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 1.1.1n and 1.1.1o [3 May 2022] 11 12 *) Fixed a bug in the c_rehash script which was not properly sanitising shell 13 metacharacters to prevent command injection. This script is distributed by 14 some operating systems in a manner where it is automatically executed. On 15 such operating systems, an attacker could execute arbitrary commands with the 16 privileges of the script. 17 18 Use of the c_rehash script is considered obsolete and should be replaced 19 by the OpenSSL rehash command line tool. 20 (CVE-2022-1292) 21 [Tomáš Mráz] 22 23 Changes between 1.1.1m and 1.1.1n [15 Mar 2022] 24 25 *) Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever 26 for non-prime moduli. 27 28 Internally this function is used when parsing certificates that contain 29 elliptic curve public keys in compressed form or explicit elliptic curve 30 parameters with a base point encoded in compressed form. 31 32 It is possible to trigger the infinite loop by crafting a certificate that 33 has invalid explicit curve parameters. 34 35 Since certificate parsing happens prior to verification of the certificate 36 signature, any process that parses an externally supplied certificate may 37 thus be subject to a denial of service attack. The infinite loop can also 38 be reached when parsing crafted private keys as they can contain explicit 39 elliptic curve parameters. 40 41 Thus vulnerable situations include: 42 43 - TLS clients consuming server certificates 44 - TLS servers consuming client certificates 45 - Hosting providers taking certificates or private keys from customers 46 - Certificate authorities parsing certification requests from subscribers 47 - Anything else which parses ASN.1 elliptic curve parameters 48 49 Also any other applications that use the BN_mod_sqrt() where the attacker 50 can control the parameter values are vulnerable to this DoS issue. 51 (CVE-2022-0778) 52 [Tomáš Mráz] 53 54 *) Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489) 55 to the list of ciphersuites providing Perfect Forward Secrecy as 56 required by SECLEVEL >= 3. 57 58 [Dmitry Belyavskiy, Nicola Tuveri] 59 60 Changes between 1.1.1l and 1.1.1m [14 Dec 2021] 61 62 *) Avoid loading of a dynamic engine twice. 63 64 [Bernd Edlinger] 65 66 *) Fixed building on Debian with kfreebsd kernels 67 68 [Mattias Ellert] 69 70 *) Prioritise DANE TLSA issuer certs over peer certs 71 72 [Viktor Dukhovni] 73 74 *) Fixed random API for MacOS prior to 10.12 75 76 These MacOS versions don't support the CommonCrypto APIs 77 78 [Lenny Primak] 79 80 Changes between 1.1.1k and 1.1.1l [24 Aug 2021] 81 82 *) Fixed an SM2 Decryption Buffer Overflow. 83 84 In order to decrypt SM2 encrypted data an application is expected to call the 85 API function EVP_PKEY_decrypt(). Typically an application will call this 86 function twice. The first time, on entry, the "out" parameter can be NULL and, 87 on exit, the "outlen" parameter is populated with the buffer size required to 88 hold the decrypted plaintext. The application can then allocate a sufficiently 89 sized buffer and call EVP_PKEY_decrypt() again, but this time passing a non-NULL 90 value for the "out" parameter. 91 92 A bug in the implementation of the SM2 decryption code means that the 93 calculation of the buffer size required to hold the plaintext returned by the 94 first call to EVP_PKEY_decrypt() can be smaller than the actual size required by 95 the second call. This can lead to a buffer overflow when EVP_PKEY_decrypt() is 96 called by the application a second time with a buffer that is too small. 97 98 A malicious attacker who is able present SM2 content for decryption to an 99 application could cause attacker chosen data to overflow the buffer by up to a 100 maximum of 62 bytes altering the contents of other data held after the 101 buffer, possibly changing application behaviour or causing the application to 102 crash. The location of the buffer is application dependent but is typically 103 heap allocated. 104 (CVE-2021-3711) 105 [Matt Caswell] 106 107 *) Fixed various read buffer overruns processing ASN.1 strings 108 109 ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING 110 structure which contains a buffer holding the string data and a field holding 111 the buffer length. This contrasts with normal C strings which are repesented as 112 a buffer for the string data which is terminated with a NUL (0) byte. 113 114 Although not a strict requirement, ASN.1 strings that are parsed using OpenSSL's 115 own "d2i" functions (and other similar parsing functions) as well as any string 116 whose value has been set with the ASN1_STRING_set() function will additionally 117 NUL terminate the byte array in the ASN1_STRING structure. 118 119 However, it is possible for applications to directly construct valid ASN1_STRING 120 structures which do not NUL terminate the byte array by directly setting the 121 "data" and "length" fields in the ASN1_STRING array. This can also happen by 122 using the ASN1_STRING_set0() function. 123 124 Numerous OpenSSL functions that print ASN.1 data have been found to assume that 125 the ASN1_STRING byte array will be NUL terminated, even though this is not 126 guaranteed for strings that have been directly constructed. Where an application 127 requests an ASN.1 structure to be printed, and where that ASN.1 structure 128 contains ASN1_STRINGs that have been directly constructed by the application 129 without NUL terminating the "data" field, then a read buffer overrun can occur. 130 131 The same thing can also occur during name constraints processing of certificates 132 (for example if a certificate has been directly constructed by the application 133 instead of loading it via the OpenSSL parsing functions, and the certificate 134 contains non NUL terminated ASN1_STRING structures). It can also occur in the 135 X509_get1_email(), X509_REQ_get1_email() and X509_get1_ocsp() functions. 136 137 If a malicious actor can cause an application to directly construct an 138 ASN1_STRING and then process it through one of the affected OpenSSL functions 139 then this issue could be hit. This might result in a crash (causing a Denial of 140 Service attack). It could also result in the disclosure of private memory 141 contents (such as private keys, or sensitive plaintext). 142 (CVE-2021-3712) 143 [Matt Caswell] 144 145 Changes between 1.1.1j and 1.1.1k [25 Mar 2021] 146 147 *) Fixed a problem with verifying a certificate chain when using the 148 X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks 149 of the certificates present in a certificate chain. It is not set by 150 default. 151 152 Starting from OpenSSL version 1.1.1h a check to disallow certificates in 153 the chain that have explicitly encoded elliptic curve parameters was added 154 as an additional strict check. 155 156 An error in the implementation of this check meant that the result of a 157 previous check to confirm that certificates in the chain are valid CA 158 certificates was overwritten. This effectively bypasses the check 159 that non-CA certificates must not be able to issue other certificates. 160 161 If a "purpose" has been configured then there is a subsequent opportunity 162 for checks that the certificate is a valid CA. All of the named "purpose" 163 values implemented in libcrypto perform this check. Therefore, where 164 a purpose is set the certificate chain will still be rejected even when the 165 strict flag has been used. A purpose is set by default in libssl client and 166 server certificate verification routines, but it can be overridden or 167 removed by an application. 168 169 In order to be affected, an application must explicitly set the 170 X509_V_FLAG_X509_STRICT verification flag and either not set a purpose 171 for the certificate verification or, in the case of TLS client or server 172 applications, override the default purpose. 173 (CVE-2021-3450) 174 [Tomáš Mráz] 175 176 *) Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously 177 crafted renegotiation ClientHello message from a client. If a TLSv1.2 178 renegotiation ClientHello omits the signature_algorithms extension (where 179 it was present in the initial ClientHello), but includes a 180 signature_algorithms_cert extension then a NULL pointer dereference will 181 result, leading to a crash and a denial of service attack. 182 183 A server is only vulnerable if it has TLSv1.2 and renegotiation enabled 184 (which is the default configuration). OpenSSL TLS clients are not impacted 185 by this issue. 186 (CVE-2021-3449) 187 [Peter Kästle and Samuel Sapalski] 188 189 Changes between 1.1.1i and 1.1.1j [16 Feb 2021] 190 191 *) Fixed the X509_issuer_and_serial_hash() function. It attempts to 192 create a unique hash value based on the issuer and serial number data 193 contained within an X509 certificate. However it was failing to correctly 194 handle any errors that may occur while parsing the issuer field (which might 195 occur if the issuer field is maliciously constructed). This may subsequently 196 result in a NULL pointer deref and a crash leading to a potential denial of 197 service attack. 198 (CVE-2021-23841) 199 [Matt Caswell] 200 201 *) Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING 202 padding mode to correctly check for rollback attacks. This is considered a 203 bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is 204 CVE-2021-23839. 205 [Matt Caswell] 206 207 *) Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate 208 functions. Previously they could overflow the output length argument in some 209 cases where the input length is close to the maximum permissable length for 210 an integer on the platform. In such cases the return value from the function 211 call would be 1 (indicating success), but the output length value would be 212 negative. This could cause applications to behave incorrectly or crash. 213 (CVE-2021-23840) 214 [Matt Caswell] 215 216 *) Fixed SRP_Calc_client_key so that it runs in constant time. The previous 217 implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This 218 could be exploited in a side channel attack to recover the password. Since 219 the attack is local host only this is outside of the current OpenSSL 220 threat model and therefore no CVE is assigned. 221 222 Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this 223 issue. 224 [Matt Caswell] 225 226 Changes between 1.1.1h and 1.1.1i [8 Dec 2020] 227 228 *) Fixed NULL pointer deref in the GENERAL_NAME_cmp function 229 This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME. 230 If an attacker can control both items being compared then this could lead 231 to a possible denial of service attack. OpenSSL itself uses the 232 GENERAL_NAME_cmp function for two purposes: 233 1) Comparing CRL distribution point names between an available CRL and a 234 CRL distribution point embedded in an X509 certificate 235 2) When verifying that a timestamp response token signer matches the 236 timestamp authority name (exposed via the API functions 237 TS_RESP_verify_response and TS_RESP_verify_token) 238 (CVE-2020-1971) 239 [Matt Caswell] 240 241 *) Add support for Apple Silicon M1 Macs with the darwin64-arm64-cc target. 242 [Stuart Carnie] 243 244 *) The security callback, which can be customised by application code, supports 245 the security operation SSL_SECOP_TMP_DH. This is defined to take an EVP_PKEY 246 in the "other" parameter. In most places this is what is passed. All these 247 places occur server side. However there was one client side call of this 248 security operation and it passed a DH object instead. This is incorrect 249 according to the definition of SSL_SECOP_TMP_DH, and is inconsistent with all 250 of the other locations. Therefore this client side call has been changed to 251 pass an EVP_PKEY instead. 252 [Matt Caswell] 253 254 *) In 1.1.1h, an expired trusted (root) certificate was not anymore rejected 255 when validating a certificate path. This check is restored in 1.1.1i. 256 [David von Oheimb] 257 258 Changes between 1.1.1g and 1.1.1h [22 Sep 2020] 259 260 *) Certificates with explicit curve parameters are now disallowed in 261 verification chains if the X509_V_FLAG_X509_STRICT flag is used. 262 [Tomas Mraz] 263 264 *) The 'MinProtocol' and 'MaxProtocol' configuration commands now silently 265 ignore TLS protocol version bounds when configuring DTLS-based contexts, and 266 conversely, silently ignore DTLS protocol version bounds when configuring 267 TLS-based contexts. The commands can be repeated to set bounds of both 268 types. The same applies with the corresponding "min_protocol" and 269 "max_protocol" command-line switches, in case some application uses both TLS 270 and DTLS. 271 272 SSL_CTX instances that are created for a fixed protocol version (e.g. 273 TLSv1_server_method()) also silently ignore version bounds. Previously 274 attempts to apply bounds to these protocol versions would result in an 275 error. Now only the "version-flexible" SSL_CTX instances are subject to 276 limits in configuration files in command-line options. 277 [Viktor Dukhovni] 278 279 *) Handshake now fails if Extended Master Secret extension is dropped 280 on renegotiation. 281 [Tomas Mraz] 282 283 *) Accidentally, an expired trusted (root) certificate is not anymore rejected 284 when validating a certificate path. 285 [David von Oheimb] 286 287 *) The Oracle Developer Studio compiler will start reporting deprecated APIs 288 289 Changes between 1.1.1f and 1.1.1g [21 Apr 2020] 290 291 *) Fixed segmentation fault in SSL_check_chain() 292 Server or client applications that call the SSL_check_chain() function 293 during or after a TLS 1.3 handshake may crash due to a NULL pointer 294 dereference as a result of incorrect handling of the 295 "signature_algorithms_cert" TLS extension. The crash occurs if an invalid 296 or unrecognised signature algorithm is received from the peer. This could 297 be exploited by a malicious peer in a Denial of Service attack. 298 (CVE-2020-1967) 299 [Benjamin Kaduk] 300 301 *) Added AES consttime code for no-asm configurations 302 an optional constant time support for AES was added 303 when building openssl for no-asm. 304 Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME 305 Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME 306 At this time this feature is by default disabled. 307 It will be enabled by default in 3.0. 308 [Bernd Edlinger] 309 310 Changes between 1.1.1e and 1.1.1f [31 Mar 2020] 311 312 *) Revert the change of EOF detection while reading in libssl to avoid 313 regressions in applications depending on the current way of reporting 314 the EOF. As the existing method is not fully accurate the change to 315 reporting the EOF via SSL_ERROR_SSL is kept on the current development 316 branch and will be present in the 3.0 release. 317 [Tomas Mraz] 318 319 *) Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1 320 when primes for RSA keys are computed. 321 Since we previously always generated primes == 2 (mod 3) for RSA keys, 322 the 2-prime and 3-prime RSA modules were easy to distinguish, since 323 N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting 324 2-prime vs. 3-prime RSA keys was possible by computing N mod 3. 325 This avoids possible fingerprinting of newly generated RSA modules. 326 [Bernd Edlinger] 327 328 Changes between 1.1.1d and 1.1.1e [17 Mar 2020] 329 *) Properly detect EOF while reading in libssl. Previously if we hit an EOF 330 while reading in libssl then we would report an error back to the 331 application (SSL_ERROR_SYSCALL) but errno would be 0. We now add 332 an error to the stack (which means we instead return SSL_ERROR_SSL) and 333 therefore give a hint as to what went wrong. 334 [Matt Caswell] 335 336 *) Check that ed25519 and ed448 are allowed by the security level. Previously 337 signature algorithms not using an MD were not being checked that they were 338 allowed by the security level. 339 [Kurt Roeckx] 340 341 *) Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername() 342 was not quite right. The behaviour was not consistent between resumption 343 and normal handshakes, and also not quite consistent with historical 344 behaviour. The behaviour in various scenarios has been clarified and 345 it has been updated to make it match historical behaviour as closely as 346 possible. 347 [Matt Caswell] 348 349 *) [VMS only] The header files that the VMS compilers include automatically, 350 __DECC_INCLUDE_PROLOGUE.H and __DECC_INCLUDE_EPILOGUE.H, use pragmas that 351 the C++ compiler doesn't understand. This is a shortcoming in the 352 compiler, but can be worked around with __cplusplus guards. 353 354 C++ applications that use OpenSSL libraries must be compiled using the 355 qualifier '/NAMES=(AS_IS,SHORTENED)' to be able to use all the OpenSSL 356 functions. Otherwise, only functions with symbols of less than 31 357 characters can be used, as the linker will not be able to successfully 358 resolve symbols with longer names. 359 [Richard Levitte] 360 361 *) Corrected the documentation of the return values from the EVP_DigestSign* 362 set of functions. The documentation mentioned negative values for some 363 errors, but this was never the case, so the mention of negative values 364 was removed. 365 366 Code that followed the documentation and thereby check with something 367 like 'EVP_DigestSignInit(...) <= 0' will continue to work undisturbed. 368 [Richard Levitte] 369 370 *) Fixed an an overflow bug in the x64_64 Montgomery squaring procedure 371 used in exponentiation with 512-bit moduli. No EC algorithms are 372 affected. Analysis suggests that attacks against 2-prime RSA1024, 373 3-prime RSA1536, and DSA1024 as a result of this defect would be very 374 difficult to perform and are not believed likely. Attacks against DH512 375 are considered just feasible. However, for an attack the target would 376 have to re-use the DH512 private key, which is not recommended anyway. 377 Also applications directly using the low level API BN_mod_exp may be 378 affected if they use BN_FLG_CONSTTIME. 379 (CVE-2019-1551) 380 [Andy Polyakov] 381 382 *) Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY. 383 The presence of this system service is determined at run-time. 384 [Richard Levitte] 385 386 *) Added newline escaping functionality to a filename when using openssl dgst. 387 This output format is to replicate the output format found in the '*sum' 388 checksum programs. This aims to preserve backward compatibility. 389 [Matt Eaton, Richard Levitte, and Paul Dale] 390 391 *) Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just 392 the first value. 393 [Jon Spillett] 394 395 Changes between 1.1.1c and 1.1.1d [10 Sep 2019] 396 397 *) Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random 398 number generator (RNG). This was intended to include protection in the 399 event of a fork() system call in order to ensure that the parent and child 400 processes did not share the same RNG state. However this protection was not 401 being used in the default case. 402 403 A partial mitigation for this issue is that the output from a high 404 precision timer is mixed into the RNG state so the likelihood of a parent 405 and child process sharing state is significantly reduced. 