375 7376 *) Convert sprintf()/BIO_puts() to BIO_printf(). 7377 [Ben Laurie (CHATS)] 7378 7379 *) buffer_gets() could terminate with the buffer only half 7380 full. Fixed. 7381 [Ben Laurie (CHATS)] 7382 7383 *) Add assertions to prevent user-supplied crypto functions from 7384 overflowing internal buffers by having large block sizes, etc. 7385 [Ben Laurie (CHATS)] 7386 7387 *) New OPENSSL_assert() macro (similar to assert(), but enabled 7388 unconditionally). 7389 [Ben Laurie (CHATS)] 7390 7391 *) Eliminate unused copy of key in RC4. 7392 [Ben Laurie (CHATS)] 7393 7394 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 7395 [Ben Laurie (CHATS)] 7396 7397 *) Fix off-by-one error in EGD path. 7398 [Ben Laurie (CHATS)] 7399 7400 *) If RANDFILE path is too long, ignore instead of truncating. 7401 [Ben Laurie (CHATS)] 7402 7403 *) Eliminate unused and incorrectly sized X.509 structure 7404 CBCParameter. 7405 [Ben Laurie (CHATS)] 7406 7407 *) Eliminate unused and dangerous function knumber(). 7408 [Ben Laurie (CHATS)] 7409 7410 *) Eliminate unused and dangerous structure, KSSL_ERR. 7411 [Ben Laurie (CHATS)] 7412 7413 *) Protect against overlong session ID context length in an encoded 7414 session object. Since these are local, this does not appear to be 7415 exploitable. 7416 [Ben Laurie (CHATS)] 7417 7418 *) Change from security patch (see 0.9.6e below) that did not affect 7419 the 0.9.6 release series: 7420 7421 Remote buffer overflow in SSL3 protocol - an attacker could 7422 supply an oversized master key in Kerberos-enabled versions. 7423 (CVE-2002-0657) 7424 [Ben Laurie (CHATS)] 7425 7426 *) Change the SSL kerb5 codes to match RFC 2712. 7427 [Richard Levitte] 7428 7429 *) Make -nameopt work fully for req and add -reqopt switch. 7430 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 7431 7432 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 7433 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 7434 7435 *) Make sure tests can be performed even if the corresponding algorithms 7436 have been removed entirely. This was also the last step to make 7437 OpenSSL compilable with DJGPP under all reasonable conditions. 7438 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 7439 7440 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 7441 to allow version independent disabling of normally unselected ciphers, 7442 which may be activated as a side-effect of selecting a single cipher. 7443 7444 (E.g., cipher list string "RSA" enables ciphersuites that are left 7445 out of "ALL" because they do not provide symmetric encryption. 7446 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 7447 [Lutz Jaenicke, Bodo Moeller] 7448 7449 *) Add appropriate support for separate platform-dependent build 7450 directories. The recommended way to make a platform-dependent 7451 build directory is the following (tested on Linux), maybe with 7452 some local tweaks: 7453 7454 # Place yourself outside of the OpenSSL source tree. In 7455 # this example, the environment variable OPENSSL_SOURCE 7456 # is assumed to contain the absolute OpenSSL source directory. 7457 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 7458 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 7459 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 7460 mkdir -p `dirname $F` 7461 ln -s $OPENSSL_SOURCE/$F $F 7462 done 7463 7464 To be absolutely sure not to disturb the source tree, a "make clean" 7465 is a good thing. If it isn't successful, don't worry about it, 7466 it probably means the source directory is very clean. 7467 [Richard Levitte] 7468 7469 *) Make sure any ENGINE control commands make local copies of string 7470 pointers passed to them whenever necessary. Otherwise it is possible 7471 the caller may have overwritten (or deallocated) the original string 7472 data when a later ENGINE operation tries to use the stored values. 7473 [Götz Babin-Ebell <babinebell@trustcenter.de>] 7474 7475 *) Improve diagnostics in file reading and command-line digests. 7476 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 7477 7478 *) Add AES modes CFB and OFB to the object database. Correct an 7479 error in AES-CFB decryption. 7480 [Richard Levitte] 7481 7482 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 7483 allows existing EVP_CIPHER_CTX structures to be reused after 7484 calling EVP_*Final(). This behaviour is used by encryption 7485 BIOs and some applications. This has the side effect that 7486 applications must explicitly clean up cipher contexts with 7487 EVP_CIPHER_CTX_cleanup() or they will leak memory. 7488 [Steve Henson] 7489 7490 *) Check the values of dna and dnb in bn_mul_recursive before calling 7491 bn_mul_comba (a non zero value means the a or b arrays do not contain 7492 n2 elements) and fallback to bn_mul_normal if either is not zero. 7493 [Steve Henson] 7494 7495 *) Fix escaping of non-ASCII characters when using the -subj option 7496 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 7497 [Lutz Jaenicke] 7498 7499 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 7500 form for "surname", serialNumber has no short form. 7501 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 7502 therefore remove "mail" short name for "internet 7". 7503 The OID for unique identifiers in X509 certificates is 7504 x500UniqueIdentifier, not uniqueIdentifier. 7505 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 7506 [Lutz Jaenicke] 7507 7508 *) Add an "init" command to the ENGINE config module and auto initialize 7509 ENGINEs. Without any "init" command the ENGINE will be initialized 7510 after all ctrl commands have been executed on it. If init=1 the 7511 ENGINE is initialized at that point (ctrls before that point are run 7512 on the uninitialized ENGINE and after on the initialized one). If 7513 init=0 then the ENGINE will not be initialized at all. 7514 [Steve Henson] 7515 7516 *) Fix the 'app_verify_callback' interface so that the user-defined 7517 argument is actually passed to the callback: In the 7518 SSL_CTX_set_cert_verify_callback() prototype, the callback 7519 declaration has been changed from 7520 int (*cb)() 7521 into 7522 int (*cb)(X509_STORE_CTX *,void *); 7523 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 7524 i=s->ctx->app_verify_callback(&ctx) 7525 has been changed into 7526 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 7527 7528 To update applications using SSL_CTX_set_cert_verify_callback(), 7529 a dummy argument can be added to their callback functions. 7530 [D. K. Smetters <smetters@parc.xerox.com>] 7531 7532 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 7533 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 7534 7535 *) Add and OPENSSL_LOAD_CONF define which will cause 7536 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 7537 This allows older applications to transparently support certain 7538 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 7539 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 7540 load the config file and OPENSSL_add_all_algorithms_conf() which will 7541 always load it have also been added. 7542 [Steve Henson] 7543 7544 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 7545 Adjust NIDs and EVP layer. 7546 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 7547 7548 *) Config modules support in openssl utility. 7549 7550 Most commands now load modules from the config file, 7551 though in a few (such as version) this isn't done 7552 because it couldn't be used for anything. 7553 7554 In the case of ca and req the config file used is 7555 the same as the utility itself: that is the -config 7556 command line option can be used to specify an 7557 alternative file. 7558 [Steve Henson] 7559 7560 *) Move default behaviour from OPENSSL_config(). If appname is NULL 7561 use "openssl_conf" if filename is NULL use default openssl config file. 7562 [Steve Henson] 7563 7564 *) Add an argument to OPENSSL_config() to allow the use of an alternative 7565 config section name. Add a new flag to tolerate a missing config file 7566 and move code to CONF_modules_load_file(). 7567 [Steve Henson] 7568 7569 *) Support for crypto accelerator cards from Accelerated Encryption 7570 Processing, www.aep.ie. (Use engine 'aep') 7571 The support was copied from 0.9.6c [engine] and adapted/corrected 7572 to work with the new engine framework. 7573 [AEP Inc. and Richard Levitte] 7574 7575 *) Support for SureWare crypto accelerator cards from Baltimore 7576 Technologies. (Use engine 'sureware') 7577 The support was copied from 0.9.6c [engine] and adapted 7578 to work with the new engine framework. 7579 [Richard Levitte] 7580 7581 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 7582 make the newer ENGINE framework commands for the CHIL engine work. 7583 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 7584 7585 *) Make it possible to produce shared libraries on ReliantUNIX. 7586 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 7587 7588 *) Add the configuration target debug-linux-ppro. 7589 Make 'openssl rsa' use the general key loading routines 7590 implemented in apps.c, and make those routines able to 7591 handle the key format FORMAT_NETSCAPE and the variant 7592 FORMAT_IISSGC. 7593 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7594 7595 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 7596 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7597 7598 *) Add -keyform to rsautl, and document -engine. 7599 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 7600 7601 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 7602 BIO_R_NO_SUCH_FILE error code rather than the generic 7603 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 7604 [Ben Laurie] 7605 7606 *) Add new functions 7607 ERR_peek_last_error 7608 ERR_peek_last_error_line 7609 ERR_peek_last_error_line_data. 7610 These are similar to 7611 ERR_peek_error 7612 ERR_peek_error_line 7613 ERR_peek_error_line_data, 7614 but report on the latest error recorded rather than the first one 7615 still in the error queue. 7616 [Ben Laurie, Bodo Moeller] 7617 7618 *) default_algorithms option in ENGINE config module. This allows things 7619 like: 7620 default_algorithms = ALL 7621 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 7622 [Steve Henson] 7623 7624 *) Preliminary ENGINE config module. 7625 [Steve Henson] 7626 7627 *) New experimental application configuration code. 7628 [Steve Henson] 7629 7630 *) Change the AES code to follow the same name structure as all other 7631 symmetric ciphers, and behave the same way. Move everything to 7632 the directory crypto/aes, thereby obsoleting crypto/rijndael. 7633 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 7634 7635 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 7636 [Ben Laurie and Theo de Raadt] 7637 7638 *) Add option to output public keys in req command. 7639 [Massimiliano Pala madwolf@openca.org] 7640 7641 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 7642 (up to about 10% better than before for P-192 and P-224). 7643 [Bodo Moeller] 7644 7645 *) New functions/macros 7646 7647 SSL_CTX_set_msg_callback(ctx, cb) 7648 SSL_CTX_set_msg_callback_arg(ctx, arg) 7649 SSL_set_msg_callback(ssl, cb) 7650 SSL_set_msg_callback_arg(ssl, arg) 7651 7652 to request calling a callback function 7653 7654 void cb(int write_p, int version, int content_type, 7655 const void *buf, size_t len, SSL *ssl, void *arg) 7656 7657 whenever a protocol message has been completely received 7658 (write_p == 0) or sent (write_p == 1). Here 'version' is the 7659 protocol version according to which the SSL library interprets 7660 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 7661 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 7662 the content type as defined in the SSL 3.0/TLS 1.0 protocol 7663 specification (change_cipher_spec(20), alert(21), handshake(22)). 7664 'buf' and 'len' point to the actual message, 'ssl' to the 7665 SSL object, and 'arg' is the application-defined value set by 7666 SSL[_CTX]_set_msg_callback_arg(). 7667 7668 'openssl s_client' and 'openssl s_server' have new '-msg' options 7669 to enable a callback that displays all protocol messages. 7670 [Bodo Moeller] 7671 7672 *) Change the shared library support so shared libraries are built as 7673 soon as the corresponding static library is finished, and thereby get 7674 openssl and the test programs linked against the shared library. 7675 This still only happens when the keyword "shard" has been given to 7676 the configuration scripts. 7677 7678 NOTE: shared library support is still an experimental thing, and 7679 backward binary compatibility is still not guaranteed. 7680 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 7681 7682 *) Add support for Subject Information Access extension. 7683 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 7684 7685 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 7686 additional bytes when new memory had to be allocated, not just 7687 when reusing an existing buffer. 7688 [Bodo Moeller] 7689 7690 *) New command line and configuration option 'utf8' for the req command. 7691 This allows field values to be specified as UTF8 strings. 7692 [Steve Henson] 7693 7694 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 7695 runs for the former and machine-readable output for the latter. 7696 [Ben Laurie] 7697 7698 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 7699 of the e-mail address in the DN (i.e., it will go into a certificate 7700 extension only). The new configuration file option 'email_in_dn = no' 7701 has the same effect. 7702 [Massimiliano Pala madwolf@openca.org] 7703 7704 *) Change all functions with names starting with des_ to be starting 7705 with DES_ instead. Add wrappers that are compatible with libdes, 7706 but are named _ossl_old_des_*. Finally, add macros that map the 7707 des_* symbols to the corresponding _ossl_old_des_* if libdes 7708 compatibility is desired. If OpenSSL 0.9.6c compatibility is 7709 desired, the des_* symbols will be mapped to DES_*, with one 7710 exception. 7711 7712 Since we provide two compatibility mappings, the user needs to 7713 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 7714 compatibility is desired. The default (i.e., when that macro 7715 isn't defined) is OpenSSL 0.9.6c compatibility. 7716 7717 There are also macros that enable and disable the support of old 7718 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 7719 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 7720 are defined, the default will apply: to support the old des routines. 7721 7722 In either case, one must include openssl/des.h to get the correct 7723 definitions. Do not try to just include openssl/des_old.h, that 7724 won't work. 7725 7726 NOTE: This is a major break of an old API into a new one. Software 7727 authors are encouraged to switch to the DES_ style functions. Some 7728 time in the future, des_old.h and the libdes compatibility functions 7729 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 7730 default), and then completely removed. 7731 [Richard Levitte] 7732 7733 *) Test for certificates which contain unsupported critical extensions. 7734 If such a certificate is found during a verify operation it is 7735 rejected by default: this behaviour can be overridden by either 7736 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 7737 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 7738 X509_supported_extension() has also been added which returns 1 if a 7739 particular extension is supported. 7740 [Steve Henson] 7741 7742 *) Modify the behaviour of EVP cipher functions in similar way to digests 7743 to retain compatibility with existing code. 7744 [Steve Henson] 7745 7746 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 7747 compatibility with existing code. In particular the 'ctx' parameter does 7748 not have to be to be initialized before the call to EVP_DigestInit() and 7749 it is tidied up after a call to EVP_DigestFinal(). New function 7750 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 7751 EVP_MD_CTX_copy() changed to not require the destination to be 7752 initialized valid and new function EVP_MD_CTX_copy_ex() added which 7753 requires the destination to be valid. 7754 7755 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 7756 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 7757 [Steve Henson] 7758 7759 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 7760 so that complete 'Handshake' protocol structures are kept in memory 7761 instead of overwriting 'msg_type' and 'length' with 'body' data. 7762 [Bodo Moeller] 7763 7764 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 7765 [Massimo Santin via Richard Levitte] 7766 7767 *) Major restructuring to the underlying ENGINE code. This includes 7768 reduction of linker bloat, separation of pure "ENGINE" manipulation 7769 (initialisation, etc) from functionality dealing with implementations 7770 of specific crypto interfaces. This change also introduces integrated 7771 support for symmetric ciphers and digest implementations - so ENGINEs 7772 can now accelerate these by providing EVP_CIPHER and EVP_MD 7773 implementations of their own. This is detailed in crypto/engine/README 7774 as it couldn't be adequately described here. However, there are a few 7775 API changes worth noting - some RSA, DSA, DH, and RAND functions that 7776 were changed in the original introduction of ENGINE code have now 7777 reverted back - the hooking from this code to ENGINE is now a good 7778 deal more passive and at run-time, operations deal directly with 7779 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 7780 dereferencing through an ENGINE pointer any more. Also, the ENGINE 7781 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 7782 they were not being used by the framework as there is no concept of a 7783 BIGNUM_METHOD and they could not be generalised to the new 7784 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 7785 ENGINE_cpy() has been removed as it cannot be consistently defined in 7786 the new code. 7787 [Geoff Thorpe] 7788 7789 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 7790 [Steve Henson] 7791 7792 *) Change mkdef.pl to sort symbols that get the same entry number, 7793 and make sure the automatically generated functions ERR_load_* 7794 become part of libeay.num as well. 7795 [Richard Levitte] 7796 7797 *) New function SSL_renegotiate_pending(). This returns true once 7798 renegotiation has been requested (either SSL_renegotiate() call 7799 or HelloRequest/ClientHello received from the peer) and becomes 7800 false once a handshake has been completed. 7801 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 7802 sends a HelloRequest, but does not ensure that a handshake takes 7803 place. SSL_renegotiate_pending() is useful for checking if the 7804 client has followed the request.) 7805 [Bodo Moeller] 7806 7807 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 7808 By default, clients may request session resumption even during 7809 renegotiation (if session ID contexts permit); with this option, 7810 session resumption is possible only in the first handshake. 7811 7812 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 7813 more bits available for options that should not be part of 7814 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 7815 [Bodo Moeller] 7816 7817 *) Add some demos for certificate and certificate request creation. 7818 [Steve Henson] 7819 7820 *) Make maximum certificate chain size accepted from the peer application 7821 settable (SSL*_get/set_max_cert_list()), as proposed by 7822 "Douglas E. Engert" <deengert@anl.gov>. 7823 [Lutz Jaenicke] 7824 7825 *) Add support for shared libraries for Unixware-7 7826 (Boyd Lynn Gerber <gerberb@zenez.com>). 7827 [Lutz Jaenicke] 7828 7829 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 7830 be done prior to destruction. Use this to unload error strings from 7831 ENGINEs that load their own error strings. NB: This adds two new API 7832 functions to "get" and "set" this destroy handler in an ENGINE. 7833 [Geoff Thorpe] 7834 7835 *) Alter all existing ENGINE implementations (except "openssl" and 7836 "openbsd") to dynamically instantiate their own error strings. This 7837 makes them more flexible to be built both as statically-linked ENGINEs 7838 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 7839 Also, add stub code to each that makes building them as self-contained 7840 shared-libraries easier (see README.ENGINE). 7841 [Geoff Thorpe] 7842 7843 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 7844 implementations into applications that are completely implemented in 7845 self-contained shared-libraries. The "dynamic" ENGINE exposes control 7846 commands that can be used to configure what shared-library to load and 7847 to control aspects of the way it is handled. Also, made an update to 7848 the README.ENGINE file that brings its information up-to-date and 7849 provides some information and instructions on the "dynamic" ENGINE 7850 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 7851 [Geoff Thorpe] 7852 7853 *) Make it possible to unload ranges of ERR strings with a new 7854 "ERR_unload_strings" function. 7855 [Geoff Thorpe] 7856 7857 *) Add a copy() function to EVP_MD. 7858 [Ben Laurie] 7859 7860 *) Make EVP_MD routines take a context pointer instead of just the 7861 md_data void pointer. 7862 [Ben Laurie] 7863 7864 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 7865 that the digest can only process a single chunk of data 7866 (typically because it is provided by a piece of 7867 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 7868 is only going to provide a single chunk of data, and hence the 7869 framework needn't accumulate the data for oneshot drivers. 7870 [Ben Laurie] 7871 7872 *) As with "ERR", make it possible to replace the underlying "ex_data" 7873 functions. This change also alters the storage and management of global 7874 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 7875 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 7876 index counters. The API functions that use this state have been changed 7877 to take a "class_index" rather than pointers to the class's local STACK 7878 and counter, and there is now an API function to dynamically create new 7879 classes. This centralisation allows us to (a) plug a lot of the 7880 thread-safety problems that existed, and (b) makes it possible to clean 7881 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 7882 such data would previously have always leaked in application code and 7883 workarounds were in place to make the memory debugging turn a blind eye 7884 to it. Application code that doesn't use this new function will still 7885 leak as before, but their memory debugging output will announce it now 7886 rather than letting it slide. 7887 7888 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 7889 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 7890 has a return value to indicate success or failure. 7891 [Geoff Thorpe] 7892 7893 *) Make it possible to replace the underlying "ERR" functions such that the 7894 global state (2 LHASH tables and 2 locks) is only used by the "default" 7895 implementation. This change also adds two functions to "get" and "set" 7896 the implementation prior to it being automatically set the first time 7897 any other ERR function takes place. Ie. an application can call "get", 7898 pass the return value to a module it has just loaded, and that module 7899 can call its own "set" function using that value. This means the 7900 module's "ERR" operations will use (and modify) the error state in the 7901 application and not in its own statically linked copy of OpenSSL code. 7902 [Geoff Thorpe] 7903 7904 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 7905 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 7906 the operation, and provides a more encapsulated way for external code 7907 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 7908 to use these functions rather than manually incrementing the counts. 7909 7910 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 7911 [Geoff Thorpe] 7912 7913 *) Add EVP test program. 7914 [Ben Laurie] 7915 7916 *) Add symmetric cipher support to ENGINE. Expect the API to change! 7917 [Ben Laurie] 7918 7919 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 7920 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 7921 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 7922 These allow a CRL to be built without having to access X509_CRL fields 7923 directly. Modify 'ca' application to use new functions. 7924 [Steve Henson] 7925 7926 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 7927 bug workarounds. Rollback attack detection is a security feature. 7928 The problem will only arise on OpenSSL servers when TLSv1 is not 7929 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 7930 Software authors not wanting to support TLSv1 will have special reasons 7931 for their choice and can explicitly enable this option. 7932 [Bodo Moeller, Lutz Jaenicke] 7933 7934 *) Rationalise EVP so it can be extended: don't include a union of 7935 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 7936 (similar to those existing for EVP_CIPHER_CTX). 7937 Usage example: 7938 7939 EVP_MD_CTX md; 7940 7941 EVP_MD_CTX_init(&md); /* new function call */ 7942 EVP_DigestInit(&md, EVP_sha1()); 7943 EVP_DigestUpdate(&md, in, len); 7944 EVP_DigestFinal(&md, out, NULL); 7945 EVP_MD_CTX_cleanup(&md); /* new function call */ 7946 7947 [Ben Laurie] 7948 7949 *) Make DES key schedule conform to the usual scheme, as well as 7950 correcting its structure. This means that calls to DES functions 7951 now have to pass a pointer to a des_key_schedule instead of a 7952 plain des_key_schedule (which was actually always a pointer 7953 anyway): E.g., 7954 7955 des_key_schedule ks; 7956 7957 des_set_key_checked(..., &ks); 7958 des_ncbc_encrypt(..., &ks, ...); 7959 7960 (Note that a later change renames 'des_...' into 'DES_...'.) 7961 [Ben Laurie] 7962 7963 *) Initial reduction of linker bloat: the use of some functions, such as 7964 PEM causes large amounts of unused functions to be linked in due to 7965 poor organisation. For example pem_all.c contains every PEM function 7966 which has a knock on effect of linking in large amounts of (unused) 7967 ASN1 code. Grouping together similar functions and splitting unrelated 7968 functions prevents this. 7969 [Steve Henson] 7970 7971 *) Cleanup of EVP macros. 7972 [Ben Laurie] 7973 7974 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 7975 correct _ecb suffix. 7976 [Ben Laurie] 7977 7978 *) Add initial OCSP responder support to ocsp application. The 7979 revocation information is handled using the text based index 7980 use by the ca application. The responder can either handle 7981 requests generated internally, supplied in files (for example 7982 via a CGI script) or using an internal minimal server. 7983 [Steve Henson] 7984 7985 *) Add configuration choices to get zlib compression for TLS. 7986 [Richard Levitte] 7987 7988 *) Changes to Kerberos SSL for RFC 2712 compliance: 7989 1. Implemented real KerberosWrapper, instead of just using 7990 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 7991 2. Implemented optional authenticator field of KerberosWrapper. 7992 7993 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 7994 and authenticator structs; see crypto/krb5/. 7995 7996 Generalized Kerberos calls to support multiple Kerberos libraries. 7997 [Vern Staats <staatsvr@asc.hpc.mil>, 7998 Jeffrey Altman <jaltman@columbia.edu> 7999 via Richard Levitte] 8000 8001 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 8002 already does with RSA. testdsa.h now has 'priv_key/pub_key' 8003 values for each of the key sizes rather than having just 8004 parameters (and 'speed' generating keys each time). 8005 [Geoff Thorpe] 8006 8007 *) Speed up EVP routines. 8008 Before: 8009encrypt 8010type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 8011des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 8012des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 8013des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 8014decrypt 8015des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 8016des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 8017des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 8018 After: 8019encrypt 8020des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 8021decrypt 8022des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 8023 [Ben Laurie] 8024 8025 *) Added the OS2-EMX target. 8026 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 8027 8028 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 8029 to support NCONF routines in extension code. New function CONF_set_nconf() 8030 to allow functions which take an NCONF to also handle the old LHASH 8031 structure: this means that the old CONF compatible routines can be 8032 retained (in particular wrt extensions) without having to duplicate the 8033 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 8034 [Steve Henson] 8035 8036 *) Enhance the general user interface with mechanisms for inner control 8037 and with possibilities to have yes/no kind of prompts. 8038 [Richard Levitte] 8039 8040 *) Change all calls to low level digest routines in the library and 8041 applications to use EVP. Add missing calls to HMAC_cleanup() and 8042 don't assume HMAC_CTX can be copied using memcpy(). 8043 [Verdon Walker <VWalker@novell.com>, Steve Henson] 8044 8045 *) Add the possibility to control engines through control names but with 8046 arbitrary arguments instead of just a string. 8047 Change the key loaders to take a UI_METHOD instead of a callback 8048 function pointer. NOTE: this breaks binary compatibility with earlier 8049 versions of OpenSSL [engine]. 8050 Adapt the nCipher code for these new conditions and add a card insertion 8051 callback. 8052 [Richard Levitte] 8053 8054 *) Enhance the general user interface with mechanisms to better support 8055 dialog box interfaces, application-defined prompts, the possibility 8056 to use defaults (for example default passwords from somewhere else) 8057 and interrupts/cancellations. 8058 [Richard Levitte] 8059 8060 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 8061 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 8062 [Steve Henson] 8063 8064 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 8065 tidy up some unnecessarily weird code in 'sk_new()'). 8066 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 8067 8068 *) Change the key loading routines for ENGINEs to use the same kind 8069 callback (pem_password_cb) as all other routines that need this 8070 kind of callback. 8071 [Richard Levitte] 8072 8073 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 8074 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 8075 than this minimum value is recommended. 8076 [Lutz Jaenicke] 8077 8078 *) New random seeder for OpenVMS, using the system process statistics 8079 that are easily reachable. 8080 [Richard Levitte] 8081 8082 *) Windows apparently can't transparently handle global 8083 variables defined in DLLs. Initialisations such as: 8084 8085 const ASN1_ITEM *it = &ASN1_INTEGER_it; 8086 8087 won't compile. This is used by the any applications that need to 8088 declare their own ASN1 modules. This was fixed by adding the option 8089 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 8090 needed for static libraries under Win32. 8091 [Steve Henson] 8092 8093 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 8094 setting of purpose and trust fields. New X509_STORE trust and 8095 purpose functions and tidy up setting in other SSL functions. 8096 [Steve Henson] 8097 8098 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 8099 structure. These are inherited by X509_STORE_CTX when it is 8100 initialised. This allows various defaults to be set in the 8101 X509_STORE structure (such as flags for CRL checking and custom 8102 purpose or trust settings) for functions which only use X509_STORE_CTX 8103 internally such as S/MIME. 8104 8105 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 8106 trust settings if they are not set in X509_STORE. This allows X509_STORE 8107 purposes and trust (in S/MIME for example) to override any set by default. 8108 8109 Add command line options for CRL checking to smime, s_client and s_server 8110 applications. 8111 [Steve Henson] 8112 8113 *) Initial CRL based revocation checking. If the CRL checking flag(s) 8114 are set then the CRL is looked up in the X509_STORE structure and 8115 its validity and signature checked, then if the certificate is found 8116 in the CRL the verify fails with a revoked error. 8117 8118 Various new CRL related callbacks added to X509_STORE_CTX structure. 8119 8120 Command line options added to 'verify' application to support this. 8121 8122 This needs some additional work, such as being able to handle multiple 8123 CRLs with different times, extension based lookup (rather than just 8124 by subject name) and ultimately more complete V2 CRL extension 8125 handling. 8126 [Steve Henson] 8127 8128 *) Add a general user interface API (crypto/ui/). This is designed 8129 to replace things like des_read_password and friends (backward 8130 compatibility functions using this new API are provided). 8131 The purpose is to remove prompting functions from the DES code 8132 section as well as provide for prompting through dialog boxes in 8133 a window system and the like. 8134 [Richard Levitte] 8135 8136 *) Add "ex_data" support to ENGINE so implementations can add state at a 8137 per-structure level rather than having to store it globally. 