406 407 If an application already calls OPENSSL_init_crypto() explicitly using 408 OPENSSL_INIT_ATFORK then this problem does not occur at all. 409 (CVE-2019-1549) 410 [Matthias St. Pierre] 411 412 *) For built-in EC curves, ensure an EC_GROUP built from the curve name is 413 used even when parsing explicit parameters, when loading a serialized key 414 or calling `EC_GROUP_new_from_ecpkparameters()`/ 415 `EC_GROUP_new_from_ecparameters()`. 416 This prevents bypass of security hardening and performance gains, 417 especially for curves with specialized EC_METHODs. 418 By default, if a key encoded with explicit parameters is loaded and later 419 serialized, the output is still encoded with explicit parameters, even if 420 internally a "named" EC_GROUP is used for computation. 421 [Nicola Tuveri] 422 423 *) Compute ECC cofactors if not provided during EC_GROUP construction. Before 424 this change, EC_GROUP_set_generator would accept order and/or cofactor as 425 NULL. After this change, only the cofactor parameter can be NULL. It also 426 does some minimal sanity checks on the passed order. 427 (CVE-2019-1547) 428 [Billy Bob Brumley] 429 430 *) Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey. 431 An attack is simple, if the first CMS_recipientInfo is valid but the 432 second CMS_recipientInfo is chosen ciphertext. If the second 433 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct 434 encryption key will be replaced by garbage, and the message cannot be 435 decoded, but if the RSA decryption fails, the correct encryption key is 436 used and the recipient will not notice the attack. 437 As a work around for this potential attack the length of the decrypted 438 key must be equal to the cipher default key length, in case the 439 certifiate is not given and all recipientInfo are tried out. 440 The old behaviour can be re-enabled in the CMS code by setting the 441 CMS_DEBUG_DECRYPT flag. 442 (CVE-2019-1563) 443 [Bernd Edlinger] 444 445 *) Early start up entropy quality from the DEVRANDOM seed source has been 446 improved for older Linux systems. The RAND subsystem will wait for 447 /dev/random to be producing output before seeding from /dev/urandom. 448 The seeded state is stored for future library initialisations using 449 a system global shared memory segment. The shared memory identifier 450 can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to 451 the desired value. The default identifier is 114. 452 [Paul Dale] 453 454 *) Correct the extended master secret constant on EBCDIC systems. Without this 455 fix TLS connections between an EBCDIC system and a non-EBCDIC system that 456 negotiate EMS will fail. Unfortunately this also means that TLS connections 457 between EBCDIC systems with this fix, and EBCDIC systems without this 458 fix will fail if they negotiate EMS. 459 [Matt Caswell] 460 461 *) Use Windows installation paths in the mingw builds 462 463 Mingw isn't a POSIX environment per se, which means that Windows 464 paths should be used for installation. 465 (CVE-2019-1552) 466 [Richard Levitte] 467 468 *) Changed DH_check to accept parameters with order q and 2q subgroups. 469 With order 2q subgroups the bit 0 of the private key is not secret 470 but DH_generate_key works around that by clearing bit 0 of the 471 private key for those. This avoids leaking bit 0 of the private key. 472 [Bernd Edlinger] 473 474 *) Significantly reduce secure memory usage by the randomness pools. 475 [Paul Dale] 476 477 *) Revert the DEVRANDOM_WAIT feature for Linux systems 478 479 The DEVRANDOM_WAIT feature added a select() call to wait for the 480 /dev/random device to become readable before reading from the 481 /dev/urandom device. 482 483 It turned out that this change had negative side effects on 484 performance which were not acceptable. After some discussion it 485 was decided to revert this feature and leave it up to the OS 486 resp. the platform maintainer to ensure a proper initialization 487 during early boot time. 488 [Matthias St. Pierre] 489 490 Changes between 1.1.1b and 1.1.1c [28 May 2019] 491 492 *) Add build tests for C++. These are generated files that only do one 493 thing, to include one public OpenSSL head file each. This tests that 494 the public header files can be usefully included in a C++ application. 495 496 This test isn't enabled by default. It can be enabled with the option 497 'enable-buildtest-c++'. 498 [Richard Levitte] 499 500 *) Enable SHA3 pre-hashing for ECDSA and DSA. 501 [Patrick Steuer] 502 503 *) Change the default RSA, DSA and DH size to 2048 bit instead of 1024. 504 This changes the size when using the genpkey app when no size is given. It 505 fixes an omission in earlier changes that changed all RSA, DSA and DH 506 generation apps to use 2048 bits by default. 507 [Kurt Roeckx] 508 509 *) Reorganize the manual pages to consistently have RETURN VALUES, 510 EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust 511 util/fix-doc-nits accordingly. 512 [Paul Yang, Joshua Lock] 513 514 *) Add the missing accessor EVP_PKEY_get0_engine() 515 [Matt Caswell] 516 517 *) Have apps like 's_client' and 's_server' output the signature scheme 518 along with other cipher suite parameters when debugging. 519 [Lorinczy Zsigmond] 520 521 *) Make OPENSSL_config() error agnostic again. 522 [Richard Levitte] 523 524 *) Do the error handling in RSA decryption constant time. 525 [Bernd Edlinger] 526 527 *) Prevent over long nonces in ChaCha20-Poly1305. 528 529 ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input 530 for every encryption operation. RFC 7539 specifies that the nonce value 531 (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length 532 and front pads the nonce with 0 bytes if it is less than 12 533 bytes. However it also incorrectly allows a nonce to be set of up to 16 534 bytes. In this case only the last 12 bytes are significant and any 535 additional leading bytes are ignored. 536 537 It is a requirement of using this cipher that nonce values are 538 unique. Messages encrypted using a reused nonce value are susceptible to 539 serious confidentiality and integrity attacks. If an application changes 540 the default nonce length to be longer than 12 bytes and then makes a 541 change to the leading bytes of the nonce expecting the new value to be a 542 new unique nonce then such an application could inadvertently encrypt 543 messages with a reused nonce. 544 545 Additionally the ignored bytes in a long nonce are not covered by the 546 integrity guarantee of this cipher. Any application that relies on the 547 integrity of these ignored leading bytes of a long nonce may be further 548 affected. Any OpenSSL internal use of this cipher, including in SSL/TLS, 549 is safe because no such use sets such a long nonce value. However user 550 applications that use this cipher directly and set a non-default nonce 551 length to be longer than 12 bytes may be vulnerable. 552 553 This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk 554 Greef of Ronomon. 555 (CVE-2019-1543) 556 [Matt Caswell] 557 558 *) Add DEVRANDOM_WAIT feature for Linux systems 559 560 On older Linux systems where the getrandom() system call is not available, 561 OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG. 562 Contrary to getrandom(), the /dev/urandom device will not block during 563 early boot when the kernel CSPRNG has not been seeded yet. 564 565 To mitigate this known weakness, use select() to wait for /dev/random to 566 become readable before reading from /dev/urandom. 567 568 *) Ensure that SM2 only uses SM3 as digest algorithm 569 [Paul Yang] 570 571 Changes between 1.1.1a and 1.1.1b [26 Feb 2019] 572 573 *) Added SCA hardening for modular field inversion in EC_GROUP through 574 a new dedicated field_inv() pointer in EC_METHOD. 575 This also addresses a leakage affecting conversions from projective 576 to affine coordinates. 577 [Billy Bob Brumley, Nicola Tuveri] 578 579 *) Change the info callback signals for the start and end of a post-handshake 580 message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START 581 and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get 582 confused by this and assume that a TLSv1.2 renegotiation has started. This 583 can break KeyUpdate handling. Instead we no longer signal the start and end 584 of a post handshake message exchange (although the messages themselves are 585 still signalled). This could break some applications that were expecting 586 the old signals. However without this KeyUpdate is not usable for many 587 applications. 588 [Matt Caswell] 589 590 *) Fix a bug in the computation of the endpoint-pair shared secret used 591 by DTLS over SCTP. This breaks interoperability with older versions 592 of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime 593 switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling 594 interoperability with such broken implementations. However, enabling 595 this switch breaks interoperability with correct implementations. 596 597 *) Fix a use after free bug in d2i_X509_PUBKEY when overwriting a 598 re-used X509_PUBKEY object if the second PUBKEY is malformed. 599 [Bernd Edlinger] 600 601 *) Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0(). 602 [Richard Levitte] 603 604 *) Remove the 'dist' target and add a tarball building script. The 605 'dist' target has fallen out of use, and it shouldn't be 606 necessary to configure just to create a source distribution. 607 [Richard Levitte] 608 609 *) Added support for Linux Kernel TLS data-path. The Linux Kernel data-path 610 improves application performance by removing data copies and providing 611 applications with zero-copy system calls such as sendfile and splice. 612 [Boris Pismenny] 613 614 Changes between 1.1.1 and 1.1.1a [20 Nov 2018] 615 616 *) Timing vulnerability in DSA signature generation 617 618 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a 619 timing side channel attack. An attacker could use variations in the signing 620 algorithm to recover the private key. 621 622 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser. 623 (CVE-2018-0734) 624 [Paul Dale] 625 626 *) Timing vulnerability in ECDSA signature generation 627 628 The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a 629 timing side channel attack. An attacker could use variations in the signing 630 algorithm to recover the private key. 631 632 This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser. 633 (CVE-2018-0735) 634 [Paul Dale] 635 636 *) Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for 637 the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names 638 are retained for backwards compatibility. 639 [Antoine Salon] 640 641 *) Fixed the issue that RAND_add()/RAND_seed() silently discards random input 642 if its length exceeds 4096 bytes. The limit has been raised to a buffer size 643 of two gigabytes and the error handling improved. 644 645 This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been 646 categorized as a normal bug, not a security issue, because the DRBG reseeds 647 automatically and is fully functional even without additional randomness 648 provided by the application. 649 650 Changes between 1.1.0i and 1.1.1 [11 Sep 2018] 651 652 *) Add a new ClientHello callback. Provides a callback interface that gives 653 the application the ability to adjust the nascent SSL object at the 654 earliest stage of ClientHello processing, immediately after extensions have 655 been collected but before they have been processed. In particular, this 656 callback can adjust the supported TLS versions in response to the contents 657 of the ClientHello 658 [Benjamin Kaduk] 659 660 *) Add SM2 base algorithm support. 661 [Jack Lloyd] 662 663 *) s390x assembly pack: add (improved) hardware-support for the following 664 cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb, 665 aes-cfb/cfb8, aes-ecb. 666 [Patrick Steuer] 667 668 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 669 parameter is no longer accepted, as it leads to a corrupt table. NULL 670 pem_str is reserved for alias entries only. 671 [Richard Levitte] 672 673 *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 674 step for prime curves. The new implementation is based on formulae from 675 differential addition-and-doubling in homogeneous projective coordinates 676 from Izu-Takagi "A fast parallel elliptic curve multiplication resistant 677 against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves 678 and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified 679 to work in projective coordinates. 680 [Billy Bob Brumley, Nicola Tuveri] 681 682 *) Change generating and checking of primes so that the error rate of not 683 being prime depends on the intended use based on the size of the input. 684 For larger primes this will result in more rounds of Miller-Rabin. 685 The maximal error rate for primes with more than 1080 bits is lowered 686 to 2^-128. 687 [Kurt Roeckx, Annie Yousar] 688 689 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 690 [Kurt Roeckx] 691 692 *) The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when 693 moving between systems, and to avoid confusion when a Windows build is 694 done with mingw vs with MSVC. For POSIX installs, there's still a 695 symlink or copy named 'tsget' to avoid that confusion as well. 696 [Richard Levitte] 697 698 *) Revert blinding in ECDSA sign and instead make problematic addition 699 length-invariant. Switch even to fixed-length Montgomery multiplication. 700 [Andy Polyakov] 701 702 *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 703 step for binary curves. The new implementation is based on formulae from 704 differential addition-and-doubling in mixed Lopez-Dahab projective 705 coordinates, modified to independently blind the operands. 706 [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri] 707 708 *) Add a scaffold to optionally enhance the Montgomery ladder implementation 709 for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing 710 EC_METHODs to implement their own specialized "ladder step", to take 711 advantage of more favorable coordinate systems or more efficient 712 differential addition-and-doubling algorithms. 713 [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri] 714 715 *) Modified the random device based seed sources to keep the relevant 716 file descriptors open rather than reopening them on each access. 717 This allows such sources to operate in a chroot() jail without 718 the associated device nodes being available. This behaviour can be 719 controlled using RAND_keep_random_devices_open(). 720 [Paul Dale] 721 722 *) Numerous side-channel attack mitigations have been applied. This may have 723 performance impacts for some algorithms for the benefit of improved 724 security. Specific changes are noted in this change log by their respective 725 authors. 726 [Matt Caswell] 727 728 *) AIX shared library support overhaul. Switch to AIX "natural" way of 729 handling shared libraries, which means collecting shared objects of 730 different versions and bitnesses in one common archive. This allows to 731 mitigate conflict between 1.0 and 1.1 side-by-side installations. It 732 doesn't affect the way 3rd party applications are linked, only how 733 multi-version installation is managed. 734 [Andy Polyakov] 735 736 *) Make ec_group_do_inverse_ord() more robust and available to other 737 EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA 738 mitigations are applied to the fallback BN_mod_inverse(). 739 When using this function rather than BN_mod_inverse() directly, new 740 EC cryptosystem implementations are then safer-by-default. 741 [Billy Bob Brumley] 742 743 *) Add coordinate blinding for EC_POINT and implement projective 744 coordinate blinding for generic prime curves as a countermeasure to 745 chosen point SCA attacks. 746 [Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley] 747 748 *) Add blinding to ECDSA and DSA signatures to protect against side channel 749 attacks discovered by Keegan Ryan (NCC Group). 750 [Matt Caswell] 751 752 *) Enforce checking in the pkeyutl command line app to ensure that the input 753 length does not exceed the maximum supported digest length when performing 754 a sign, verify or verifyrecover operation. 755 [Matt Caswell] 756 757 *) SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking 758 I/O in combination with something like select() or poll() will hang. This 759 can be turned off again using SSL_CTX_clear_mode(). 760 Many applications do not properly handle non-application data records, and 761 TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works 762 around the problems in those applications, but can also break some. 763 It's recommended to read the manpages about SSL_read(), SSL_write(), 764 SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and 765 SSL_CTX_set_read_ahead() again. 766 [Kurt Roeckx] 767 768 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 769 now allow empty (zero character) pass phrases. 770 [Richard Levitte] 771 772 *) Apply blinding to binary field modular inversion and remove patent 773 pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation. 774 [Billy Bob Brumley] 775 776 *) Deprecate ec2_mult.c and unify scalar multiplication code paths for 777 binary and prime elliptic curves. 778 [Billy Bob Brumley] 779 780 *) Remove ECDSA nonce padding: EC_POINT_mul is now responsible for 781 constant time fixed point multiplication. 782 [Billy Bob Brumley] 783 784 *) Revise elliptic curve scalar multiplication with timing attack 785 defenses: ec_wNAF_mul redirects to a constant time implementation 786 when computing fixed point and variable point multiplication (which 787 in OpenSSL are mostly used with secret scalars in keygen, sign, 788 ECDH derive operations). 789 [Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García, 790 Sohaib ul Hassan] 791 792 *) Updated CONTRIBUTING 793 [Rich Salz] 794 795 *) Updated DRBG / RAND to request nonce and additional low entropy 796 randomness from the system. 797 [Matthias St. Pierre] 798 799 *) Updated 'openssl rehash' to use OpenSSL consistent default. 800 [Richard Levitte] 801 802 *) Moved the load of the ssl_conf module to libcrypto, which helps 803 loading engines that libssl uses before libssl is initialised. 804 [Matt Caswell] 805 806 *) Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA 807 [Matt Caswell] 808 809 *) Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases. 810 [Ingo Schwarze, Rich Salz] 811 812 *) Added output of accepting IP address and port for 'openssl s_server' 813 [Richard Levitte] 814 815 *) Added a new API for TLSv1.3 ciphersuites: 816 SSL_CTX_set_ciphersuites() 817 SSL_set_ciphersuites() 818 [Matt Caswell] 819 820 *) Memory allocation failures consistently add an error to the error 821 stack. 822 [Rich Salz] 823 824 *) Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values 825 in libcrypto when run as setuid/setgid. 826 [Bernd Edlinger] 827 828 *) Load any config file by default when libssl is used. 829 [Matt Caswell] 830 831 *) Added new public header file <openssl/rand_drbg.h> and documentation 832 for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview. 833 [Matthias St. Pierre] 834 835 *) QNX support removed (cannot find contributors to get their approval 836 for the license change). 837 [Rich Salz] 838 839 *) TLSv1.3 replay protection for early data has been implemented. See the 840 SSL_read_early_data() man page for further details. 841 [Matt Caswell] 842 843 *) Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite 844 configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and 845 below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3. 846 In order to avoid issues where legacy TLSv1.2 ciphersuite configuration 847 would otherwise inadvertently disable all TLSv1.3 ciphersuites the 848 configuration has been separated out. See the ciphers man page or the 849 SSL_CTX_set_ciphersuites() man page for more information. 850 [Matt Caswell] 851 852 *) On POSIX (BSD, Linux, ...) systems the ocsp(1) command running 853 in responder mode now supports the new "-multi" option, which 854 spawns the specified number of child processes to handle OCSP 855 requests. The "-timeout" option now also limits the OCSP 856 responder's patience to wait to receive the full client request 857 on a newly accepted connection. Child processes are respawned 858 as needed, and the CA index file is automatically reloaded 859 when changed. This makes it possible to run the "ocsp" responder 860 as a long-running service, making the OpenSSL CA somewhat more 861 feature-complete. In this mode, most diagnostic messages logged 862 after entering the event loop are logged via syslog(3) rather than 863 written to stderr. 864 [Viktor Dukhovni] 865 866 *) Added support for X448 and Ed448. Heavily based on original work by 867 Mike Hamburg. 868 [Matt Caswell] 869 870 *) Extend OSSL_STORE with capabilities to search and to narrow the set of 871 objects loaded. This adds the functions OSSL_STORE_expect() and 872 OSSL_STORE_find() as well as needed tools to construct searches and 873 get the search data out of them. 874 [Richard Levitte] 875 876 *) Support for TLSv1.3 added. Note that users upgrading from an earlier 877 version of OpenSSL should review their configuration settings to ensure 878 that they are still appropriate for TLSv1.3. For further information see: 879 https://wiki.openssl.org/index.php/TLS1.3 880 [Matt Caswell] 881 882 *) Grand redesign of the OpenSSL random generator 883 884 The default RAND method now utilizes an AES-CTR DRBG according to 885 NIST standard SP 800-90Ar1. The new random generator is essentially 886 a port of the default random generator from the OpenSSL FIPS 2.0 887 object module. It is a hybrid deterministic random bit generator 888 using an AES-CTR bit stream and which seeds and reseeds itself 889 automatically using trusted system entropy sources. 890 891 Some of its new features are: 892 o Support for multiple DRBG instances with seed chaining. 893 o The default RAND method makes use of a DRBG. 894 o There is a public and private DRBG instance. 895 o The DRBG instances are fork-safe. 896 o Keep all global DRBG instances on the secure heap if it is enabled. 897 o The public and private DRBG instance are per thread for lock free 898 operation 899 [Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre] 900 901 *) Changed Configure so it only says what it does and doesn't dump 902 so much data. Instead, ./configdata.pm should be used as a script 903 to display all sorts of configuration data. 904 [Richard Levitte] 905 906 *) Added processing of "make variables" to Configure. 907 [Richard Levitte] 908 909 *) Added SHA512/224 and SHA512/256 algorithm support. 910 [Paul Dale] 911 912 *) The last traces of Netware support, first removed in 1.1.0, have 913 now been removed. 914 [Rich Salz] 915 916 *) Get rid of Makefile.shared, and in the process, make the processing 917 of certain files (rc.obj, or the .def/.map/.opt files produced from 918 the ordinal files) more visible and hopefully easier to trace and 919 debug (or make silent). 920 [Richard Levitte] 921 922 *) Make it possible to have environment variable assignments as 923 arguments to config / Configure. 924 [Richard Levitte] 925 926 *) Add multi-prime RSA (RFC 8017) support. 927 [Paul Yang] 928 929 *) Add SM3 implemented according to GB/T 32905-2016 930 [ Jack Lloyd <[email protected]>, 931 Ronald Tse <[email protected]>, 932 Erick Borsboom <[email protected]> ] 933 934 *) Add 'Maximum Fragment Length' TLS extension negotiation and support 935 as documented in RFC6066. 936 Based on a patch from Tomasz Moń 937 [Filipe Raimundo da Silva] 938 939 *) Add SM4 implemented according to GB/T 32907-2016. 940 [ Jack Lloyd <[email protected]>, 941 Ronald Tse <[email protected]>, 942 Erick Borsboom <[email protected]> ] 943 944 *) Reimplement -newreq-nodes and ERR_error_string_n; the 945 original author does not agree with the license change. 946 [Rich Salz] 947 948 *) Add ARIA AEAD TLS support. 949 [Jon Spillett] 950 951 *) Some macro definitions to support VS6 have been removed. Visual 952 Studio 6 has not worked since 1.1.0 953 [Rich Salz] 954 955 *) Add ERR_clear_last_mark(), to allow callers to clear the last mark 956 without clearing the errors. 957 [Richard Levitte] 958 959 *) Add "atfork" functions. If building on a system that without 960 pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application 961 requirements. The RAND facility now uses/requires this. 962 [Rich Salz] 963 964 *) Add SHA3. 965 [Andy Polyakov] 966 967 *) The UI API becomes a permanent and integral part of libcrypto, i.e. 968 not possible to disable entirely. However, it's still possible to 969 disable the console reading UI method, UI_OpenSSL() (use UI_null() 970 as a fallback). 971 972 To disable, configure with 'no-ui-console'. 'no-ui' is still 973 possible to use as an alias. Check at compile time with the 974 macro OPENSSL_NO_UI_CONSOLE. The macro OPENSSL_NO_UI is still 975 possible to check and is an alias for OPENSSL_NO_UI_CONSOLE. 976 [Richard Levitte] 977 978 *) Add a STORE module, which implements a uniform and URI based reader of 979 stores that can contain keys, certificates, CRLs and numerous other 980 objects. The main API is loosely based on a few stdio functions, 981 and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof, 982 OSSL_STORE_error and OSSL_STORE_close. 983 The implementation uses backends called "loaders" to implement arbitrary 984 URI schemes. There is one built in "loader" for the 'file' scheme. 985 [Richard Levitte] 986 987 *) Add devcrypto engine. This has been implemented against cryptodev-linux, 988 then adjusted to work on FreeBSD 8.4 as well. 989 Enable by configuring with 'enable-devcryptoeng'. This is done by default 990 on BSD implementations, as cryptodev.h is assumed to exist on all of them. 991 [Richard Levitte] 992 993 *) Module names can prefixed with OSSL_ or OPENSSL_. This affects 994 util/mkerr.pl, which is adapted to allow those prefixes, leading to 995 error code calls like this: 996 997 OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER); 998 999 With this change, we claim the namespaces OSSL and OPENSSL in a manner 1000 that can be encoded in C. For the foreseeable future, this will only 1001 affect new modules. 1002 [Richard Levitte and Tim Hudson] 1003 1004 *) Removed BSD cryptodev engine. 1005 [Rich Salz] 1006 1007 *) Add a build target 'build_all_generated', to build all generated files 1008 and only that. This can be used to prepare everything that requires 1009 things like perl for a system that lacks perl and then move everything 1010 to that system and do the rest of the build there. 1011 [Richard Levitte] 1012 1013 *) In the UI interface, make it possible to duplicate the user data. This 1014 can be used by engines that need to retain the data for a longer time 1015 than just the call where this user data is passed. 1016 [Richard Levitte] 1017 1018 *) Ignore the '-named_curve auto' value for compatibility of applications 1019 with OpenSSL 1.0.2. 1020 [Tomas Mraz <[email protected]>] 1021 1022 *) Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2 1023 bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such 1024 alerts across multiple records (some of which could be empty). In practice 1025 it make no sense to send an empty alert record, or to fragment one. TLSv1.3 1026 prohibits this altogether and other libraries (BoringSSL, NSS) do not 1027 support this at all. Supporting it adds significant complexity to the 1028 record layer, and its removal is unlikely to cause interoperability 1029 issues. 1030 [Matt Caswell] 1031 1032 *) Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed 1033 with Z. These are meant to replace LONG and ZLONG and to be size safe. 1034 The use of LONG and ZLONG is discouraged and scheduled for deprecation 1035 in OpenSSL 1.2.0. 1036 [Richard Levitte] 1037 1038 *) Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string, 1039 'z' is to be used for [s]size_t, and 'j' - with [u]int64_t. 1040 [Richard Levitte, Andy Polyakov] 1041 1042 *) Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine() 1043 does for RSA, etc. 1044 [Richard Levitte] 1045 1046 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 1047 platform rather than 'mingw'. 1048 [Richard Levitte] 1049 1050 *) The functions X509_STORE_add_cert and X509_STORE_add_crl return 1051 success if they are asked to add an object which already exists 1052 in the store. This change cascades to other functions which load 1053 certificates and CRLs. 1054 [Paul Dale] 1055 1056 *) x86_64 assembly pack: annotate code with DWARF CFI directives to 1057 facilitate stack unwinding even from assembly subroutines. 1058 [Andy Polyakov] 1059 1060 *) Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN. 1061 Also remove OPENSSL_GLOBAL entirely, as it became a no-op. 1062 [Richard Levitte] 1063 1064 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 1065 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 1066 which is the minimum version we support. 1067 [Richard Levitte] 1068 1069 *) Certificate time validation (X509_cmp_time) enforces stricter 1070 compliance with RFC 5280. Fractional seconds and timezone offsets 1071 are no longer allowed. 1072 [Emilia Käsper] 1073 1074 *) Add support for ARIA 1075 [Paul Dale] 1076 1077 *) s_client will now send the Server Name Indication (SNI) extension by 1078 default unless the new "-noservername" option is used. The server name is 1079 based on the host provided to the "-connect" option unless overridden by 1080 using "-servername". 1081 [Matt Caswell] 1082 1083 *) Add support for SipHash 1084 [Todd Short] 1085 1086 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0 1087 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 1088 prevent issues where no progress is being made and the peer continually 1089 sends unrecognised record types, using up resources processing them. 1090 [Matt Caswell] 1091 1092 *) 'openssl passwd' can now produce SHA256 and SHA512 based output, 1093 using the algorithm defined in 1094 https://www.akkadia.org/drepper/SHA-crypt.txt 1095 [Richard Levitte] 1096 1097 *) Heartbeat support has been removed; the ABI is changed for now. 1098 [Richard Levitte, Rich Salz] 1099 1100 *) Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd. 1101 [Emilia Käsper] 1102 1103 *) The RSA "null" method, which was partially supported to avoid patent 1104 issues, has been replaced to always returns NULL. 1105 [Rich Salz] 1106 1107 1108 Changes between 1.1.0h and 1.1.0i [xx XXX xxxx] 1109 1110 *) Client DoS due to large DH parameter 1111 1112 During key agreement in a TLS handshake using a DH(E) based ciphersuite a 1113 malicious server can send a very large prime value to the client. This will 1114 cause the client to spend an unreasonably long period of time generating a 1115 key for this prime resulting in a hang until the client has finished. This 1116 could be exploited in a Denial Of Service attack. 1117 1118 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken 1119 (CVE-2018-0732) 1120 [Guido Vranken] 1121 1122 *) Cache timing vulnerability in RSA Key Generation 1123 1124 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to 1125 a cache timing side channel attack. An attacker with sufficient access to 1126 mount cache timing attacks during the RSA key generation process could 1127 recover the private key. 1128 1129 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera 1130 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia. 1131 (CVE-2018-0737) 1132 [Billy Brumley] 1133 1134 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 1135 parameter is no longer accepted, as it leads to a corrupt table. NULL 1136 pem_str is reserved for alias entries only. 1137 [Richard Levitte] 1138 1139 *) Revert blinding in ECDSA sign and instead make problematic addition 1140 length-invariant. Switch even to fixed-length Montgomery multiplication. 1141 [Andy Polyakov] 1142 1143 *) Change generating and checking of primes so that the error rate of not 1144 being prime depends on the intended use based on the size of the input. 1145 For larger primes this will result in more rounds of Miller-Rabin. 1146 The maximal error rate for primes with more than 1080 bits is lowered 1147 to 2^-128. 1148 [Kurt Roeckx, Annie Yousar] 1149 1150 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 1151 [Kurt Roeckx] 1152 1153 *) Add blinding to ECDSA and DSA signatures to protect against side channel 1154 attacks discovered by Keegan Ryan (NCC Group). 1155 [Matt Caswell] 1156 1157 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 1158 now allow empty (zero character) pass phrases. 1159 [Richard Levitte] 1160 1161 *) Certificate time validation (X509_cmp_time) enforces stricter 1162 compliance with RFC 5280. Fractional seconds and timezone offsets 1163 are no longer allowed. 1164 [Emilia Käsper] 1165 1166 *) Fixed a text canonicalisation bug in CMS 1167 1168 Where a CMS detached signature is used with text content the text goes 1169 through a canonicalisation process first prior to signing or verifying a 1170 signature. This process strips trailing space at the end of lines, converts 1171 line terminators to CRLF and removes additional trailing line terminators 1172 at the end of a file. A bug in the canonicalisation process meant that 1173 some characters, such as form-feed, were incorrectly treated as whitespace 1174 and removed. This is contrary to the specification (RFC5485). This fix 1175 could mean that detached text data signed with an earlier version of 1176 OpenSSL 1.1.0 may fail to verify using the fixed version, or text data 1177 signed with a fixed OpenSSL may fail to verify with an earlier version of 1178 OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data 1179 and use the "-binary" flag (for the "cms" command line application) or set 1180 the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()). 1181 [Matt Caswell] 1182 1183 Changes between 1.1.0g and 1.1.0h [27 Mar 2018] 1184 1185 *) Constructed ASN.1 types with a recursive definition could exceed the stack 1186 1187 Constructed ASN.1 types with a recursive definition (such as can be found 1188 in PKCS7) could eventually exceed the stack given malicious input with 1189 excessive recursion. This could result in a Denial Of Service attack. There 1190 are no such structures used within SSL/TLS that come from untrusted sources 1191 so this is considered safe. 1192 1193 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz 1194 project. 1195 (CVE-2018-0739) 1196 [Matt Caswell] 1197 1198 *) Incorrect CRYPTO_memcmp on HP-UX PA-RISC 1199 1200 Because of an implementation bug the PA-RISC CRYPTO_memcmp function is 1201 effectively reduced to only comparing the least significant bit of each 1202 byte. This allows an attacker to forge messages that would be considered as 1203 authenticated in an amount of tries lower than that guaranteed by the 1204 security claims of the scheme. The module can only be compiled by the 1205 HP-UX assembler, so that only HP-UX PA-RISC targets are affected. 1206 1207 This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg 1208 (IBM). 1209 (CVE-2018-0733) 1210 [Andy Polyakov] 1211 1212 *) Add a build target 'build_all_generated', to build all generated files 1213 and only that. This can be used to prepare everything that requires 1214 things like perl for a system that lacks perl and then move everything 1215 to that system and do the rest of the build there. 1216 [Richard Levitte] 1217 1218 *) Backport SSL_OP_NO_RENGOTIATION 1219 1220 OpenSSL 1.0.2 and below had the ability to disable renegotiation using the 1221 (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity 1222 changes this is no longer possible in 1.1.0. Therefore the new 1223 SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to 1224 1.1.0 to provide equivalent functionality. 1225 1226 Note that if an application built against 1.1.0h headers (or above) is run 1227 using an older version of 1.1.0 (prior to 1.1.0h) then the option will be 1228 accepted but nothing will happen, i.e. renegotiation will not be prevented. 1229 [Matt Caswell] 1230 1231 *) Removed the OS390-Unix config target. It relied on a script that doesn't 1232 exist. 1233 [Rich Salz] 1234 1235 *) rsaz_1024_mul_avx2 overflow bug on x86_64 1236 1237 There is an overflow bug in the AVX2 Montgomery multiplication procedure 1238 used in exponentiation with 1024-bit moduli. No EC algorithms are affected. 1239 Analysis suggests that attacks against RSA and DSA as a result of this 1240 defect would be very difficult to perform and are not believed likely. 1241 Attacks against DH1024 are considered just feasible, because most of the 1242 work necessary to deduce information about a private key may be performed 1243 offline. The amount of resources required for such an attack would be 1244 significant. However, for an attack on TLS to be meaningful, the server 1245 would have to share the DH1024 private key among multiple clients, which is 1246 no longer an option since CVE-2016-0701. 1247 1248 This only affects processors that support the AVX2 but not ADX extensions 1249 like Intel Haswell (4th generation). 1250 1251 This issue was reported to OpenSSL by David Benjamin (Google). The issue 1252 was originally found via the OSS-Fuzz project. 1253 (CVE-2017-3738) 1254 [Andy Polyakov] 1255 1256 Changes between 1.1.0f and 1.1.0g [2 Nov 2017] 1257 1258 *) bn_sqrx8x_internal carry bug on x86_64 1259 1260 There is a carry propagating bug in the x86_64 Montgomery squaring 1261 procedure. No EC algorithms are affected. Analysis suggests that attacks 1262 against RSA and DSA as a result of this defect would be very difficult to 1263 perform and are not believed likely. Attacks against DH are considered just 1264 feasible (although very difficult) because most of the work necessary to 1265 deduce information about a private key may be performed offline. The amount 1266 of resources required for such an attack would be very significant and 1267 likely only accessible to a limited number of attackers. An attacker would 1268 additionally need online access to an unpatched system using the target 1269 private key in a scenario with persistent DH parameters and a private 1270 key that is shared between multiple clients. 1271 1272 This only affects processors that support the BMI1, BMI2 and ADX extensions 1273 like Intel Broadwell (5th generation) and later or AMD Ryzen. 