8138 [Geoff] 8139 8140 *) Make it possible for ENGINE structures to be copied when retrieved by 8141 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 8142 This causes the "original" ENGINE structure to act like a template, 8143 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 8144 operational state can be localised to each ENGINE structure, despite the 8145 fact they all share the same "methods". New ENGINE structures returned in 8146 this case have no functional references and the return value is the single 8147 structural reference. This matches the single structural reference returned 8148 by ENGINE_by_id() normally, when it is incremented on the pre-existing 8149 ENGINE structure. 8150 [Geoff] 8151 8152 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 8153 needs to match any other type at all we need to manually clear the 8154 tag cache. 8155 [Steve Henson] 8156 8157 *) Changes to the "openssl engine" utility to include; 8158 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 8159 about an ENGINE's available control commands. 8160 - executing control commands from command line arguments using the 8161 '-pre' and '-post' switches. '-post' is only used if '-t' is 8162 specified and the ENGINE is successfully initialised. The syntax for 8163 the individual commands are colon-separated, for example; 8164 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 8165 [Geoff] 8166 8167 *) New dynamic control command support for ENGINEs. ENGINEs can now 8168 declare their own commands (numbers), names (strings), descriptions, 8169 and input types for run-time discovery by calling applications. A 8170 subset of these commands are implicitly classed as "executable" 8171 depending on their input type, and only these can be invoked through 8172 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 8173 can be based on user input, config files, etc). The distinction is 8174 that "executable" commands cannot return anything other than a boolean 8175 result and can only support numeric or string input, whereas some 8176 discoverable commands may only be for direct use through 8177 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 8178 pointers, or other custom uses. The "executable" commands are to 8179 support parameterisations of ENGINE behaviour that can be 8180 unambiguously defined by ENGINEs and used consistently across any 8181 OpenSSL-based application. Commands have been added to all the 8182 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 8183 control over shared-library paths without source code alterations. 8184 [Geoff] 8185 8186 *) Changed all ENGINE implementations to dynamically allocate their 8187 ENGINEs rather than declaring them statically. Apart from this being 8188 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 8189 this also allows the implementations to compile without using the 8190 internal engine_int.h header. 8191 [Geoff] 8192 8193 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 8194 'const' value. Any code that should be able to modify a RAND_METHOD 8195 should already have non-const pointers to it (ie. they should only 8196 modify their own ones). 8197 [Geoff] 8198 8199 *) Made a variety of little tweaks to the ENGINE code. 8200 - "atalla" and "ubsec" string definitions were moved from header files 8201 to C code. "nuron" string definitions were placed in variables 8202 rather than hard-coded - allowing parameterisation of these values 8203 later on via ctrl() commands. 8204 - Removed unused "#if 0"'d code. 8205 - Fixed engine list iteration code so it uses ENGINE_free() to release 8206 structural references. 8207 - Constified the RAND_METHOD element of ENGINE structures. 8208 - Constified various get/set functions as appropriate and added 8209 missing functions (including a catch-all ENGINE_cpy that duplicates 8210 all ENGINE values onto a new ENGINE except reference counts/state). 8211 - Removed NULL parameter checks in get/set functions. Setting a method 8212 or function to NULL is a way of cancelling out a previously set 8213 value. Passing a NULL ENGINE parameter is just plain stupid anyway 8214 and doesn't justify the extra error symbols and code. 8215 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 8216 flags from engine_int.h to engine.h. 8217 - Changed prototypes for ENGINE handler functions (init(), finish(), 8218 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 8219 [Geoff] 8220 8221 *) Implement binary inversion algorithm for BN_mod_inverse in addition 8222 to the algorithm using long division. The binary algorithm can be 8223 used only if the modulus is odd. On 32-bit systems, it is faster 8224 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 8225 roughly 5-15% for 256-bit moduli), so we use it only for moduli 8226 up to 450 bits. In 64-bit environments, the binary algorithm 8227 appears to be advantageous for much longer moduli; here we use it 8228 for moduli up to 2048 bits. 8229 [Bodo Moeller] 8230 8231 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 8232 could not support the combine flag in choice fields. 8233 [Steve Henson] 8234 8235 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 8236 extensions from a certificate request to the certificate. 8237 [Steve Henson] 8238 8239 *) Allow multiple 'certopt' and 'nameopt' options to be separated 8240 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 8241 file: this allows the display of the certificate about to be 8242 signed to be customised, to allow certain fields to be included 8243 or excluded and extension details. The old system didn't display 8244 multicharacter strings properly, omitted fields not in the policy 8245 and couldn't display additional details such as extensions. 8246 [Steve Henson] 8247 8248 *) Function EC_POINTs_mul for multiple scalar multiplication 8249 of an arbitrary number of elliptic curve points 8250 \sum scalars[i]*points[i], 8251 optionally including the generator defined for the EC_GROUP: 8252 scalar*generator + \sum scalars[i]*points[i]. 8253 8254 EC_POINT_mul is a simple wrapper function for the typical case 8255 that the point list has just one item (besides the optional 8256 generator). 8257 [Bodo Moeller] 8258 8259 *) First EC_METHODs for curves over GF(p): 8260 8261 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 8262 operations and provides various method functions that can also 8263 operate with faster implementations of modular arithmetic. 8264 8265 EC_GFp_mont_method() reuses most functions that are part of 8266 EC_GFp_simple_method, but uses Montgomery arithmetic. 8267 8268 [Bodo Moeller; point addition and point doubling 8269 implementation directly derived from source code provided by 8270 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 8271 8272 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 8273 crypto/ec/ec_lib.c): 8274 8275 Curves are EC_GROUP objects (with an optional group generator) 8276 based on EC_METHODs that are built into the library. 8277 8278 Points are EC_POINT objects based on EC_GROUP objects. 8279 8280 Most of the framework would be able to handle curves over arbitrary 8281 finite fields, but as there are no obvious types for fields other 8282 than GF(p), some functions are limited to that for now. 8283 [Bodo Moeller] 8284 8285 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 8286 that the file contains a complete HTTP response. 8287 [Richard Levitte] 8288 8289 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 8290 change the def and num file printf format specifier from "%-40sXXX" 8291 to "%-39s XXX". The latter will always guarantee a space after the 8292 field while the former will cause them to run together if the field 8293 is 40 of more characters long. 8294 [Steve Henson] 8295 8296 *) Constify the cipher and digest 'method' functions and structures 8297 and modify related functions to take constant EVP_MD and EVP_CIPHER 8298 pointers. 8299 [Steve Henson] 8300 8301 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 8302 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 8303 [Bodo Moeller] 8304 8305 *) Modify EVP_Digest*() routines so they now return values. Although the 8306 internal software routines can never fail additional hardware versions 8307 might. 8308 [Steve Henson] 8309 8310 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 8311 8312 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 8313 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 8314 8315 ASN1 error codes 8316 ERR_R_NESTED_ASN1_ERROR 8317 ... 8318 ERR_R_MISSING_ASN1_EOS 8319 were 4 .. 9, conflicting with 8320 ERR_LIB_RSA (= ERR_R_RSA_LIB) 8321 ... 8322 ERR_LIB_PEM (= ERR_R_PEM_LIB). 8323 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 8324 8325 Add new error code 'ERR_R_INTERNAL_ERROR'. 8326 [Bodo Moeller] 8327 8328 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 8329 suffices. 8330 [Bodo Moeller] 8331 8332 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 8333 sets the subject name for a new request or supersedes the 8334 subject name in a given request. Formats that can be parsed are 8335 'CN=Some Name, OU=myOU, C=IT' 8336 and 8337 'CN=Some Name/OU=myOU/C=IT'. 8338 8339 Add options '-batch' and '-verbose' to 'openssl req'. 8340 [Massimiliano Pala <madwolf@hackmasters.net>] 8341 8342 *) Introduce the possibility to access global variables through 8343 functions on platform were that's the best way to handle exporting 8344 global variables in shared libraries. To enable this functionality, 8345 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 8346 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 8347 is normally done by Configure or something similar). 8348 8349 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 8350 in the source file (foo.c) like this: 8351 8352 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 8353 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 8354 8355 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 8356 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 8357 8358 OPENSSL_DECLARE_GLOBAL(int,foo); 8359 #define foo OPENSSL_GLOBAL_REF(foo) 8360 OPENSSL_DECLARE_GLOBAL(double,bar); 8361 #define bar OPENSSL_GLOBAL_REF(bar) 8362 8363 The #defines are very important, and therefore so is including the 8364 header file everywhere where the defined globals are used. 8365 8366 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 8367 of ASN.1 items, but that structure is a bit different. 8368 8369 The largest change is in util/mkdef.pl which has been enhanced with 8370 better and easier to understand logic to choose which symbols should 8371 go into the Windows .def files as well as a number of fixes and code 8372 cleanup (among others, algorithm keywords are now sorted 8373 lexicographically to avoid constant rewrites). 8374 [Richard Levitte] 8375 8376 *) In BN_div() keep a copy of the sign of 'num' before writing the 8377 result to 'rm' because if rm==num the value will be overwritten 8378 and produce the wrong result if 'num' is negative: this caused 8379 problems with BN_mod() and BN_nnmod(). 8380 [Steve Henson] 8381 8382 *) Function OCSP_request_verify(). This checks the signature on an 8383 OCSP request and verifies the signer certificate. The signer 8384 certificate is just checked for a generic purpose and OCSP request 8385 trust settings. 8386 [Steve Henson] 8387 8388 *) Add OCSP_check_validity() function to check the validity of OCSP 8389 responses. OCSP responses are prepared in real time and may only 8390 be a few seconds old. Simply checking that the current time lies 8391 between thisUpdate and nextUpdate max reject otherwise valid responses 8392 caused by either OCSP responder or client clock inaccuracy. Instead 8393 we allow thisUpdate and nextUpdate to fall within a certain period of 8394 the current time. The age of the response can also optionally be 8395 checked. Two new options -validity_period and -status_age added to 8396 ocsp utility. 8397 [Steve Henson] 8398 8399 *) If signature or public key algorithm is unrecognized print out its 8400 OID rather that just UNKNOWN. 8401 [Steve Henson] 8402 8403 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 8404 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 8405 ID to be generated from the issuer certificate alone which can then be 8406 passed to OCSP_id_issuer_cmp(). 8407 [Steve Henson] 8408 8409 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 8410 ASN1 modules to export functions returning ASN1_ITEM pointers 8411 instead of the ASN1_ITEM structures themselves. This adds several 8412 new macros which allow the underlying ASN1 function/structure to 8413 be accessed transparently. As a result code should not use ASN1_ITEM 8414 references directly (such as &X509_it) but instead use the relevant 8415 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 8416 use of the new ASN1 code on platforms where exporting structures 8417 is problematical (for example in shared libraries) but exporting 8418 functions returning pointers to structures is not. 8419 [Steve Henson] 8420 8421 *) Add support for overriding the generation of SSL/TLS session IDs. 8422 These callbacks can be registered either in an SSL_CTX or per SSL. 8423 The purpose of this is to allow applications to control, if they wish, 8424 the arbitrary values chosen for use as session IDs, particularly as it 8425 can be useful for session caching in multiple-server environments. A 8426 command-line switch for testing this (and any client code that wishes 8427 to use such a feature) has been added to "s_server". 8428 [Geoff Thorpe, Lutz Jaenicke] 8429 8430 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 8431 of the form '#if defined(...) || defined(...) || ...' and 8432 '#if !defined(...) && !defined(...) && ...'. This also avoids 8433 the growing number of special cases it was previously handling. 8434 [Richard Levitte] 8435 8436 *) Make all configuration macros available for application by making 8437 sure they are available in opensslconf.h, by giving them names starting 8438 with "OPENSSL_" to avoid conflicts with other packages and by making 8439 sure e_os2.h will cover all platform-specific cases together with 8440 opensslconf.h. 8441 Additionally, it is now possible to define configuration/platform- 8442 specific names (called "system identities"). In the C code, these 8443 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 8444 macro with the name beginning with "OPENSSL_SYS_", which is determined 8445 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 8446 what is available. 8447 [Richard Levitte] 8448 8449 *) New option -set_serial to 'req' and 'x509' this allows the serial 8450 number to use to be specified on the command line. Previously self 8451 signed certificates were hard coded with serial number 0 and the 8452 CA options of 'x509' had to use a serial number in a file which was 8453 auto incremented. 8454 [Steve Henson] 8455 8456 *) New options to 'ca' utility to support V2 CRL entry extensions. 8457 Currently CRL reason, invalidity date and hold instruction are 8458 supported. Add new CRL extensions to V3 code and some new objects. 8459 [Steve Henson] 8460 8461 *) New function EVP_CIPHER_CTX_set_padding() this is used to 8462 disable standard block padding (aka PKCS#5 padding) in the EVP 8463 API, which was previously mandatory. This means that the data is 8464 not padded in any way and so the total length much be a multiple 8465 of the block size, otherwise an error occurs. 8466 [Steve Henson] 8467 8468 *) Initial (incomplete) OCSP SSL support. 8469 [Steve Henson] 8470 8471 *) New function OCSP_parse_url(). This splits up a URL into its host, 8472 port and path components: primarily to parse OCSP URLs. New -url 8473 option to ocsp utility. 8474 [Steve Henson] 8475 8476 *) New nonce behavior. The return value of OCSP_check_nonce() now 8477 reflects the various checks performed. Applications can decide 8478 whether to tolerate certain situations such as an absent nonce 8479 in a response when one was present in a request: the ocsp application 8480 just prints out a warning. New function OCSP_add1_basic_nonce() 8481 this is to allow responders to include a nonce in a response even if 8482 the request is nonce-less. 8483 [Steve Henson] 8484 8485 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 8486 skipped when using openssl x509 multiple times on a single input file, 8487 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 8488 [Bodo Moeller] 8489 8490 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 8491 set string type: to handle setting ASN1_TIME structures. Fix ca 8492 utility to correctly initialize revocation date of CRLs. 8493 [Steve Henson] 8494 8495 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 8496 the clients preferred ciphersuites and rather use its own preferences. 8497 Should help to work around M$ SGC (Server Gated Cryptography) bug in 8498 Internet Explorer by ensuring unchanged hash method during stepup. 8499 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 8500 [Lutz Jaenicke] 8501 8502 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 8503 to aes and add a new 'exist' option to print out symbols that don't 8504 appear to exist. 8505 [Steve Henson] 8506 8507 *) Additional options to ocsp utility to allow flags to be set and 8508 additional certificates supplied. 8509 [Steve Henson] 8510 8511 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 8512 OCSP client a number of certificate to only verify the response 8513 signature against. 8514 [Richard Levitte] 8515 8516 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 8517 handle the new API. Currently only ECB, CBC modes supported. Add new 8518 AES OIDs. 8519 8520 Add TLS AES ciphersuites as described in RFC3268, "Advanced 8521 Encryption Standard (AES) Ciphersuites for Transport Layer 8522 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 8523 not enabled by default and were not part of the "ALL" ciphersuite 8524 alias because they were not yet official; they could be 8525 explicitly requested by specifying the "AESdraft" ciphersuite 8526 group alias. In the final release of OpenSSL 0.9.7, the group 8527 alias is called "AES" and is part of "ALL".) 8528 [Ben Laurie, Steve Henson, Bodo Moeller] 8529 8530 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 8531 request to response. 8532 [Steve Henson] 8533 8534 *) Functions for OCSP responders. OCSP_request_onereq_count(), 8535 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 8536 extract information from a certificate request. OCSP_response_create() 8537 creates a response and optionally adds a basic response structure. 8538 OCSP_basic_add1_status() adds a complete single response to a basic 8539 response and returns the OCSP_SINGLERESP structure just added (to allow 8540 extensions to be included for example). OCSP_basic_add1_cert() adds a 8541 certificate to a basic response and OCSP_basic_sign() signs a basic 8542 response with various flags. New helper functions ASN1_TIME_check() 8543 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 8544 (converts ASN1_TIME to GeneralizedTime). 8545 [Steve Henson] 8546 8547 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 8548 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 8549 structure from a certificate. X509_pubkey_digest() digests the public_key 8550 contents: this is used in various key identifiers. 8551 [Steve Henson] 8552 8553 *) Make sk_sort() tolerate a NULL argument. 8554 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 8555 8556 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 8557 passed by the function are trusted implicitly. If any of them signed the 8558 response then it is assumed to be valid and is not verified. 8559 [Steve Henson] 8560 8561 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 8562 to data. This was previously part of the PKCS7 ASN1 code. This 8563 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 8564 [Steve Henson, reported by Kenneth R. Robinette 8565 <support@securenetterm.com>] 8566 8567 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 8568 routines: without these tracing memory leaks is very painful. 8569 Fix leaks in PKCS12 and PKCS7 routines. 8570 [Steve Henson] 8571 8572 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 8573 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 8574 effectively meant GeneralizedTime would never be used. Now it 8575 is initialised to -1 but X509_time_adj() now has to check the value 8576 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 8577 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 8578 [Steve Henson, reported by Kenneth R. Robinette 8579 <support@securenetterm.com>] 8580 8581 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 8582 result in a zero length in the ASN1_INTEGER structure which was 8583 not consistent with the structure when d2i_ASN1_INTEGER() was used 8584 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 8585 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 8586 where it did not print out a minus for negative ASN1_INTEGER. 8587 [Steve Henson] 8588 8589 *) Add summary printout to ocsp utility. The various functions which 8590 convert status values to strings have been renamed to: 8591 OCSP_response_status_str(), OCSP_cert_status_str() and 8592 OCSP_crl_reason_str() and are no longer static. New options 8593 to verify nonce values and to disable verification. OCSP response 8594 printout format cleaned up. 8595 [Steve Henson] 8596 8597 *) Add additional OCSP certificate checks. These are those specified 8598 in RFC2560. This consists of two separate checks: the CA of the 8599 certificate being checked must either be the OCSP signer certificate 8600 or the issuer of the OCSP signer certificate. In the latter case the 8601 OCSP signer certificate must contain the OCSP signing extended key 8602 usage. This check is performed by attempting to match the OCSP 8603 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 8604 in the OCSP_CERTID structures of the response. 8605 [Steve Henson] 8606 8607 *) Initial OCSP certificate verification added to OCSP_basic_verify() 8608 and related routines. This uses the standard OpenSSL certificate 8609 verify routines to perform initial checks (just CA validity) and 8610 to obtain the certificate chain. Then additional checks will be 8611 performed on the chain. Currently the root CA is checked to see 8612 if it is explicitly trusted for OCSP signing. This is used to set 8613 a root CA as a global signing root: that is any certificate that 8614 chains to that CA is an acceptable OCSP signing certificate. 8615 [Steve Henson] 8616 8617 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 8618 extensions from a separate configuration file. 8619 As when reading extensions from the main configuration file, 8620 the '-extensions ...' option may be used for specifying the 8621 section to use. 8622 [Massimiliano Pala <madwolf@comune.modena.it>] 8623 8624 *) New OCSP utility. Allows OCSP requests to be generated or 8625 read. The request can be sent to a responder and the output 8626 parsed, outputted or printed in text form. Not complete yet: 8627 still needs to check the OCSP response validity. 8628 [Steve Henson] 8629 8630 *) New subcommands for 'openssl ca': 8631 'openssl ca -status <serial>' prints the status of the cert with 8632 the given serial number (according to the index file). 8633 'openssl ca -updatedb' updates the expiry status of certificates 8634 in the index file. 8635 [Massimiliano Pala <madwolf@comune.modena.it>] 8636 8637 *) New '-newreq-nodes' command option to CA.pl. This is like 8638 '-newreq', but calls 'openssl req' with the '-nodes' option 8639 so that the resulting key is not encrypted. 8640 [Damien Miller <djm@mindrot.org>] 8641 8642 *) New configuration for the GNU Hurd. 8643 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 8644 8645 *) Initial code to implement OCSP basic response verify. This 8646 is currently incomplete. Currently just finds the signer's 8647 certificate and verifies the signature on the response. 8648 [Steve Henson] 8649 8650 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 8651 value of OPENSSLDIR. This is available via the new '-d' option 8652 to 'openssl version', and is also included in 'openssl version -a'. 8653 [Bodo Moeller] 8654 8655 *) Allowing defining memory allocation callbacks that will be given 8656 file name and line number information in additional arguments 8657 (a const char* and an int). The basic functionality remains, as 8658 well as the original possibility to just replace malloc(), 8659 realloc() and free() by functions that do not know about these 8660 additional arguments. To register and find out the current 8661 settings for extended allocation functions, the following 8662 functions are provided: 8663 8664 CRYPTO_set_mem_ex_functions 8665 CRYPTO_set_locked_mem_ex_functions 8666 CRYPTO_get_mem_ex_functions 8667 CRYPTO_get_locked_mem_ex_functions 8668 8669 These work the same way as CRYPTO_set_mem_functions and friends. 8670 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 8671 extended allocation function is enabled. 8672 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 8673 a conventional allocation function is enabled. 8674 [Richard Levitte, Bodo Moeller] 8675 8676 *) Finish off removing the remaining LHASH function pointer casts. 8677 There should no longer be any prototype-casting required when using 8678 the LHASH abstraction, and any casts that remain are "bugs". See 8679 the callback types and macros at the head of lhash.h for details 8680 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 8681 [Geoff Thorpe] 8682 8683 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 8684 If /dev/[u]random devices are not available or do not return enough 8685 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 8686 be queried. 8687 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 8688 /etc/entropy will be queried once each in this sequence, querying stops 8689 when enough entropy was collected without querying more sockets. 8690 [Lutz Jaenicke] 8691 8692 *) Change the Unix RAND_poll() variant to be able to poll several 8693 random devices, as specified by DEVRANDOM, until a sufficient amount 8694 of data has been collected. We spend at most 10 ms on each file 8695 (select timeout) and read in non-blocking mode. DEVRANDOM now 8696 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 8697 (previously it was just the string "/dev/urandom"), so on typical 8698 platforms the 10 ms delay will never occur. 8699 Also separate out the Unix variant to its own file, rand_unix.c. 8700 For VMS, there's a currently-empty rand_vms.c. 8701 [Richard Levitte] 8702 8703 *) Move OCSP client related routines to ocsp_cl.c. These 8704 provide utility functions which an application needing 8705 to issue a request to an OCSP responder and analyse the 8706 response will typically need: as opposed to those which an 8707 OCSP responder itself would need which will be added later. 8708 8709 OCSP_request_sign() signs an OCSP request with an API similar 8710 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 8711 response. OCSP_response_get1_basic() extracts basic response 8712 from response. OCSP_resp_find_status(): finds and extracts status 8713 information from an OCSP_CERTID structure (which will be created 8714 when the request structure is built). These are built from lower 8715 level functions which work on OCSP_SINGLERESP structures but 8716 won't normally be used unless the application wishes to examine 8717 extensions in the OCSP response for example. 8718 8719 Replace nonce routines with a pair of functions. 8720 OCSP_request_add1_nonce() adds a nonce value and optionally 8721 generates a random value. OCSP_check_nonce() checks the 8722 validity of the nonce in an OCSP response. 8723 [Steve Henson] 8724 8725 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 8726 This doesn't copy the supplied OCSP_CERTID and avoids the 8727 need to free up the newly created id. Change return type 8728 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 8729 This can then be used to add extensions to the request. 8730 Deleted OCSP_request_new(), since most of its functionality 8731 is now in OCSP_REQUEST_new() (and the case insensitive name 8732 clash) apart from the ability to set the request name which 8733 will be added elsewhere. 8734 [Steve Henson] 8735 8736 *) Update OCSP API. Remove obsolete extensions argument from 8737 various functions. Extensions are now handled using the new 8738 OCSP extension code. New simple OCSP HTTP function which 8739 can be used to send requests and parse the response. 8740 [Steve Henson] 8741 8742 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 8743 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 8744 uses the special reorder version of SET OF to sort the attributes 8745 and reorder them to match the encoded order. This resolves a long 8746 standing problem: a verify on a PKCS7 structure just after signing 8747 it used to fail because the attribute order did not match the 8748 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 8749 it uses the received order. This is necessary to tolerate some broken 8750 software that does not order SET OF. This is handled by encoding 8751 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 8752 to produce the required SET OF. 8753 [Steve Henson] 8754 8755 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 8756 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 8757 files to get correct declarations of the ASN.1 item variables. 8758 [Richard Levitte] 8759 8760 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 8761 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 8762 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 8763 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 8764 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 8765 ASN1_ITEM and no wrapper functions. 8766 [Steve Henson] 8767 8768 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 8769 replace the old function pointer based I/O routines. Change most of 8770 the *_d2i_bio() and *_d2i_fp() functions to use these. 8771 [Steve Henson] 8772 8773 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 8774 lines, recognize more "algorithms" that can be deselected, and make 8775 it complain about algorithm deselection that isn't recognised. 8776 [Richard Levitte] 8777 8778 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 8779 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 8780 to use new functions. Add NO_ASN1_OLD which can be set to remove 8781 some old style ASN1 functions: this can be used to determine if old 8782 code will still work when these eventually go away. 8783 [Steve Henson] 8784 8785 *) New extension functions for OCSP structures, these follow the 8786 same conventions as certificates and CRLs. 8787 [Steve Henson] 8788 8789 *) New function X509V3_add1_i2d(). This automatically encodes and 8790 adds an extension. Its behaviour can be customised with various 8791 flags to append, replace or delete. Various wrappers added for 8792 certificates and CRLs. 8793 [Steve Henson] 8794 8795 *) Fix to avoid calling the underlying ASN1 print routine when 8796 an extension cannot be parsed. Correct a typo in the 8797 OCSP_SERVICELOC extension. Tidy up print OCSP format. 8798 [Steve Henson] 8799 8800 *) Make mkdef.pl parse some of the ASN1 macros and add appropriate 8801 entries for variables. 8802 [Steve Henson] 8803 8804 *) Add functionality to apps/openssl.c for detecting locking 8805 problems: As the program is single-threaded, all we have 8806 to do is register a locking callback using an array for 8807 storing which locks are currently held by the program. 8808 [Bodo Moeller] 8809 8810 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 8811 SSL_get_ex_data_X509_STORE_idx(), which is used in 8812 ssl_verify_cert_chain() and thus can be called at any time 8813 during TLS/SSL handshakes so that thread-safety is essential. 8814 Unfortunately, the ex_data design is not at all suited 8815 for multi-threaded use, so it probably should be abolished. 8816 [Bodo Moeller] 8817 8818 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 8819 [Broadcom, tweaked and integrated by Geoff Thorpe] 8820 8821 *) Move common extension printing code to new function 8822 X509V3_print_extensions(). Reorganise OCSP print routines and 8823 implement some needed OCSP ASN1 functions. Add OCSP extensions. 8824 [Steve Henson] 8825 8826 *) New function X509_signature_print() to remove duplication in some 8827 print routines. 8828 [Steve Henson] 8829 8830 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 8831 set (this was treated exactly the same as SET OF previously). This 8832 is used to reorder the STACK representing the structure to match the 8833 encoding. This will be used to get round a problem where a PKCS7 8834 structure which was signed could not be verified because the STACK 8835 order did not reflect the encoded order. 8836 [Steve Henson] 8837 8838 *) Reimplement the OCSP ASN1 module using the new code. 8839 [Steve Henson] 8840 8841 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 8842 for its ASN1 operations. The old style function pointers still exist 8843 for now but they will eventually go away. 8844 [Steve Henson] 8845 8846 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 8847 completely replaces the old ASN1 functionality with a table driven 8848 encoder and decoder which interprets an ASN1_ITEM structure describing 8849 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 8850 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 8851 has also been converted to the new form. 8852 [Steve Henson] 8853 8854 *) Change BN_mod_exp_recp so that negative moduli are tolerated 8855 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 8856 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 8857 for negative moduli. 8858 [Bodo Moeller] 8859 8860 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 8861 of not touching the result's sign bit. 8862 [Bodo Moeller] 8863 8864 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 8865 set. 8866 [Bodo Moeller] 8867 8868 *) Changed the LHASH code to use prototypes for callbacks, and created 8869 macros to declare and implement thin (optionally static) functions 8870 that provide type-safety and avoid function pointer casting for the 8871 type-specific callbacks. 