1274 1275 This issue was reported to OpenSSL by the OSS-Fuzz project. 1276 (CVE-2017-3736) 1277 [Andy Polyakov] 1278 1279 *) Malformed X.509 IPAddressFamily could cause OOB read 1280 1281 If an X.509 certificate has a malformed IPAddressFamily extension, 1282 OpenSSL could do a one-byte buffer overread. The most likely result 1283 would be an erroneous display of the certificate in text format. 1284 1285 This issue was reported to OpenSSL by the OSS-Fuzz project. 1286 (CVE-2017-3735) 1287 [Rich Salz] 1288 1289 Changes between 1.1.0e and 1.1.0f [25 May 2017] 1290 1291 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 1292 platform rather than 'mingw'. 1293 [Richard Levitte] 1294 1295 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 1296 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 1297 which is the minimum version we support. 1298 [Richard Levitte] 1299 1300 Changes between 1.1.0d and 1.1.0e [16 Feb 2017] 1301 1302 *) Encrypt-Then-Mac renegotiation crash 1303 1304 During a renegotiation handshake if the Encrypt-Then-Mac extension is 1305 negotiated where it was not in the original handshake (or vice-versa) then 1306 this can cause OpenSSL to crash (dependant on ciphersuite). Both clients 1307 and servers are affected. 1308 1309 This issue was reported to OpenSSL by Joe Orton (Red Hat). 1310 (CVE-2017-3733) 1311 [Matt Caswell] 1312 1313 Changes between 1.1.0c and 1.1.0d [26 Jan 2017] 1314 1315 *) Truncated packet could crash via OOB read 1316 1317 If one side of an SSL/TLS path is running on a 32-bit host and a specific 1318 cipher is being used, then a truncated packet can cause that host to 1319 perform an out-of-bounds read, usually resulting in a crash. 1320 1321 This issue was reported to OpenSSL by Robert Święcki of Google. 1322 (CVE-2017-3731) 1323 [Andy Polyakov] 1324 1325 *) Bad (EC)DHE parameters cause a client crash 1326 1327 If a malicious server supplies bad parameters for a DHE or ECDHE key 1328 exchange then this can result in the client attempting to dereference a 1329 NULL pointer leading to a client crash. This could be exploited in a Denial 1330 of Service attack. 1331 1332 This issue was reported to OpenSSL by Guido Vranken. 1333 (CVE-2017-3730) 1334 [Matt Caswell] 1335 1336 *) BN_mod_exp may produce incorrect results on x86_64 1337 1338 There is a carry propagating bug in the x86_64 Montgomery squaring 1339 procedure. No EC algorithms are affected. Analysis suggests that attacks 1340 against RSA and DSA as a result of this defect would be very difficult to 1341 perform and are not believed likely. Attacks against DH are considered just 1342 feasible (although very difficult) because most of the work necessary to 1343 deduce information about a private key may be performed offline. The amount 1344 of resources required for such an attack would be very significant and 1345 likely only accessible to a limited number of attackers. An attacker would 1346 additionally need online access to an unpatched system using the target 1347 private key in a scenario with persistent DH parameters and a private 1348 key that is shared between multiple clients. For example this can occur by 1349 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 1350 similar to CVE-2015-3193 but must be treated as a separate problem. 1351 1352 This issue was reported to OpenSSL by the OSS-Fuzz project. 1353 (CVE-2017-3732) 1354 [Andy Polyakov] 1355 1356 Changes between 1.1.0b and 1.1.0c [10 Nov 2016] 1357 1358 *) ChaCha20/Poly1305 heap-buffer-overflow 1359 1360 TLS connections using *-CHACHA20-POLY1305 ciphersuites are susceptible to 1361 a DoS attack by corrupting larger payloads. This can result in an OpenSSL 1362 crash. This issue is not considered to be exploitable beyond a DoS. 1363 1364 This issue was reported to OpenSSL by Robert Święcki (Google Security Team) 1365 (CVE-2016-7054) 1366 [Richard Levitte] 1367 1368 *) CMS Null dereference 1369 1370 Applications parsing invalid CMS structures can crash with a NULL pointer 1371 dereference. This is caused by a bug in the handling of the ASN.1 CHOICE 1372 type in OpenSSL 1.1.0 which can result in a NULL value being passed to the 1373 structure callback if an attempt is made to free certain invalid encodings. 1374 Only CHOICE structures using a callback which do not handle NULL value are 1375 affected. 1376 1377 This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure. 1378 (CVE-2016-7053) 1379 [Stephen Henson] 1380 1381 *) Montgomery multiplication may produce incorrect results 1382 1383 There is a carry propagating bug in the Broadwell-specific Montgomery 1384 multiplication procedure that handles input lengths divisible by, but 1385 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 1386 and DH private keys are impossible. This is because the subroutine in 1387 question is not used in operations with the private key itself and an input 1388 of the attacker's direct choice. Otherwise the bug can manifest itself as 1389 transient authentication and key negotiation failures or reproducible 1390 erroneous outcome of public-key operations with specially crafted input. 1391 Among EC algorithms only Brainpool P-512 curves are affected and one 1392 presumably can attack ECDH key negotiation. Impact was not analyzed in 1393 detail, because pre-requisites for attack are considered unlikely. Namely 1394 multiple clients have to choose the curve in question and the server has to 1395 share the private key among them, neither of which is default behaviour. 1396 Even then only clients that chose the curve will be affected. 1397 1398 This issue was publicly reported as transient failures and was not 1399 initially recognized as a security issue. Thanks to Richard Morgan for 1400 providing reproducible case. 1401 (CVE-2016-7055) 1402 [Andy Polyakov] 1403 1404 *) Removed automatic addition of RPATH in shared libraries and executables, 1405 as this was a remainder from OpenSSL 1.0.x and isn't needed any more. 1406 [Richard Levitte] 1407 1408 Changes between 1.1.0a and 1.1.0b [26 Sep 2016] 1409 1410 *) Fix Use After Free for large message sizes 1411 1412 The patch applied to address CVE-2016-6307 resulted in an issue where if a 1413 message larger than approx 16k is received then the underlying buffer to 1414 store the incoming message is reallocated and moved. Unfortunately a 1415 dangling pointer to the old location is left which results in an attempt to 1416 write to the previously freed location. This is likely to result in a 1417 crash, however it could potentially lead to execution of arbitrary code. 1418 1419 This issue only affects OpenSSL 1.1.0a. 1420 1421 This issue was reported to OpenSSL by Robert Święcki. 1422 (CVE-2016-6309) 1423 [Matt Caswell] 1424 1425 Changes between 1.1.0 and 1.1.0a [22 Sep 2016] 1426 1427 *) OCSP Status Request extension unbounded memory growth 1428 1429 A malicious client can send an excessively large OCSP Status Request 1430 extension. If that client continually requests renegotiation, sending a 1431 large OCSP Status Request extension each time, then there will be unbounded 1432 memory growth on the server. This will eventually lead to a Denial Of 1433 Service attack through memory exhaustion. Servers with a default 1434 configuration are vulnerable even if they do not support OCSP. Builds using 1435 the "no-ocsp" build time option are not affected. 1436 1437 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 1438 (CVE-2016-6304) 1439 [Matt Caswell] 1440 1441 *) SSL_peek() hang on empty record 1442 1443 OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer 1444 sends an empty record. This could be exploited by a malicious peer in a 1445 Denial Of Service attack. 1446 1447 This issue was reported to OpenSSL by Alex Gaynor. 1448 (CVE-2016-6305) 1449 [Matt Caswell] 1450 1451 *) Excessive allocation of memory in tls_get_message_header() and 1452 dtls1_preprocess_fragment() 1453 1454 A (D)TLS message includes 3 bytes for its length in the header for the 1455 message. This would allow for messages up to 16Mb in length. Messages of 1456 this length are excessive and OpenSSL includes a check to ensure that a 1457 peer is sending reasonably sized messages in order to avoid too much memory 1458 being consumed to service a connection. A flaw in the logic of version 1459 1.1.0 means that memory for the message is allocated too early, prior to 1460 the excessive message length check. Due to way memory is allocated in 1461 OpenSSL this could mean an attacker could force up to 21Mb to be allocated 1462 to service a connection. This could lead to a Denial of Service through 1463 memory exhaustion. However, the excessive message length check still takes 1464 place, and this would cause the connection to immediately fail. Assuming 1465 that the application calls SSL_free() on the failed connection in a timely 1466 manner then the 21Mb of allocated memory will then be immediately freed 1467 again. Therefore the excessive memory allocation will be transitory in 1468 nature. This then means that there is only a security impact if: 1469 1470 1) The application does not call SSL_free() in a timely manner in the event 1471 that the connection fails 1472 or 1473 2) The application is working in a constrained environment where there is 1474 very little free memory 1475 or 1476 3) The attacker initiates multiple connection attempts such that there are 1477 multiple connections in a state where memory has been allocated for the 1478 connection; SSL_free() has not yet been called; and there is insufficient 1479 memory to service the multiple requests. 1480 1481 Except in the instance of (1) above any Denial Of Service is likely to be 1482 transitory because as soon as the connection fails the memory is 1483 subsequently freed again in the SSL_free() call. However there is an 1484 increased risk during this period of application crashes due to the lack of 1485 memory - which would then mean a more serious Denial of Service. 1486 1487 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 1488 (CVE-2016-6307 and CVE-2016-6308) 1489 [Matt Caswell] 1490 1491 *) solaris-x86-cc, i.e. 32-bit configuration with vendor compiler, 1492 had to be removed. Primary reason is that vendor assembler can't 1493 assemble our modules with -KPIC flag. As result it, assembly 1494 support, was not even available as option. But its lack means 1495 lack of side-channel resistant code, which is incompatible with 1496 security by todays standards. Fortunately gcc is readily available 1497 prepackaged option, which we firmly point at... 1498 [Andy Polyakov] 1499 1500 Changes between 1.0.2h and 1.1.0 [25 Aug 2016] 1501 1502 *) Windows command-line tool supports UTF-8 opt-in option for arguments 1503 and console input. Setting OPENSSL_WIN32_UTF8 environment variable 1504 (to any value) allows Windows user to access PKCS#12 file generated 1505 with Windows CryptoAPI and protected with non-ASCII password, as well 1506 as files generated under UTF-8 locale on Linux also protected with 1507 non-ASCII password. 1508 [Andy Polyakov] 1509 1510 *) To mitigate the SWEET32 attack (CVE-2016-2183), 3DES cipher suites 1511 have been disabled by default and removed from DEFAULT, just like RC4. 1512 See the RC4 item below to re-enable both. 1513 [Rich Salz] 1514 1515 *) The method for finding the storage location for the Windows RAND seed file 1516 has changed. First we check %RANDFILE%. If that is not set then we check 1517 the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If 1518 all else fails we fall back to C:\. 1519 [Matt Caswell] 1520 1521 *) The EVP_EncryptUpdate() function has had its return type changed from void 1522 to int. A return of 0 indicates and error while a return of 1 indicates 1523 success. 1524 [Matt Caswell] 1525 1526 *) The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and 1527 DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch 1528 off the constant time implementation for RSA, DSA and DH have been made 1529 no-ops and deprecated. 1530 [Matt Caswell] 1531 1532 *) Windows RAND implementation was simplified to only get entropy by 1533 calling CryptGenRandom(). Various other RAND-related tickets 1534 were also closed. 1535 [Joseph Wylie Yandle, Rich Salz] 1536 1537 *) The stack and lhash API's were renamed to start with OPENSSL_SK_ 1538 and OPENSSL_LH_, respectively. The old names are available 1539 with API compatibility. They new names are now completely documented. 1540 [Rich Salz] 1541 1542 *) Unify TYPE_up_ref(obj) methods signature. 1543 SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(), 1544 X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an 1545 int (instead of void) like all others TYPE_up_ref() methods. 1546 So now these methods also check the return value of CRYPTO_atomic_add(), 1547 and the validity of object reference counter. 1548 [[email protected]] 1549 1550 *) With Windows Visual Studio builds, the .pdb files are installed 1551 alongside the installed libraries and executables. For a static 1552 library installation, ossl_static.pdb is the associate compiler 1553 generated .pdb file to be used when linking programs. 1554 [Richard Levitte] 1555 1556 *) Remove openssl.spec. Packaging files belong with the packagers. 1557 [Richard Levitte] 1558 1559 *) Automatic Darwin/OSX configuration has had a refresh, it will now 1560 recognise x86_64 architectures automatically. You can still decide 1561 to build for a different bitness with the environment variable 1562 KERNEL_BITS (can be 32 or 64), for example: 1563 1564 KERNEL_BITS=32 ./config 1565 1566 [Richard Levitte] 1567 1568 *) Change default algorithms in pkcs8 utility to use PKCS#5 v2.0, 1569 256 bit AES and HMAC with SHA256. 1570 [Steve Henson] 1571 1572 *) Remove support for MIPS o32 ABI on IRIX (and IRIX only). 1573 [Andy Polyakov] 1574 1575 *) Triple-DES ciphers have been moved from HIGH to MEDIUM. 1576 [Rich Salz] 1577 1578 *) To enable users to have their own config files and build file templates, 1579 Configure looks in the directory indicated by the environment variable 1580 OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/ 1581 directory. On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical 1582 name and is used as is. 1583 [Richard Levitte] 1584 1585 *) The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX, 1586 X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD. The unused type 1587 X509_CERT_FILE_CTX was removed. 1588 [Rich Salz] 1589 1590 *) "shared" builds are now the default. To create only static libraries use 1591 the "no-shared" Configure option. 1592 [Matt Caswell] 1593 1594 *) Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options. 1595 All of these option have not worked for some while and are fundamental 1596 algorithms. 1597 [Matt Caswell] 1598 1599 *) Make various cleanup routines no-ops and mark them as deprecated. Most 1600 global cleanup functions are no longer required because they are handled 1601 via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages). 1602 Explicitly de-initing can cause problems (e.g. where a library that uses 1603 OpenSSL de-inits, but an application is still using it). The affected 1604 functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(), 1605 EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(), 1606 RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and 1607 COMP_zlib_cleanup(). 1608 [Matt Caswell] 1609 1610 *) --strict-warnings no longer enables runtime debugging options 1611 such as REF_DEBUG. Instead, debug options are automatically 1612 enabled with '--debug' builds. 1613 [Andy Polyakov, Emilia Käsper] 1614 1615 *) Made DH and DH_METHOD opaque. The structures for managing DH objects 1616 have been moved out of the public header files. New functions for managing 1617 these have been added. 1618 [Matt Caswell] 1619 1620 *) Made RSA and RSA_METHOD opaque. The structures for managing RSA 1621 objects have been moved out of the public header files. New 1622 functions for managing these have been added. 1623 [Richard Levitte] 1624 1625 *) Made DSA and DSA_METHOD opaque. The structures for managing DSA objects 1626 have been moved out of the public header files. New functions for managing 1627 these have been added. 1628 [Matt Caswell] 1629 1630 *) Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been 1631 moved out of the public header files. New functions for managing these 1632 have been added. 1633 [Matt Caswell] 1634 1635 *) Removed no-rijndael as a config option. Rijndael is an old name for AES. 1636 [Matt Caswell] 1637 1638 *) Removed the mk1mf build scripts. 1639 [Richard Levitte] 1640 1641 *) Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so 1642 it is always safe to #include a header now. 1643 [Rich Salz] 1644 1645 *) Removed the aged BC-32 config and all its supporting scripts 1646 [Richard Levitte] 1647 1648 *) Removed support for Ultrix, Netware, and OS/2. 1649 [Rich Salz] 1650 1651 *) Add support for HKDF. 1652 [Alessandro Ghedini] 1653 1654 *) Add support for blake2b and blake2s 1655 [Bill Cox] 1656 1657 *) Added support for "pipelining". Ciphers that have the 1658 EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple 1659 encryptions/decryptions simultaneously. There are currently no built-in 1660 ciphers with this property but the expectation is that engines will be able 1661 to offer it to significantly improve throughput. Support has been extended 1662 into libssl so that multiple records for a single connection can be 1663 processed in one go (for >=TLS 1.1). 1664 [Matt Caswell] 1665 1666 *) Added the AFALG engine. This is an async capable engine which is able to 1667 offload work to the Linux kernel. In this initial version it only supports 1668 AES128-CBC. The kernel must be version 4.1.0 or greater. 1669 [Catriona Lucey] 1670 1671 *) OpenSSL now uses a new threading API. It is no longer necessary to 1672 set locking callbacks to use OpenSSL in a multi-threaded environment. There 1673 are two supported threading models: pthreads and windows threads. It is 1674 also possible to configure OpenSSL at compile time for "no-threads". The 1675 old threading API should no longer be used. The functions have been 1676 replaced with "no-op" compatibility macros. 1677 [Alessandro Ghedini, Matt Caswell] 1678 1679 *) Modify behavior of ALPN to invoke callback after SNI/servername 1680 callback, such that updates to the SSL_CTX affect ALPN. 1681 [Todd Short] 1682 1683 *) Add SSL_CIPHER queries for authentication and key-exchange. 1684 [Todd Short] 1685 1686 *) Changes to the DEFAULT cipherlist: 1687 - Prefer (EC)DHE handshakes over plain RSA. 1688 - Prefer AEAD ciphers over legacy ciphers. 1689 - Prefer ECDSA over RSA when both certificates are available. 1690 - Prefer TLSv1.2 ciphers/PRF. 1691 - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the 1692 default cipherlist. 1693 [Emilia Käsper] 1694 1695 *) Change the ECC default curve list to be this, in order: x25519, 1696 secp256r1, secp521r1, secp384r1. 1697 [Rich Salz] 1698 1699 *) RC4 based libssl ciphersuites are now classed as "weak" ciphers and are 1700 disabled by default. They can be re-enabled using the 1701 enable-weak-ssl-ciphers option to Configure. 1702 [Matt Caswell] 1703 1704 *) If the server has ALPN configured, but supports no protocols that the 1705 client advertises, send a fatal "no_application_protocol" alert. 1706 This behaviour is SHALL in RFC 7301, though it isn't universally 1707 implemented by other servers. 1708 [Emilia Käsper] 1709 1710 *) Add X25519 support. 1711 Add ASN.1 and EVP_PKEY methods for X25519. This includes support 1712 for public and private key encoding using the format documented in 1713 draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports 1714 key generation and key derivation. 1715 1716 TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses 1717 X25519(29). 1718 [Steve Henson] 1719 1720 *) Deprecate SRP_VBASE_get_by_user. 1721 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 1722 In order to fix an unavoidable memory leak (CVE-2016-0798), 1723 SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP 1724 seed, even if the seed is configured. 1725 1726 Users should use SRP_VBASE_get1_by_user instead. Note that in 1727 SRP_VBASE_get1_by_user, caller must free the returned value. Note 1728 also that even though configuring the SRP seed attempts to hide 1729 invalid usernames by continuing the handshake with fake 1730 credentials, this behaviour is not constant time and no strong 1731 guarantees are made that the handshake is indistinguishable from 1732 that of a valid user. 1733 [Emilia Käsper] 1734 1735 *) Configuration change; it's now possible to build dynamic engines 1736 without having to build shared libraries and vice versa. This 1737 only applies to the engines in engines/, those in crypto/engine/ 1738 will always be built into libcrypto (i.e. "static"). 1739 1740 Building dynamic engines is enabled by default; to disable, use 1741 the configuration option "disable-dynamic-engine". 1742 1743 The only requirements for building dynamic engines are the 1744 presence of the DSO module and building with position independent 1745 code, so they will also automatically be disabled if configuring 1746 with "disable-dso" or "disable-pic". 1747 1748 The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE 1749 are also taken away from openssl/opensslconf.h, as they are 1750 irrelevant. 1751 [Richard Levitte] 1752 1753 *) Configuration change; if there is a known flag to compile 1754 position independent code, it will always be applied on the 1755 libcrypto and libssl object files, and never on the application 1756 object files. This means other libraries that use routines from 1757 libcrypto / libssl can be made into shared libraries regardless 1758 of how OpenSSL was configured. 1759 1760 If this isn't desirable, the configuration options "disable-pic" 1761 or "no-pic" can be used to disable the use of PIC. This will 1762 also disable building shared libraries and dynamic engines. 1763 [Richard Levitte] 1764 1765 *) Removed JPAKE code. It was experimental and has no wide use. 1766 [Rich Salz] 1767 1768 *) The INSTALL_PREFIX Makefile variable has been renamed to 1769 DESTDIR. That makes for less confusion on what this variable 1770 is for. Also, the configuration option --install_prefix is 1771 removed. 1772 [Richard Levitte] 1773 1774 *) Heartbeat for TLS has been removed and is disabled by default 1775 for DTLS; configure with enable-heartbeats. Code that uses the 1776 old #define's might need to be updated. 1777 [Emilia Käsper, Rich Salz] 1778 1779 *) Rename REF_CHECK to REF_DEBUG. 1780 [Rich Salz] 1781 1782 *) New "unified" build system 1783 1784 The "unified" build system is aimed to be a common system for all 1785 platforms we support. With it comes new support for VMS. 1786 1787 This system builds supports building in a different directory tree 1788 than the source tree. It produces one Makefile (for unix family 1789 or lookalikes), or one descrip.mms (for VMS). 1790 1791 The source of information to make the Makefile / descrip.mms is 1792 small files called 'build.info', holding the necessary 1793 information for each directory with source to compile, and a 1794 template in Configurations, like unix-Makefile.tmpl or 1795 descrip.mms.tmpl. 1796 1797 With this change, the library names were also renamed on Windows 1798 and on VMS. They now have names that are closer to the standard 1799 on Unix, and include the major version number, and in certain 1800 cases, the architecture they are built for. See "Notes on shared 1801 libraries" in INSTALL. 1802 1803 We rely heavily on the perl module Text::Template. 1804 [Richard Levitte] 1805 1806 *) Added support for auto-initialisation and de-initialisation of the library. 1807 OpenSSL no longer requires explicit init or deinit routines to be called, 1808 except in certain circumstances. See the OPENSSL_init_crypto() and 1809 OPENSSL_init_ssl() man pages for further information. 1810 [Matt Caswell] 1811 1812 *) The arguments to the DTLSv1_listen function have changed. Specifically the 1813 "peer" argument is now expected to be a BIO_ADDR object. 1814 1815 *) Rewrite of BIO networking library. The BIO library lacked consistent 1816 support of IPv6, and adding it required some more extensive 1817 modifications. This introduces the BIO_ADDR and BIO_ADDRINFO types, 1818 which hold all types of addresses and chains of address information. 1819 It also introduces a new API, with functions like BIO_socket, 1820 BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept. 1821 The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram 1822 have been adapted accordingly. 1823 [Richard Levitte] 1824 1825 *) RSA_padding_check_PKCS1_type_1 now accepts inputs with and without 1826 the leading 0-byte. 1827 [Emilia Käsper] 1828 1829 *) CRIME protection: disable compression by default, even if OpenSSL is 1830 compiled with zlib enabled. Applications can still enable compression 1831 by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by 1832 using the SSL_CONF library to configure compression. 1833 [Emilia Käsper] 1834 1835 *) The signature of the session callback configured with 1836 SSL_CTX_sess_set_get_cb was changed. The read-only input buffer 1837 was explicitly marked as 'const unsigned char*' instead of 1838 'unsigned char*'. 1839 [Emilia Käsper] 1840 1841 *) Always DPURIFY. Remove the use of uninitialized memory in the 1842 RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op. 1843 [Emilia Käsper] 1844 1845 *) Removed many obsolete configuration items, including 1846 DES_PTR, DES_RISC1, DES_RISC2, DES_INT 1847 MD2_CHAR, MD2_INT, MD2_LONG 1848 BF_PTR, BF_PTR2 1849 IDEA_SHORT, IDEA_LONG 1850 RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX 1851 [Rich Salz, with advice from Andy Polyakov] 1852 1853 *) Many BN internals have been moved to an internal header file. 1854 [Rich Salz with help from Andy Polyakov] 1855 1856 *) Configuration and writing out the results from it has changed. 1857 Files such as Makefile include/openssl/opensslconf.h and are now 1858 produced through general templates, such as Makefile.in and 1859 crypto/opensslconf.h.in and some help from the perl module 1860 Text::Template. 1861 1862 Also, the center of configuration information is no longer 1863 Makefile. Instead, Configure produces a perl module in 1864 configdata.pm which holds most of the config data (in the hash 1865 table %config), the target data that comes from the target 1866 configuration in one of the Configurations/*.conf files (in 1867 %target). 1868 [Richard Levitte] 1869 1870 *) To clarify their intended purposes, the Configure options 1871 --prefix and --openssldir change their semantics, and become more 1872 straightforward and less interdependent. 1873 1874 --prefix shall be used exclusively to give the location INSTALLTOP 1875 where programs, scripts, libraries, include files and manuals are 1876 going to be installed. The default is now /usr/local. 1877 1878 --openssldir shall be used exclusively to give the default 1879 location OPENSSLDIR where certificates, private keys, CRLs are 1880 managed. This is also where the default openssl.cnf gets 1881 installed. 1882 If the directory given with this option is a relative path, the 1883 values of both the --prefix value and the --openssldir value will 1884 be combined to become OPENSSLDIR. 1885 The default for --openssldir is INSTALLTOP/ssl. 1886 1887 Anyone who uses --openssldir to specify where OpenSSL is to be 1888 installed MUST change to use --prefix instead. 1889 [Richard Levitte] 1890 1891 *) The GOST engine was out of date and therefore it has been removed. An up 1892 to date GOST engine is now being maintained in an external repository. 1893 See: https://wiki.openssl.org/index.php/Binaries. Libssl still retains 1894 support for GOST ciphersuites (these are only activated if a GOST engine 1895 is present). 1896 [Matt Caswell] 1897 1898 *) EGD is no longer supported by default; use enable-egd when 1899 configuring. 1900 [Ben Kaduk and Rich Salz] 1901 1902 *) The distribution now has Makefile.in files, which are used to 1903 create Makefile's when Configure is run. *Configure must be run 1904 before trying to build now.* 1905 [Rich Salz] 1906 1907 *) The return value for SSL_CIPHER_description() for error conditions 1908 has changed. 1909 [Rich Salz] 1910 1911 *) Support for RFC6698/RFC7671 DANE TLSA peer authentication. 1912 1913 Obtaining and performing DNSSEC validation of TLSA records is 1914 the application's responsibility. The application provides 1915 the TLSA records of its choice to OpenSSL, and these are then 1916 used to authenticate the peer. 1917 1918 The TLSA records need not even come from DNS. They can, for 1919 example, be used to implement local end-entity certificate or 1920 trust-anchor "pinning", where the "pin" data takes the form 1921 of TLSA records, which can augment or replace verification 1922 based on the usual WebPKI public certification authorities. 1923 [Viktor Dukhovni] 1924 1925 *) Revert default OPENSSL_NO_DEPRECATED setting. Instead OpenSSL 1926 continues to support deprecated interfaces in default builds. 1927 However, applications are strongly advised to compile their 1928 source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides 1929 the declarations of all interfaces deprecated in 0.9.8, 1.0.0 1930 or the 1.1.0 releases. 1931 1932 In environments in which all applications have been ported to 1933 not use any deprecated interfaces OpenSSL's Configure script 1934 should be used with the --api=1.1.0 option to entirely remove 1935 support for the deprecated features from the library and 1936 unconditionally disable them in the installed headers. 1937 Essentially the same effect can be achieved with the "no-deprecated" 1938 argument to Configure, except that this will always restrict 1939 the build to just the latest API, rather than a fixed API 1940 version. 1941 1942 As applications are ported to future revisions of the API, 1943 they should update their compile-time OPENSSL_API_COMPAT define 1944 accordingly, but in most cases should be able to continue to 1945 compile with later releases. 1946 1947 The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are 1948 0x10000000L and 0x00908000L, respectively. However those 1949 versions did not support the OPENSSL_API_COMPAT feature, and 1950 so applications are not typically tested for explicit support 1951 of just the undeprecated features of either release. 1952 [Viktor Dukhovni] 1953 1954 *) Add support for setting the minimum and maximum supported protocol. 1955 It can bet set via the SSL_set_min_proto_version() and 1956 SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and 1957 MaxProtocol. It's recommended to use the new APIs to disable 1958 protocols instead of disabling individual protocols using 1959 SSL_set_options() or SSL_CONF's Protocol. This change also 1960 removes support for disabling TLS 1.2 in the OpenSSL TLS 1961 client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT. 1962 [Kurt Roeckx] 1963 1964 *) Support for ChaCha20 and Poly1305 added to libcrypto and libssl. 1965 [Andy Polyakov] 1966 1967 *) New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD 1968 and integrates ECDSA and ECDH functionality into EC. Implementations can 1969 now redirect key generation and no longer need to convert to or from 1970 ECDSA_SIG format. 1971 1972 Note: the ecdsa.h and ecdh.h headers are now no longer needed and just 1973 include the ec.h header file instead. 1974 [Steve Henson] 1975 1976 *) Remove support for all 40 and 56 bit ciphers. This includes all the export 1977 ciphers who are no longer supported and drops support the ephemeral RSA key 1978 exchange. The LOW ciphers currently doesn't have any ciphers in it. 1979 [Kurt Roeckx] 1980 1981 *) Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX 1982 opaque. For HMAC_CTX, the following constructors and destructors 1983 were added: 1984 1985 HMAC_CTX *HMAC_CTX_new(void); 1986 void HMAC_CTX_free(HMAC_CTX *ctx); 1987 1988 For EVP_MD and EVP_CIPHER, complete APIs to create, fill and 1989 destroy such methods has been added. See EVP_MD_meth_new(3) and 1990 EVP_CIPHER_meth_new(3) for documentation. 1991 1992 Additional changes: 1993 1) EVP_MD_CTX_cleanup(), EVP_CIPHER_CTX_cleanup() and 1994 HMAC_CTX_cleanup() were removed. HMAC_CTX_reset() and 1995 EVP_MD_CTX_reset() should be called instead to reinitialise 1996 an already created structure. 1997 2) For consistency with the majority of our object creators and 1998 destructors, EVP_MD_CTX_(create|destroy) were renamed to 1999 EVP_MD_CTX_(new|free). The old names are retained as macros 2000 for deprecated builds. 2001 [Richard Levitte] 2002 2003 *) Added ASYNC support. Libcrypto now includes the async sub-library to enable 2004 cryptographic operations to be performed asynchronously as long as an 2005 asynchronous capable engine is used. See the ASYNC_start_job() man page for 2006 further details. Libssl has also had this capability integrated with the 2007 introduction of the new mode SSL_MODE_ASYNC and associated error 2008 SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man 2009 pages. This work was developed in partnership with Intel Corp. 2010 [Matt Caswell] 2011 2012 *) SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is 2013 always enabled now. If you want to disable the support you should 2014 exclude it using the list of supported ciphers. This also means that the 2015 "-no_ecdhe" option has been removed from s_server. 2016 [Kurt Roeckx] 2017 2018 *) SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls 2019 SSL_{CTX_}set1_curves() which can set a list. 2020 [Kurt Roeckx] 2021 2022 *) Remove support for SSL_{CTX_}set_tmp_ecdh_callback(). You should set the 2023 curve you want to support using SSL_{CTX_}set1_curves(). 2024 [Kurt Roeckx] 2025 2026 *) State machine rewrite. The state machine code has been significantly 2027 refactored in order to remove much duplication of code and solve issues 2028 with the old code (see ssl/statem/README for further details). This change 2029 does have some associated API changes. Notably the SSL_state() function 2030 has been removed and replaced by SSL_get_state which now returns an 2031 "OSSL_HANDSHAKE_STATE" instead of an int. SSL_set_state() has been removed 2032 altogether. The previous handshake states defined in ssl.h and ssl3.h have 2033 also been removed. 2034 [Matt Caswell] 2035 2036 *) All instances of the string "ssleay" in the public API were replaced 2037 with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's) 2038 Some error codes related to internal RSA_eay API's were renamed. 2039 [Rich Salz] 2040 2041 *) The demo files in crypto/threads were moved to demo/threads. 2042 [Rich Salz] 2043 2044 *) Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp, 2045 sureware and ubsec. 2046 [Matt Caswell, Rich Salz] 2047 2048 *) New ASN.1 embed macro. 2049 2050 New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the 2051 structure is not allocated: it is part of the parent. That is instead of 2052 2053 FOO *x; 2054 2055 it must be: 2056 2057 FOO x; 2058 2059 This reduces memory fragmentation and make it impossible to accidentally 2060 set a mandatory field to NULL. 2061 2062 This currently only works for some fields specifically a SEQUENCE, CHOICE, 2063 or ASN1_STRING type which is part of a parent SEQUENCE. Since it is 2064 equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or 2065 SEQUENCE OF. 2066 [Steve Henson] 2067 2068 *) Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled. 2069 [Emilia Käsper] 2070 2071 *) Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although 2072 in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also 2073 an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add 2074 DES and RC4 ciphersuites. 2075 [Matt Caswell] 2076 2077 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 2078 This changes the decoding behaviour for some invalid messages, 2079 though the change is mostly in the more lenient direction, and 2080 legacy behaviour is preserved as much as possible. 2081 [Emilia Käsper] 2082 2083 *) Fix no-stdio build. 2084 [ David Woodhouse <[email protected]> and also 2085 Ivan Nestlerode <[email protected]> ] 2086 2087 *) New testing framework 2088 The testing framework has been largely rewritten and is now using 2089 perl and the perl modules Test::Harness and an extended variant of 2090 Test::More called OpenSSL::Test to do its work. All test scripts in 2091 test/ have been rewritten into test recipes, and all direct calls to 2092 executables in test/Makefile have become individual recipes using the 2093 simplified testing OpenSSL::Test::Simple. 2094 2095 For documentation on our testing modules, do: 2096 2097 perldoc test/testlib/OpenSSL/Test/Simple.pm 2098 perldoc test/testlib/OpenSSL/Test.pm 2099 2100 [Richard Levitte] 2101 2102 *) Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT 2103 are used; the latter aborts on memory leaks (usually checked on exit). 2104 Some undocumented "set malloc, etc., hooks" functions were removed 2105 and others were changed. All are now documented. 2106 [Rich Salz] 2107 2108 *) In DSA_generate_parameters_ex, if the provided seed is too short, 2109 return an error 2110 [Rich Salz and Ismo Puustinen <[email protected]>] 2111 2112 *) Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites 2113 from RFC4279, RFC4785, RFC5487, RFC5489. 2114 2115 Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the 2116 original RSA_PSK patch. 2117 [Steve Henson] 2118 2119 *) Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay 2120 era flag was never set throughout the codebase (only read). Also removed 2121 SSL3_FLAGS_POP_BUFFER which was only used if 2122 SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set. 2123 [Matt Caswell] 2124 2125 *) Changed the default name options in the "ca", "crl", "req" and "x509" 2126 to be "oneline" instead of "compat". 2127 [Richard Levitte] 2128 2129 *) Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're 2130 not aware of clients that still exhibit this bug, and the workaround 2131 hasn't been working properly for a while. 2132 [Emilia Käsper] 2133 2134 *) The return type of BIO_number_read() and BIO_number_written() as well as 2135 the corresponding num_read and num_write members in the BIO structure has 2136 changed from unsigned long to uint64_t. On platforms where an unsigned 2137 long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is 2138 transferred. 