8872 [Geoff Thorpe] 8873 8874 *) Added Kerberos Cipher Suites to be used with TLS, as written in 8875 RFC 2712. 8876 [Veers Staats <staatsvr@asc.hpc.mil>, 8877 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 8878 8879 *) Reformat the FAQ so the different questions and answers can be divided 8880 in sections depending on the subject. 8881 [Richard Levitte] 8882 8883 *) Have the zlib compression code load ZLIB.DLL dynamically under 8884 Windows. 8885 [Richard Levitte] 8886 8887 *) New function BN_mod_sqrt for computing square roots modulo a prime 8888 (using the probabilistic Tonelli-Shanks algorithm unless 8889 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 8890 be handled deterministically). 8891 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 8892 8893 *) Make BN_mod_inverse faster by explicitly handling small quotients 8894 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 8895 512 bits], about 30% for larger ones [1024 or 2048 bits].) 8896 [Bodo Moeller] 8897 8898 *) New function BN_kronecker. 8899 [Bodo Moeller] 8900 8901 *) Fix BN_gcd so that it works on negative inputs; the result is 8902 positive unless both parameters are zero. 8903 Previously something reasonably close to an infinite loop was 8904 possible because numbers could be growing instead of shrinking 8905 in the implementation of Euclid's algorithm. 8906 [Bodo Moeller] 8907 8908 *) Fix BN_is_word() and BN_is_one() macros to take into account the 8909 sign of the number in question. 8910 8911 Fix BN_is_word(a,w) to work correctly for w == 0. 8912 8913 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 8914 because its test if the absolute value of 'a' equals 'w'. 8915 Note that BN_abs_is_word does *not* handle w == 0 reliably; 8916 it exists mostly for use in the implementations of BN_is_zero(), 8917 BN_is_one(), and BN_is_word(). 8918 [Bodo Moeller] 8919 8920 *) New function BN_swap. 8921 [Bodo Moeller] 8922 8923 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 8924 the exponentiation functions are more likely to produce reasonable 8925 results on negative inputs. 8926 [Bodo Moeller] 8927 8928 *) Change BN_mod_mul so that the result is always non-negative. 8929 Previously, it could be negative if one of the factors was negative; 8930 I don't think anyone really wanted that behaviour. 8931 [Bodo Moeller] 8932 8933 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 8934 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 8935 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 8936 and add new functions: 8937 8938 BN_nnmod 8939 BN_mod_sqr 8940 BN_mod_add 8941 BN_mod_add_quick 8942 BN_mod_sub 8943 BN_mod_sub_quick 8944 BN_mod_lshift1 8945 BN_mod_lshift1_quick 8946 BN_mod_lshift 8947 BN_mod_lshift_quick 8948 8949 These functions always generate non-negative results. 8950 8951 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 8952 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 8953 8954 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 8955 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 8956 be reduced modulo m. 8957 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 8958 8959#if 0 8960 The following entry accidentally appeared in the CHANGES file 8961 distributed with OpenSSL 0.9.7. The modifications described in 8962 it do *not* apply to OpenSSL 0.9.7. 8963 8964 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 8965 was actually never needed) and in BN_mul(). The removal in BN_mul() 8966 required a small change in bn_mul_part_recursive() and the addition 8967 of the functions bn_cmp_part_words(), bn_sub_part_words() and 8968 bn_add_part_words(), which do the same thing as bn_cmp_words(), 8969 bn_sub_words() and bn_add_words() except they take arrays with 8970 differing sizes. 8971 [Richard Levitte] 8972#endif 8973 8974 *) In 'openssl passwd', verify passwords read from the terminal 8975 unless the '-salt' option is used (which usually means that 8976 verification would just waste user's time since the resulting 8977 hash is going to be compared with some given password hash) 8978 or the new '-noverify' option is used. 8979 8980 This is an incompatible change, but it does not affect 8981 non-interactive use of 'openssl passwd' (passwords on the command 8982 line, '-stdin' option, '-in ...' option) and thus should not 8983 cause any problems. 8984 [Bodo Moeller] 8985 8986 *) Remove all references to RSAref, since there's no more need for it. 8987 [Richard Levitte] 8988 8989 *) Make DSO load along a path given through an environment variable 8990 (SHLIB_PATH) with shl_load(). 8991 [Richard Levitte] 8992 8993 *) Constify the ENGINE code as a result of BIGNUM constification. 8994 Also constify the RSA code and most things related to it. In a 8995 few places, most notable in the depth of the ASN.1 code, ugly 8996 casts back to non-const were required (to be solved at a later 8997 time) 8998 [Richard Levitte] 8999 9000 *) Make it so the openssl application has all engines loaded by default. 9001 [Richard Levitte] 9002 9003 *) Constify the BIGNUM routines a little more. 9004 [Richard Levitte] 9005 9006 *) Add the following functions: 9007 9008 ENGINE_load_cswift() 9009 ENGINE_load_chil() 9010 ENGINE_load_atalla() 9011 ENGINE_load_nuron() 9012 ENGINE_load_builtin_engines() 9013 9014 That way, an application can itself choose if external engines that 9015 are built-in in OpenSSL shall ever be used or not. The benefit is 9016 that applications won't have to be linked with libdl or other dso 9017 libraries unless it's really needed. 9018 9019 Changed 'openssl engine' to load all engines on demand. 9020 Changed the engine header files to avoid the duplication of some 9021 declarations (they differed!). 9022 [Richard Levitte] 9023 9024 *) 'openssl engine' can now list capabilities. 9025 [Richard Levitte] 9026 9027 *) Better error reporting in 'openssl engine'. 9028 [Richard Levitte] 9029 9030 *) Never call load_dh_param(NULL) in s_server. 9031 [Bodo Moeller] 9032 9033 *) Add engine application. It can currently list engines by name and 9034 identity, and test if they are actually available. 9035 [Richard Levitte] 9036 9037 *) Improve RPM specification file by forcing symbolic linking and making 9038 sure the installed documentation is also owned by root.root. 9039 [Damien Miller <djm@mindrot.org>] 9040 9041 *) Give the OpenSSL applications more possibilities to make use of 9042 keys (public as well as private) handled by engines. 9043 [Richard Levitte] 9044 9045 *) Add OCSP code that comes from CertCo. 9046 [Richard Levitte] 9047 9048 *) Add VMS support for the Rijndael code. 9049 [Richard Levitte] 9050 9051 *) Added untested support for Nuron crypto accelerator. 9052 [Ben Laurie] 9053 9054 *) Add support for external cryptographic devices. This code was 9055 previously distributed separately as the "engine" branch. 9056 [Geoff Thorpe, Richard Levitte] 9057 9058 *) Rework the filename-translation in the DSO code. It is now possible to 9059 have far greater control over how a "name" is turned into a filename 9060 depending on the operating environment and any oddities about the 9061 different shared library filenames on each system. 9062 [Geoff Thorpe] 9063 9064 *) Support threads on FreeBSD-elf in Configure. 9065 [Richard Levitte] 9066 9067 *) Fix for SHA1 assembly problem with MASM: it produces 9068 warnings about corrupt line number information when assembling 9069 with debugging information. This is caused by the overlapping 9070 of two sections. 9071 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 9072 9073 *) NCONF changes. 9074 NCONF_get_number() has no error checking at all. As a replacement, 9075 NCONF_get_number_e() is defined (_e for "error checking") and is 9076 promoted strongly. The old NCONF_get_number is kept around for 9077 binary backward compatibility. 9078 Make it possible for methods to load from something other than a BIO, 9079 by providing a function pointer that is given a name instead of a BIO. 9080 For example, this could be used to load configuration data from an 9081 LDAP server. 9082 [Richard Levitte] 9083 9084 *) Fix for non blocking accept BIOs. Added new I/O special reason 9085 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 9086 with non blocking I/O was not possible because no retry code was 9087 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 9088 this case. 9089 [Steve Henson] 9090 9091 *) Added the beginnings of Rijndael support. 9092 [Ben Laurie] 9093 9094 *) Fix for bug in DirectoryString mask setting. Add support for 9095 X509_NAME_print_ex() in 'req' and X509_print_ex() function 9096 to allow certificate printing to more controllable, additional 9097 'certopt' option to 'x509' to allow new printing options to be 9098 set. 9099 [Steve Henson] 9100 9101 *) Clean old EAY MD5 hack from e_os.h. 9102 [Richard Levitte] 9103 9104 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 9105 9106 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 9107 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 9108 [Joe Orton, Steve Henson] 9109 9110 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 9111 9112 *) Fix additional bug revealed by the NISCC test suite: 9113 9114 Stop bug triggering large recursion when presented with 9115 certain ASN.1 tags (CVE-2003-0851) 9116 [Steve Henson] 9117 9118 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 9119 9120 *) Fix various bugs revealed by running the NISCC test suite: 9121 9122 Stop out of bounds reads in the ASN1 code when presented with 9123 invalid tags (CVE-2003-0543 and CVE-2003-0544). 9124 9125 If verify callback ignores invalid public key errors don't try to check 9126 certificate signature with the NULL public key. 9127 9128 [Steve Henson] 9129 9130 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 9131 if the server requested one: as stated in TLS 1.0 and SSL 3.0 9132 specifications. 9133 [Steve Henson] 9134 9135 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 9136 extra data after the compression methods not only for TLS 1.0 9137 but also for SSL 3.0 (as required by the specification). 9138 [Bodo Moeller; problem pointed out by Matthias Loepfe] 9139 9140 *) Change X509_certificate_type() to mark the key as exported/exportable 9141 when it's 512 *bits* long, not 512 bytes. 9142 [Richard Levitte] 9143 9144 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 9145 9146 *) Countermeasure against the Klima-Pokorny-Rosa extension of 9147 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 9148 a protocol version number mismatch like a decryption error 9149 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 9150 [Bodo Moeller] 9151 9152 *) Turn on RSA blinding by default in the default implementation 9153 to avoid a timing attack. Applications that don't want it can call 9154 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 9155 They would be ill-advised to do so in most cases. 9156 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 9157 9158 *) Change RSA blinding code so that it works when the PRNG is not 9159 seeded (in this case, the secret RSA exponent is abused as 9160 an unpredictable seed -- if it is not unpredictable, there 9161 is no point in blinding anyway). Make RSA blinding thread-safe 9162 by remembering the creator's thread ID in rsa->blinding and 9163 having all other threads use local one-time blinding factors 9164 (this requires more computation than sharing rsa->blinding, but 9165 avoids excessive locking; and if an RSA object is not shared 9166 between threads, blinding will still be very fast). 9167 [Bodo Moeller] 9168 9169 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 9170 9171 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 9172 via timing by performing a MAC computation even if incorrect 9173 block cipher padding has been found. This is a countermeasure 9174 against active attacks where the attacker has to distinguish 9175 between bad padding and a MAC verification error. (CVE-2003-0078) 9176 9177 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 9178 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 9179 Martin Vuagnoux (EPFL, Ilion)] 9180 9181 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 9182 9183 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 9184 memory from its contents. This is done with a counter that will 9185 place alternating values in each byte. This can be used to solve 9186 two issues: 1) the removal of calls to memset() by highly optimizing 9187 compilers, and 2) cleansing with other values than 0, since those can 9188 be read through on certain media, for example a swap space on disk. 9189 [Geoff Thorpe] 9190 9191 *) Bugfix: client side session caching did not work with external caching, 9192 because the session->cipher setting was not restored when reloading 9193 from the external cache. This problem was masked, when 9194 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 9195 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 9196 [Lutz Jaenicke] 9197 9198 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 9199 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 9200 [Zeev Lieber <zeev-l@yahoo.com>] 9201 9202 *) Undo an undocumented change introduced in 0.9.6e which caused 9203 repeated calls to OpenSSL_add_all_ciphers() and 9204 OpenSSL_add_all_digests() to be ignored, even after calling 9205 EVP_cleanup(). 9206 [Richard Levitte] 9207 9208 *) Change the default configuration reader to deal with last line not 9209 being properly terminated. 9210 [Richard Levitte] 9211 9212 *) Change X509_NAME_cmp() so it applies the special rules on handling 9213 DN values that are of type PrintableString, as well as RDNs of type 9214 emailAddress where the value has the type ia5String. 9215 [stefank@valicert.com via Richard Levitte] 9216 9217 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 9218 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 9219 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 9220 the bitwise-OR of the two for use by the majority of applications 9221 wanting this behaviour, and update the docs. The documented 9222 behaviour and actual behaviour were inconsistent and had been 9223 changing anyway, so this is more a bug-fix than a behavioural 9224 change. 9225 [Geoff Thorpe, diagnosed by Nadav Har'El] 9226 9227 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 9228 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 9229 [Bodo Moeller] 9230 9231 *) Fix initialization code race conditions in 9232 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 9233 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 9234 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 9235 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 9236 ssl2_get_cipher_by_char(), 9237 ssl3_get_cipher_by_char(). 9238 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 9239 9240 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 9241 the cached sessions are flushed, as the remove_cb() might use ex_data 9242 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 9243 (see [openssl.org #212]). 9244 [Geoff Thorpe, Lutz Jaenicke] 9245 9246 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 9247 length, instead of the encoding length to d2i_ASN1_OBJECT. 9248 [Steve Henson] 9249 9250 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 9251 9252 *) [In 0.9.6g-engine release:] 9253 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 9254 [Lynn Gazis <lgazis@rainbow.com>] 9255 9256 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 9257 9258 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 9259 and get fix the header length calculation. 9260 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 9261 Alon Kantor <alonk@checkpoint.com> (and others), 9262 Steve Henson] 9263 9264 *) Use proper error handling instead of 'assertions' in buffer 9265 overflow checks added in 0.9.6e. This prevents DoS (the 9266 assertions could call abort()). 9267 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 9268 9269 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 9270 9271 *) Add various sanity checks to asn1_get_length() to reject 9272 the ASN1 length bytes if they exceed sizeof(long), will appear 9273 negative or the content length exceeds the length of the 9274 supplied buffer. 9275 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 9276 9277 *) Fix cipher selection routines: ciphers without encryption had no flags 9278 for the cipher strength set and where therefore not handled correctly 9279 by the selection routines (PR #130). 9280 [Lutz Jaenicke] 9281 9282 *) Fix EVP_dsa_sha macro. 9283 [Nils Larsch] 9284 9285 *) New option 9286 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 9287 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 9288 that was added in OpenSSL 0.9.6d. 9289 9290 As the countermeasure turned out to be incompatible with some 9291 broken SSL implementations, the new option is part of SSL_OP_ALL. 9292 SSL_OP_ALL is usually employed when compatibility with weird SSL 9293 implementations is desired (e.g. '-bugs' option to 's_client' and 9294 's_server'), so the new option is automatically set in many 9295 applications. 9296 [Bodo Moeller] 9297 9298 *) Changes in security patch: 9299 9300 Changes marked "(CHATS)" were sponsored by the Defense Advanced 9301 Research Projects Agency (DARPA) and Air Force Research Laboratory, 9302 Air Force Materiel Command, USAF, under agreement number 9303 F30602-01-2-0537. 9304 9305 *) Add various sanity checks to asn1_get_length() to reject 9306 the ASN1 length bytes if they exceed sizeof(long), will appear 9307 negative or the content length exceeds the length of the 9308 supplied buffer. (CVE-2002-0659) 9309 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 9310 9311 *) Assertions for various potential buffer overflows, not known to 9312 happen in practice. 9313 [Ben Laurie (CHATS)] 9314 9315 *) Various temporary buffers to hold ASCII versions of integers were 9316 too small for 64 bit platforms. (CVE-2002-0655) 9317 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 9318 9319 *) Remote buffer overflow in SSL3 protocol - an attacker could 9320 supply an oversized session ID to a client. (CVE-2002-0656) 9321 [Ben Laurie (CHATS)] 9322 9323 *) Remote buffer overflow in SSL2 protocol - an attacker could 9324 supply an oversized client master key. (CVE-2002-0656) 9325 [Ben Laurie (CHATS)] 9326 9327 Changes between 0.9.6c and 0.9.6d [9 May 2002] 9328 9329 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 9330 encoded as NULL) with id-dsa-with-sha1. 9331 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 9332 9333 *) Check various X509_...() return values in apps/req.c. 9334 [Nils Larsch <nla@trustcenter.de>] 9335 9336 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 9337 an end-of-file condition would erroneously be flagged, when the CRLF 9338 was just at the end of a processed block. The bug was discovered when 9339 processing data through a buffering memory BIO handing the data to a 9340 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 9341 <ptsekov@syntrex.com> and Nedelcho Stanev. 9342 [Lutz Jaenicke] 9343 9344 *) Implement a countermeasure against a vulnerability recently found 9345 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 9346 before application data chunks to avoid the use of known IVs 9347 with data potentially chosen by the attacker. 9348 [Bodo Moeller] 9349 9350 *) Fix length checks in ssl3_get_client_hello(). 9351 [Bodo Moeller] 9352 9353 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 9354 to prevent ssl3_read_internal() from incorrectly assuming that 9355 ssl3_read_bytes() found application data while handshake 9356 processing was enabled when in fact s->s3->in_read_app_data was 9357 merely automatically cleared during the initial handshake. 9358 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 9359 9360 *) Fix object definitions for Private and Enterprise: they were not 9361 recognized in their shortname (=lowercase) representation. Extend 9362 obj_dat.pl to issue an error when using undefined keywords instead 9363 of silently ignoring the problem (Svenning Sorensen 9364 <sss@sss.dnsalias.net>). 9365 [Lutz Jaenicke] 9366 9367 *) Fix DH_generate_parameters() so that it works for 'non-standard' 9368 generators, i.e. generators other than 2 and 5. (Previously, the 9369 code did not properly initialise the 'add' and 'rem' values to 9370 BN_generate_prime().) 9371 9372 In the new general case, we do not insist that 'generator' is 9373 actually a primitive root: This requirement is rather pointless; 9374 a generator of the order-q subgroup is just as good, if not 9375 better. 9376 [Bodo Moeller] 9377 9378 *) Map new X509 verification errors to alerts. Discovered and submitted by 9379 Tom Wu <tom@arcot.com>. 9380 [Lutz Jaenicke] 9381 9382 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 9383 returning non-zero before the data has been completely received 9384 when using non-blocking I/O. 9385 [Bodo Moeller; problem pointed out by John Hughes] 9386 9387 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 9388 [Ben Laurie, Lutz Jaenicke] 9389 9390 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 9391 Yoram Zahavi <YoramZ@gilian.com>). 9392 [Lutz Jaenicke] 9393 9394 *) Add information about CygWin 1.3 and on, and preserve proper 9395 configuration for the versions before that. 9396 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 9397 9398 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 9399 check whether we deal with a copy of a session and do not delete from 9400 the cache in this case. Problem reported by "Izhar Shoshani Levi" 9401 <izhar@checkpoint.com>. 9402 [Lutz Jaenicke] 9403 9404 *) Do not store session data into the internal session cache, if it 9405 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 9406 flag is set). Proposed by Aslam <aslam@funk.com>. 9407 [Lutz Jaenicke] 9408 9409 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 9410 value is 0. 9411 [Richard Levitte] 9412 9413 *) [In 0.9.6d-engine release:] 9414 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 9415 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 9416 9417 *) Add the configuration target linux-s390x. 9418 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 9419 9420 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 9421 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 9422 variable as an indication that a ClientHello message has been 9423 received. As the flag value will be lost between multiple 9424 invocations of ssl3_accept when using non-blocking I/O, the 9425 function may not be aware that a handshake has actually taken 9426 place, thus preventing a new session from being added to the 9427 session cache. 9428 9429 To avoid this problem, we now set s->new_session to 2 instead of 9430 using a local variable. 9431 [Lutz Jaenicke, Bodo Moeller] 9432 9433 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 9434 if the SSL_R_LENGTH_MISMATCH error is detected. 9435 [Geoff Thorpe, Bodo Moeller] 9436 9437 *) New 'shared_ldflag' column in Configure platform table. 9438 [Richard Levitte] 9439 9440 *) Fix EVP_CIPHER_mode macro. 9441 ["Dan S. Camper" <dan@bti.net>] 9442 9443 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 9444 type, we must throw them away by setting rr->length to 0. 9445 [D P Chang <dpc@qualys.com>] 9446 9447 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 9448 9449 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 9450 <Dominikus.Scherkl@biodata.com>. (The previous implementation 9451 worked incorrectly for those cases where range = 10..._2 and 9452 3*range is two bits longer than range.) 9453 [Bodo Moeller] 9454 9455 *) Only add signing time to PKCS7 structures if it is not already 9456 present. 9457 [Steve Henson] 9458 9459 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 9460 OBJ_ld_ce should be OBJ_id_ce. 9461 Also some ip-pda OIDs in crypto/objects/objects.txt were 9462 incorrect (cf. RFC 3039). 9463 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 9464 9465 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 9466 returns early because it has nothing to do. 9467 [Andy Schneider <andy.schneider@bjss.co.uk>] 9468 9469 *) [In 0.9.6c-engine release:] 9470 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 9471 [Andy Schneider <andy.schneider@bjss.co.uk>] 9472 9473 *) [In 0.9.6c-engine release:] 9474 Add support for Cryptographic Appliance's keyserver technology. 9475 (Use engine 'keyclient') 9476 [Cryptographic Appliances and Geoff Thorpe] 9477 9478 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 9479 is called via tools/c89.sh because arguments have to be 9480 rearranged (all '-L' options must appear before the first object 9481 modules). 9482 [Richard Shapiro <rshapiro@abinitio.com>] 9483 9484 *) [In 0.9.6c-engine release:] 9485 Add support for Broadcom crypto accelerator cards, backported 9486 from 0.9.7. 9487 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 9488 9489 *) [In 0.9.6c-engine release:] 9490 Add support for SureWare crypto accelerator cards from 9491 Baltimore Technologies. (Use engine 'sureware') 9492 [Baltimore Technologies and Mark Cox] 9493 9494 *) [In 0.9.6c-engine release:] 9495 Add support for crypto accelerator cards from Accelerated 9496 Encryption Processing, www.aep.ie. (Use engine 'aep') 9497 [AEP Inc. and Mark Cox] 9498 9499 *) Add a configuration entry for gcc on UnixWare. 9500 [Gary Benson <gbenson@redhat.com>] 9501 9502 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 9503 messages are stored in a single piece (fixed-length part and 9504 variable-length part combined) and fix various bugs found on the way. 9505 [Bodo Moeller] 9506 9507 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 9508 instead. BIO_gethostbyname() does not know what timeouts are 9509 appropriate, so entries would stay in cache even when they have 9510 become invalid. 9511 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 9512 9513 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 9514 faced with a pathologically small ClientHello fragment that does 9515 not contain client_version: Instead of aborting with an error, 9516 simply choose the highest available protocol version (i.e., 9517 TLS 1.0 unless it is disabled). In practice, ClientHello 9518 messages are never sent like this, but this change gives us 9519 strictly correct behaviour at least for TLS. 9520 [Bodo Moeller] 9521 9522 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 9523 never resets s->method to s->ctx->method when called from within 9524 one of the SSL handshake functions. 9525 [Bodo Moeller; problem pointed out by Niko Baric] 9526 9527 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 9528 (sent using the client's version number) if client_version is 9529 smaller than the protocol version in use. Also change 9530 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 9531 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 9532 the client will at least see that alert. 9533 [Bodo Moeller] 9534 9535 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 9536 correctly. 9537 [Bodo Moeller] 9538 9539 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 9540 client receives HelloRequest while in a handshake. 9541 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 9542 9543 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 9544 should end in 'break', not 'goto end' which circumvents various 9545 cleanups done in state SSL_ST_OK. But session related stuff 9546 must be disabled for SSL_ST_OK in the case that we just sent a 9547 HelloRequest. 9548 9549 Also avoid some overhead by not calling ssl_init_wbio_buffer() 9550 before just sending a HelloRequest. 9551 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 9552 9553 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 9554 reveal whether illegal block cipher padding was found or a MAC 9555 verification error occurred. (Neither SSLerr() codes nor alerts 9556 are directly visible to potential attackers, but the information 9557 may leak via logfiles.) 9558 9559 Similar changes are not required for the SSL 2.0 implementation 9560 because the number of padding bytes is sent in clear for SSL 2.0, 9561 and the extra bytes are just ignored. However ssl/s2_pkt.c 9562 failed to verify that the purported number of padding bytes is in 9563 the legal range. 9564 [Bodo Moeller] 9565 9566 *) Add OpenUNIX-8 support including shared libraries 9567 (Boyd Lynn Gerber <gerberb@zenez.com>). 9568 [Lutz Jaenicke] 9569 9570 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 9571 'wristwatch attack' using huge encoding parameters (cf. 9572 James H. Manger's CRYPTO 2001 paper). Note that the 9573 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 9574 encoding parameters and hence was not vulnerable. 9575 [Bodo Moeller] 9576 9577 *) BN_sqr() bug fix. 9578 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>] 9579 9580 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 9581 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 9582 followed by modular reduction. 9583 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 9584 9585 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 9586 equivalent based on BN_pseudo_rand() instead of BN_rand(). 9587 [Bodo Moeller] 9588 9589 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 9590 This function was broken, as the check for a new client hello message 9591 to handle SGC did not allow these large messages. 9592 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 9593 [Lutz Jaenicke] 9594 9595 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 9596 [Lutz Jaenicke] 9597 9598 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 9599 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 9600 [Lutz Jaenicke] 9601 9602 *) Rework the configuration and shared library support for Tru64 Unix. 9603 The configuration part makes use of modern compiler features and 9604 still retains old compiler behavior for those that run older versions 9605 of the OS. The shared library support part includes a variant that 9606 uses the RPATH feature, and is available through the special 9607 configuration target "alpha-cc-rpath", which will never be selected 9608 automatically. 9609 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 9610 9611 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 9612 with the same message size as in ssl3_get_certificate_request(). 9613 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 9614 messages might inadvertently be reject as too long. 9615 [Petr Lampa <lampa@fee.vutbr.cz>] 9616 9617 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 9618 [Andy Polyakov] 9619 9620 *) Modified SSL library such that the verify_callback that has been set 9621 specifically for an SSL object with SSL_set_verify() is actually being 9622 used. Before the change, a verify_callback set with this function was 9623 ignored and the verify_callback() set in the SSL_CTX at the time of 9624 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 9625 to allow the necessary settings. 9626 [Lutz Jaenicke] 9627 9628 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 9629 explicitly to NULL, as at least on Solaris 8 this seems not always to be 9630 done automatically (in contradiction to the requirements of the C 9631 standard). This made problems when used from OpenSSH. 9632 [Lutz Jaenicke] 9633 9634 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 9635 dh->length and always used 9636 9637 BN_rand_range(priv_key, dh->p). 9638 9639 BN_rand_range() is not necessary for Diffie-Hellman, and this 9640 specific range makes Diffie-Hellman unnecessarily inefficient if 9641 dh->length (recommended exponent length) is much smaller than the 9642 length of dh->p. We could use BN_rand_range() if the order of 9643 the subgroup was stored in the DH structure, but we only have 9644 dh->length. 9645 9646 So switch back to 9647 9648 BN_rand(priv_key, l, ...) 9649 9650 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 9651 otherwise. 9652 [Bodo Moeller] 9653 9654 *) In 9655 9656 RSA_eay_public_encrypt 9657 RSA_eay_private_decrypt 9658 RSA_eay_private_encrypt (signing) 9659 RSA_eay_public_decrypt (signature verification) 9660 9661 (default implementations for RSA_public_encrypt, 9662 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 9663 always reject numbers >= n. 9664 [Bodo Moeller] 9665 9666 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 9667 to synchronize access to 'locking_thread'. This is necessary on 9668 systems where access to 'locking_thread' (an 'unsigned long' 9669 variable) is not atomic. 9670 [Bodo Moeller] 9671 9672 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 9673 *before* setting the 'crypto_lock_rand' flag. The previous code had 9674 a race condition if 0 is a valid thread ID. 9675 [Travis Vitek <vitek@roguewave.com>] 9676 9677 *) Add support for shared libraries under Irix. 9678 [Albert Chin-A-Young <china@thewrittenword.com>] 9679 9680 *) Add configuration option to build on Linux on both big-endian and 9681 little-endian MIPS. 9682 [Ralf Baechle <ralf@uni-koblenz.de>] 9683 9684 *) Add the possibility to create shared libraries on HP-UX. 9685 [Richard Levitte] 9686 9687 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 9688 9689 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 9690 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 9691 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 9692 PRNG state recovery was possible based on the output of 9693 one PRNG request appropriately sized to gain knowledge on 9694 'md' followed by enough consecutive 1-byte PRNG requests 9695 to traverse all of 'state'. 9696 9697 1. When updating 'md_local' (the current thread's copy of 'md') 9698 during PRNG output generation, hash all of the previous 9699 'md_local' value, not just the half used for PRNG output. 