2139 [Matt Caswell] 2140 2141 *) Given the pervasive nature of TLS extensions it is inadvisable to run 2142 OpenSSL without support for them. It also means that maintaining 2143 the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably 2144 not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed. 2145 [Matt Caswell] 2146 2147 *) Removed support for the two export grade static DH ciphersuites 2148 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 2149 were newly added (along with a number of other static DH ciphersuites) to 2150 1.0.2. However the two export ones have *never* worked since they were 2151 introduced. It seems strange in any case to be adding new export 2152 ciphersuites, and given "logjam" it also does not seem correct to fix them. 2153 [Matt Caswell] 2154 2155 *) Version negotiation has been rewritten. In particular SSLv23_method(), 2156 SSLv23_client_method() and SSLv23_server_method() have been deprecated, 2157 and turned into macros which simply call the new preferred function names 2158 TLS_method(), TLS_client_method() and TLS_server_method(). All new code 2159 should use the new names instead. Also as part of this change the ssl23.h 2160 header file has been removed. 2161 [Matt Caswell] 2162 2163 *) Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This 2164 code and the associated standard is no longer considered fit-for-purpose. 2165 [Matt Caswell] 2166 2167 *) RT2547 was closed. When generating a private key, try to make the 2168 output file readable only by the owner. This behavior change might 2169 be noticeable when interacting with other software. 2170 2171 *) Documented all exdata functions. Added CRYPTO_free_ex_index. 2172 Added a test. 2173 [Rich Salz] 2174 2175 *) Added HTTP GET support to the ocsp command. 2176 [Rich Salz] 2177 2178 *) Changed default digest for the dgst and enc commands from MD5 to 2179 sha256 2180 [Rich Salz] 2181 2182 *) RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead. 2183 [Matt Caswell] 2184 2185 *) Added support for TLS extended master secret from 2186 draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an 2187 initial patch which was a great help during development. 2188 [Steve Henson] 2189 2190 *) All libssl internal structures have been removed from the public header 2191 files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is 2192 now redundant). Users should not attempt to access internal structures 2193 directly. Instead they should use the provided API functions. 2194 [Matt Caswell] 2195 2196 *) config has been changed so that by default OPENSSL_NO_DEPRECATED is used. 2197 Access to deprecated functions can be re-enabled by running config with 2198 "enable-deprecated". In addition applications wishing to use deprecated 2199 functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour 2200 will, by default, disable some transitive includes that previously existed 2201 in the header files (e.g. ec.h will no longer, by default, include bn.h) 2202 [Matt Caswell] 2203 2204 *) Added support for OCB mode. OpenSSL has been granted a patent license 2205 compatible with the OpenSSL license for use of OCB. Details are available 2206 at https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf. Support 2207 for OCB can be removed by calling config with no-ocb. 2208 [Matt Caswell] 2209 2210 *) SSLv2 support has been removed. It still supports receiving a SSLv2 2211 compatible client hello. 2212 [Kurt Roeckx] 2213 2214 *) Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz], 2215 done while fixing the error code for the key-too-small case. 2216 [Annie Yousar <[email protected]>] 2217 2218 *) CA.sh has been removed; use CA.pl instead. 2219 [Rich Salz] 2220 2221 *) Removed old DES API. 2222 [Rich Salz] 2223 2224 *) Remove various unsupported platforms: 2225 Sony NEWS4 2226 BEOS and BEOS_R5 2227 NeXT 2228 SUNOS 2229 MPE/iX 2230 Sinix/ReliantUNIX RM400 2231 DGUX 2232 NCR 2233 Tandem 2234 Cray 2235 16-bit platforms such as WIN16 2236 [Rich Salz] 2237 2238 *) Clean up OPENSSL_NO_xxx #define's 2239 Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF 2240 Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx 2241 OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC 2242 OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160 2243 OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO 2244 Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY 2245 OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP 2246 OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK 2247 OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY 2248 Remove MS_STATIC; it's a relic from platforms <32 bits. 2249 [Rich Salz] 2250 2251 *) Cleaned up dead code 2252 Remove all but one '#ifdef undef' which is to be looked at. 2253 [Rich Salz] 2254 2255 *) Clean up calling of xxx_free routines. 2256 Just like free(), fix most of the xxx_free routines to accept 2257 NULL. Remove the non-null checks from callers. Save much code. 2258 [Rich Salz] 2259 2260 *) Add secure heap for storage of private keys (when possible). 2261 Add BIO_s_secmem(), CBIGNUM, etc. 2262 Contributed by Akamai Technologies under our Corporate CLA. 2263 [Rich Salz] 2264 2265 *) Experimental support for a new, fast, unbiased prime candidate generator, 2266 bn_probable_prime_dh_coprime(). Not currently used by any prime generator. 2267 [Felix Laurie von Massenbach <[email protected]>] 2268 2269 *) New output format NSS in the sess_id command line tool. This allows 2270 exporting the session id and the master key in NSS keylog format. 2271 [Martin Kaiser <[email protected]>] 2272 2273 *) Harmonize version and its documentation. -f flag is used to display 2274 compilation flags. 2275 [mancha <[email protected]>] 2276 2277 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 2278 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 2279 [mancha <[email protected]>] 2280 2281 *) Fix some double frees. These are not thought to be exploitable. 2282 [mancha <[email protected]>] 2283 2284 *) A missing bounds check in the handling of the TLS heartbeat extension 2285 can be used to reveal up to 64k of memory to a connected client or 2286 server. 2287 2288 Thanks for Neel Mehta of Google Security for discovering this bug and to 2289 Adam Langley <[email protected]> and Bodo Moeller <[email protected]> for 2290 preparing the fix (CVE-2014-0160) 2291 [Adam Langley, Bodo Moeller] 2292 2293 *) Fix for the attack described in the paper "Recovering OpenSSL 2294 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 2295 by Yuval Yarom and Naomi Benger. Details can be obtained from: 2296 http://eprint.iacr.org/2014/140 2297 2298 Thanks to Yuval Yarom and Naomi Benger for discovering this 2299 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 2300 [Yuval Yarom and Naomi Benger] 2301 2302 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 2303 this fixes a limitation in previous versions of OpenSSL. 2304 [Steve Henson] 2305 2306 *) Experimental encrypt-then-mac support. 2307 2308 Experimental support for encrypt then mac from 2309 draft-gutmann-tls-encrypt-then-mac-02.txt 2310 2311 To enable it set the appropriate extension number (0x42 for the test 2312 server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42 2313 2314 For non-compliant peers (i.e. just about everything) this should have no 2315 effect. 2316 2317 WARNING: EXPERIMENTAL, SUBJECT TO CHANGE. 2318 2319 [Steve Henson] 2320 2321 *) Add EVP support for key wrapping algorithms, to avoid problems with 2322 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 2323 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 2324 algorithms and include tests cases. 2325 [Steve Henson] 2326 2327 *) Extend CMS code to support RSA-PSS signatures and RSA-OAEP for 2328 enveloped data. 2329 [Steve Henson] 2330 2331 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 2332 MGF1 digest and OAEP label. 2333 [Steve Henson] 2334 2335 *) Make openssl verify return errors. 2336 [Chris Palmer <[email protected]> and Ben Laurie] 2337 2338 *) New function ASN1_TIME_diff to calculate the difference between two 2339 ASN1_TIME structures or one structure and the current time. 2340 [Steve Henson] 2341 2342 *) Update fips_test_suite to support multiple command line options. New 2343 test to induce all self test errors in sequence and check expected 2344 failures. 2345 [Steve Henson] 2346 2347 *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and 2348 sign or verify all in one operation. 2349 [Steve Henson] 2350 2351 *) Add fips_algvs: a multicall fips utility incorporating all the algorithm 2352 test programs and fips_test_suite. Includes functionality to parse 2353 the minimal script output of fipsalgest.pl directly. 2354 [Steve Henson] 2355 2356 *) Add authorisation parameter to FIPS_module_mode_set(). 2357 [Steve Henson] 2358 2359 *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves. 2360 [Steve Henson] 2361 2362 *) Use separate DRBG fields for internal and external flags. New function 2363 FIPS_drbg_health_check() to perform on demand health checking. Add 2364 generation tests to fips_test_suite with reduced health check interval to 2365 demonstrate periodic health checking. Add "nodh" option to 2366 fips_test_suite to skip very slow DH test. 2367 [Steve Henson] 2368 2369 *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers 2370 based on NID. 2371 [Steve Henson] 2372 2373 *) More extensive health check for DRBG checking many more failure modes. 2374 New function FIPS_selftest_drbg_all() to handle every possible DRBG 2375 combination: call this in fips_test_suite. 2376 [Steve Henson] 2377 2378 *) Add support for canonical generation of DSA parameter 'g'. See 2379 FIPS 186-3 A.2.3. 2380 2381 *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and 2382 POST to handle HMAC cases. 2383 [Steve Henson] 2384 2385 *) Add functions FIPS_module_version() and FIPS_module_version_text() 2386 to return numerical and string versions of the FIPS module number. 2387 [Steve Henson] 2388 2389 *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and 2390 FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented 2391 outside the validated module in the FIPS capable OpenSSL. 2392 [Steve Henson] 2393 2394 *) Minor change to DRBG entropy callback semantics. In some cases 2395 there is no multiple of the block length between min_len and 2396 max_len. Allow the callback to return more than max_len bytes 2397 of entropy but discard any extra: it is the callback's responsibility 2398 to ensure that the extra data discarded does not impact the 2399 requested amount of entropy. 2400 [Steve Henson] 2401 2402 *) Add PRNG security strength checks to RSA, DSA and ECDSA using 2403 information in FIPS186-3, SP800-57 and SP800-131A. 2404 [Steve Henson] 2405 2406 *) CCM support via EVP. Interface is very similar to GCM case except we 2407 must supply all data in one chunk (i.e. no update, final) and the 2408 message length must be supplied if AAD is used. Add algorithm test 2409 support. 2410 [Steve Henson] 2411 2412 *) Initial version of POST overhaul. Add POST callback to allow the status 2413 of POST to be monitored and/or failures induced. Modify fips_test_suite 2414 to use callback. Always run all selftests even if one fails. 2415 [Steve Henson] 2416 2417 *) XTS support including algorithm test driver in the fips_gcmtest program. 2418 Note: this does increase the maximum key length from 32 to 64 bytes but 2419 there should be no binary compatibility issues as existing applications 2420 will never use XTS mode. 2421 [Steve Henson] 2422 2423 *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies 2424 to OpenSSL RAND code and replace with a tiny FIPS RAND API which also 2425 performs algorithm blocking for unapproved PRNG types. Also do not 2426 set PRNG type in FIPS_mode_set(): leave this to the application. 2427 Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with 2428 the standard OpenSSL PRNG: set additional data to a date time vector. 2429 [Steve Henson] 2430 2431 *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*. 2432 This shouldn't present any incompatibility problems because applications 2433 shouldn't be using these directly and any that are will need to rethink 2434 anyway as the X9.31 PRNG is now deprecated by FIPS 140-2 2435 [Steve Henson] 2436 2437 *) Extensive self tests and health checking required by SP800-90 DRBG. 2438 Remove strength parameter from FIPS_drbg_instantiate and always 2439 instantiate at maximum supported strength. 2440 [Steve Henson] 2441 2442 *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing. 2443 [Steve Henson] 2444 2445 *) New algorithm test program fips_dhvs to handle DH primitives only testing. 2446 [Steve Henson] 2447 2448 *) New function DH_compute_key_padded() to compute a DH key and pad with 2449 leading zeroes if needed: this complies with SP800-56A et al. 2450 [Steve Henson] 2451 2452 *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by 2453 anything, incomplete, subject to change and largely untested at present. 2454 [Steve Henson] 2455 2456 *) Modify fipscanisteronly build option to only build the necessary object 2457 files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile. 2458 [Steve Henson] 2459 2460 *) Add experimental option FIPSSYMS to give all symbols in 2461 fipscanister.o and FIPS or fips prefix. This will avoid 2462 conflicts with future versions of OpenSSL. Add perl script 2463 util/fipsas.pl to preprocess assembly language source files 2464 and rename any affected symbols. 2465 [Steve Henson] 2466 2467 *) Add selftest checks and algorithm block of non-fips algorithms in 2468 FIPS mode. Remove DES2 from selftests. 2469 [Steve Henson] 2470 2471 *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just 2472 return internal method without any ENGINE dependencies. Add new 2473 tiny fips sign and verify functions. 2474 [Steve Henson] 2475 2476 *) New build option no-ec2m to disable characteristic 2 code. 2477 [Steve Henson] 2478 2479 *) New build option "fipscanisteronly". This only builds fipscanister.o 2480 and (currently) associated fips utilities. Uses the file Makefile.fips 2481 instead of Makefile.org as the prototype. 2482 [Steve Henson] 2483 2484 *) Add some FIPS mode restrictions to GCM. Add internal IV generator. 2485 Update fips_gcmtest to use IV generator. 2486 [Steve Henson] 2487 2488 *) Initial, experimental EVP support for AES-GCM. AAD can be input by 2489 setting output buffer to NULL. The *Final function must be 2490 called although it will not retrieve any additional data. The tag 2491 can be set or retrieved with a ctrl. The IV length is by default 12 2492 bytes (96 bits) but can be set to an alternative value. If the IV 2493 length exceeds the maximum IV length (currently 16 bytes) it cannot be 2494 set before the key. 2495 [Steve Henson] 2496 2497 *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the 2498 underlying do_cipher function handles all cipher semantics itself 2499 including padding and finalisation. This is useful if (for example) 2500 an ENGINE cipher handles block padding itself. The behaviour of 2501 do_cipher is subtly changed if this flag is set: the return value 2502 is the number of characters written to the output buffer (zero is 2503 no longer an error code) or a negative error code. Also if the 2504 input buffer is NULL and length 0 finalisation should be performed. 2505 [Steve Henson] 2506 2507 *) If a candidate issuer certificate is already part of the constructed 2508 path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case. 2509 [Steve Henson] 2510 2511 *) Improve forward-security support: add functions 2512 2513 void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure)) 2514 void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure)) 2515 2516 for use by SSL/TLS servers; the callback function will be called whenever a 2517 new session is created, and gets to decide whether the session may be 2518 cached to make it resumable (return 0) or not (return 1). (As by the 2519 SSL/TLS protocol specifications, the session_id sent by the server will be 2520 empty to indicate that the session is not resumable; also, the server will 2521 not generate RFC 4507 (RFC 5077) session tickets.) 2522 2523 A simple reasonable callback implementation is to return is_forward_secure. 2524 This parameter will be set to 1 or 0 depending on the ciphersuite selected 2525 by the SSL/TLS server library, indicating whether it can provide forward 2526 security. 2527 [Emilia Käsper <[email protected]> (Google)] 2528 2529 *) New -verify_name option in command line utilities to set verification 2530 parameters by name. 2531 [Steve Henson] 2532 2533 *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE. 2534 Add CMAC pkey methods. 2535 [Steve Henson] 2536 2537 *) Experimental renegotiation in s_server -www mode. If the client 2538 browses /reneg connection is renegotiated. If /renegcert it is 2539 renegotiated requesting a certificate. 2540 [Steve Henson] 2541 2542 *) Add an "external" session cache for debugging purposes to s_server. This 2543 should help trace issues which normally are only apparent in deployed 2544 multi-process servers. 2545 [Steve Henson] 2546 2547 *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where 2548 return value is ignored. NB. The functions RAND_add(), RAND_seed(), 2549 BIO_set_cipher() and some obscure PEM functions were changed so they 2550 can now return an error. The RAND changes required a change to the 2551 RAND_METHOD structure. 2552 [Steve Henson] 2553 2554 *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of 2555 a gcc attribute to warn if the result of a function is ignored. This 2556 is enable if DEBUG_UNUSED is set. Add to several functions in evp.h 2557 whose return value is often ignored. 2558 [Steve Henson] 2559 2560 *) New -noct, -requestct, -requirect and -ctlogfile options for s_client. 2561 These allow SCTs (signed certificate timestamps) to be requested and 2562 validated when establishing a connection. 2563 [Rob Percival <[email protected]>] 2564 2565 Changes between 1.0.2g and 1.0.2h [3 May 2016] 2566 2567 *) Prevent padding oracle in AES-NI CBC MAC check 2568 2569 A MITM attacker can use a padding oracle attack to decrypt traffic 2570 when the connection uses an AES CBC cipher and the server support 2571 AES-NI. 2572 2573 This issue was introduced as part of the fix for Lucky 13 padding 2574 attack (CVE-2013-0169). The padding check was rewritten to be in 2575 constant time by making sure that always the same bytes are read and 2576 compared against either the MAC or padding bytes. But it no longer 2577 checked that there was enough data to have both the MAC and padding 2578 bytes. 2579 2580 This issue was reported by Juraj Somorovsky using TLS-Attacker. 2581 (CVE-2016-2107) 2582 [Kurt Roeckx] 2583 2584 *) Fix EVP_EncodeUpdate overflow 2585 2586 An overflow can occur in the EVP_EncodeUpdate() function which is used for 2587 Base64 encoding of binary data. If an attacker is able to supply very large 2588 amounts of input data then a length check can overflow resulting in a heap 2589 corruption. 2590 2591 Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by 2592 the PEM_write_bio* family of functions. These are mainly used within the 2593 OpenSSL command line applications, so any application which processes data 2594 from an untrusted source and outputs it as a PEM file should be considered 2595 vulnerable to this issue. User applications that call these APIs directly 2596 with large amounts of untrusted data may also be vulnerable. 2597 2598 This issue was reported by Guido Vranken. 2599 (CVE-2016-2105) 2600 [Matt Caswell] 2601 2602 *) Fix EVP_EncryptUpdate overflow 2603 2604 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 2605 is able to supply very large amounts of input data after a previous call to 2606 EVP_EncryptUpdate() with a partial block then a length check can overflow 2607 resulting in a heap corruption. Following an analysis of all OpenSSL 2608 internal usage of the EVP_EncryptUpdate() function all usage is one of two 2609 forms. The first form is where the EVP_EncryptUpdate() call is known to be 2610 the first called function after an EVP_EncryptInit(), and therefore that 2611 specific call must be safe. The second form is where the length passed to 2612 EVP_EncryptUpdate() can be seen from the code to be some small value and 2613 therefore there is no possibility of an overflow. Since all instances are 2614 one of these two forms, it is believed that there can be no overflows in 2615 internal code due to this problem. It should be noted that 2616 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 2617 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 2618 of these calls have also been analysed too and it is believed there are no 2619 instances in internal usage where an overflow could occur. 2620 2621 This issue was reported by Guido Vranken. 2622 (CVE-2016-2106) 2623 [Matt Caswell] 2624 2625 *) Prevent ASN.1 BIO excessive memory allocation 2626 2627 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 2628 a short invalid encoding can cause allocation of large amounts of memory 2629 potentially consuming excessive resources or exhausting memory. 2630 2631 Any application parsing untrusted data through d2i BIO functions is 2632 affected. The memory based functions such as d2i_X509() are *not* affected. 2633 Since the memory based functions are used by the TLS library, TLS 2634 applications are not affected. 2635 2636 This issue was reported by Brian Carpenter. 2637 (CVE-2016-2109) 2638 [Stephen Henson] 2639 2640 *) EBCDIC overread 2641 2642 ASN1 Strings that are over 1024 bytes can cause an overread in applications 2643 using the X509_NAME_oneline() function on EBCDIC systems. This could result 2644 in arbitrary stack data being returned in the buffer. 2645 2646 This issue was reported by Guido Vranken. 2647 (CVE-2016-2176) 2648 [Matt Caswell] 2649 2650 *) Modify behavior of ALPN to invoke callback after SNI/servername 2651 callback, such that updates to the SSL_CTX affect ALPN. 2652 [Todd Short] 2653 2654 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the 2655 default. 2656 [Kurt Roeckx] 2657 2658 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the 2659 methods are enabled and ssl2 is disabled the methods return NULL. 2660 [Kurt Roeckx] 2661 2662 Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 2663 2664 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 2665 Builds that are not configured with "enable-weak-ssl-ciphers" will not 2666 provide any "EXPORT" or "LOW" strength ciphers. 2667 [Viktor Dukhovni] 2668 2669 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 2670 is by default disabled at build-time. Builds that are not configured with 2671 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 2672 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 2673 will need to explicitly call either of: 2674 2675 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 2676 or 2677 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 2678 2679 as appropriate. Even if either of those is used, or the application 2680 explicitly uses the version-specific SSLv2_method() or its client and 2681 server variants, SSLv2 ciphers vulnerable to exhaustive search key 2682 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 2683 ciphers, and SSLv2 56-bit DES are no longer available. 2684 (CVE-2016-0800) 2685 [Viktor Dukhovni] 2686 2687 *) Fix a double-free in DSA code 2688 2689 A double free bug was discovered when OpenSSL parses malformed DSA private 2690 keys and could lead to a DoS attack or memory corruption for applications 2691 that receive DSA private keys from untrusted sources. This scenario is 2692 considered rare. 2693 2694 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 2695 libFuzzer. 2696 (CVE-2016-0705) 2697 [Stephen Henson] 2698 2699 *) Disable SRP fake user seed to address a server memory leak. 2700 2701 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 2702 2703 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 2704 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 2705 was changed to ignore the "fake user" SRP seed, even if the seed 2706 is configured. 2707 2708 Users should use SRP_VBASE_get1_by_user instead. Note that in 2709 SRP_VBASE_get1_by_user, caller must free the returned value. Note 2710 also that even though configuring the SRP seed attempts to hide 2711 invalid usernames by continuing the handshake with fake 2712 credentials, this behaviour is not constant time and no strong 2713 guarantees are made that the handshake is indistinguishable from 2714 that of a valid user. 2715 (CVE-2016-0798) 2716 [Emilia Käsper] 2717 2718 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 2719 2720 In the BN_hex2bn function the number of hex digits is calculated using an 2721 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 2722 large values of |i| this can result in |bn_expand| not allocating any 2723 memory because |i * 4| is negative. This can leave the internal BIGNUM data 2724 field as NULL leading to a subsequent NULL ptr deref. For very large values 2725 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 2726 In this case memory is allocated to the internal BIGNUM data field, but it 2727 is insufficiently sized leading to heap corruption. A similar issue exists 2728 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 2729 is ever called by user applications with very large untrusted hex/dec data. 2730 This is anticipated to be a rare occurrence. 2731 2732 All OpenSSL internal usage of these functions use data that is not expected 2733 to be untrusted, e.g. config file data or application command line 2734 arguments. If user developed applications generate config file data based 2735 on untrusted data then it is possible that this could also lead to security 2736 consequences. This is also anticipated to be rare. 2737 2738 This issue was reported to OpenSSL by Guido Vranken. 2739 (CVE-2016-0797) 2740 [Matt Caswell] 2741 2742 *) Fix memory issues in BIO_*printf functions 2743 2744 The internal |fmtstr| function used in processing a "%s" format string in 2745 the BIO_*printf functions could overflow while calculating the length of a 2746 string and cause an OOB read when printing very long strings. 2747 2748 Additionally the internal |doapr_outch| function can attempt to write to an 2749 OOB memory location (at an offset from the NULL pointer) in the event of a 2750 memory allocation failure. In 1.0.2 and below this could be caused where 2751 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 2752 could be in processing a very long "%s" format string. Memory leaks can 2753 also occur. 2754 2755 The first issue may mask the second issue dependent on compiler behaviour. 2756 These problems could enable attacks where large amounts of untrusted data 2757 is passed to the BIO_*printf functions. If applications use these functions 2758 in this way then they could be vulnerable. OpenSSL itself uses these 2759 functions when printing out human-readable dumps of ASN.1 data. Therefore 2760 applications that print this data could be vulnerable if the data is from 2761 untrusted sources. OpenSSL command line applications could also be 2762 vulnerable where they print out ASN.1 data, or if untrusted data is passed 2763 as command line arguments. 2764 2765 Libssl is not considered directly vulnerable. Additionally certificates etc 2766 received via remote connections via libssl are also unlikely to be able to 2767 trigger these issues because of message size limits enforced within libssl. 2768 2769 This issue was reported to OpenSSL Guido Vranken. 2770 (CVE-2016-0799) 2771 [Matt Caswell] 2772 2773 *) Side channel attack on modular exponentiation 2774 2775 A side-channel attack was found which makes use of cache-bank conflicts on 2776 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 2777 of RSA keys. The ability to exploit this issue is limited as it relies on 2778 an attacker who has control of code in a thread running on the same 2779 hyper-threaded core as the victim thread which is performing decryptions. 2780 2781 This issue was reported to OpenSSL by Yuval Yarom, The University of 2782 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 2783 Nadia Heninger, University of Pennsylvania with more information at 2784 http://cachebleed.info. 2785 (CVE-2016-0702) 2786 [Andy Polyakov] 2787 2788 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 2789 if no keysize is specified with default_bits. This fixes an 2790 omission in an earlier change that changed all RSA/DSA key generation 2791 apps to use 2048 bits by default. 2792 [Emilia Käsper] 2793 2794 Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 2795 *) DH small subgroups 2796 2797 Historically OpenSSL only ever generated DH parameters based on "safe" 2798 primes. More recently (in version 1.0.2) support was provided for 2799 generating X9.42 style parameter files such as those required for RFC 5114 2800 support. The primes used in such files may not be "safe". Where an 2801 application is using DH configured with parameters based on primes that are 2802 not "safe" then an attacker could use this fact to find a peer's private 2803 DH exponent. This attack requires that the attacker complete multiple 2804 handshakes in which the peer uses the same private DH exponent. For example 2805 this could be used to discover a TLS server's private DH exponent if it's 2806 reusing the private DH exponent or it's using a static DH ciphersuite. 2807 2808 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 2809 TLS. It is not on by default. If the option is not set then the server 2810 reuses the same private DH exponent for the life of the server process and 2811 would be vulnerable to this attack. It is believed that many popular 2812 applications do set this option and would therefore not be at risk. 2813 2814 The fix for this issue adds an additional check where a "q" parameter is 2815 available (as is the case in X9.42 based parameters). This detects the 2816 only known attack, and is the only possible defense for static DH 2817 ciphersuites. This could have some performance impact. 2818 2819 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 2820 default and cannot be disabled. This could have some performance impact. 2821 2822 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 2823 (CVE-2016-0701) 2824 [Matt Caswell] 2825 2826 *) SSLv2 doesn't block disabled ciphers 2827 2828 A malicious client can negotiate SSLv2 ciphers that have been disabled on 2829 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 2830 been disabled, provided that the SSLv2 protocol was not also disabled via 2831 SSL_OP_NO_SSLv2. 2832 2833 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 2834 and Sebastian Schinzel. 2835 (CVE-2015-3197) 2836 [Viktor Dukhovni] 2837 2838 Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 2839 2840 *) BN_mod_exp may produce incorrect results on x86_64 2841 2842 There is a carry propagating bug in the x86_64 Montgomery squaring 2843 procedure. No EC algorithms are affected. Analysis suggests that attacks 2844 against RSA and DSA as a result of this defect would be very difficult to 2845 perform and are not believed likely. Attacks against DH are considered just 2846 feasible (although very difficult) because most of the work necessary to 2847 deduce information about a private key may be performed offline. The amount 2848 of resources required for such an attack would be very significant and 2849 likely only accessible to a limited number of attackers. An attacker would 2850 additionally need online access to an unpatched system using the target 2851 private key in a scenario with persistent DH parameters and a private 2852 key that is shared between multiple clients. For example this can occur by 2853 default in OpenSSL DHE based SSL/TLS ciphersuites. 2854 2855 This issue was reported to OpenSSL by Hanno Böck. 2856 (CVE-2015-3193) 2857 [Andy Polyakov] 2858 2859 *) Certificate verify crash with missing PSS parameter 2860 2861 The signature verification routines will crash with a NULL pointer 2862 dereference if presented with an ASN.1 signature using the RSA PSS 2863 algorithm and absent mask generation function parameter. Since these 2864 routines are used to verify certificate signature algorithms this can be 2865 used to crash any certificate verification operation and exploited in a 2866 DoS attack. Any application which performs certificate verification is 2867 vulnerable including OpenSSL clients and servers which enable client 2868 authentication. 2869 2870 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG). 2871 (CVE-2015-3194) 2872 [Stephen Henson] 2873 2874 *) X509_ATTRIBUTE memory leak 2875 2876 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 2877 memory. This structure is used by the PKCS#7 and CMS routines so any 2878 application which reads PKCS#7 or CMS data from untrusted sources is 2879 affected. SSL/TLS is not affected. 2880 2881 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 2882 libFuzzer. 2883 (CVE-2015-3195) 2884 [Stephen Henson] 2885 2886 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 2887 This changes the decoding behaviour for some invalid messages, 2888 though the change is mostly in the more lenient direction, and 2889 legacy behaviour is preserved as much as possible. 2890 [Emilia Käsper] 2891 2892 *) In DSA_generate_parameters_ex, if the provided seed is too short, 2893 return an error 2894 [Rich Salz and Ismo Puustinen <[email protected]>] 2895 2896 Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 2897 2898 *) Alternate chains certificate forgery 2899 2900 During certificate verification, OpenSSL will attempt to find an 2901 alternative certificate chain if the first attempt to build such a chain 2902 fails. An error in the implementation of this logic can mean that an 2903 attacker could cause certain checks on untrusted certificates to be 2904 bypassed, such as the CA flag, enabling them to use a valid leaf 2905 certificate to act as a CA and "issue" an invalid certificate. 2906 2907 This issue was reported to OpenSSL by Adam Langley/David Benjamin 2908 (Google/BoringSSL). 2909 [Matt Caswell] 2910 2911 Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 2912 2913 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 2914 incompatibility in the handling of HMAC. The previous ABI has now been 2915 restored. 2916 [Matt Caswell] 2917 2918 Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 2919 2920 *) Malformed ECParameters causes infinite loop 2921 2922 When processing an ECParameters structure OpenSSL enters an infinite loop 2923 if the curve specified is over a specially malformed binary polynomial 2924 field. 2925 2926 This can be used to perform denial of service against any 2927 system which processes public keys, certificate requests or 2928 certificates. This includes TLS clients and TLS servers with 2929 client authentication enabled. 2930 2931 This issue was reported to OpenSSL by Joseph Barr-Pixton. 2932 (CVE-2015-1788) 2933 [Andy Polyakov] 2934 2935 *) Exploitable out-of-bounds read in X509_cmp_time 2936 2937 X509_cmp_time does not properly check the length of the ASN1_TIME 2938 string and can read a few bytes out of bounds. In addition, 2939 X509_cmp_time accepts an arbitrary number of fractional seconds in the 2940 time string. 2941 2942 An attacker can use this to craft malformed certificates and CRLs of 2943 various sizes and potentially cause a segmentation fault, resulting in 2944 a DoS on applications that verify certificates or CRLs. TLS clients 2945 that verify CRLs are affected. TLS clients and servers with client 2946 authentication enabled may be affected if they use custom verification 2947 callbacks. 2948 2949 This issue was reported to OpenSSL by Robert Swiecki (Google), and 2950 independently by Hanno Böck. 2951 (CVE-2015-1789) 2952 [Emilia Käsper] 2953 2954 *) PKCS7 crash with missing EnvelopedContent 2955 2956 The PKCS#7 parsing code does not handle missing inner EncryptedContent 2957 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 2958 with missing content and trigger a NULL pointer dereference on parsing. 2959 2960 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 2961 structures from untrusted sources are affected. OpenSSL clients and 2962 servers are not affected. 2963 2964 This issue was reported to OpenSSL by Michal Zalewski (Google). 2965 (CVE-2015-1790) 2966 [Emilia Käsper] 2967 2968 *) CMS verify infinite loop with unknown hash function 2969 2970 When verifying a signedData message the CMS code can enter an infinite loop 2971 if presented with an unknown hash function OID. This can be used to perform 2972 denial of service against any system which verifies signedData messages using 2973 the CMS code. 2974 This issue was reported to OpenSSL by Johannes Bauer. 2975 (CVE-2015-1792) 2976 [Stephen Henson] 2977 2978 *) Race condition handling NewSessionTicket 2979 2980 If a NewSessionTicket is received by a multi-threaded client when attempting to 2981 reuse a previous ticket then a race condition can occur potentially leading to 2982 a double free of the ticket data. 2983 (CVE-2015-1791) 2984 [Matt Caswell] 2985 2986 *) Only support 256-bit or stronger elliptic curves with the 2987 'ecdh_auto' setting (server) or by default (client). Of supported 2988 curves, prefer P-256 (both). 2989 [Emilia Kasper] 2990 2991 Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 2992 2993 *) ClientHello sigalgs DoS fix 2994 2995 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 2996 invalid signature algorithms extension a NULL pointer dereference will 2997 occur. This can be exploited in a DoS attack against the server. 2998 2999 This issue was was reported to OpenSSL by David Ramos of Stanford 3000 University. 3001 (CVE-2015-0291) 3002 [Stephen Henson and Matt Caswell] 3003 3004 *) Multiblock corrupted pointer fix 3005 3006 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 3007 feature only applies on 64 bit x86 architecture platforms that support AES 3008 NI instructions. A defect in the implementation of "multiblock" can cause 3009 OpenSSL's internal write buffer to become incorrectly set to NULL when 3010 using non-blocking IO. Typically, when the user application is using a 3011 socket BIO for writing, this will only result in a failed connection. 