9700 9701 2. Make the number of bytes from 'state' included into the hash 9702 independent from the number of PRNG bytes requested. 9703 9704 The first measure alone would be sufficient to avoid 9705 Markku-Juhani's attack. (Actually it had never occurred 9706 to me that the half of 'md_local' used for chaining was the 9707 half from which PRNG output bytes were taken -- I had always 9708 assumed that the secret half would be used.) The second 9709 measure makes sure that additional data from 'state' is never 9710 mixed into 'md_local' in small portions; this heuristically 9711 further strengthens the PRNG. 9712 [Bodo Moeller] 9713 9714 *) Fix crypto/bn/asm/mips3.s. 9715 [Andy Polyakov] 9716 9717 *) When only the key is given to "enc", the IV is undefined. Print out 9718 an error message in this case. 9719 [Lutz Jaenicke] 9720 9721 *) Handle special case when X509_NAME is empty in X509 printing routines. 9722 [Steve Henson] 9723 9724 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 9725 positive and less than q. 9726 [Bodo Moeller] 9727 9728 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 9729 used: it isn't thread safe and the add_lock_callback should handle 9730 that itself. 9731 [Paul Rose <Paul.Rose@bridge.com>] 9732 9733 *) Verify that incoming data obeys the block size in 9734 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 9735 [Bodo Moeller] 9736 9737 *) Fix OAEP check. 9738 [Ulf Möller, Bodo Möller] 9739 9740 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 9741 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 9742 when fixing the server behaviour for backwards-compatible 'client 9743 hello' messages. (Note that the attack is impractical against 9744 SSL 3.0 and TLS 1.0 anyway because length and version checking 9745 means that the probability of guessing a valid ciphertext is 9746 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 9747 paper.) 9748 9749 Before 0.9.5, the countermeasure (hide the error by generating a 9750 random 'decryption result') did not work properly because 9751 ERR_clear_error() was missing, meaning that SSL_get_error() would 9752 detect the supposedly ignored error. 9753 9754 Both problems are now fixed. 9755 [Bodo Moeller] 9756 9757 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 9758 (previously it was 1024). 9759 [Bodo Moeller] 9760 9761 *) Fix for compatibility mode trust settings: ignore trust settings 9762 unless some valid trust or reject settings are present. 9763 [Steve Henson] 9764 9765 *) Fix for blowfish EVP: its a variable length cipher. 9766 [Steve Henson] 9767 9768 *) Fix various bugs related to DSA S/MIME verification. Handle missing 9769 parameters in DSA public key structures and return an error in the 9770 DSA routines if parameters are absent. 9771 [Steve Henson] 9772 9773 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 9774 in the current directory if neither $RANDFILE nor $HOME was set. 9775 RAND_file_name() in 0.9.6a returned NULL in this case. This has 9776 caused some confusion to Windows users who haven't defined $HOME. 9777 Thus RAND_file_name() is changed again: e_os.h can define a 9778 DEFAULT_HOME, which will be used if $HOME is not set. 9779 For Windows, we use "C:"; on other platforms, we still require 9780 environment variables. 9781 9782 *) Move 'if (!initialized) RAND_poll()' into regions protected by 9783 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 9784 having multiple threads call RAND_poll() concurrently. 9785 [Bodo Moeller] 9786 9787 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 9788 combination of a flag and a thread ID variable. 9789 Otherwise while one thread is in ssleay_rand_bytes (which sets the 9790 flag), *other* threads can enter ssleay_add_bytes without obeying 9791 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 9792 that they do not hold after the first thread unsets add_do_not_lock). 9793 [Bodo Moeller] 9794 9795 *) Change bctest again: '-x' expressions are not available in all 9796 versions of 'test'. 9797 [Bodo Moeller] 9798 9799 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 9800 9801 *) Fix a couple of memory leaks in PKCS7_dataDecode() 9802 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 9803 9804 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 9805 the default extension for executables, if any. Also, make the perl 9806 scripts that use symlink() to test if it really exists and use "cp" 9807 if it doesn't. All this made OpenSSL compilable and installable in 9808 CygWin. 9809 [Richard Levitte] 9810 9811 *) Fix for asn1_GetSequence() for indefinite length constructed data. 9812 If SEQUENCE is length is indefinite just set c->slen to the total 9813 amount of data available. 9814 [Steve Henson, reported by shige@FreeBSD.org] 9815 [This change does not apply to 0.9.7.] 9816 9817 *) Change bctest to avoid here-documents inside command substitution 9818 (workaround for FreeBSD /bin/sh bug). 9819 For compatibility with Ultrix, avoid shell functions (introduced 9820 in the bctest version that searches along $PATH). 9821 [Bodo Moeller] 9822 9823 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 9824 with des_encrypt() defined on some operating systems, like Solaris 9825 and UnixWare. 9826 [Richard Levitte] 9827 9828 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 9829 On the Importance of Eliminating Errors in Cryptographic 9830 Computations, J. Cryptology 14 (2001) 2, 101-119, 9831 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 9832 [Ulf Moeller] 9833 9834 *) MIPS assembler BIGNUM division bug fix. 9835 [Andy Polyakov] 9836 9837 *) Disabled incorrect Alpha assembler code. 9838 [Richard Levitte] 9839 9840 *) Fix PKCS#7 decode routines so they correctly update the length 9841 after reading an EOC for the EXPLICIT tag. 9842 [Steve Henson] 9843 [This change does not apply to 0.9.7.] 9844 9845 *) Fix bug in PKCS#12 key generation routines. This was triggered 9846 if a 3DES key was generated with a 0 initial byte. Include 9847 PKCS12_BROKEN_KEYGEN compilation option to retain the old 9848 (but broken) behaviour. 9849 [Steve Henson] 9850 9851 *) Enhance bctest to search for a working bc along $PATH and print 9852 it when found. 9853 [Tim Rice <tim@multitalents.net> via Richard Levitte] 9854 9855 *) Fix memory leaks in err.c: free err_data string if necessary; 9856 don't write to the wrong index in ERR_set_error_data. 9857 [Bodo Moeller] 9858 9859 *) Implement ssl23_peek (analogous to ssl23_read), which previously 9860 did not exist. 9861 [Bodo Moeller] 9862 9863 *) Replace rdtsc with _emit statements for VC++ version 5. 9864 [Jeremy Cooper <jeremy@baymoo.org>] 9865 9866 *) Make it possible to reuse SSLv2 sessions. 9867 [Richard Levitte] 9868 9869 *) In copy_email() check for >= 0 as a return value for 9870 X509_NAME_get_index_by_NID() since 0 is a valid index. 9871 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 9872 9873 *) Avoid coredump with unsupported or invalid public keys by checking if 9874 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 9875 PKCS7_verify() fails with non detached data. 9876 [Steve Henson] 9877 9878 *) Don't use getenv in library functions when run as setuid/setgid. 9879 New function OPENSSL_issetugid(). 9880 [Ulf Moeller] 9881 9882 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 9883 due to incorrect handling of multi-threading: 9884 9885 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 9886 9887 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 9888 9889 3. Count how many times MemCheck_off() has been called so that 9890 nested use can be treated correctly. This also avoids 9891 inband-signalling in the previous code (which relied on the 9892 assumption that thread ID 0 is impossible). 9893 [Bodo Moeller] 9894 9895 *) Add "-rand" option also to s_client and s_server. 9896 [Lutz Jaenicke] 9897 9898 *) Fix CPU detection on Irix 6.x. 9899 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 9900 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 9901 9902 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 9903 was empty. 9904 [Steve Henson] 9905 [This change does not apply to 0.9.7.] 9906 9907 *) Use the cached encoding of an X509_NAME structure rather than 9908 copying it. This is apparently the reason for the libsafe "errors" 9909 but the code is actually correct. 9910 [Steve Henson] 9911 9912 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 9913 Bleichenbacher's DSA attack. 9914 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 9915 to be set and top=0 forces the highest bit to be set; top=-1 is new 9916 and leaves the highest bit random. 9917 [Ulf Moeller, Bodo Moeller] 9918 9919 *) In the NCONF_...-based implementations for CONF_... queries 9920 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 9921 a temporary CONF structure with the data component set to NULL 9922 (which gives segmentation faults in lh_retrieve). 9923 Instead, use NULL for the CONF pointer in CONF_get_string and 9924 CONF_get_number (which may use environment variables) and directly 9925 return NULL from CONF_get_section. 9926 [Bodo Moeller] 9927 9928 *) Fix potential buffer overrun for EBCDIC. 9929 [Ulf Moeller] 9930 9931 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 9932 keyUsage if basicConstraints absent for a CA. 9933 [Steve Henson] 9934 9935 *) Make SMIME_write_PKCS7() write mail header values with a format that 9936 is more generally accepted (no spaces before the semicolon), since 9937 some programs can't parse those values properly otherwise. Also make 9938 sure BIO's that break lines after each write do not create invalid 9939 headers. 9940 [Richard Levitte] 9941 9942 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 9943 macros previously used would not encode an empty SEQUENCE OF 9944 and break the signature. 9945 [Steve Henson] 9946 [This change does not apply to 0.9.7.] 9947 9948 *) Zero the premaster secret after deriving the master secret in 9949 DH ciphersuites. 9950 [Steve Henson] 9951 9952 *) Add some EVP_add_digest_alias registrations (as found in 9953 OpenSSL_add_all_digests()) to SSL_library_init() 9954 aka OpenSSL_add_ssl_algorithms(). This provides improved 9955 compatibility with peers using X.509 certificates 9956 with unconventional AlgorithmIdentifier OIDs. 9957 [Bodo Moeller] 9958 9959 *) Fix for Irix with NO_ASM. 9960 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 9961 9962 *) ./config script fixes. 9963 [Ulf Moeller, Richard Levitte] 9964 9965 *) Fix 'openssl passwd -1'. 9966 [Bodo Moeller] 9967 9968 *) Change PKCS12_key_gen_asc() so it can cope with non null 9969 terminated strings whose length is passed in the passlen 9970 parameter, for example from PEM callbacks. This was done 9971 by adding an extra length parameter to asc2uni(). 9972 [Steve Henson, reported by <oddissey@samsung.co.kr>] 9973 9974 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 9975 call failed, free the DSA structure. 9976 [Bodo Moeller] 9977 9978 *) Fix to uni2asc() to cope with zero length Unicode strings. 9979 These are present in some PKCS#12 files. 9980 [Steve Henson] 9981 9982 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 9983 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 9984 when writing a 32767 byte record. 9985 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 9986 9987 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 9988 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 9989 9990 (RSA objects have a reference count access to which is protected 9991 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 9992 so they are meant to be shared between threads.) 9993 [Bodo Moeller, Geoff Thorpe; original patch submitted by 9994 "Reddie, Steven" <Steven.Reddie@ca.com>] 9995 9996 *) Fix a deadlock in CRYPTO_mem_leaks(). 9997 [Bodo Moeller] 9998 9999 *) Use better test patterns in bntest. 10000 [Ulf Möller] 10001 10002 *) rand_win.c fix for Borland C. 10003 [Ulf Möller] 10004 10005 *) BN_rshift bugfix for n == 0. 10006 [Bodo Moeller] 10007 10008 *) Add a 'bctest' script that checks for some known 'bc' bugs 10009 so that 'make test' does not abort just because 'bc' is broken. 10010 [Bodo Moeller] 10011 10012 *) Store verify_result within SSL_SESSION also for client side to 10013 avoid potential security hole. (Re-used sessions on the client side 10014 always resulted in verify_result==X509_V_OK, not using the original 10015 result of the server certificate verification.) 10016 [Lutz Jaenicke] 10017 10018 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 10019 SSL3_RT_APPLICATION_DATA, return 0. 10020 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 10021 [Bodo Moeller] 10022 10023 *) Fix SSL_peek: 10024 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 10025 releases, have been re-implemented by renaming the previous 10026 implementations of ssl2_read and ssl3_read to ssl2_read_internal 10027 and ssl3_read_internal, respectively, and adding 'peek' parameters 10028 to them. The new ssl[23]_{read,peek} functions are calls to 10029 ssl[23]_read_internal with the 'peek' flag set appropriately. 10030 A 'peek' parameter has also been added to ssl3_read_bytes, which 10031 does the actual work for ssl3_read_internal. 10032 [Bodo Moeller] 10033 10034 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 10035 the method-specific "init()" handler. Also clean up ex_data after 10036 calling the method-specific "finish()" handler. Previously, this was 10037 happening the other way round. 10038 [Geoff Thorpe] 10039 10040 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 10041 The previous value, 12, was not always sufficient for BN_mod_exp(). 10042 [Bodo Moeller] 10043 10044 *) Make sure that shared libraries get the internal name engine with 10045 the full version number and not just 0. This should mark the 10046 shared libraries as not backward compatible. Of course, this should 10047 be changed again when we can guarantee backward binary compatibility. 10048 [Richard Levitte] 10049 10050 *) Fix typo in get_cert_by_subject() in by_dir.c 10051 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 10052 10053 *) Rework the system to generate shared libraries: 10054 10055 - Make note of the expected extension for the shared libraries and 10056 if there is a need for symbolic links from for example libcrypto.so.0 10057 to libcrypto.so.0.9.7. There is extended info in Configure for 10058 that. 10059 10060 - Make as few rebuilds of the shared libraries as possible. 10061 10062 - Still avoid linking the OpenSSL programs with the shared libraries. 10063 10064 - When installing, install the shared libraries separately from the 10065 static ones. 10066 [Richard Levitte] 10067 10068 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 10069 10070 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 10071 and not in SSL_clear because the latter is also used by the 10072 accept/connect functions; previously, the settings made by 10073 SSL_set_read_ahead would be lost during the handshake. 10074 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 10075 10076 *) Correct util/mkdef.pl to be selective about disabled algorithms. 10077 Previously, it would create entries for disabled algorithms no 10078 matter what. 10079 [Richard Levitte] 10080 10081 *) Added several new manual pages for SSL_* function. 10082 [Lutz Jaenicke] 10083 10084 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 10085 10086 *) In ssl23_get_client_hello, generate an error message when faced 10087 with an initial SSL 3.0/TLS record that is too small to contain the 10088 first two bytes of the ClientHello message, i.e. client_version. 10089 (Note that this is a pathologic case that probably has never happened 10090 in real life.) The previous approach was to use the version number 10091 from the record header as a substitute; but our protocol choice 10092 should not depend on that one because it is not authenticated 10093 by the Finished messages. 10094 [Bodo Moeller] 10095 10096 *) More robust randomness gathering functions for Windows. 10097 [Jeffrey Altman <jaltman@columbia.edu>] 10098 10099 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 10100 not set then we don't setup the error code for issuer check errors 10101 to avoid possibly overwriting other errors which the callback does 10102 handle. If an application does set the flag then we assume it knows 10103 what it is doing and can handle the new informational codes 10104 appropriately. 10105 [Steve Henson] 10106 10107 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 10108 a general "ANY" type, as such it should be able to decode anything 10109 including tagged types. However it didn't check the class so it would 10110 wrongly interpret tagged types in the same way as their universal 10111 counterpart and unknown types were just rejected. Changed so that the 10112 tagged and unknown types are handled in the same way as a SEQUENCE: 10113 that is the encoding is stored intact. There is also a new type 10114 "V_ASN1_OTHER" which is used when the class is not universal, in this 10115 case we have no idea what the actual type is so we just lump them all 10116 together. 10117 [Steve Henson] 10118 10119 *) On VMS, stdout may very well lead to a file that is written to 10120 in a record-oriented fashion. That means that every write() will 10121 write a separate record, which will be read separately by the 10122 programs trying to read from it. This can be very confusing. 10123 10124 The solution is to put a BIO filter in the way that will buffer 10125 text until a linefeed is reached, and then write everything a 10126 line at a time, so every record written will be an actual line, 10127 not chunks of lines and not (usually doesn't happen, but I've 10128 seen it once) several lines in one record. BIO_f_linebuffer() is 10129 the answer. 10130 10131 Currently, it's a VMS-only method, because that's where it has 10132 been tested well enough. 10133 [Richard Levitte] 10134 10135 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 10136 it can return incorrect results. 10137 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 10138 but it was in 0.9.6-beta[12].) 10139 [Bodo Moeller] 10140 10141 *) Disable the check for content being present when verifying detached 10142 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 10143 include zero length content when signing messages. 10144 [Steve Henson] 10145 10146 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 10147 BIO_ctrl (for BIO pairs). 10148 [Bodo Möller] 10149 10150 *) Add DSO method for VMS. 10151 [Richard Levitte] 10152 10153 *) Bug fix: Montgomery multiplication could produce results with the 10154 wrong sign. 10155 [Ulf Möller] 10156 10157 *) Add RPM specification openssl.spec and modify it to build three 10158 packages. The default package contains applications, application 10159 documentation and run-time libraries. The devel package contains 10160 include files, static libraries and function documentation. The 10161 doc package contains the contents of the doc directory. The original 10162 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 10163 [Richard Levitte] 10164 10165 *) Add a large number of documentation files for many SSL routines. 10166 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 10167 10168 *) Add a configuration entry for Sony News 4. 10169 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 10170 10171 *) Don't set the two most significant bits to one when generating a 10172 random number < q in the DSA library. 10173 [Ulf Möller] 10174 10175 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 10176 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 10177 the underlying transport is blocking) if a handshake took place. 10178 (The default behaviour is needed by applications such as s_client 10179 and s_server that use select() to determine when to use SSL_read; 10180 but for applications that know in advance when to expect data, it 10181 just makes things more complicated.) 10182 [Bodo Moeller] 10183 10184 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 10185 from EGD. 10186 [Ben Laurie] 10187 10188 *) Add a few more EBCDIC conditionals that make `req' and `x509' 10189 work better on such systems. 10190 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10191 10192 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 10193 Update PKCS12_parse() so it copies the friendlyName and the 10194 keyid to the certificates aux info. 10195 [Steve Henson] 10196 10197 *) Fix bug in PKCS7_verify() which caused an infinite loop 10198 if there was more than one signature. 10199 [Sven Uszpelkat <su@celocom.de>] 10200 10201 *) Major change in util/mkdef.pl to include extra information 10202 about each symbol, as well as presenting variables as well 10203 as functions. This change means that there's n more need 10204 to rebuild the .num files when some algorithms are excluded. 10205 [Richard Levitte] 10206 10207 *) Allow the verify time to be set by an application, 10208 rather than always using the current time. 10209 [Steve Henson] 10210 10211 *) Phase 2 verify code reorganisation. The certificate 10212 verify code now looks up an issuer certificate by a 10213 number of criteria: subject name, authority key id 10214 and key usage. It also verifies self signed certificates 10215 by the same criteria. The main comparison function is 10216 X509_check_issued() which performs these checks. 10217 10218 Lot of changes were necessary in order to support this 10219 without completely rewriting the lookup code. 10220 10221 Authority and subject key identifier are now cached. 10222 10223 The LHASH 'certs' is X509_STORE has now been replaced 10224 by a STACK_OF(X509_OBJECT). This is mainly because an 10225 LHASH can't store or retrieve multiple objects with 10226 the same hash value. 10227 10228 As a result various functions (which were all internal 10229 use only) have changed to handle the new X509_STORE 10230 structure. This will break anything that messed round 10231 with X509_STORE internally. 10232 10233 The functions X509_STORE_add_cert() now checks for an 10234 exact match, rather than just subject name. 10235 10236 The X509_STORE API doesn't directly support the retrieval 10237 of multiple certificates matching a given criteria, however 10238 this can be worked round by performing a lookup first 10239 (which will fill the cache with candidate certificates) 10240 and then examining the cache for matches. This is probably 10241 the best we can do without throwing out X509_LOOKUP 10242 entirely (maybe later...). 10243 10244 The X509_VERIFY_CTX structure has been enhanced considerably. 10245 10246 All certificate lookup operations now go via a get_issuer() 10247 callback. Although this currently uses an X509_STORE it 10248 can be replaced by custom lookups. This is a simple way 10249 to bypass the X509_STORE hackery necessary to make this 10250 work and makes it possible to use more efficient techniques 10251 in future. A very simple version which uses a simple 10252 STACK for its trusted certificate store is also provided 10253 using X509_STORE_CTX_trusted_stack(). 10254 10255 The verify_cb() and verify() callbacks now have equivalents 10256 in the X509_STORE_CTX structure. 10257 10258 X509_STORE_CTX also has a 'flags' field which can be used 10259 to customise the verify behaviour. 10260 [Steve Henson] 10261 10262 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 10263 excludes S/MIME capabilities. 10264 [Steve Henson] 10265 10266 *) When a certificate request is read in keep a copy of the 10267 original encoding of the signed data and use it when outputting 10268 again. Signatures then use the original encoding rather than 10269 a decoded, encoded version which may cause problems if the 10270 request is improperly encoded. 10271 [Steve Henson] 10272 10273 *) For consistency with other BIO_puts implementations, call 10274 buffer_write(b, ...) directly in buffer_puts instead of calling 10275 BIO_write(b, ...). 10276 10277 In BIO_puts, increment b->num_write as in BIO_write. 10278 [Peter.Sylvester@EdelWeb.fr] 10279 10280 *) Fix BN_mul_word for the case where the word is 0. (We have to use 10281 BN_zero, we may not return a BIGNUM with an array consisting of 10282 words set to zero.) 10283 [Bodo Moeller] 10284 10285 *) Avoid calling abort() from within the library when problems are 10286 detected, except if preprocessor symbols have been defined 10287 (such as REF_CHECK, BN_DEBUG etc.). 10288 [Bodo Moeller] 10289 10290 *) New openssl application 'rsautl'. This utility can be 10291 used for low level RSA operations. DER public key 10292 BIO/fp routines also added. 10293 [Steve Henson] 10294 10295 *) New Configure entry and patches for compiling on QNX 4. 10296 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 10297 10298 *) A demo state-machine implementation was sponsored by 10299 Nuron (http://www.nuron.com/) and is now available in 10300 demos/state_machine. 10301 [Ben Laurie] 10302 10303 *) New options added to the 'dgst' utility for signature 10304 generation and verification. 10305 [Steve Henson] 10306 10307 *) Unrecognized PKCS#7 content types are now handled via a 10308 catch all ASN1_TYPE structure. This allows unsupported 10309 types to be stored as a "blob" and an application can 10310 encode and decode it manually. 10311 [Steve Henson] 10312 10313 *) Fix various signed/unsigned issues to make a_strex.c 10314 compile under VC++. 10315 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 10316 10317 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 10318 length if passed a buffer. ASN1_INTEGER_to_BN failed 10319 if passed a NULL BN and its argument was negative. 10320 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 10321 10322 *) Modification to PKCS#7 encoding routines to output definite 10323 length encoding. Since currently the whole structures are in 10324 memory there's not real point in using indefinite length 10325 constructed encoding. However if OpenSSL is compiled with 10326 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 10327 [Steve Henson] 10328 10329 *) Added BIO_vprintf() and BIO_vsnprintf(). 10330 [Richard Levitte] 10331 10332 *) Added more prefixes to parse for in the strings written 10333 through a logging bio, to cover all the levels that are available 10334 through syslog. The prefixes are now: 10335 10336 PANIC, EMERG, EMR => LOG_EMERG 10337 ALERT, ALR => LOG_ALERT 10338 CRIT, CRI => LOG_CRIT 10339 ERROR, ERR => LOG_ERR 10340 WARNING, WARN, WAR => LOG_WARNING 10341 NOTICE, NOTE, NOT => LOG_NOTICE 10342 INFO, INF => LOG_INFO 10343 DEBUG, DBG => LOG_DEBUG 10344 10345 and as before, if none of those prefixes are present at the 10346 beginning of the string, LOG_ERR is chosen. 10347 10348 On Win32, the LOG_* levels are mapped according to this: 10349 10350 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 10351 LOG_WARNING => EVENTLOG_WARNING_TYPE 10352 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 10353 10354 [Richard Levitte] 10355 10356 *) Made it possible to reconfigure with just the configuration 10357 argument "reconf" or "reconfigure". The command line arguments 10358 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 10359 and are retrieved from there when reconfiguring. 10360 [Richard Levitte] 10361 10362 *) MD4 implemented. 10363 [Assar Westerlund <assar@sics.se>, Richard Levitte] 10364 10365 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 10366 [Richard Levitte] 10367 10368 *) The obj_dat.pl script was messing up the sorting of object 10369 names. The reason was that it compared the quoted version 10370 of strings as a result "OCSP" > "OCSP Signing" because 10371 " > SPACE. Changed script to store unquoted versions of 10372 names and add quotes on output. It was also omitting some 10373 names from the lookup table if they were given a default 10374 value (that is if SN is missing it is given the same 10375 value as LN and vice versa), these are now added on the 10376 grounds that if an object has a name we should be able to 10377 look it up. Finally added warning output when duplicate 10378 short or long names are found. 10379 [Steve Henson] 10380 10381 *) Changes needed for Tandem NSK. 10382 [Scott Uroff <scott@xypro.com>] 10383 10384 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 10385 RSA_padding_check_SSLv23(), special padding was never detected 10386 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 10387 version rollback attacks was not effective. 10388 10389 In s23_clnt.c, don't use special rollback-attack detection padding 10390 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 10391 client; similarly, in s23_srvr.c, don't do the rollback check if 10392 SSL 2.0 is the only protocol enabled in the server. 10393 [Bodo Moeller] 10394 10395 *) Make it possible to get hexdumps of unprintable data with 'openssl 10396 asn1parse'. By implication, the functions ASN1_parse_dump() and 10397 BIO_dump_indent() are added. 10398 [Richard Levitte] 10399 10400 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 10401 these print out strings and name structures based on various 10402 flags including RFC2253 support and proper handling of 10403 multibyte characters. Added options to the 'x509' utility 10404 to allow the various flags to be set. 10405 [Steve Henson] 10406 10407 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 10408 Also change the functions X509_cmp_current_time() and 10409 X509_gmtime_adj() work with an ASN1_TIME structure, 10410 this will enable certificates using GeneralizedTime in validity 10411 dates to be checked. 10412 [Steve Henson] 10413 10414 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 10415 negative public key encodings) on by default, 10416 NO_NEG_PUBKEY_BUG can be set to disable it. 10417 [Steve Henson] 10418 10419 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 10420 content octets. An i2c_ASN1_OBJECT is unnecessary because 10421 the encoding can be trivially obtained from the structure. 10422 [Steve Henson] 10423 10424 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 10425 not read locks (CRYPTO_r_[un]lock). 10426 [Bodo Moeller] 10427 10428 *) A first attempt at creating official support for shared 10429 libraries through configuration. I've kept it so the 10430 default is static libraries only, and the OpenSSL programs 10431 are always statically linked for now, but there are 10432 preparations for dynamic linking in place. 10433 This has been tested on Linux and Tru64. 10434 [Richard Levitte] 10435 10436 *) Randomness polling function for Win9x, as described in: 10437 Peter Gutmann, Software Generation of Practically Strong 10438 Random Numbers. 10439 [Ulf Möller] 10440 10441 *) Fix so PRNG is seeded in req if using an already existing 10442 DSA key. 10443 [Steve Henson] 10444 10445 *) New options to smime application. -inform and -outform 10446 allow alternative formats for the S/MIME message including 10447 PEM and DER. The -content option allows the content to be 10448 specified separately. This should allow things like Netscape 10449 form signing output easier to verify. 10450 [Steve Henson] 10451 10452 *) Fix the ASN1 encoding of tags using the 'long form'. 10453 [Steve Henson] 10454 10455 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 10456 STRING types. These convert content octets to and from the 10457 underlying type. The actual tag and length octets are 10458 already assumed to have been read in and checked. These 10459 are needed because all other string types have virtually 10460 identical handling apart from the tag. By having versions 10461 of the ASN1 functions that just operate on content octets 10462 IMPLICIT tagging can be handled properly. It also allows 10463 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 10464 and ASN1_INTEGER are identical apart from the tag. 10465 [Steve Henson] 10466 10467 *) Change the handling of OID objects as follows: 10468 10469 - New object identifiers are inserted in objects.txt, following 10470 the syntax given in objects.README. 10471 - objects.pl is used to process obj_mac.num and create a new 10472 obj_mac.h. 10473 - obj_dat.pl is used to create a new obj_dat.h, using the data in 10474 obj_mac.h. 10475 10476 This is currently kind of a hack, and the perl code in objects.pl 10477 isn't very elegant, but it works as I intended. The simplest way 10478 to check that it worked correctly is to look in obj_dat.h and 10479 check the array nid_objs and make sure the objects haven't moved 10480 around (this is important!). Additions are OK, as well as 10481 consistent name changes. 10482 [Richard Levitte] 10483 10484 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 10485 [Bodo Moeller] 10486 10487 *) Addition of the command line parameter '-rand file' to 'openssl req'. 10488 The given file adds to whatever has already been seeded into the 10489 random pool through the RANDFILE configuration file option or 10490 environment variable, or the default random state file. 10491 [Richard Levitte] 10492 10493 *) mkstack.pl now sorts each macro group into lexical order. 10494 Previously the output order depended on the order the files 10495 appeared in the directory, resulting in needless rewriting 10496 of safestack.h . 10497 [Steve Henson] 10498 10499 *) Patches to make OpenSSL compile under Win32 again. Mostly 10500 work arounds for the VC++ problem that it treats func() as 10501 func(void). Also stripped out the parts of mkdef.pl that 10502 added extra typesafe functions: these no longer exist. 10503 [Steve Henson] 10504 10505 *) Reorganisation of the stack code. The macros are now all 10506 collected in safestack.h . Each macro is defined in terms of 10507 a "stack macro" of the form SKM_<name>(type, a, b). The 10508 DEBUG_SAFESTACK is now handled in terms of function casts, 10509 this has the advantage of retaining type safety without the 10510 use of additional functions. If DEBUG_SAFESTACK is not defined 10511 then the non typesafe macros are used instead. Also modified the 10512 mkstack.pl script to handle the new form. Needs testing to see 10513 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 10514 the default if no major problems. Similar behaviour for ASN1_SET_OF 10515 and PKCS12_STACK_OF. 10516 [Steve Henson] 10517 10518 *) When some versions of IIS use the 'NET' form of private key the 10519 key derivation algorithm is different. Normally MD5(password) is 10520 used as a 128 bit RC4 key. In the modified case 10521 MD5(MD5(password) + "SGCKEYSALT") is used instead. Added some 10522 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 10523 as the old Netscape_RSA functions except they have an additional 10524 'sgckey' parameter which uses the modified algorithm. Also added 10525 an -sgckey command line option to the rsa utility. Thanks to 10526 Adrian Peck <bertie@ncipher.com> for posting details of the modified 10527 algorithm to openssl-dev. 10528 [Steve Henson] 10529 10530 *) The evp_local.h macros were using 'c.##kname' which resulted in 10531 invalid expansion on some systems (SCO 5.0.5 for example). 10532 Corrected to 'c.kname'. 10533 [Phillip Porch <root@theporch.com>] 10534 10535 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 10536 a STACK of email addresses from a certificate or request, these look 10537 in the subject name and the subject alternative name extensions and 10538 omit any duplicate addresses. 10539 [Steve Henson] 10540 10541 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 10542 This makes DSA verification about 2 % faster. 10543 [Bodo Moeller] 10544 10545 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 10546 (meaning that now 2^5 values will be precomputed, which is only 4 KB 10547 plus overhead for 1024 bit moduli). 10548 This makes exponentiations about 0.5 % faster for 1024 bit 10549 exponents (as measured by "openssl speed rsa2048"). 10550 [Bodo Moeller] 10551 10552 *) Rename memory handling macros to avoid conflicts with other 10553 software: 10554 Malloc => OPENSSL_malloc 10555 Malloc_locked => OPENSSL_malloc_locked 10556 Realloc => OPENSSL_realloc 10557 Free => OPENSSL_free 10558 [Richard Levitte] 10559 10560 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 10561 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 10562 [Bodo Moeller] 10563 10564 *) CygWin32 support. 10565 [John Jarvie <jjarvie@newsguy.com>] 10566 10567 *) The type-safe stack code has been rejigged. It is now only compiled 10568 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 10569 by default all type-specific stack functions are "#define"d back to 10570 standard stack functions. This results in more streamlined output 10571 but retains the type-safety checking possibilities of the original 10572 approach. 10573 [Geoff Thorpe] 10574 10575 *) The STACK code has been cleaned up, and certain type declarations 10576 that didn't make a lot of sense have been brought in line. This has 10577 also involved a cleanup of sorts in safestack.h to more correctly 10578 map type-safe stack functions onto their plain stack counterparts. 10579 This work has also resulted in a variety of "const"ifications of 10580 lots of the code, especially "_cmp" operations which should normally 10581 be prototyped with "const" parameters anyway. 10582 [Geoff Thorpe] 10583 10584 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 10585 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 10586 (The PRNG state consists of two parts, the large pool 'state' and 'md', 10587 where all of 'md' is used each time the PRNG is used, but 'state' 10588 is used only indexed by a cyclic counter. As entropy may not be 10589 well distributed from the beginning, 'md' is important as a 10590 chaining variable. However, the output function chains only half 10591 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 10592 all of 'md', and seeding with STATE_SIZE dummy bytes will result 10593 in all of 'state' being rewritten, with the new values depending 10594 on virtually all of 'md'. This overcomes the 80 bit limitation.) 10595 [Bodo Moeller] 10596 10597 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 10598 the handshake is continued after ssl_verify_cert_chain(); 10599 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 10600 can lead to 'unexplainable' connection aborts later. 10601 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 10602 10603 *) Major EVP API cipher revision. 10604 Add hooks for extra EVP features. This allows various cipher 10605 parameters to be set in the EVP interface. Support added for variable 10606 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 10607 setting of RC2 and RC5 parameters. 10608 10609 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 10610 ciphers. 10611 10612 Remove lots of duplicated code from the EVP library. For example *every* 10613 cipher init() function handles the 'iv' in the same way according to the 10614 cipher mode. They also all do nothing if the 'key' parameter is NULL and 10615 for CFB and OFB modes they zero ctx->num. 10616 10617 New functionality allows removal of S/MIME code RC2 hack. 10618 10619 Most of the routines have the same form and so can be declared in terms 10620 of macros. 10621 10622 By shifting this to the top level EVP_CipherInit() it can be removed from 10623 all individual ciphers. If the cipher wants to handle IVs or keys 10624 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 10625 flags. 10626 10627 Change lots of functions like EVP_EncryptUpdate() to now return a 10628 value: although software versions of the algorithms cannot fail 10629 any installed hardware versions can. 10630 [Steve Henson] 10631 10632 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 10633 this option is set, tolerate broken clients that send the negotiated 10634 protocol version number instead of the requested protocol version 10635 number. 10636 [Bodo Moeller] 10637 10638 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 10639 i.e. non-zero for export ciphersuites, zero otherwise. 10640 Previous versions had this flag inverted, inconsistent with 10641 rsa_tmp_cb (..._TMP_RSA_CB). 10642 [Bodo Moeller; problem reported by Amit Chopra] 10643 10644 *) Add missing DSA library text string. Work around for some IIS 10645 key files with invalid SEQUENCE encoding. 10646 [Steve Henson] 10647 10648 *) Add a document (doc/standards.txt) that list all kinds of standards 10649 and so on that are implemented in OpenSSL. 10650 [Richard Levitte] 10651 10652 *) Enhance c_rehash script. Old version would mishandle certificates 10653 with the same subject name hash and wouldn't handle CRLs at all. 10654 Added -fingerprint option to crl utility, to support new c_rehash 10655 features. 10656 [Steve Henson] 10657 10658 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 10659 [Ulf Möller] 10660 10661 *) Fix for SSL server purpose checking. Server checking was 10662 rejecting certificates which had extended key usage present 10663 but no ssl client purpose. 10664 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 10665 10666 *) Make PKCS#12 code work with no password. The PKCS#12 spec 10667 is a little unclear about how a blank password is handled. 10668 Since the password in encoded as a BMPString with terminating 10669 double NULL a zero length password would end up as just the 10670 double NULL. However no password at all is different and is 10671 handled differently in the PKCS#12 key generation code. NS 10672 treats a blank password as zero length. MSIE treats it as no 10673 password on export: but it will try both on import. We now do 10674 the same: PKCS12_parse() tries zero length and no password if 10675 the password is set to "" or NULL (NULL is now a valid password: 10676 it wasn't before) as does the pkcs12 application. 10677 [Steve Henson] 10678 10679 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 10680 perror when PEM_read_bio_X509_REQ fails, the error message must 10681 be obtained from the error queue. 10682 [Bodo Moeller] 10683 10684 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 10685 it in ERR_remove_state if appropriate, and change ERR_get_state 10686 accordingly to avoid race conditions (this is necessary because 10687 thread_hash is no longer constant once set). 10688 [Bodo Moeller] 10689 10690 *) Bugfix for linux-elf makefile.one. 10691 [Ulf Möller] 10692 10693 *) RSA_get_default_method() will now cause a default 10694 RSA_METHOD to be chosen if one doesn't exist already. 10695 Previously this was only set during a call to RSA_new() 10696 or RSA_new_method(NULL) meaning it was possible for 10697 RSA_get_default_method() to return NULL. 10698 [Geoff Thorpe] 10699 10700 *) Added native name translation to the existing DSO code 10701 that will convert (if the flag to do so is set) filenames 10702 that are sufficiently small and have no path information 10703 into a canonical native form. Eg. "blah" converted to 10704 "libblah.so" or "blah.dll" etc. 10705 [Geoff Thorpe] 10706 10707 *) New function ERR_error_string_n(e, buf, len) which is like 10708 ERR_error_string(e, buf), but writes at most 'len' bytes 10709 including the 0 terminator. For ERR_error_string_n, 'buf' 10710 may not be NULL. 10711 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 10712 10713 *) CONF library reworked to become more general. A new CONF 10714 configuration file reader "class" is implemented as well as a 10715 new functions (NCONF_*, for "New CONF") to handle it. The now 10716 old CONF_* functions are still there, but are reimplemented to 10717 work in terms of the new functions. Also, a set of functions 10718 to handle the internal storage of the configuration data is 10719 provided to make it easier to write new configuration file 10720 reader "classes" (I can definitely see something reading a 10721 configuration file in XML format, for example), called _CONF_*, 10722 or "the configuration storage API"... 10723 10724 The new configuration file reading functions are: 10725 10726 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 10727 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 10728 10729 NCONF_default, NCONF_WIN32 10730 10731 NCONF_dump_fp, NCONF_dump_bio 10732 10733 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 10734 NCONF_new creates a new CONF object. This works in the same way 10735 as other interfaces in OpenSSL, like the BIO interface. 10736 NCONF_dump_* dump the internal storage of the configuration file, 10737 which is useful for debugging. All other functions take the same 10738 arguments as the old CONF_* functions with the exception of the 10739 first that must be a `CONF *' instead of a `LHASH *'. 10740 10741 To make it easier to use the new classes with the old CONF_* functions, 10742 the function CONF_set_default_method is provided. 10743 [Richard Levitte] 10744 10745 *) Add '-tls1' option to 'openssl ciphers', which was already 10746 mentioned in the documentation but had not been implemented. 10747 (This option is not yet really useful because even the additional 10748 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 10749 [Bodo Moeller] 10750 10751 *) Initial DSO code added into libcrypto for letting OpenSSL (and 10752 OpenSSL-based applications) load shared libraries and bind to 10753 them in a portable way. 10754 [Geoff Thorpe, with contributions from Richard Levitte] 10755 10756 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 10757 10758 *) Make sure _lrotl and _lrotr are only used with MSVC. 10759 10760 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 10761 (the default implementation of RAND_status). 10762 10763 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 10764 to '-clrext' (= clear extensions), as intended and documented. 10765 [Bodo Moeller; inconsistency pointed out by Michael Attili 10766 <attili@amaxo.com>] 10767 10768 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 10769 was larger than the MD block size. 10770 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 10771 10772 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 10773 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 10774 using the passed key: if the passed key was a private key the result 10775 of X509_print(), for example, would be to print out all the private key 10776 components. 10777 [Steve Henson] 10778 10779 *) des_quad_cksum() byte order bug fix. 10780 [Ulf Möller, using the problem description in krb4-0.9.7, where 10781 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 10782 10783 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 10784 discouraged. 10785 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 10786 10787 *) For easily testing in shell scripts whether some command 10788 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 10789 returns with exit code 0 iff no command of the given name is available. 10790 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 10791 the output goes to stdout and nothing is printed to stderr. 10792 Additional arguments are always ignored. 10793 10794 Since for each cipher there is a command of the same name, 10795 the 'no-cipher' compilation switches can be tested this way. 10796 10797 ('openssl no-XXX' is not able to detect pseudo-commands such 10798 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 10799 [Bodo Moeller] 10800 10801 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 10802 [Bodo Moeller] 10803 10804 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 10805 is set; it will be thrown away anyway because each handshake creates 10806 its own key. 10807 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 10808 to parameters -- in previous versions (since OpenSSL 0.9.3) the 10809 'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning 10810 you effectively got SSL_OP_SINGLE_DH_USE when using this macro. 10811 [Bodo Moeller] 10812 10813 *) New s_client option -ign_eof: EOF at stdin is ignored, and 10814 'Q' and 'R' lose their special meanings (quit/renegotiate). 10815 This is part of what -quiet does; unlike -quiet, -ign_eof 10816 does not suppress any output. 10817 [Richard Levitte] 10818 10819 *) Add compatibility options to the purpose and trust code. The 10820 purpose X509_PURPOSE_ANY is "any purpose" which automatically 10821 accepts a certificate or CA, this was the previous behaviour, 10822 with all the associated security issues. 10823 10824 X509_TRUST_COMPAT is the old trust behaviour: only and 10825 automatically trust self signed roots in certificate store. A 10826 new trust setting X509_TRUST_DEFAULT is used to specify that 10827 a purpose has no associated trust setting and it should instead 10828 use the value in the default purpose. 10829 [Steve Henson] 10830 10831 *) Fix the PKCS#8 DSA private key code so it decodes keys again 10832 and fix a memory leak. 10833 [Steve Henson] 10834 10835 *) In util/mkerr.pl (which implements 'make errors'), preserve 10836 reason strings from the previous version of the .c file, as 10837 the default to have only downcase letters (and digits) in 10838 automatically generated reasons codes is not always appropriate. 10839 [Bodo Moeller] 10840 10841 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 10842 using strerror. Previously, ERR_reason_error_string() returned 10843 library names as reason strings for SYSerr; but SYSerr is a special 10844 case where small numbers are errno values, not library numbers. 10845 [Bodo Moeller] 10846 10847 *) Add '-dsaparam' option to 'openssl dhparam' application. This 10848 converts DSA parameters into DH parameters. (When creating parameters, 10849 DSA_generate_parameters is used.) 10850 [Bodo Moeller] 10851 10852 *) Include 'length' (recommended exponent length) in C code generated 10853 by 'openssl dhparam -C'. 10854 [Bodo Moeller] 10855 10856 *) The second argument to set_label in perlasm was already being used 10857 so couldn't be used as a "file scope" flag. Moved to third argument 10858 which was free. 10859 [Steve Henson] 10860 10861 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 10862 instead of RAND_bytes for encryption IVs and salts. 10863 [Bodo Moeller] 10864 10865 *) Include RAND_status() into RAND_METHOD instead of implementing 10866 it only for md_rand.c Otherwise replacing the PRNG by calling 10867 RAND_set_rand_method would be impossible. 10868 [Bodo Moeller] 10869 10870 *) Don't let DSA_generate_key() enter an infinite loop if the random 10871 number generation fails. 10872 [Bodo Moeller] 10873 10874 *) New 'rand' application for creating pseudo-random output. 10875 [Bodo Moeller] 10876 10877 *) Added configuration support for Linux/IA64 10878 [Rolf Haberrecker <rolf@suse.de>] 10879 10880 *) Assembler module support for Mingw32. 10881 [Ulf Möller] 10882 10883 *) Shared library support for HPUX (in shlib/). 10884 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 10885 10886 *) Shared library support for Solaris gcc. 10887 [Lutz Behnke <behnke@trustcenter.de>] 10888 10889 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 10890 10891 *) PKCS7_encrypt() was adding text MIME headers twice because they 10892 were added manually and by SMIME_crlf_copy(). 10893 [Steve Henson] 10894 10895 *) In bntest.c don't call BN_rand with zero bits argument. 10896 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 10897 10898 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 10899 case was implemented. This caused BN_div_recp() to fail occasionally. 10900 [Ulf Möller] 10901 10902 *) Add an optional second argument to the set_label() in the perl 10903 assembly language builder. If this argument exists and is set 10904 to 1 it signals that the assembler should use a symbol whose 10905 scope is the entire file, not just the current function. This 10906 is needed with MASM which uses the format label:: for this scope. 10907 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 10908 10909 *) Change the ASN1 types so they are typedefs by default. Before 10910 almost all types were #define'd to ASN1_STRING which was causing 10911 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 10912 for example. 10913 [Steve Henson] 10914 10915 *) Change names of new functions to the new get1/get0 naming 10916 convention: After 'get1', the caller owns a reference count 10917 and has to call ..._free; 'get0' returns a pointer to some 10918 data structure without incrementing reference counters. 10919 (Some of the existing 'get' functions increment a reference 10920 counter, some don't.) 10921 Similarly, 'set1' and 'add1' functions increase reference 10922 counters or duplicate objects. 10923 [Steve Henson] 10924 10925 *) Allow for the possibility of temp RSA key generation failure: 10926 the code used to assume it always worked and crashed on failure. 10927 [Steve Henson] 10928 10929 *) Fix potential buffer overrun problem in BIO_printf(). 10930 [Ulf Möller, using public domain code by Patrick Powell; problem 10931 pointed out by David Sacerdote <das33@cornell.edu>] 10932 10933 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 10934 RAND_egd() and RAND_status(). In the command line application, 10935 the EGD socket can be specified like a seed file using RANDFILE 10936 or -rand. 10937 [Ulf Möller] 10938 10939 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 10940 Some CAs (e.g. Verisign) distribute certificates in this form. 10941 [Steve Henson] 10942 10943 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 10944 list to exclude them. This means that no special compilation option 10945 is needed to use anonymous DH: it just needs to be included in the 10946 cipher list. 10947 [Steve Henson] 10948 10949 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 10950 EVP_MD_type. The old functionality is available in a new macro called 10951 EVP_MD_md(). Change code that uses it and update docs. 10952 [Steve Henson] 10953 10954 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 10955 where the 'void *' argument is replaced by a function pointer argument. 10956 Previously 'void *' was abused to point to functions, which works on 10957 many platforms, but is not correct. As these functions are usually 10958 called by macros defined in OpenSSL header files, most source code 10959 should work without changes. 10960 [Richard Levitte] 10961 10962 *) <openssl/opensslconf.h> (which is created by Configure) now contains 10963 sections with information on -D... compiler switches used for 10964 compiling the library so that applications can see them. To enable 10965 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 10966 must be defined. E.g., 10967 #define OPENSSL_ALGORITHM_DEFINES 10968 #include <openssl/opensslconf.h> 10969 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 10970 [Richard Levitte, Ulf and Bodo Möller] 10971 10972 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 10973 record layer. 10974 [Bodo Moeller] 10975 10976 *) Change the 'other' type in certificate aux info to a STACK_OF 10977 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 10978 the required ASN1 format: arbitrary types determined by an OID. 10979 [Steve Henson] 10980 10981 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 10982 argument to 'req'. This is not because the function is newer or 10983 better than others it just uses the work 'NEW' in the certificate 10984 request header lines. Some software needs this. 10985 [Steve Henson] 10986 10987 *) Reorganise password command line arguments: now passwords can be 10988 obtained from various sources. Delete the PEM_cb function and make 10989 it the default behaviour: i.e. if the callback is NULL and the 10990 usrdata argument is not NULL interpret it as a null terminated pass 10991 phrase. If usrdata and the callback are NULL then the pass phrase 10992 is prompted for as usual. 10993 [Steve Henson] 10994 10995 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 10996 the support is automatically enabled. The resulting binaries will 10997 autodetect the card and use it if present. 10998 [Ben Laurie and Compaq Inc.] 10999 11000 *) Work around for Netscape hang bug. This sends certificate request 11001 and server done in one record. Since this is perfectly legal in the 11002 SSL/TLS protocol it isn't a "bug" option and is on by default. See 11003 the bugs/SSLv3 entry for more info. 11004 [Steve Henson] 11005 11006 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 11007 [Andy Polyakov] 11008 11009 *) Add -rand argument to smime and pkcs12 applications and read/write 11010 of seed file. 11011 [Steve Henson] 11012 11013 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 11014 [Bodo Moeller] 11015 11016 *) Add command line password options to the remaining applications. 11017 [Steve Henson] 11018 11019 *) Bug fix for BN_div_recp() for numerators with an even number of 11020 bits. 11021 [Ulf Möller] 11022 11023 *) More tests in bntest.c, and changed test_bn output. 11024 [Ulf Möller] 11025 11026 *) ./config recognizes MacOS X now. 11027 [Andy Polyakov] 11028 11029 *) Bug fix for BN_div() when the first words of num and divisor are 11030 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 11031 [Ulf Möller] 11032 11033 *) Add support for various broken PKCS#8 formats, and command line 11034 options to produce them. 11035 [Steve Henson] 11036 11037 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 11038 get temporary BIGNUMs from a BN_CTX. 11039 [Ulf Möller] 11040 11041 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 11042 for p == 0. 11043 [Ulf Möller] 11044 11045 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 11046 include a #define from the old name to the new. The original intent 11047 was that statically linked binaries could for example just call 11048 SSLeay_add_all_ciphers() to just add ciphers to the table and not 11049 link with digests. This never worked because SSLeay_add_all_digests() 11050 and SSLeay_add_all_ciphers() were in the same source file so calling 11051 one would link with the other. They are now in separate source files. 11052 [Steve Henson] 11053 11054 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 11055 [Steve Henson] 11056 11057 *) Use a less unusual form of the Miller-Rabin primality test (it used 11058 a binary algorithm for exponentiation integrated into the Miller-Rabin 11059 loop, our standard modexp algorithms are faster). 11060 [Bodo Moeller] 11061 11062 *) Support for the EBCDIC character set completed. 11063 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 11064 11065 *) Source code cleanups: use const where appropriate, eliminate casts, 11066 use void * instead of char * in lhash. 11067 [Ulf Möller] 11068 11069 *) Bugfix: ssl3_send_server_key_exchange was not restartable 11070 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 11071 this the server could overwrite ephemeral keys that the client 11072 has already seen). 11073 [Bodo Moeller] 11074 11075 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 11076 using 50 iterations of the Rabin-Miller test. 11077 11078 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 11079 iterations of the Rabin-Miller test as required by the appendix 11080 to FIPS PUB 186[-1]) instead of DSA_is_prime. 11081 As BN_is_prime_fasttest includes trial division, DSA parameter 11082 generation becomes much faster. 11083 11084 This implies a change for the callback functions in DSA_is_prime 11085 and DSA_generate_parameters: The callback function is called once 11086 for each positive witness in the Rabin-Miller test, not just 11087 occasionally in the inner loop; and the parameters to the 11088 callback function now provide an iteration count for the outer 11089 loop rather than for the current invocation of the inner loop. 11090 DSA_generate_parameters additionally can call the callback 11091 function with an 'iteration count' of -1, meaning that a 11092 candidate has passed the trial division test (when q is generated 11093 from an application-provided seed, trial division is skipped). 11094 [Bodo Moeller] 11095 11096 *) New function BN_is_prime_fasttest that optionally does trial 11097 division before starting the Rabin-Miller test and has 11098 an additional BN_CTX * argument (whereas BN_is_prime always 11099 has to allocate at least one BN_CTX). 11100 'callback(1, -1, cb_arg)' is called when a number has passed the 11101 trial division stage. 11102 [Bodo Moeller] 11103 11104 *) Fix for bug in CRL encoding. The validity dates weren't being handled 11105 as ASN1_TIME. 11106 [Steve Henson] 11107 11108 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 11109 [Steve Henson] 11110 11111 *) New function BN_pseudo_rand(). 11112 [Ulf Möller] 11113 11114 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 11115 bignum version of BN_from_montgomery() with the working code from 11116 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 11117 the comments. 11118 [Ulf Möller] 11119 11120 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 11121 made it impossible to use the same SSL_SESSION data structure in 11122 SSL2 clients in multiple threads. 11123 [Bodo Moeller] 11124 11125 *) The return value of RAND_load_file() no longer counts bytes obtained 11126 by stat(). RAND_load_file(..., -1) is new and uses the complete file 11127 to seed the PRNG (previously an explicit byte count was required). 11128 [Ulf Möller, Bodo Möller] 11129 11130 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 11131 used (char *) instead of (void *) and had casts all over the place. 11132 [Steve Henson] 11133 11134 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 11135 [Ulf Möller] 11136 11137 *) Retain source code compatibility for BN_prime_checks macro: 11138 BN_is_prime(..., BN_prime_checks, ...) now uses 11139 BN_prime_checks_for_size to determine the appropriate number of 11140 Rabin-Miller iterations. 11141 [Ulf Möller] 11142 11143 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 11144 DH_CHECK_P_NOT_SAFE_PRIME. 11145 (Check if this is true? OpenPGP calls them "strong".) 11146 [Ulf Möller] 11147 11148 *) Merge the functionality of "dh" and "gendh" programs into a new program 11149 "dhparam". The old programs are retained for now but will handle DH keys 11150 (instead of parameters) in future. 11151 [Steve Henson] 11152 11153 *) Make the ciphers, s_server and s_client programs check the return values 11154 when a new cipher list is set. 11155 [Steve Henson] 11156 11157 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 11158 ciphers. Before when the 56bit ciphers were enabled the sorting was 11159 wrong. 11160 11161 The syntax for the cipher sorting has been extended to support sorting by 11162 cipher-strength (using the strength_bits hard coded in the tables). 11163 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 11164 11165 Fix a bug in the cipher-command parser: when supplying a cipher command 11166 string with an "undefined" symbol (neither command nor alphanumeric 11167 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 11168 an error is flagged. 11169 11170 Due to the strength-sorting extension, the code of the 11171 ssl_create_cipher_list() function was completely rearranged. I hope that 11172 the readability was also increased :-) 11173 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 11174 11175 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 11176 for the first serial number and places 2 in the serial number file. This 11177 avoids problems when the root CA is created with serial number zero and 11178 the first user certificate has the same issuer name and serial number 11179 as the root CA. 11180 [Steve Henson] 11181 11182 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 11183 the new code. Add documentation for this stuff. 11184 [Steve Henson] 11185 11186 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 11187 X509_*() to X509at_*() on the grounds that they don't handle X509 11188 structures and behave in an analogous way to the X509v3 functions: 11189 they shouldn't be called directly but wrapper functions should be used 11190 instead. 11191 11192 So we also now have some wrapper functions that call the X509at functions 11193 when passed certificate requests. (TO DO: similar things can be done with 11194 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 11195 things. Some of these need some d2i or i2d and print functionality 11196 because they handle more complex structures.) 11197 [Steve Henson] 11198 11199 *) Add missing #ifndefs that caused missing symbols when building libssl 11200 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 11201 NO_RSA in ssl/s2*.c. 11202 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller] 11203 11204 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 11205 has a return value which indicates the quality of the random data 11206 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 11207 error queue. New function RAND_pseudo_bytes() generates output that is 11208 guaranteed to be unique but not unpredictable. RAND_add is like 11209 RAND_seed, but takes an extra argument for an entropy estimate 11210 (RAND_seed always assumes full entropy). 11211 [Ulf Möller] 11212 11213 *) Do more iterations of Rabin-Miller probable prime test (specifically, 11214 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 11215 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 11216 in crypto/bn/bn_prime.c for the complete table). This guarantees a 11217 false-positive rate of at most 2^-80 for random input. 11218 [Bodo Moeller] 11219 11220 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 11221 [Bodo Moeller] 11222 11223 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 11224 in the 0.9.5 release), this returns the chain 11225 from an X509_CTX structure with a dup of the stack and all 11226 the X509 reference counts upped: so the stack will exist 11227 after X509_CTX_cleanup() has been called. Modify pkcs12.c 11228 to use this. 11229 11230 Also make SSL_SESSION_print() print out the verify return 11231 code. 11232 [Steve Henson] 11233 11234 *) Add manpage for the pkcs12 command. Also change the default 11235 behaviour so MAC iteration counts are used unless the new 11236 -nomaciter option is used. This improves file security and 11237 only older versions of MSIE (4.0 for example) need it. 11238 [Steve Henson] 11239 11240 *) Honor the no-xxx Configure options when creating .DEF files. 11241 [Ulf Möller] 11242 11243 *) Add PKCS#10 attributes to field table: challengePassword, 11244 unstructuredName and unstructuredAddress. These are taken from 11245 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 11246 international characters are used. 11247 11248 More changes to X509_ATTRIBUTE code: allow the setting of types 11249 based on strings. Remove the 'loc' parameter when adding 11250 attributes because these will be a SET OF encoding which is sorted 11251 in ASN1 order. 11252 [Steve Henson] 11253 11254 *) Initial changes to the 'req' utility to allow request generation 11255 automation. This will allow an application to just generate a template 11256 file containing all the field values and have req construct the 11257 request. 11258 11259 Initial support for X509_ATTRIBUTE handling. Stacks of these are 11260 used all over the place including certificate requests and PKCS#7 11261 structures. They are currently handled manually where necessary with 11262 some primitive wrappers for PKCS#7. The new functions behave in a 11263 manner analogous to the X509 extension functions: they allow 11264 attributes to be looked up by NID and added. 11265 11266 Later something similar to the X509V3 code would be desirable to 11267 automatically handle the encoding, decoding and printing of the 11268 more complex types. The string types like challengePassword can 11269 be handled by the string table functions. 11270 11271 Also modified the multi byte string table handling. Now there is 11272 a 'global mask' which masks out certain types. The table itself 11273 can use the flag STABLE_NO_MASK to ignore the mask setting: this 11274 is useful when for example there is only one permissible type 11275 (as in countryName) and using the mask might result in no valid 11276 types at all. 11277 [Steve Henson] 11278 11279 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 11280 SSL_get_peer_finished to allow applications to obtain the latest 11281 Finished messages sent to the peer or expected from the peer, 11282 respectively. (SSL_get_peer_finished is usually the Finished message 11283 actually received from the peer, otherwise the protocol will be aborted.) 11284 11285 As the Finished message are message digests of the complete handshake 11286 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 11287 be used for external authentication procedures when the authentication 11288 provided by SSL/TLS is not desired or is not enough. 11289 [Bodo Moeller] 11290 11291 *) Enhanced support for Alpha Linux is added. Now ./config checks if 11292 the host supports BWX extension and if Compaq C is present on the 11293 $PATH. Just exploiting of the BWX extension results in 20-30% 11294 performance kick for some algorithms, e.g. DES and RC4 to mention 11295 a couple. Compaq C in turn generates ~20% faster code for MD5 and 11296 SHA1. 11297 [Andy Polyakov] 11298 11299 *) Add support for MS "fast SGC". This is arguably a violation of the 11300 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 11301 weak crypto and after checking the certificate is SGC a second one 11302 with strong crypto. MS SGC stops the first handshake after receiving 11303 the server certificate message and sends a second client hello. Since 11304 a server will typically do all the time consuming operations before 11305 expecting any further messages from the client (server key exchange 11306 is the most expensive) there is little difference between the two. 11307 11308 To get OpenSSL to support MS SGC we have to permit a second client 11309 hello message after we have sent server done. In addition we have to 11310 reset the MAC if we do get this second client hello. 11311 [Steve Henson] 11312 11313 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 11314 if a DER encoded private key is RSA or DSA traditional format. Changed 11315 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 11316 format DER encoded private key. Newer code should use PKCS#8 format which 11317 has the key type encoded in the ASN1 structure. Added DER private key 11318 support to pkcs8 application. 11319 [Steve Henson] 11320 11321 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 11322 ciphersuites has been selected (as required by the SSL 3/TLS 1 11323 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 11324 is set, we interpret this as a request to violate the specification 11325 (the worst that can happen is a handshake failure, and 'correct' 11326 behaviour would result in a handshake failure anyway). 11327 [Bodo Moeller] 11328 11329 *) In SSL_CTX_add_session, take into account that there might be multiple 11330 SSL_SESSION structures with the same session ID (e.g. when two threads 11331 concurrently obtain them from an external cache). 11332 The internal cache can handle only one SSL_SESSION with a given ID, 11333 so if there's a conflict, we now throw out the old one to achieve 11334 consistency. 11335 [Bodo Moeller] 11336 11337 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 11338 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 11339 some routines that use cipher OIDs: some ciphers do not have OIDs 11340 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 11341 example. 11342 [Steve Henson] 11343 11344 *) Simplify the trust setting structure and code. Now we just have 11345 two sequences of OIDs for trusted and rejected settings. These will 11346 typically have values the same as the extended key usage extension 11347 and any application specific purposes. 11348 11349 The trust checking code now has a default behaviour: it will just 11350 check for an object with the same NID as the passed id. Functions can 11351 be provided to override either the default behaviour or the behaviour 11352 for a given id. SSL client, server and email already have functions 11353 in place for compatibility: they check the NID and also return "trusted" 11354 if the certificate is self signed. 11355 [Steve Henson] 11356 11357 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 11358 traditional format into an EVP_PKEY structure. 11359 [Steve Henson] 11360 11361 *) Add a password callback function PEM_cb() which either prompts for 11362 a password if usr_data is NULL or otherwise assumes it is a null 11363 terminated password. Allow passwords to be passed on command line 11364 environment or config files in a few more utilities. 11365 [Steve Henson] 11366 11367 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 11368 keys. Add some short names for PKCS#8 PBE algorithms and allow them 11369 to be specified on the command line for the pkcs8 and pkcs12 utilities. 11370 Update documentation. 11371 [Steve Henson] 11372 11373 *) Support for ASN1 "NULL" type. This could be handled before by using 11374 ASN1_TYPE but there wasn't any function that would try to read a NULL 11375 and produce an error if it couldn't. For compatibility we also have 11376 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 11377 don't allocate anything because they don't need to. 11378 [Steve Henson] 11379 11380 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 11381 for details. 11382 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 11383 11384 *) Rebuild of the memory allocation routines used by OpenSSL code and 11385 possibly others as well. The purpose is to make an interface that 11386 provide hooks so anyone can build a separate set of allocation and 11387 deallocation routines to be used by OpenSSL, for example memory 11388 pool implementations, or something else, which was previously hard 11389 since Malloc(), Realloc() and Free() were defined as macros having 11390 the values malloc, realloc and free, respectively (except for Win32 11391 compilations). The same is provided for memory debugging code. 11392 OpenSSL already comes with functionality to find memory leaks, but 11393 this gives people a chance to debug other memory problems. 11394 11395 With these changes, a new set of functions and macros have appeared: 11396 11397 CRYPTO_set_mem_debug_functions() [F] 11398 CRYPTO_get_mem_debug_functions() [F] 11399 CRYPTO_dbg_set_options() [F] 11400 CRYPTO_dbg_get_options() [F] 11401 CRYPTO_malloc_debug_init() [M] 11402 11403 The memory debug functions are NULL by default, unless the library 11404 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 11405 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 11406 gives the standard debugging functions that come with OpenSSL) or 11407 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 11408 provided by the library user) must be used. When the standard 11409 debugging functions are used, CRYPTO_dbg_set_options can be used to 11410 request additional information: 11411 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 11412 the CRYPTO_MDEBUG_xxx macro when compiling the library. 11413 11414 Also, things like CRYPTO_set_mem_functions will always give the 11415 expected result (the new set of functions is used for allocation 11416 and deallocation) at all times, regardless of platform and compiler 11417 options. 11418 11419 To finish it up, some functions that were never use in any other 11420 way than through macros have a new API and new semantic: 11421 11422 CRYPTO_dbg_malloc() 11423 CRYPTO_dbg_realloc() 11424 CRYPTO_dbg_free() 11425 11426 All macros of value have retained their old syntax. 11427 [Richard Levitte and Bodo Moeller] 11428 11429 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 11430 ordering of SMIMECapabilities wasn't in "strength order" and there 11431 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 11432 algorithm. 11433 [Steve Henson] 11434 11435 *) Some ASN1 types with illegal zero length encoding (INTEGER, 11436 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 11437 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 11438 11439 *) Merge in my S/MIME library for OpenSSL. This provides a simple 11440 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 11441 functionality to handle multipart/signed properly) and a utility 11442 called 'smime' to call all this stuff. This is based on code I 11443 originally wrote for Celo who have kindly allowed it to be 11444 included in OpenSSL. 11445 [Steve Henson] 11446 11447 *) Add variants des_set_key_checked and des_set_key_unchecked of 11448 des_set_key (aka des_key_sched). Global variable des_check_key 11449 decides which of these is called by des_set_key; this way 11450 des_check_key behaves as it always did, but applications and 11451 the library itself, which was buggy for des_check_key == 1, 11452 have a cleaner way to pick the version they need. 11453 [Bodo Moeller] 11454 11455 *) New function PKCS12_newpass() which changes the password of a 11456 PKCS12 structure. 11457 [Steve Henson] 11458 11459 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 11460 dynamic mix. In both cases the ids can be used as an index into the 11461 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 11462 functions so they accept a list of the field values and the 11463 application doesn't need to directly manipulate the X509_TRUST 11464 structure. 11465 [Steve Henson] 11466 11467 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 11468 need initialising. 11469 [Steve Henson] 11470 11471 *) Modify the way the V3 extension code looks up extensions. This now 11472 works in a similar way to the object code: we have some "standard" 11473 extensions in a static table which is searched with OBJ_bsearch() 11474 and the application can add dynamic ones if needed. The file 11475 crypto/x509v3/ext_dat.h now has the info: this file needs to be 11476 updated whenever a new extension is added to the core code and kept 11477 in ext_nid order. There is a simple program 'tabtest.c' which checks 11478 this. New extensions are not added too often so this file can readily 11479 be maintained manually. 11480 11481 There are two big advantages in doing things this way. The extensions 11482 can be looked up immediately and no longer need to be "added" using 11483 X509V3_add_standard_extensions(): this function now does nothing. 11484 [Side note: I get *lots* of email saying the extension code doesn't 11485 work because people forget to call this function] 11486 Also no dynamic allocation is done unless new extensions are added: 11487 so if we don't add custom extensions there is no need to call 11488 X509V3_EXT_cleanup(). 11489 [Steve Henson] 11490 11491 *) Modify enc utility's salting as follows: make salting the default. Add a 11492 magic header, so unsalted files fail gracefully instead of just decrypting 11493 to garbage. This is because not salting is a big security hole, so people 11494 should be discouraged from doing it. 11495 [Ben Laurie] 11496 11497 *) Fixes and enhancements to the 'x509' utility. It allowed a message 11498 digest to be passed on the command line but it only used this 11499 parameter when signing a certificate. Modified so all relevant 11500 operations are affected by the digest parameter including the 11501 -fingerprint and -x509toreq options. Also -x509toreq choked if a 11502 DSA key was used because it didn't fix the digest. 11503 [Steve Henson] 11504 11505 *) Initial certificate chain verify code. Currently tests the untrusted 11506 certificates for consistency with the verify purpose (which is set 11507 when the X509_STORE_CTX structure is set up) and checks the pathlength. 11508 11509 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 11510 this is because it will reject chains with invalid extensions whereas 11511 every previous version of OpenSSL and SSLeay made no checks at all. 11512 11513 Trust code: checks the root CA for the relevant trust settings. Trust 11514 settings have an initial value consistent with the verify purpose: e.g. 11515 if the verify purpose is for SSL client use it expects the CA to be 11516 trusted for SSL client use. However the default value can be changed to 11517 permit custom trust settings: one example of this would be to only trust 11518 certificates from a specific "secure" set of CAs. 11519 11520 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 11521 which should be used for version portability: especially since the 11522 verify structure is likely to change more often now. 11523 11524 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 11525 to set them. If not set then assume SSL clients will verify SSL servers 11526 and vice versa. 11527 11528 Two new options to the verify program: -untrusted allows a set of 11529 untrusted certificates to be passed in and -purpose which sets the 11530 intended purpose of the certificate. If a purpose is set then the 11531 new chain verify code is used to check extension consistency. 11532 [Steve Henson] 11533 11534 *) Support for the authority information access extension. 11535 [Steve Henson] 11536 11537 *) Modify RSA and DSA PEM read routines to transparently handle 11538 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 11539 public keys in a format compatible with certificate 11540 SubjectPublicKeyInfo structures. Unfortunately there were already 11541 functions called *_PublicKey_* which used various odd formats so 11542 these are retained for compatibility: however the DSA variants were 11543 never in a public release so they have been deleted. Changed dsa/rsa 11544 utilities to handle the new format: note no releases ever handled public 11545 keys so we should be OK. 11546 11547 The primary motivation for this change is to avoid the same fiasco 11548 that dogs private keys: there are several incompatible private key 11549 formats some of which are standard and some OpenSSL specific and 11550 require various evil hacks to allow partial transparent handling and 11551 even then it doesn't work with DER formats. Given the option anything 11552 other than PKCS#8 should be dumped: but the other formats have to 11553 stay in the name of compatibility. 11554 11555 With public keys and the benefit of hindsight one standard format 11556 is used which works with EVP_PKEY, RSA or DSA structures: though 11557 it clearly returns an error if you try to read the wrong kind of key. 11558 11559 Added a -pubkey option to the 'x509' utility to output the public key. 11560 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 11561 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 11562 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 11563 that do the same as the EVP_PKEY_assign_*() except they up the 11564 reference count of the added key (they don't "swallow" the 11565 supplied key). 11566 [Steve Henson] 11567 11568 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 11569 CRLs would fail if the file contained no certificates or no CRLs: 11570 added a new function to read in both types and return the number 11571 read: this means that if none are read it will be an error. The 11572 DER versions of the certificate and CRL reader would always fail 11573 because it isn't possible to mix certificates and CRLs in DER format 11574 without choking one or the other routine. Changed this to just read 11575 a certificate: this is the best we can do. Also modified the code 11576 in apps/verify.c to take notice of return codes: it was previously 11577 attempting to read in certificates from NULL pointers and ignoring 11578 any errors: this is one reason why the cert and CRL reader seemed 11579 to work. It doesn't check return codes from the default certificate 11580 routines: these may well fail if the certificates aren't installed. 11581 [Steve Henson] 11582 11583 *) Code to support otherName option in GeneralName. 11584 [Steve Henson] 11585 11586 *) First update to verify code. Change the verify utility 11587 so it warns if it is passed a self signed certificate: 11588 for consistency with the normal behaviour. X509_verify 11589 has been modified to it will now verify a self signed 11590 certificate if *exactly* the same certificate appears 11591 in the store: it was previously impossible to trust a 11592 single self signed certificate. This means that: 11593 openssl verify ss.pem 11594 now gives a warning about a self signed certificate but 11595 openssl verify -CAfile ss.pem ss.pem 11596 is OK. 11597 [Steve Henson] 11598 11599 *) For servers, store verify_result in SSL_SESSION data structure 11600 (and add it to external session representation). 11601 This is needed when client certificate verifications fails, 11602 but an application-provided verification callback (set by 11603 SSL_CTX_set_cert_verify_callback) allows accepting the session 11604 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 11605 but returns 1): When the session is reused, we have to set 11606 ssl->verify_result to the appropriate error code to avoid 11607 security holes. 11608 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 11609 11610 *) Fix a bug in the new PKCS#7 code: it didn't consider the 11611 case in PKCS7_dataInit() where the signed PKCS7 structure 11612 didn't contain any existing data because it was being created. 11613 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 11614 11615 *) Add a salt to the key derivation routines in enc.c. This 11616 forms the first 8 bytes of the encrypted file. Also add a 11617 -S option to allow a salt to be input on the command line. 11618 [Steve Henson] 11619 11620 *) New function X509_cmp(). Oddly enough there wasn't a function 11621 to compare two certificates. We do this by working out the SHA1 11622 hash and comparing that. X509_cmp() will be needed by the trust 11623 code. 11624 [Steve Henson] 11625 11626 *) SSL_get1_session() is like SSL_get_session(), but increments 11627 the reference count in the SSL_SESSION returned. 11628 [Geoff Thorpe <geoff@eu.c2.net>] 11629 11630 *) Fix for 'req': it was adding a null to request attributes. 11631 Also change the X509_LOOKUP and X509_INFO code to handle 11632 certificate auxiliary information. 11633 [Steve Henson] 11634 11635 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 11636 the 'enc' command. 11637 [Steve Henson] 11638 11639 *) Add the possibility to add extra information to the memory leak 11640 detecting output, to form tracebacks, showing from where each 11641 allocation was originated: CRYPTO_push_info("constant string") adds 11642 the string plus current file name and line number to a per-thread 11643 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 11644 is like calling CYRPTO_pop_info() until the stack is empty. 11645 Also updated memory leak detection code to be multi-thread-safe. 11646 [Richard Levitte] 11647 11648 *) Add options -text and -noout to pkcs7 utility and delete the 11649 encryption options which never did anything. Update docs. 11650 [Steve Henson] 11651 11652 *) Add options to some of the utilities to allow the pass phrase 11653 to be included on either the command line (not recommended on 11654 OSes like Unix) or read from the environment. Update the 11655 manpages and fix a few bugs. 11656 [Steve Henson] 11657 11658 *) Add a few manpages for some of the openssl commands. 11659 [Steve Henson] 11660 11661 *) Fix the -revoke option in ca. It was freeing up memory twice, 11662 leaking and not finding already revoked certificates. 11663 [Steve Henson] 11664 11665 *) Extensive changes to support certificate auxiliary information. 11666 This involves the use of X509_CERT_AUX structure and X509_AUX 11667 functions. An X509_AUX function such as PEM_read_X509_AUX() 11668 can still read in a certificate file in the usual way but it 11669 will also read in any additional "auxiliary information". By 11670 doing things this way a fair degree of compatibility can be 11671 retained: existing certificates can have this information added 11672 using the new 'x509' options. 11673 11674 Current auxiliary information includes an "alias" and some trust 11675 settings. The trust settings will ultimately be used in enhanced 11676 certificate chain verification routines: currently a certificate 11677 can only be trusted if it is self signed and then it is trusted 11678 for all purposes. 11679 [Steve Henson] 11680 11681 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 11682 The problem was that one of the replacement routines had not been working 11683 since SSLeay releases. For now the offending routine has been replaced 11684 with non-optimised assembler. Even so, this now gives around 95% 11685 performance improvement for 1024 bit RSA signs. 11686 [Mark Cox] 11687 11688 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 11689 handling. Most clients have the effective key size in bits equal to 11690 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 11691 A few however don't do this and instead use the size of the decrypted key 11692 to determine the RC2 key length and the AlgorithmIdentifier to determine 11693 the effective key length. In this case the effective key length can still 11694 be 40 bits but the key length can be 168 bits for example. This is fixed 11695 by manually forcing an RC2 key into the EVP_PKEY structure because the 11696 EVP code can't currently handle unusual RC2 key sizes: it always assumes 11697 the key length and effective key length are equal. 11698 [Steve Henson] 11699 11700 *) Add a bunch of functions that should simplify the creation of 11701 X509_NAME structures. Now you should be able to do: 11702 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 11703 and have it automatically work out the correct field type and fill in 11704 the structures. The more adventurous can try: 11705 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 11706 and it will (hopefully) work out the correct multibyte encoding. 11707 [Steve Henson] 11708 11709 *) Change the 'req' utility to use the new field handling and multibyte 11710 copy routines. Before the DN field creation was handled in an ad hoc 11711 way in req, ca, and x509 which was rather broken and didn't support 11712 BMPStrings or UTF8Strings. Since some software doesn't implement 11713 BMPStrings or UTF8Strings yet, they can be enabled using the config file 11714 using the dirstring_type option. See the new comment in the default 11715 openssl.cnf for more info. 11716 [Steve Henson] 11717 11718 *) Make crypto/rand/md_rand.c more robust: 11719 - Assure unique random numbers after fork(). 11720 - Make sure that concurrent threads access the global counter and 11721 md serializably so that we never lose entropy in them 11722 or use exactly the same state in multiple threads. 11723 Access to the large state is not always serializable because 11724 the additional locking could be a performance killer, and 11725 md should be large enough anyway. 11726 [Bodo Moeller] 11727 11728 *) New file apps/app_rand.c with commonly needed functionality 11729 for handling the random seed file. 11730 11731 Use the random seed file in some applications that previously did not: 11732 ca, 11733 dsaparam -genkey (which also ignored its '-rand' option), 11734 s_client, 11735 s_server, 11736 x509 (when signing). 11737 Except on systems with /dev/urandom, it is crucial to have a random 11738 seed file at least for key creation, DSA signing, and for DH exchanges; 11739 for RSA signatures we could do without one. 11740 11741 gendh and gendsa (unlike genrsa) used to read only the first byte 11742 of each file listed in the '-rand' option. The function as previously 11743 found in genrsa is now in app_rand.c and is used by all programs 11744 that support '-rand'. 11745 [Bodo Moeller] 11746 11747 *) In RAND_write_file, use mode 0600 for creating files; 11748 don't just chmod when it may be too late. 11749 [Bodo Moeller] 11750 11751 *) Report an error from X509_STORE_load_locations 11752 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 11753 [Bill Perry] 11754 11755 *) New function ASN1_mbstring_copy() this copies a string in either 11756 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 11757 into an ASN1_STRING type. A mask of permissible types is passed 11758 and it chooses the "minimal" type to use or an error if not type 11759 is suitable. 11760 [Steve Henson] 11761 11762 *) Add function equivalents to the various macros in asn1.h. The old 11763 macros are retained with an M_ prefix. Code inside the library can 11764 use the M_ macros. External code (including the openssl utility) 11765 should *NOT* in order to be "shared library friendly". 11766 [Steve Henson] 11767 11768 *) Add various functions that can check a certificate's extensions 11769 to see if it usable for various purposes such as SSL client, 11770 server or S/MIME and CAs of these types. This is currently 11771 VERY EXPERIMENTAL but will ultimately be used for certificate chain 11772 verification. Also added a -purpose flag to x509 utility to 11773 print out all the purposes. 11774 [Steve Henson] 11775 11776 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 11777 functions. 11778 [Steve Henson] 11779 11780 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 11781 for, obtain and decode and extension and obtain its critical flag. 11782 This allows all the necessary extension code to be handled in a 11783 single function call. 11784 [Steve Henson] 11785 11786 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 11787 platforms. See crypto/rc4/rc4_enc.c for further details. 11788 [Andy Polyakov] 11789 11790 *) New -noout option to asn1parse. This causes no output to be produced 11791 its main use is when combined with -strparse and -out to extract data 11792 from a file (which may not be in ASN.1 format). 11793 [Steve Henson] 11794 11795 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 11796 when producing the local key id. 11797 [Richard Levitte <levitte@stacken.kth.se>] 11798 11799 *) New option -dhparam in s_server. This allows a DH parameter file to be 11800 stated explicitly. If it is not stated then it tries the first server 11801 certificate file. The previous behaviour hard coded the filename 11802 "server.pem". 11803 [Steve Henson] 11804 11805 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 11806 a public key to be input or output. For example: 11807 openssl rsa -in key.pem -pubout -out pubkey.pem 11808 Also added necessary DSA public key functions to handle this. 11809 [Steve Henson] 11810 11811 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 11812 in the message. This was handled by allowing 11813 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 11814 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 11815 11816 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 11817 to the end of the strings whereas this didn't. This would cause problems 11818 if strings read with d2i_ASN1_bytes() were later modified. 11819 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 11820 11821 *) Fix for base64 decode bug. When a base64 bio reads only one line of 11822 data and it contains EOF it will end up returning an error. This is 11823 caused by input 46 bytes long. The cause is due to the way base64 11824 BIOs find the start of base64 encoded data. They do this by trying a 11825 trial decode on each line until they find one that works. When they 11826 do a flag is set and it starts again knowing it can pass all the 11827 data directly through the decoder. Unfortunately it doesn't reset 11828 the context it uses. This means that if EOF is reached an attempt 11829 is made to pass two EOFs through the context and this causes the 11830 resulting error. This can also cause other problems as well. As is 11831 usual with these problems it takes *ages* to find and the fix is 11832 trivial: move one line. 11833 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 11834 11835 *) Ugly workaround to get s_client and s_server working under Windows. The 11836 old code wouldn't work because it needed to select() on sockets and the 11837 tty (for keypresses and to see if data could be written). Win32 only 11838 supports select() on sockets so we select() with a 1s timeout on the 11839 sockets and then see if any characters are waiting to be read, if none 11840 are present then we retry, we also assume we can always write data to 11841 the tty. This isn't nice because the code then blocks until we've 11842 received a complete line of data and it is effectively polling the 11843 keyboard at 1s intervals: however it's quite a bit better than not 11844 working at all :-) A dedicated Windows application might handle this 11845 with an event loop for example. 11846 [Steve Henson] 11847 11848 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 11849 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 11850 will be called when RSA_sign() and RSA_verify() are used. This is useful 11851 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 11852 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 11853 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 11854 This necessitated the support of an extra signature type NID_md5_sha1 11855 for SSL signatures and modifications to the SSL library to use it instead 11856 of calling RSA_public_decrypt() and RSA_private_encrypt(). 11857 [Steve Henson] 11858 11859 *) Add new -verify -CAfile and -CApath options to the crl program, these 11860 will lookup a CRL issuers certificate and verify the signature in a 11861 similar way to the verify program. Tidy up the crl program so it 11862 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 11863 less strict. It will now permit CRL extensions even if it is not 11864 a V2 CRL: this will allow it to tolerate some broken CRLs. 11865 [Steve Henson] 11866 11867 *) Initialize all non-automatic variables each time one of the openssl 11868 sub-programs is started (this is necessary as they may be started 11869 multiple times from the "OpenSSL>" prompt). 11870 [Lennart Bang, Bodo Moeller] 11871 11872 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 11873 removing all other RSA functionality (this is what NO_RSA does). This 11874 is so (for example) those in the US can disable those operations covered 11875 by the RSA patent while allowing storage and parsing of RSA keys and RSA 11876 key generation. 11877 [Steve Henson] 11878 11879 *) Non-copying interface to BIO pairs. 11880 (still largely untested) 11881 [Bodo Moeller] 11882 11883 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 11884 ASCII string. This was handled independently in various places before. 11885 [Steve Henson] 11886 11887 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 11888 UTF8 strings a character at a time. 11889 [Steve Henson] 11890 11891 *) Use client_version from client hello to select the protocol 11892 (s23_srvr.c) and for RSA client key exchange verification 11893 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 11894 [Bodo Moeller] 11895 11896 *) Add various utility functions to handle SPKACs, these were previously 11897 handled by poking round in the structure internals. Added new function 11898 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 11899 print, verify and generate SPKACs. Based on an original idea from 11900 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 11901 [Steve Henson] 11902 11903 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 11904 [Andy Polyakov] 11905 11906 *) Allow the config file extension section to be overwritten on the 11907 command line. Based on an original idea from Massimiliano Pala 11908 <madwolf@comune.modena.it>. The new option is called -extensions 11909 and can be applied to ca, req and x509. Also -reqexts to override 11910 the request extensions in req and -crlexts to override the crl extensions 11911 in ca. 11912 [Steve Henson] 11913 11914 *) Add new feature to the SPKAC handling in ca. Now you can include 11915 the same field multiple times by preceding it by "XXXX." for example: 11916 1.OU="Unit name 1" 11917 2.OU="Unit name 2" 11918 this is the same syntax as used in the req config file. 11919 [Steve Henson] 11920 11921 *) Allow certificate extensions to be added to certificate requests. These 11922 are specified in a 'req_extensions' option of the req section of the 11923 config file. They can be printed out with the -text option to req but 11924 are otherwise ignored at present. 11925 [Steve Henson] 11926 11927 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 11928 data read consists of only the final block it would not decrypted because 11929 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 11930 A misplaced 'break' also meant the decrypted final block might not be 11931 copied until the next read. 11932 [Steve Henson] 11933 11934 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 11935 a few extra parameters to the DH structure: these will be useful if 11936 for example we want the value of 'q' or implement X9.42 DH. 11937 [Steve Henson] 11938 11939 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 11940 provides hooks that allow the default DSA functions or functions on a 11941 "per key" basis to be replaced. This allows hardware acceleration and 11942 hardware key storage to be handled without major modification to the 11943 library. Also added low level modexp hooks and CRYPTO_EX structure and 11944 associated functions. 11945 [Steve Henson] 11946 11947 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 11948 as "read only": it can't be written to and the buffer it points to will 11949 not be freed. Reading from a read only BIO is much more efficient than 11950 a normal memory BIO. This was added because there are several times when 11951 an area of memory needs to be read from a BIO. The previous method was 11952 to create a memory BIO and write the data to it, this results in two 11953 copies of the data and an O(n^2) reading algorithm. There is a new 11954 function BIO_new_mem_buf() which creates a read only memory BIO from 11955 an area of memory. Also modified the PKCS#7 routines to use read only 11956 memory BIOs. 11957 [Steve Henson] 11958 11959 *) Bugfix: ssl23_get_client_hello did not work properly when called in 11960 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 11961 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 11962 but a retry condition occurred while trying to read the rest. 11963 [Bodo Moeller] 11964 11965 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 11966 NID_pkcs7_encrypted by default: this was wrong since this should almost 11967 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 11968 the encrypted data type: this is a more sensible place to put it and it 11969 allows the PKCS#12 code to be tidied up that duplicated this 11970 functionality. 11971 [Steve Henson] 11972 11973 *) Changed obj_dat.pl script so it takes its input and output files on 11974 the command line. This should avoid shell escape redirection problems 11975 under Win32. 11976 [Steve Henson] 11977 11978 *) Initial support for certificate extension requests, these are included 11979 in things like Xenroll certificate requests. Included functions to allow 11980 extensions to be obtained and added. 11981 [Steve Henson] 11982 11983 *) -crlf option to s_client and s_server for sending newlines as 11984 CRLF (as required by many protocols). 11985 [Bodo Moeller] 11986 11987 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 11988 11989 *) Install libRSAglue.a when OpenSSL is built with RSAref. 11990 [Ralf S. Engelschall] 11991 11992 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 11993 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 11994 11995 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 11996 program. 11997 [Steve Henson] 11998 11999 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 12000 DH parameters/keys (q is lost during that conversion, but the resulting 12001 DH parameters contain its length). 12002 12003 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 12004 much faster than DH_generate_parameters (which creates parameters 12005 where p = 2*q + 1), and also the smaller q makes DH computations 12006 much more efficient (160-bit exponentiation instead of 1024-bit 12007 exponentiation); so this provides a convenient way to support DHE 12008 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 12009 utter importance to use 12010 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 12011 or 12012 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 12013 when such DH parameters are used, because otherwise small subgroup 12014 attacks may become possible! 12015 [Bodo Moeller] 12016 12017 *) Avoid memory leak in i2d_DHparams. 12018 [Bodo Moeller] 12019 12020 *) Allow the -k option to be used more than once in the enc program: 12021 this allows the same encrypted message to be read by multiple recipients. 12022 [Steve Henson] 12023 12024 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 12025 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 12026 it will always use the numerical form of the OID, even if it has a short 12027 or long name. 12028 [Steve Henson] 12029 12030 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 12031 method only got called if p,q,dmp1,dmq1,iqmp components were present, 12032 otherwise bn_mod_exp was called. In the case of hardware keys for example 12033 no private key components need be present and it might store extra data 12034 in the RSA structure, which cannot be accessed from bn_mod_exp. 12035 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 12036 private key operations. 12037 [Steve Henson] 12038 12039 *) Added support for SPARC Linux. 12040 [Andy Polyakov] 12041 12042 *) pem_password_cb function type incompatibly changed from 12043 typedef int pem_password_cb(char *buf, int size, int rwflag); 12044 to 12045 ....(char *buf, int size, int rwflag, void *userdata); 12046 so that applications can pass data to their callbacks: 12047 The PEM[_ASN1]_{read,write}... functions and macros now take an 12048 additional void * argument, which is just handed through whenever 12049 the password callback is called. 12050 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 12051 12052 New function SSL_CTX_set_default_passwd_cb_userdata. 12053 12054 Compatibility note: As many C implementations push function arguments 12055 onto the stack in reverse order, the new library version is likely to 12056 interoperate with programs that have been compiled with the old 12057 pem_password_cb definition (PEM_whatever takes some data that 12058 happens to be on the stack as its last argument, and the callback 12059 just ignores this garbage); but there is no guarantee whatsoever that 12060 this will work. 12061 12062 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 12063 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 12064 problems not only on Windows, but also on some Unix platforms. 12065 To avoid problematic command lines, these definitions are now in an 12066 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 12067 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 12068 [Bodo Moeller] 12069 12070 *) MIPS III/IV assembler module is reimplemented. 12071 [Andy Polyakov] 12072 12073 *) More DES library cleanups: remove references to srand/rand and 12074 delete an unused file. 12075 [Ulf Möller] 12076 12077 *) Add support for the free Netwide assembler (NASM) under Win32, 12078 since not many people have MASM (ml) and it can be hard to obtain. 12079 This is currently experimental but it seems to work OK and pass all 12080 the tests. Check out INSTALL.W32 for info. 12081 [Steve Henson] 12082 12083 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 12084 without temporary keys kept an extra copy of the server key, 12085 and connections with temporary keys did not free everything in case 12086 of an error. 12087 [Bodo Moeller] 12088 12089 *) New function RSA_check_key and new openssl rsa option -check 12090 for verifying the consistency of RSA keys. 12091 [Ulf Moeller, Bodo Moeller] 12092 12093 *) Various changes to make Win32 compile work: 12094 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 12095 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 12096 comparison" warnings. 12097 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 12098 [Steve Henson] 12099 12100 *) Add a debugging option to PKCS#5 v2 key generation function: when 12101 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 12102 derived keys are printed to stderr. 12103 [Steve Henson] 12104 12105 *) Copy the flags in ASN1_STRING_dup(). 12106 [Roman E. Pavlov <pre@mo.msk.ru>] 12107 12108 *) The x509 application mishandled signing requests containing DSA 12109 keys when the signing key was also DSA and the parameters didn't match. 12110 12111 It was supposed to omit the parameters when they matched the signing key: 12112 the verifying software was then supposed to automatically use the CA's 12113 parameters if they were absent from the end user certificate. 12114 12115 Omitting parameters is no longer recommended. The test was also 12116 the wrong way round! This was probably due to unusual behaviour in 12117 EVP_cmp_parameters() which returns 1 if the parameters match. 12118 This meant that parameters were omitted when they *didn't* match and 12119 the certificate was useless. Certificates signed with 'ca' didn't have 12120 this bug. 12121 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 12122 12123 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 12124 The interface is as follows: 12125 Applications can use 12126 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 12127 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 12128 "off" is now the default. 12129 The library internally uses 12130 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 12131 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 12132 to disable memory-checking temporarily. 12133 12134 Some inconsistent states that previously were possible (and were 12135 even the default) are now avoided. 12136 12137 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 12138 with each memory chunk allocated; this is occasionally more helpful 12139 than just having a counter. 12140 12141 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 12142 12143 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 12144 extensions. 12145 [Bodo Moeller] 12146 12147 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 12148 which largely parallels "options", but is for changing API behaviour, 12149 whereas "options" are about protocol behaviour. 12150 Initial "mode" flags are: 12151 12152 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 12153 a single record has been written. 12154 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 12155 retries use the same buffer location. 12156 (But all of the contents must be 12157 copied!) 12158 [Bodo Moeller] 12159 12160 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 12161 worked. 12162 12163 *) Fix problems with no-hmac etc. 12164 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>] 12165 12166 *) New functions RSA_get_default_method(), RSA_set_method() and 12167 RSA_get_method(). These allows replacement of RSA_METHODs without having 12168 to mess around with the internals of an RSA structure. 12169 [Steve Henson] 12170 12171 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 12172 Also really enable memory leak checks in openssl.c and in some 12173 test programs. 12174 [Chad C. Mulligan, Bodo Moeller] 12175 12176 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 12177 up the length of negative integers. This has now been simplified to just 12178 store the length when it is first determined and use it later, rather 12179 than trying to keep track of where data is copied and updating it to 12180 point to the end. 12181 [Steve Henson, reported by Brien Wheeler 12182 <bwheeler@authentica-security.com>] 12183 12184 *) Add a new function PKCS7_signatureVerify. This allows the verification 12185 of a PKCS#7 signature but with the signing certificate passed to the 12186 function itself. This contrasts with PKCS7_dataVerify which assumes the 12187 certificate is present in the PKCS#7 structure. This isn't always the 12188 case: certificates can be omitted from a PKCS#7 structure and be 12189 distributed by "out of band" means (such as a certificate database). 12190 [Steve Henson] 12191 12192 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 12193 function prototypes in pem.h, also change util/mkdef.pl to add the 12194 necessary function names. 12195 [Steve Henson] 12196 12197 *) mk1mf.pl (used by Windows builds) did not properly read the 12198 options set by Configure in the top level Makefile, and Configure 12199 was not even able to write more than one option correctly. 12200 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 12201 [Bodo Moeller] 12202 12203 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 12204 file to be loaded from a BIO or FILE pointer. The BIO version will 12205 for example allow memory BIOs to contain config info. 12206 [Steve Henson] 12207 12208 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 12209 Whoever hopes to achieve shared-library compatibility across versions 12210 must use this, not the compile-time macro. 12211 (Exercise 0.9.4: Which is the minimum library version required by 12212 such programs?) 12213 Note: All this applies only to multi-threaded programs, others don't 12214 need locks. 12215 [Bodo Moeller] 12216 12217 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 12218 through a BIO pair triggered the default case, i.e. 12219 SSLerr(...,SSL_R_UNKNOWN_STATE). 12220 [Bodo Moeller] 12221 12222 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 12223 can use the SSL library even if none of the specific BIOs is 12224 appropriate. 12225 [Bodo Moeller] 12226 12227 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 12228 for the encoded length. 12229 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 12230 12231 *) Add initial documentation of the X509V3 functions. 12232 [Steve Henson] 12233 12234 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 12235 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 12236 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 12237 secure PKCS#8 private key format with a high iteration count. 12238 [Steve Henson] 12239 12240 *) Fix determination of Perl interpreter: A perl or perl5 12241 _directory_ in $PATH was also accepted as the interpreter. 12242 [Ralf S. Engelschall] 12243 12244 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 12245 wrong with it but it was very old and did things like calling 12246 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 12247 unusual formatting. 12248 [Steve Henson] 12249 12250 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 12251 to use the new extension code. 12252 [Steve Henson] 12253 12254 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 12255 with macros. This should make it easier to change their form, add extra 12256 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 12257 constant. 12258 [Steve Henson] 12259 12260 *) Add to configuration table a new entry that can specify an alternative 12261 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 12262 according to Mark Crispin <MRC@Panda.COM>. 12263 [Bodo Moeller] 12264 12265#if 0 12266 *) DES CBC did not update the IV. Weird. 12267 [Ben Laurie] 12268#else 12269 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 12270 Changing the behaviour of the former might break existing programs -- 12271 where IV updating is needed, des_ncbc_encrypt can be used. 12272#endif 12273 12274 *) When bntest is run from "make test" it drives bc to check its 12275 calculations, as well as internally checking them. If an internal check 12276 fails, it needs to cause bc to give a non-zero result or make test carries 12277 on without noticing the failure. Fixed. 12278 [Ben Laurie] 12279 12280 *) DES library cleanups. 12281 [Ulf Möller] 12282 12283 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 12284 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 12285 ciphers. NOTE: although the key derivation function has been verified 12286 against some published test vectors it has not been extensively tested 12287 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 12288 of v2.0. 12289 [Steve Henson] 12290 12291 *) Instead of "mkdir -p", which is not fully portable, use new 12292 Perl script "util/mkdir-p.pl". 12293 [Bodo Moeller] 12294 12295 *) Rewrite the way password based encryption (PBE) is handled. It used to 12296 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 12297 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 12298 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 12299 the 'parameter' field of the AlgorithmIdentifier is passed to the 12300 underlying key generation function so it must do its own ASN1 parsing. 12301 This has also changed the EVP_PBE_CipherInit() function which now has a 12302 'parameter' argument instead of literal salt and iteration count values 12303 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 12304 [Steve Henson] 12305 12306 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 12307 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 12308 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 12309 KEY" because this clashed with PKCS#8 unencrypted string. Since this 12310 value was just used as a "magic string" and not used directly its 12311 value doesn't matter. 12312 [Steve Henson] 12313 12314 *) Introduce some semblance of const correctness to BN. Shame C doesn't 12315 support mutable. 12316 [Ben Laurie] 12317 12318 *) "linux-sparc64" configuration (ultrapenguin). 12319 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 12320 "linux-sparc" configuration. 12321 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 12322 12323 *) config now generates no-xxx options for missing ciphers. 12324 [Ulf Möller] 12325 12326 *) Support the EBCDIC character set (work in progress). 12327 File ebcdic.c not yet included because it has a different license. 12328 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 12329 12330 *) Support BS2000/OSD-POSIX. 12331 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 12332 12333 *) Make callbacks for key generation use void * instead of char *. 12334 [Ben Laurie] 12335 12336 *) Make S/MIME samples compile (not yet tested). 12337 [Ben Laurie] 12338 12339 *) Additional typesafe stacks. 12340 [Ben Laurie] 12341 12342 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 12343 [Bodo Moeller] 12344 12345 12346 Changes between 0.9.3 and 0.9.3a [29 May 1999] 12347 12348 *) New configuration variant "sco5-gcc". 12349 12350 *) Updated some demos. 12351 [Sean O Riordain, Wade Scholine] 12352 12353 *) Add missing BIO_free at exit of pkcs12 application. 12354 [Wu Zhigang] 12355 12356 *) Fix memory leak in conf.c. 12357 [Steve Henson] 12358 12359 *) Updates for Win32 to assembler version of MD5. 12360 [Steve Henson] 12361 12362 *) Set #! path to perl in apps/der_chop to where we found it 12363 instead of using a fixed path. 12364 [Bodo Moeller] 12365 12366 *) SHA library changes for irix64-mips4-cc. 12367 [Andy Polyakov] 12368 12369 *) Improvements for VMS support. 12370 [Richard Levitte] 12371 12372 12373 Changes between 0.9.2b and 0.9.3 [24 May 1999] 12374 12375 *) Bignum library bug fix. IRIX 6 passes "make test" now! 12376 This also avoids the problems with SC4.2 and unpatched SC5. 12377 [Andy Polyakov <appro@fy.chalmers.se>] 12378 12379 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 12380 These are required because of the typesafe stack would otherwise break 12381 existing code. If old code used a structure member which used to be STACK 12382 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 12383 sk_num or sk_value it would produce an error because the num, data members 12384 are not present in STACK_OF. Now it just produces a warning. sk_set 12385 replaces the old method of assigning a value to sk_value 12386 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 12387 that does this will no longer work (and should use sk_set instead) but 12388 this could be regarded as a "questionable" behaviour anyway. 12389 [Steve Henson] 12390 12391 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 12392 correctly handle encrypted S/MIME data. 12393 [Steve Henson] 12394 12395 *) Change type of various DES function arguments from des_cblock 12396 (which means, in function argument declarations, pointer to char) 12397 to des_cblock * (meaning pointer to array with 8 char elements), 12398 which allows the compiler to do more typechecking; it was like 12399 that back in SSLeay, but with lots of ugly casts. 12400 12401 Introduce new type const_des_cblock. 12402 [Bodo Moeller] 12403 12404 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 12405 problems: find RecipientInfo structure that matches recipient certificate 12406 and initialise the ASN1 structures properly based on passed cipher. 12407 [Steve Henson] 12408 12409 *) Belatedly make the BN tests actually check the results. 12410 [Ben Laurie] 12411 12412 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 12413 to and from BNs: it was completely broken. New compilation option 12414 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 12415 key elements as negative integers. 12416 [Steve Henson] 12417 12418 *) Reorganize and speed up MD5. 12419 [Andy Polyakov <appro@fy.chalmers.se>] 12420 12421 *) VMS support. 12422 [Richard Levitte <richard@levitte.org>] 12423 12424 *) New option -out to asn1parse to allow the parsed structure to be 12425 output to a file. This is most useful when combined with the -strparse 12426 option to examine the output of things like OCTET STRINGS. 12427 [Steve Henson] 12428 12429 *) Make SSL library a little more fool-proof by not requiring any longer 12430 that SSL_set_{accept,connect}_state be called before 12431 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 12432 in many applications because usually everything *appeared* to work as 12433 intended anyway -- now it really works as intended). 12434 [Bodo Moeller] 12435 12436 *) Move openssl.cnf out of lib/. 12437 [Ulf Möller] 12438 12439 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 12440 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 12441 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 12442 [Ralf S. Engelschall] 12443 12444 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 12445 handle PKCS#7 enveloped data properly. 12446 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 12447 12448 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 12449 copying pointers. The cert_st handling is changed by this in 12450 various ways (and thus what used to be known as ctx->default_cert 12451 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 12452 any longer when s->cert does not give us what we need). 12453 ssl_cert_instantiate becomes obsolete by this change. 12454 As soon as we've got the new code right (possibly it already is?), 12455 we have solved a couple of bugs of the earlier code where s->cert 12456 was used as if it could not have been shared with other SSL structures. 12457 12458 Note that using the SSL API in certain dirty ways now will result 12459 in different behaviour than observed with earlier library versions: 12460 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 12461 does not influence s as it used to. 12462 12463 In order to clean up things more thoroughly, inside SSL_SESSION 12464 we don't use CERT any longer, but a new structure SESS_CERT 12465 that holds per-session data (if available); currently, this is 12466 the peer's certificate chain and, for clients, the server's certificate 12467 and temporary key. CERT holds only those values that can have 12468 meaningful defaults in an SSL_CTX. 12469 [Bodo Moeller] 12470 12471 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 12472 from the internal representation. Various PKCS#7 fixes: remove some 12473 evil casts and set the enc_dig_alg field properly based on the signing 12474 key type. 12475 [Steve Henson] 12476 12477 *) Allow PKCS#12 password to be set from the command line or the 12478 environment. Let 'ca' get its config file name from the environment 12479 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 12480 and 'x509'). 12481 [Steve Henson] 12482 12483 *) Allow certificate policies extension to use an IA5STRING for the 12484 organization field. This is contrary to the PKIX definition but 12485 VeriSign uses it and IE5 only recognises this form. Document 'x509' 12486 extension option. 12487 [Steve Henson] 12488 12489 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 12490 without disallowing inline assembler and the like for non-pedantic builds. 12491 [Ben Laurie] 12492 12493 *) Support Borland C++ builder. 12494 [Janez Jere <jj@void.si>, modified by Ulf Möller] 12495 12496 *) Support Mingw32. 12497 [Ulf Möller] 12498 12499 *) SHA-1 cleanups and performance enhancements. 12500 [Andy Polyakov <appro@fy.chalmers.se>] 12501 12502 *) Sparc v8plus assembler for the bignum library. 12503 [Andy Polyakov <appro@fy.chalmers.se>] 12504 12505 *) Accept any -xxx and +xxx compiler options in Configure. 12506 [Ulf Möller] 12507 12508 *) Update HPUX configuration. 12509 [Anonymous] 12510 12511 *) Add missing sk_<type>_unshift() function to safestack.h 12512 [Ralf S. Engelschall] 12513 12514 *) New function SSL_CTX_use_certificate_chain_file that sets the 12515 "extra_cert"s in addition to the certificate. (This makes sense 12516 only for "PEM" format files, as chains as a whole are not 12517 DER-encoded.) 12518 [Bodo Moeller] 12519 12520 *) Support verify_depth from the SSL API. 12521 x509_vfy.c had what can be considered an off-by-one-error: 12522 Its depth (which was not part of the external interface) 12523 was actually counting the number of certificates in a chain; 12524 now it really counts the depth. 12525 [Bodo Moeller] 12526 12527 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 12528 instead of X509err, which often resulted in confusing error 12529 messages since the error codes are not globally unique 12530 (e.g. an alleged error in ssl3_accept when a certificate 12531 didn't match the private key). 12532 12533 *) New function SSL_CTX_set_session_id_context that allows to set a default 12534 value (so that you don't need SSL_set_session_id_context for each 12535 connection using the SSL_CTX). 12536 [Bodo Moeller] 12537 12538 *) OAEP decoding bug fix. 12539 [Ulf Möller] 12540 12541 *) Support INSTALL_PREFIX for package builders, as proposed by 12542 David Harris. 12543 [Bodo Moeller] 12544 12545 *) New Configure options "threads" and "no-threads". For systems 12546 where the proper compiler options are known (currently Solaris 12547 and Linux), "threads" is the default. 12548 [Bodo Moeller] 12549 12550 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 12551 [Bodo Moeller] 12552 12553 *) Install various scripts to $(OPENSSLDIR)/misc, not to 12554 $(INSTALLTOP)/bin -- they shouldn't clutter directories 12555 such as /usr/local/bin. 12556 [Bodo Moeller] 12557 12558 *) "make linux-shared" to build shared libraries. 12559 [Niels Poppe <niels@netbox.org>] 12560 12561 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 12562 [Ulf Möller] 12563 12564 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 12565 extension adding in x509 utility. 12566 [Steve Henson] 12567 12568 *) Remove NOPROTO sections and error code comments. 12569 [Ulf Möller] 12570 12571 *) Partial rewrite of the DEF file generator to now parse the ANSI 12572 prototypes. 12573 [Steve Henson] 12574 12575 *) New Configure options --prefix=DIR and --openssldir=DIR. 12576 [Ulf Möller] 12577 12578 *) Complete rewrite of the error code script(s). It is all now handled 12579 by one script at the top level which handles error code gathering, 12580 header rewriting and C source file generation. It should be much better 12581 than the old method: it now uses a modified version of Ulf's parser to 12582 read the ANSI prototypes in all header files (thus the old K&R definitions 12583 aren't needed for error creation any more) and do a better job of 12584 translating function codes into names. The old 'ASN1 error code embedded 12585 in a comment' is no longer necessary and it doesn't use .err files which 12586 have now been deleted. Also the error code call doesn't have to appear all 12587 on one line (which resulted in some large lines...). 12588 [Steve Henson] 12589 12590 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 12591 [Bodo Moeller] 12592 12593 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 12594 0 (which usually indicates a closed connection), but continue reading. 12595 [Bodo Moeller] 12596 12597 *) Fix some race conditions. 12598 [Bodo Moeller] 12599 12600 *) Add support for CRL distribution points extension. Add Certificate 12601 Policies and CRL distribution points documentation. 12602 [Steve Henson] 12603 12604 *) Move the autogenerated header file parts to crypto/opensslconf.h. 12605 [Ulf Möller] 12606 12607 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 12608 8 of keying material. Merlin has also confirmed interop with this fix 12609 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 12610 [Merlin Hughes <merlin@baltimore.ie>] 12611 12612 *) Fix lots of warnings. 12613 [Richard Levitte <levitte@stacken.kth.se>] 12614 12615 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 12616 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 12617 [Richard Levitte <levitte@stacken.kth.se>] 12618 12619 *) Fix problems with sizeof(long) == 8. 12620 [Andy Polyakov <appro@fy.chalmers.se>] 12621 12622 *) Change functions to ANSI C. 12623 [Ulf Möller] 12624 12625 *) Fix typos in error codes. 12626 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller] 12627 12628 *) Remove defunct assembler files from Configure. 12629 [Ulf Möller] 12630 12631 *) SPARC v8 assembler BIGNUM implementation. 12632 [Andy Polyakov <appro@fy.chalmers.se>] 12633 12634 *) Support for Certificate Policies extension: both print and set. 12635 Various additions to support the r2i method this uses. 12636 [Steve Henson] 12637 12638 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 12639 return a const string when you are expecting an allocated buffer. 12640 [Ben Laurie] 12641 12642 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 12643 types DirectoryString and DisplayText. 12644 [Steve Henson] 12645 12646 *) Add code to allow r2i extensions to access the configuration database, 12647 add an LHASH database driver and add several ctx helper functions. 12648 [Steve Henson] 12649 12650 *) Fix an evil bug in bn_expand2() which caused various BN functions to 12651 fail when they extended the size of a BIGNUM. 12652 [Steve Henson] 12653 12654 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 12655 support typesafe stack. 12656 [Steve Henson] 12657 12658 *) Fix typo in SSL_[gs]et_options(). 12659 [Nils Frostberg <nils@medcom.se>] 12660 12661 *) Delete various functions and files that belonged to the (now obsolete) 12662 old X509V3 handling code. 12663 [Steve Henson] 12664 12665 *) New Configure option "rsaref". 12666 [Ulf Möller] 12667 12668 *) Don't auto-generate pem.h. 12669 [Bodo Moeller] 12670 12671 *) Introduce type-safe ASN.1 SETs. 12672 [Ben Laurie] 12673 12674 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 12675 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 12676 12677 *) Introduce type-safe STACKs. This will almost certainly break lots of code 12678 that links with OpenSSL (well at least cause lots of warnings), but fear 12679 not: the conversion is trivial, and it eliminates loads of evil casts. A 12680 few STACKed things have been converted already. Feel free to convert more. 12681 In the fullness of time, I'll do away with the STACK type altogether. 12682 [Ben Laurie] 12683 12684 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 12685 specified in <certfile> by updating the entry in the index.txt file. 12686 This way one no longer has to edit the index.txt file manually for 12687 revoking a certificate. The -revoke option does the gory details now. 12688 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 12689 12690 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 12691 `-text' option at all and this way the `-noout -text' combination was 12692 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 12693 [Ralf S. Engelschall] 12694 12695 *) Make sure a corresponding plain text error message exists for the 12696 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 12697 verify callback function determined that a certificate was revoked. 12698 [Ralf S. Engelschall] 12699 12700 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 12701 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 12702 all available ciphers including rc5, which was forgotten until now. 12703 In order to let the testing shell script know which algorithms 12704 are available, a new (up to now undocumented) command 12705 "openssl list-cipher-commands" is used. 12706 [Bodo Moeller] 12707 12708 *) Bugfix: s_client occasionally would sleep in select() when 12709 it should have checked SSL_pending() first. 12710 [Bodo Moeller] 12711 12712 *) New functions DSA_do_sign and DSA_do_verify to provide access to 12713 the raw DSA values prior to ASN.1 encoding. 12714 [Ulf Möller] 12715 12716 *) Tweaks to Configure 12717 [Niels Poppe <niels@netbox.org>] 12718 12719 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 12720 yet... 12721 [Steve Henson] 12722 12723 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 12724 [Ulf Möller] 12725 12726 *) New config option to avoid instructions that are illegal on the 80386. 12727 The default code is faster, but requires at least a 486. 12728 [Ulf Möller] 12729 12730 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 12731 SSL2_SERVER_VERSION (not used at all) macros, which are now the 12732 same as SSL2_VERSION anyway. 12733 [Bodo Moeller] 12734 12735 *) New "-showcerts" option for s_client. 12736 [Bodo Moeller] 12737 12738 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 12739 application. Various cleanups and fixes. 12740 [Steve Henson] 12741 12742 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 12743 modify error routines to work internally. Add error codes and PBE init 12744 to library startup routines. 12745 [Steve Henson] 12746 12747 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 12748 packing functions to asn1 and evp. Changed function names and error 12749 codes along the way. 12750 [Steve Henson] 12751 12752 *) PKCS12 integration: and so it begins... First of several patches to 12753 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 12754 objects to objects.h 12755 [Steve Henson] 12756 12757 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 12758 and display support for Thawte strong extranet extension. 12759 [Steve Henson] 12760 12761 *) Add LinuxPPC support. 12762 [Jeff Dubrule <igor@pobox.org>] 12763 12764 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 12765 bn_div_words in alpha.s. 12766 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 12767 12768 *) Make sure the RSA OAEP test is skipped under -DRSAref because 12769 OAEP isn't supported when OpenSSL is built with RSAref. 12770 [Ulf Moeller <ulf@fitug.de>] 12771 12772 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 12773 so they no longer are missing under -DNOPROTO. 12774 [Soren S. Jorvang <soren@t.dk>] 12775 12776 12777 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 12778 12779 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 12780 doesn't work when the session is reused. Coming soon! 12781 [Ben Laurie] 12782 12783 *) Fix a security hole, that allows sessions to be reused in the wrong 12784 context thus bypassing client cert protection! All software that uses 12785 client certs and session caches in multiple contexts NEEDS PATCHING to 12786 allow session reuse! A fuller solution is in the works. 12787 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 12788 12789 *) Some more source tree cleanups (removed obsolete files 12790 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 12791 permission on "config" script to be executable) and a fix for the INSTALL 12792 document. 12793 [Ulf Moeller <ulf@fitug.de>] 12794 12795 *) Remove some legacy and erroneous uses of malloc, free instead of 12796 Malloc, Free. 12797 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 12798 12799 *) Make rsa_oaep_test return non-zero on error. 12800 [Ulf Moeller <ulf@fitug.de>] 12801 12802 *) Add support for native Solaris shared libraries. Configure 12803 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 12804 if someone would make that last step automatic. 12805 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 12806 12807 *) ctx_size was not built with the right compiler during "make links". Fixed. 12808 [Ben Laurie] 12809 12810 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 12811 except NULL ciphers". This means the default cipher list will no longer 12812 enable NULL ciphers. They need to be specifically enabled e.g. with 12813 the string "DEFAULT:eNULL". 12814 [Steve Henson] 12815 12816 *) Fix to RSA private encryption routines: if p < q then it would 12817 occasionally produce an invalid result. This will only happen with 12818 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 12819 [Steve Henson] 12820 12821 *) Be less restrictive and allow also `perl util/perlpath.pl 12822 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 12823 because this way one can also use an interpreter named `perl5' (which is 12824 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 12825 installed as `perl'). 12826 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 12827 12828 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 12829 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 12830 12831 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 12832 advapi32.lib to Win32 build and change the pem test comparison 12833 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 12834 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 12835 and crypto/des/ede_cbcm_enc.c. 12836 [Steve Henson] 12837 12838 *) DES quad checksum was broken on big-endian architectures. Fixed. 12839 [Ben Laurie] 12840 12841 *) Comment out two functions in bio.h that aren't implemented. Fix up the 12842 Win32 test batch file so it (might) work again. The Win32 test batch file 12843 is horrible: I feel ill.... 12844 [Steve Henson] 12845 12846 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 12847 in e_os.h. Audit of header files to check ANSI and non ANSI 12848 sections: 10 functions were absent from non ANSI section and not exported 12849 from Windows DLLs. Fixed up libeay.num for new functions. 12850 [Steve Henson] 12851 12852 *) Make `openssl version' output lines consistent. 12853 [Ralf S. Engelschall] 12854 12855 *) Fix Win32 symbol export lists for BIO functions: Added 12856 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 12857 to ms/libeay{16,32}.def. 12858 [Ralf S. Engelschall] 12859 12860 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 12861 fine under Unix and passes some trivial tests I've now added. But the 12862 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 12863 added to make sure no one expects that this stuff really works in the 12864 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 12865 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 12866 openssl_bio.xs. 12867 [Ralf S. Engelschall] 12868 12869 *) Fix the generation of two part addresses in perl. 12870 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 12871 12872 *) Add config entry for Linux on MIPS. 12873 [John Tobey <jtobey@channel1.com>] 12874 12875 *) Make links whenever Configure is run, unless we are on Windoze. 12876 [Ben Laurie] 12877 12878 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 12879 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 12880 in CRLs. 12881 [Steve Henson] 12882 12883 *) Add a useful kludge to allow package maintainers to specify compiler and 12884 other platforms details on the command line without having to patch the 12885 Configure script every time: One now can use ``perl Configure 12886 <id>:<details>'', i.e. platform ids are allowed to have details appended 12887 to them (separated by colons). This is treated as there would be a static 12888 pre-configured entry in Configure's %table under key <id> with value 12889 <details> and ``perl Configure <id>'' is called. So, when you want to 12890 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 12891 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 12892 now, which overrides the FreeBSD-elf entry on-the-fly. 12893 [Ralf S. Engelschall] 12894 12895 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 12896 [Ben Laurie] 12897 12898 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 12899 on the `perl Configure ...' command line. This way one can compile 12900 OpenSSL libraries with Position Independent Code (PIC) which is needed 12901 for linking it into DSOs. 12902 [Ralf S. Engelschall] 12903 12904 *) Remarkably, export ciphers were totally broken and no-one had noticed! 12905 Fixed. 12906 [Ben Laurie] 12907 12908 *) Cleaned up the LICENSE document: The official contact for any license 12909 questions now is the OpenSSL core team under openssl-core@openssl.org. 12910 And add a paragraph about the dual-license situation to make sure people 12911 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 12912 to the OpenSSL toolkit. 12913 [Ralf S. Engelschall] 12914 12915 *) General source tree makefile cleanups: Made `making xxx in yyy...' 12916 display consistent in the source tree and replaced `/bin/rm' by `rm'. 12917 Additionally cleaned up the `make links' target: Remove unnecessary 12918 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 12919 to speed processing and no longer clutter the display with confusing 12920 stuff. Instead only the actually done links are displayed. 12921 [Ralf S. Engelschall] 12922 12923 *) Permit null encryption ciphersuites, used for authentication only. It used 12924 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 12925 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 12926 encryption. 12927 [Ben Laurie] 12928 12929 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 12930 signed attributes when verifying signatures (this would break them), 12931 the detached data encoding was wrong and public keys obtained using 12932 X509_get_pubkey() weren't freed. 12933 [Steve Henson] 12934 12935 *) Add text documentation for the BUFFER functions. Also added a work around 12936 to a Win95 console bug. This was triggered by the password read stuff: the 12937 last character typed gets carried over to the next fread(). If you were 12938 generating a new cert request using 'req' for example then the last 12939 character of the passphrase would be CR which would then enter the first 12940 field as blank. 12941 [Steve Henson] 12942 12943 *) Added the new `Includes OpenSSL Cryptography Software' button as 12944 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 12945 button and can be used by applications based on OpenSSL to show the 12946 relationship to the OpenSSL project. 12947 [Ralf S. Engelschall] 12948 12949 *) Remove confusing variables in function signatures in files 12950 ssl/ssl_lib.c and ssl/ssl.h. 12951 [Lennart Bong <lob@kulthea.stacken.kth.se>] 12952 12953 *) Don't install bss_file.c under PREFIX/include/ 12954 [Lennart Bong <lob@kulthea.stacken.kth.se>] 12955 12956 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 12957 functions that return function pointers and has support for NT specific 12958 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 12959 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 12960 unsigned to signed types: this was killing the Win32 compile. 12961 [Steve Henson] 12962 12963 *) Add new certificate file to stack functions, 12964 SSL_add_dir_cert_subjects_to_stack() and 12965 SSL_add_file_cert_subjects_to_stack(). These largely supplant 12966 SSL_load_client_CA_file(), and can be used to add multiple certs easily 12967 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 12968 This means that Apache-SSL and similar packages don't have to mess around 12969 to add as many CAs as they want to the preferred list. 12970 [Ben Laurie] 12971 12972 *) Experiment with doxygen documentation. Currently only partially applied to 12973 ssl/ssl_lib.c. 12974 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 12975 openssl.doxy as the configuration file. 12976 [Ben Laurie] 12977 12978 *) Get rid of remaining C++-style comments which strict C compilers hate. 12979 [Ralf S. Engelschall, pointed out by Carlos Amengual] 12980 12981 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 12982 compiled in by default: it has problems with large keys. 12983 [Steve Henson] 12984 12985 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 12986 DH private keys and/or callback functions which directly correspond to 12987 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 12988 is needed for applications which have to configure certificates on a 12989 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 12990 (e.g. s_server). 12991 For the RSA certificate situation is makes no difference, but 12992 for the DSA certificate situation this fixes the "no shared cipher" 12993 problem where the OpenSSL cipher selection procedure failed because the 12994 temporary keys were not overtaken from the context and the API provided 12995 no way to reconfigure them. 12996 The new functions now let applications reconfigure the stuff and they 12997 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 12998 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 12999 non-public-API function ssl_cert_instantiate() is used as a helper 13000 function and also to reduce code redundancy inside ssl_rsa.c. 13001 [Ralf S. Engelschall] 13002 13003 *) Move s_server -dcert and -dkey options out of the undocumented feature 13004 area because they are useful for the DSA situation and should be 13005 recognized by the users. 13006 [Ralf S. Engelschall] 13007 13008 *) Fix the cipher decision scheme for export ciphers: the export bits are 13009 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 13010 SSL_EXP_MASK. So, the original variable has to be used instead of the 13011 already masked variable. 13012 [Richard Levitte <levitte@stacken.kth.se>] 13013 13014 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 13015 [Richard Levitte <levitte@stacken.kth.se>] 13016 13017 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 13018 from `int' to `unsigned int' because it's a length and initialized by 13019 EVP_DigestFinal() which expects an `unsigned int *'. 13020 [Richard Levitte <levitte@stacken.kth.se>] 13021 13022 *) Don't hard-code path to Perl interpreter on shebang line of Configure 13023 script. Instead use the usual Shell->Perl transition trick. 13024 [Ralf S. Engelschall] 13025 13026 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 13027 (in addition to RSA certificates) to match the behaviour of `openssl dsa 13028 -noout -modulus' as it's already the case for `openssl rsa -noout 13029 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 13030 currently the public key is printed (a decision which was already done by 13031 `openssl dsa -modulus' in the past) which serves a similar purpose. 13032 Additionally the NO_RSA no longer completely removes the whole -modulus 13033 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 13034 now, too. 13035 [Ralf S. Engelschall] 13036 13037 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 13038 BIO. See the source (crypto/evp/bio_ok.c) for more info. 13039 [Arne Ansper <arne@ats.cyber.ee>] 13040 13041 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 13042 to be added. Now both 'req' and 'ca' can use new objects defined in the 13043 config file. 13044 [Steve Henson] 13045 13046 *) Add cool BIO that does syslog (or event log on NT). 13047 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 13048 13049 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 13050 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 13051 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 13052 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 13053 [Ben Laurie] 13054 13055 *) Add preliminary config info for new extension code. 13056 [Steve Henson] 13057 13058 *) Make RSA_NO_PADDING really use no padding. 13059 [Ulf Moeller <ulf@fitug.de>] 13060 13061 *) Generate errors when private/public key check is done. 13062 [Ben Laurie] 13063 13064 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 13065 for some CRL extensions and new objects added. 13066 [Steve Henson] 13067 13068 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 13069 key usage extension and fuller support for authority key id. 13070 [Steve Henson] 13071 13072 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 13073 padding method for RSA, which is recommended for new applications in PKCS 13074 #1 v2.0 (RFC 2437, October 1998). 13075 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 13076 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 13077 against Bleichbacher's attack on RSA. 13078 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 13079 Ben Laurie] 13080 13081 *) Updates to the new SSL compression code 13082 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13083 13084 *) Fix so that the version number in the master secret, when passed 13085 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 13086 (because the server will not accept higher), that the version number 13087 is 0x03,0x01, not 0x03,0x00 13088 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13089 13090 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 13091 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 13092 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 13093 [Steve Henson] 13094 13095 *) Support for RAW extensions where an arbitrary extension can be 13096 created by including its DER encoding. See apps/openssl.cnf for 13097 an example. 13098 [Steve Henson] 13099 13100 *) Make sure latest Perl versions don't interpret some generated C array 13101 code as Perl array code in the crypto/err/err_genc.pl script. 13102 [Lars Weber <3weber@informatik.uni-hamburg.de>] 13103 13104 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 13105 not many people have the assembler. Various Win32 compilation fixes and 13106 update to the INSTALL.W32 file with (hopefully) more accurate Win32 13107 build instructions. 13108 [Steve Henson] 13109 13110 *) Modify configure script 'Configure' to automatically create crypto/date.h 13111 file under Win32 and also build pem.h from pem.org. New script 13112 util/mkfiles.pl to create the MINFO file on environments that can't do a 13113 'make files': perl util/mkfiles.pl >MINFO should work. 13114 [Steve Henson] 13115 13116 *) Major rework of DES function declarations, in the pursuit of correctness 13117 and purity. As a result, many evil casts evaporated, and some weirdness, 13118 too. You may find this causes warnings in your code. Zapping your evil 13119 casts will probably fix them. Mostly. 13120 [Ben Laurie] 13121 13122 *) Fix for a typo in asn1.h. Bug fix to object creation script 13123 obj_dat.pl. It considered a zero in an object definition to mean 13124 "end of object": none of the objects in objects.h have any zeros 13125 so it wasn't spotted. 13126 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 13127 13128 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 13129 Masking (CBCM). In the absence of test vectors, the best I have been able 13130 to do is check that the decrypt undoes the encrypt, so far. Send me test 13131 vectors if you have them. 13132 [Ben Laurie] 13133 13134 *) Correct calculation of key length for export ciphers (too much space was 13135 allocated for null ciphers). This has not been tested! 13136 [Ben Laurie] 13137 13138 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 13139 message is now correct (it understands "crypto" and "ssl" on its 13140 command line). There is also now an "update" option. This will update 13141 the util/ssleay.num and util/libeay.num files with any new functions. 13142 If you do a: 13143 perl util/mkdef.pl crypto ssl update 13144 it will update them. 13145 [Steve Henson] 13146 13147 *) Overhauled the Perl interface (perl/*): 13148 - ported BN stuff to OpenSSL's different BN library 13149 - made the perl/ source tree CVS-aware 13150 - renamed the package from SSLeay to OpenSSL (the files still contain 13151 their history because I've copied them in the repository) 13152 - removed obsolete files (the test scripts will be replaced 13153 by better Test::Harness variants in the future) 13154 [Ralf S. Engelschall] 13155 13156 *) First cut for a very conservative source tree cleanup: 13157 1. merge various obsolete readme texts into doc/ssleay.txt 13158 where we collect the old documents and readme texts. 13159 2. remove the first part of files where I'm already sure that we no 13160 longer need them because of three reasons: either they are just temporary 13161 files which were left by Eric or they are preserved original files where 13162 I've verified that the diff is also available in the CVS via "cvs diff 13163 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 13164 the crypto/md/ stuff). 13165 [Ralf S. Engelschall] 13166 13167 *) More extension code. Incomplete support for subject and issuer alt 13168 name, issuer and authority key id. Change the i2v function parameters 13169 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 13170 what that's for :-) Fix to ASN1 macro which messed up 13171 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 13172 [Steve Henson] 13173 13174 *) Preliminary support for ENUMERATED type. This is largely copied from the 13175 INTEGER code. 13176 [Steve Henson] 13177 13178 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 13179 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13180 13181 *) Make sure `make rehash' target really finds the `openssl' program. 13182 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 13183 13184 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 13185 like to hear about it if this slows down other processors. 13186 [Ben Laurie] 13187 13188 *) Add CygWin32 platform information to Configure script. 13189 [Alan Batie <batie@aahz.jf.intel.com>] 13190 13191 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 13192 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 13193 13194 *) New program nseq to manipulate netscape certificate sequences 13195 [Steve Henson] 13196 13197 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 13198 few typos. 13199 [Steve Henson] 13200 13201 *) Fixes to BN code. Previously the default was to define BN_RECURSION 13202 but the BN code had some problems that would cause failures when 13203 doing certificate verification and some other functions. 13204 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13205 13206 *) Add ASN1 and PEM code to support netscape certificate sequences. 13207 [Steve Henson] 13208 13209 *) Add ASN1 and PEM code to support netscape certificate sequences. 13210 [Steve Henson] 13211 13212 *) Add several PKIX and private extended key usage OIDs. 13213 [Steve Henson] 13214 13215 *) Modify the 'ca' program to handle the new extension code. Modify 13216 openssl.cnf for new extension format, add comments. 13217 [Steve Henson] 13218 13219 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 13220 and add a sample to openssl.cnf so req -x509 now adds appropriate 13221 CA extensions. 13222 [Steve Henson] 13223 13224 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 13225 error code, add initial support to X509_print() and x509 application. 13226 [Steve Henson] 13227 13228 *) Takes a deep breath and start adding X509 V3 extension support code. Add 13229 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 13230 stuff is currently isolated and isn't even compiled yet. 13231 [Steve Henson] 13232 13233 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 13234 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 13235 Removed the versions check from X509 routines when loading extensions: 13236 this allows certain broken certificates that don't set the version 13237 properly to be processed. 13238 [Steve Henson] 13239 13240 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 13241 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 13242 can still be regenerated with "make depend". 13243 [Ben Laurie] 13244 13245 *) Spelling mistake in C version of CAST-128. 13246 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 13247 13248 *) Changes to the error generation code. The perl script err-code.pl 13249 now reads in the old error codes and retains the old numbers, only 13250 adding new ones if necessary. It also only changes the .err files if new 13251 codes are added. The makefiles have been modified to only insert errors 13252 when needed (to avoid needlessly modifying header files). This is done 13253 by only inserting errors if the .err file is newer than the auto generated 13254 C file. To rebuild all the error codes from scratch (the old behaviour) 13255 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 13256 or delete all the .err files. 13257 [Steve Henson] 13258 13259 *) CAST-128 was incorrectly implemented for short keys. The C version has 13260 been fixed, but is untested. The assembler versions are also fixed, but 13261 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 13262 to regenerate it if needed. 13263 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 13264 Hagino <itojun@kame.net>] 13265 13266 *) File was opened incorrectly in randfile.c. 13267 [Ulf Möller <ulf@fitug.de>] 13268 13269 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 13270 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 13271 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 13272 al: it's just almost always a UTCTime. Note this patch adds new error 13273 codes so do a "make errors" if there are problems. 13274 [Steve Henson] 13275 13276 *) Correct Linux 1 recognition in config. 13277 [Ulf Möller <ulf@fitug.de>] 13278 13279 *) Remove pointless MD5 hash when using DSA keys in ca. 13280 [Anonymous <nobody@replay.com>] 13281 13282 *) Generate an error if given an empty string as a cert directory. Also 13283 generate an error if handed NULL (previously returned 0 to indicate an 13284 error, but didn't set one). 13285 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 13286 13287 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 13288 [Ben Laurie] 13289 13290 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 13291 parameters. This was causing a warning which killed off the Win32 compile. 13292 [Steve Henson] 13293 13294 *) Remove C++ style comments from crypto/bn/bn_local.h. 13295 [Neil Costigan <neil.costigan@celocom.com>] 13296 13297 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 13298 based on a text string, looking up short and long names and finally 13299 "dot" format. The "dot" format stuff didn't work. Added new function 13300 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 13301 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 13302 OID is not part of the table. 13303 [Steve Henson] 13304 13305 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 13306 X509_LOOKUP_by_alias(). 13307 [Ben Laurie] 13308 13309 *) Sort openssl functions by name. 13310 [Ben Laurie] 13311 13312 *) Get the gendsa program working (hopefully) and add it to app list. Remove 13313 encryption from sample DSA keys (in case anyone is interested the password 13314 was "1234"). 13315 [Steve Henson] 13316 13317 *) Make _all_ *_free functions accept a NULL pointer. 13318 [Frans Heymans <fheymans@isaserver.be>] 13319 13320 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 13321 NULL pointers. 13322 [Anonymous <nobody@replay.com>] 13323 13324 *) s_server should send the CAfile as acceptable CAs, not its own cert. 13325 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 13326 13327 *) Don't blow it for numeric -newkey arguments to apps/req. 13328 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 13329 13330 *) Temp key "for export" tests were wrong in s3_srvr.c. 13331 [Anonymous <nobody@replay.com>] 13332 13333 *) Add prototype for temp key callback functions 13334 SSL_CTX_set_tmp_{rsa,dh}_callback(). 13335 [Ben Laurie] 13336 13337 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 13338 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 13339 [Steve Henson] 13340 13341 *) X509_name_add_entry() freed the wrong thing after an error. 13342 [Arne Ansper <arne@ats.cyber.ee>] 13343 13344 *) rsa_eay.c would attempt to free a NULL context. 13345 [Arne Ansper <arne@ats.cyber.ee>] 13346 13347 *) BIO_s_socket() had a broken should_retry() on Windoze. 13348 [Arne Ansper <arne@ats.cyber.ee>] 13349 13350 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 13351 [Arne Ansper <arne@ats.cyber.ee>] 13352 13353 *) Make sure the already existing X509_STORE->depth variable is initialized 13354 in X509_STORE_new(), but document the fact that this variable is still 13355 unused in the certificate verification process. 13356 [Ralf S. Engelschall] 13357 13358 *) Fix the various library and apps files to free up pkeys obtained from 13359 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 13360 [Steve Henson] 13361 13362 *) Fix reference counting in X509_PUBKEY_get(). This makes 13363 demos/maurice/example2.c work, amongst others, probably. 13364 [Steve Henson and Ben Laurie] 13365 13366 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 13367 `openssl' and second, the shortcut symlinks for the `openssl <command>' 13368 are no longer created. This way we have a single and consistent command 13369 line interface `openssl <command>', similar to `cvs <command>'. 13370 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 13371 13372 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 13373 BIT STRING wrapper always have zero unused bits. 13374 [Steve Henson] 13375 13376 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 13377 [Steve Henson] 13378 13379 *) Make the top-level INSTALL documentation easier to understand. 13380 [Paul Sutton] 13381 13382 *) Makefiles updated to exit if an error occurs in a sub-directory 13383 make (including if user presses ^C) [Paul Sutton] 13384 13385 *) Make Montgomery context stuff explicit in RSA data structure. 13386 [Ben Laurie] 13387 13388 *) Fix build order of pem and err to allow for generated pem.h. 13389 [Ben Laurie] 13390 13391 *) Fix renumbering bug in X509_NAME_delete_entry(). 13392 [Ben Laurie] 13393 13394 *) Enhanced the err-ins.pl script so it makes the error library number 13395 global and can add a library name. This is needed for external ASN1 and 13396 other error libraries. 13397 [Steve Henson] 13398 13399 *) Fixed sk_insert which never worked properly. 13400 [Steve Henson] 13401 13402 *) Fix ASN1 macros so they can handle indefinite length constructed 13403 EXPLICIT tags. Some non standard certificates use these: they can now 13404 be read in. 13405 [Steve Henson] 13406 13407 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 13408 into a single doc/ssleay.txt bundle. This way the information is still 13409 preserved but no longer messes up this directory. Now it's new room for 13410 the new set of documentation files. 13411 [Ralf S. Engelschall] 13412 13413 *) SETs were incorrectly DER encoded. This was a major pain, because they 13414 shared code with SEQUENCEs, which aren't coded the same. This means that 13415 almost everything to do with SETs or SEQUENCEs has either changed name or 13416 number of arguments. 13417 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 13418 13419 *) Fix test data to work with the above. 13420 [Ben Laurie] 13421 13422 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 13423 was already fixed by Eric for 0.9.1 it seems. 13424 [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>] 13425 13426 *) Autodetect FreeBSD3. 13427 [Ben Laurie] 13428 13429 *) Fix various bugs in Configure. This affects the following platforms: 13430 nextstep 13431 ncr-scde 13432 unixware-2.0 13433 unixware-2.0-pentium 13434 sco5-cc. 13435 [Ben Laurie] 13436 13437 *) Eliminate generated files from CVS. Reorder tests to regenerate files 13438 before they are needed. 13439 [Ben Laurie] 13440 13441 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 13442 [Ben Laurie] 13443 13444 13445 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 13446 13447 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 13448 changed SSLeay to OpenSSL in version strings. 13449 [Ralf S. Engelschall] 13450 13451 *) Some fixups to the top-level documents. 13452 [Paul Sutton] 13453 13454 *) Fixed the nasty bug where rsaref.h was not found under compile-time 13455 because the symlink to include/ was missing. 13456 [Ralf S. Engelschall] 13457 13458 *) Incorporated the popular no-RSA/DSA-only patches 13459 which allow to compile a RSA-free SSLeay. 13460 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 13461 13462 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 13463 when "ssleay" is still not found. 13464 [Ralf S. Engelschall] 13465 13466 *) Added more platforms to Configure: Cray T3E, HPUX 11, 13467 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 13468 13469 *) Updated the README file. 13470 [Ralf S. Engelschall] 13471 13472 *) Added various .cvsignore files in the CVS repository subdirs 13473 to make a "cvs update" really silent. 13474 [Ralf S. Engelschall] 13475 13476 *) Recompiled the error-definition header files and added 13477 missing symbols to the Win32 linker tables. 13478 [Ralf S. Engelschall] 13479 13480 *) Cleaned up the top-level documents; 13481 o new files: CHANGES and LICENSE 13482 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 13483 o merged COPYRIGHT into LICENSE 13484 o removed obsolete TODO file 13485 o renamed MICROSOFT to INSTALL.W32 13486 [Ralf S. Engelschall] 13487 13488 *) Removed dummy files from the 0.9.1b source tree: 13489 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 13490 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 13491 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 13492 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 13493 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 13494 [Ralf S. Engelschall] 13495 13496 *) Added various platform portability fixes. 13497 [Mark J. Cox] 13498 13499 *) The Genesis of the OpenSSL rpject: 13500 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 13501 Young and Tim J. Hudson created while they were working for C2Net until 13502 summer 1998. 13503 [The OpenSSL Project] 13504 13505 13506 Changes between 0.9.0b and 0.9.1b [not released] 13507 13508 *) Updated a few CA certificates under certs/ 13509 [Eric A. Young] 13510 13511 *) Changed some BIGNUM api stuff. 13512 [Eric A. Young] 13513 13514 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 13515 DGUX x86, Linux Alpha, etc. 13516 [Eric A. Young] 13517 13518 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 13519 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 13520 available). 13521 [Eric A. Young] 13522 13523 *) Add -strparse option to asn1pars program which parses nested 13524 binary structures 13525 [Dr Stephen Henson <shenson@bigfoot.com>] 13526 13527 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 13528 [Eric A. Young] 13529 13530 *) DSA fix for "ca" program. 13531 [Eric A. Young] 13532 13533 *) Added "-genkey" option to "dsaparam" program. 13534 [Eric A. Young] 13535 13536 *) Added RIPE MD160 (rmd160) message digest. 13537 [Eric A. Young] 13538 13539 *) Added -a (all) option to "ssleay version" command. 13540 [Eric A. Young] 13541 13542 *) Added PLATFORM define which is the id given to Configure. 13543 [Eric A. Young] 13544 13545 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 13546 [Eric A. Young] 13547 13548 *) Extended the ASN.1 parser routines. 13549 [Eric A. Young] 13550 13551 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 13552 [Eric A. Young] 13553 13554 *) Added a BN_CTX to the BN library. 13555 [Eric A. Young] 13556 13557 *) Fixed the weak key values in DES library 13558 [Eric A. Young] 13559 13560 *) Changed API in EVP library for cipher aliases. 13561 [Eric A. Young] 13562 13563 *) Added support for RC2/64bit cipher. 13564 [Eric A. Young] 13565 13566 *) Converted the lhash library to the crypto/mem.c functions. 13567 [Eric A. Young] 13568 13569 *) Added more recognized ASN.1 object ids. 13570 [Eric A. Young] 13571 13572 *) Added more RSA padding checks for SSL/TLS. 13573 [Eric A. Young] 13574 13575 *) Added BIO proxy/filter functionality. 13576 [Eric A. Young] 13577 13578 *) Added extra_certs to SSL_CTX which can be used 13579 send extra CA certificates to the client in the CA cert chain sending 13580 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 13581 [Eric A. Young] 13582 13583 *) Now Fortezza is denied in the authentication phase because 13584 this is key exchange mechanism is not supported by SSLeay at all. 13585 [Eric A. Young] 13586 13587 *) Additional PKCS1 checks. 13588 [Eric A. Young] 13589 13590 *) Support the string "TLSv1" for all TLS v1 ciphers. 13591 [Eric A. Young] 13592 13593 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 13594 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 13595 [Eric A. Young] 13596 13597 *) Fixed a few memory leaks. 13598 [Eric A. Young] 13599 13600 *) Fixed various code and comment typos. 13601 [Eric A. Young] 13602 13603 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 13604 bytes sent in the client random. 13605 [Edward Bishop <ebishop@spyglass.com>] 13606