3012 However if some other BIO is used then it is likely that a segmentation 3013 fault will be triggered, thus enabling a potential DoS attack. 3014 3015 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 3016 (CVE-2015-0290) 3017 [Matt Caswell] 3018 3019 *) Segmentation fault in DTLSv1_listen fix 3020 3021 The DTLSv1_listen function is intended to be stateless and processes the 3022 initial ClientHello from many peers. It is common for user code to loop 3023 over the call to DTLSv1_listen until a valid ClientHello is received with 3024 an associated cookie. A defect in the implementation of DTLSv1_listen means 3025 that state is preserved in the SSL object from one invocation to the next 3026 that can lead to a segmentation fault. Errors processing the initial 3027 ClientHello can trigger this scenario. An example of such an error could be 3028 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 3029 server. 3030 3031 This issue was reported to OpenSSL by Per Allansson. 3032 (CVE-2015-0207) 3033 [Matt Caswell] 3034 3035 *) Segmentation fault in ASN1_TYPE_cmp fix 3036 3037 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 3038 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 3039 certificate signature algorithm consistency this can be used to crash any 3040 certificate verification operation and exploited in a DoS attack. Any 3041 application which performs certificate verification is vulnerable including 3042 OpenSSL clients and servers which enable client authentication. 3043 (CVE-2015-0286) 3044 [Stephen Henson] 3045 3046 *) Segmentation fault for invalid PSS parameters fix 3047 3048 The signature verification routines will crash with a NULL pointer 3049 dereference if presented with an ASN.1 signature using the RSA PSS 3050 algorithm and invalid parameters. Since these routines are used to verify 3051 certificate signature algorithms this can be used to crash any 3052 certificate verification operation and exploited in a DoS attack. Any 3053 application which performs certificate verification is vulnerable including 3054 OpenSSL clients and servers which enable client authentication. 3055 3056 This issue was was reported to OpenSSL by Brian Carpenter. 3057 (CVE-2015-0208) 3058 [Stephen Henson] 3059 3060 *) ASN.1 structure reuse memory corruption fix 3061 3062 Reusing a structure in ASN.1 parsing may allow an attacker to cause 3063 memory corruption via an invalid write. Such reuse is and has been 3064 strongly discouraged and is believed to be rare. 3065 3066 Applications that parse structures containing CHOICE or ANY DEFINED BY 3067 components may be affected. Certificate parsing (d2i_X509 and related 3068 functions) are however not affected. OpenSSL clients and servers are 3069 not affected. 3070 (CVE-2015-0287) 3071 [Stephen Henson] 3072 3073 *) PKCS7 NULL pointer dereferences fix 3074 3075 The PKCS#7 parsing code does not handle missing outer ContentInfo 3076 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 3077 missing content and trigger a NULL pointer dereference on parsing. 3078 3079 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 3080 otherwise parse PKCS#7 structures from untrusted sources are 3081 affected. OpenSSL clients and servers are not affected. 3082 3083 This issue was reported to OpenSSL by Michal Zalewski (Google). 3084 (CVE-2015-0289) 3085 [Emilia Käsper] 3086 3087 *) DoS via reachable assert in SSLv2 servers fix 3088 3089 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 3090 servers that both support SSLv2 and enable export cipher suites by sending 3091 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 3092 3093 This issue was discovered by Sean Burford (Google) and Emilia Käsper 3094 (OpenSSL development team). 3095 (CVE-2015-0293) 3096 [Emilia Käsper] 3097 3098 *) Empty CKE with client auth and DHE fix 3099 3100 If client auth is used then a server can seg fault in the event of a DHE 3101 ciphersuite being selected and a zero length ClientKeyExchange message 3102 being sent by the client. This could be exploited in a DoS attack. 3103 (CVE-2015-1787) 3104 [Matt Caswell] 3105 3106 *) Handshake with unseeded PRNG fix 3107 3108 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 3109 with an unseeded PRNG. The conditions are: 3110 - The client is on a platform where the PRNG has not been seeded 3111 automatically, and the user has not seeded manually 3112 - A protocol specific client method version has been used (i.e. not 3113 SSL_client_methodv23) 3114 - A ciphersuite is used that does not require additional random data from 3115 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 3116 3117 If the handshake succeeds then the client random that has been used will 3118 have been generated from a PRNG with insufficient entropy and therefore the 3119 output may be predictable. 3120 3121 For example using the following command with an unseeded openssl will 3122 succeed on an unpatched platform: 3123 3124 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 3125 (CVE-2015-0285) 3126 [Matt Caswell] 3127 3128 *) Use After Free following d2i_ECPrivatekey error fix 3129 3130 A malformed EC private key file consumed via the d2i_ECPrivateKey function 3131 could cause a use after free condition. This, in turn, could cause a double 3132 free in several private key parsing functions (such as d2i_PrivateKey 3133 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 3134 for applications that receive EC private keys from untrusted 3135 sources. This scenario is considered rare. 3136 3137 This issue was discovered by the BoringSSL project and fixed in their 3138 commit 517073cd4b. 3139 (CVE-2015-0209) 3140 [Matt Caswell] 3141 3142 *) X509_to_X509_REQ NULL pointer deref fix 3143 3144 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 3145 the certificate key is invalid. This function is rarely used in practice. 3146 3147 This issue was discovered by Brian Carpenter. 3148 (CVE-2015-0288) 3149 [Stephen Henson] 3150 3151 *) Removed the export ciphers from the DEFAULT ciphers 3152 [Kurt Roeckx] 3153 3154 Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 3155 3156 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 3157 ARMv5 through ARMv8, as opposite to "locking" it to single one. 3158 So far those who have to target multiple platforms would compromise 3159 and argue that binary targeting say ARMv5 would still execute on 3160 ARMv8. "Universal" build resolves this compromise by providing 3161 near-optimal performance even on newer platforms. 3162 [Andy Polyakov] 3163 3164 *) Accelerated NIST P-256 elliptic curve implementation for x86_64 3165 (other platforms pending). 3166 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov] 3167 3168 *) Add support for the SignedCertificateTimestampList certificate and 3169 OCSP response extensions from RFC6962. 3170 [Rob Stradling] 3171 3172 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 3173 for corner cases. (Certain input points at infinity could lead to 3174 bogus results, with non-infinity inputs mapped to infinity too.) 3175 [Bodo Moeller] 3176 3177 *) Initial support for PowerISA 2.0.7, first implemented in POWER8. 3178 This covers AES, SHA256/512 and GHASH. "Initial" means that most 3179 common cases are optimized and there still is room for further 3180 improvements. Vector Permutation AES for Altivec is also added. 3181 [Andy Polyakov] 3182 3183 *) Add support for little-endian ppc64 Linux target. 3184 [Marcelo Cerri (IBM)] 3185 3186 *) Initial support for AMRv8 ISA crypto extensions. This covers AES, 3187 SHA1, SHA256 and GHASH. "Initial" means that most common cases 3188 are optimized and there still is room for further improvements. 3189 Both 32- and 64-bit modes are supported. 3190 [Andy Polyakov, Ard Biesheuvel (Linaro)] 3191 3192 *) Improved ARMv7 NEON support. 3193 [Andy Polyakov] 3194 3195 *) Support for SPARC Architecture 2011 crypto extensions, first 3196 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 3197 SHA256/512, MD5, GHASH and modular exponentiation. 3198 [Andy Polyakov, David Miller] 3199 3200 *) Accelerated modular exponentiation for Intel processors, a.k.a. 3201 RSAZ. 3202 [Shay Gueron & Vlad Krasnov (Intel Corp)] 3203 3204 *) Support for new and upcoming Intel processors, including AVX2, 3205 BMI and SHA ISA extensions. This includes additional "stitched" 3206 implementations, AESNI-SHA256 and GCM, and multi-buffer support 3207 for TLS encrypt. 3208 3209 This work was sponsored by Intel Corp. 3210 [Andy Polyakov] 3211 3212 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 3213 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 3214 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 3215 [Steve Henson] 3216 3217 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 3218 this fixes a limitation in previous versions of OpenSSL. 3219 [Steve Henson] 3220 3221 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 3222 MGF1 digest and OAEP label. 3223 [Steve Henson] 3224 3225 *) Add EVP support for key wrapping algorithms, to avoid problems with 3226 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 3227 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 3228 algorithms and include tests cases. 3229 [Steve Henson] 3230 3231 *) Add functions to allocate and set the fields of an ECDSA_METHOD 3232 structure. 3233 [Douglas E. Engert, Steve Henson] 3234 3235 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 3236 difference in days and seconds between two tm or ASN1_TIME structures. 3237 [Steve Henson] 3238 3239 *) Add -rev test option to s_server to just reverse order of characters 3240 received by client and send back to server. Also prints an abbreviated 3241 summary of the connection parameters. 3242 [Steve Henson] 3243 3244 *) New option -brief for s_client and s_server to print out a brief summary 3245 of connection parameters. 3246 [Steve Henson] 3247 3248 *) Add callbacks for arbitrary TLS extensions. 3249 [Trevor Perrin <[email protected]> and Ben Laurie] 3250 3251 *) New option -crl_download in several openssl utilities to download CRLs 3252 from CRLDP extension in certificates. 3253 [Steve Henson] 3254 3255 *) New options -CRL and -CRLform for s_client and s_server for CRLs. 3256 [Steve Henson] 3257 3258 *) New function X509_CRL_diff to generate a delta CRL from the difference 3259 of two full CRLs. Add support to "crl" utility. 3260 [Steve Henson] 3261 3262 *) New functions to set lookup_crls function and to retrieve 3263 X509_STORE from X509_STORE_CTX. 3264 [Steve Henson] 3265 3266 *) Print out deprecated issuer and subject unique ID fields in 3267 certificates. 3268 [Steve Henson] 3269 3270 *) Extend OCSP I/O functions so they can be used for simple general purpose 3271 HTTP as well as OCSP. New wrapper function which can be used to download 3272 CRLs using the OCSP API. 3273 [Steve Henson] 3274 3275 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs. 3276 [Steve Henson] 3277 3278 *) SSL_CONF* functions. These provide a common framework for application 3279 configuration using configuration files or command lines. 3280 [Steve Henson] 3281 3282 *) SSL/TLS tracing code. This parses out SSL/TLS records using the 3283 message callback and prints the results. Needs compile time option 3284 "enable-ssl-trace". New options to s_client and s_server to enable 3285 tracing. 3286 [Steve Henson] 3287 3288 *) New ctrl and macro to retrieve supported points extensions. 3289 Print out extension in s_server and s_client. 3290 [Steve Henson] 3291 3292 *) New functions to retrieve certificate signature and signature 3293 OID NID. 3294 [Steve Henson] 3295 3296 *) Add functions to retrieve and manipulate the raw cipherlist sent by a 3297 client to OpenSSL. 3298 [Steve Henson] 3299 3300 *) New Suite B modes for TLS code. These use and enforce the requirements 3301 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 3302 only use Suite B curves. The Suite B modes can be set by using the 3303 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 3304 [Steve Henson] 3305 3306 *) New chain verification flags for Suite B levels of security. Check 3307 algorithms are acceptable when flags are set in X509_verify_cert. 3308 [Steve Henson] 3309 3310 *) Make tls1_check_chain return a set of flags indicating checks passed 3311 by a certificate chain. Add additional tests to handle client 3312 certificates: checks for matching certificate type and issuer name 3313 comparison. 3314 [Steve Henson] 3315 3316 *) If an attempt is made to use a signature algorithm not in the peer 3317 preference list abort the handshake. If client has no suitable 3318 signature algorithms in response to a certificate request do not 3319 use the certificate. 3320 [Steve Henson] 3321 3322 *) If server EC tmp key is not in client preference list abort handshake. 3323 [Steve Henson] 3324 3325 *) Add support for certificate stores in CERT structure. This makes it 3326 possible to have different stores per SSL structure or one store in 3327 the parent SSL_CTX. Include distinct stores for certificate chain 3328 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 3329 to build and store a certificate chain in CERT structure: returning 3330 an error if the chain cannot be built: this will allow applications 3331 to test if a chain is correctly configured. 3332 3333 Note: if the CERT based stores are not set then the parent SSL_CTX 3334 store is used to retain compatibility with existing behaviour. 3335 3336 [Steve Henson] 3337 3338 *) New function ssl_set_client_disabled to set a ciphersuite disabled 3339 mask based on the current session, check mask when sending client 3340 hello and checking the requested ciphersuite. 3341 [Steve Henson] 3342 3343 *) New ctrls to retrieve and set certificate types in a certificate 3344 request message. Print out received values in s_client. If certificate 3345 types is not set with custom values set sensible values based on 3346 supported signature algorithms. 3347 [Steve Henson] 3348 3349 *) Support for distinct client and server supported signature algorithms. 3350 [Steve Henson] 3351 3352 *) Add certificate callback. If set this is called whenever a certificate 3353 is required by client or server. An application can decide which 3354 certificate chain to present based on arbitrary criteria: for example 3355 supported signature algorithms. Add very simple example to s_server. 3356 This fixes many of the problems and restrictions of the existing client 3357 certificate callback: for example you can now clear an existing 3358 certificate and specify the whole chain. 3359 [Steve Henson] 3360 3361 *) Add new "valid_flags" field to CERT_PKEY structure which determines what 3362 the certificate can be used for (if anything). Set valid_flags field 3363 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 3364 to have similar checks in it. 3365 3366 Add new "cert_flags" field to CERT structure and include a "strict mode". 3367 This enforces some TLS certificate requirements (such as only permitting 3368 certificate signature algorithms contained in the supported algorithms 3369 extension) which some implementations ignore: this option should be used 3370 with caution as it could cause interoperability issues. 3371 [Steve Henson] 3372 3373 *) Update and tidy signature algorithm extension processing. Work out 3374 shared signature algorithms based on preferences and peer algorithms 3375 and print them out in s_client and s_server. Abort handshake if no 3376 shared signature algorithms. 3377 [Steve Henson] 3378 3379 *) Add new functions to allow customised supported signature algorithms 3380 for SSL and SSL_CTX structures. Add options to s_client and s_server 3381 to support them. 3382 [Steve Henson] 3383 3384 *) New function SSL_certs_clear() to delete all references to certificates 3385 from an SSL structure. Before this once a certificate had been added 3386 it couldn't be removed. 3387 [Steve Henson] 3388 3389 *) Integrate hostname, email address and IP address checking with certificate 3390 verification. New verify options supporting checking in openssl utility. 3391 [Steve Henson] 3392 3393 *) Fixes and wildcard matching support to hostname and email checking 3394 functions. Add manual page. 3395 [Florian Weimer (Red Hat Product Security Team)] 3396 3397 *) New functions to check a hostname email or IP address against a 3398 certificate. Add options x509 utility to print results of checks against 3399 a certificate. 3400 [Steve Henson] 3401 3402 *) Fix OCSP checking. 3403 [Rob Stradling <[email protected]> and Ben Laurie] 3404 3405 *) Initial experimental support for explicitly trusted non-root CAs. 3406 OpenSSL still tries to build a complete chain to a root but if an 3407 intermediate CA has a trust setting included that is used. The first 3408 setting is used: whether to trust (e.g., -addtrust option to the x509 3409 utility) or reject. 3410 [Steve Henson] 3411 3412 *) Add -trusted_first option which attempts to find certificates in the 3413 trusted store even if an untrusted chain is also supplied. 3414 [Steve Henson] 3415 3416 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 3417 platform support for Linux and Android. 3418 [Andy Polyakov] 3419 3420 *) Support for linux-x32, ILP32 environment in x86_64 framework. 3421 [Andy Polyakov] 3422 3423 *) Experimental multi-implementation support for FIPS capable OpenSSL. 3424 When in FIPS mode the approved implementations are used as normal, 3425 when not in FIPS mode the internal unapproved versions are used instead. 3426 This means that the FIPS capable OpenSSL isn't forced to use the 3427 (often lower performance) FIPS implementations outside FIPS mode. 3428 [Steve Henson] 3429 3430 *) Transparently support X9.42 DH parameters when calling 3431 PEM_read_bio_DHparameters. This means existing applications can handle 3432 the new parameter format automatically. 3433 [Steve Henson] 3434 3435 *) Initial experimental support for X9.42 DH parameter format: mainly 3436 to support use of 'q' parameter for RFC5114 parameters. 3437 [Steve Henson] 3438 3439 *) Add DH parameters from RFC5114 including test data to dhtest. 3440 [Steve Henson] 3441 3442 *) Support for automatic EC temporary key parameter selection. If enabled 3443 the most preferred EC parameters are automatically used instead of 3444 hardcoded fixed parameters. Now a server just has to call: 3445 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 3446 support ECDH and use the most appropriate parameters. 3447 [Steve Henson] 3448 3449 *) Enhance and tidy EC curve and point format TLS extension code. Use 3450 static structures instead of allocation if default values are used. 3451 New ctrls to set curves we wish to support and to retrieve shared curves. 3452 Print out shared curves in s_server. New options to s_server and s_client 3453 to set list of supported curves. 3454 [Steve Henson] 3455 3456 *) New ctrls to retrieve supported signature algorithms and 3457 supported curve values as an array of NIDs. Extend openssl utility 3458 to print out received values. 3459 [Steve Henson] 3460 3461 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 3462 between NIDs and the more common NIST names such as "P-256". Enhance 3463 ecparam utility and ECC method to recognise the NIST names for curves. 3464 [Steve Henson] 3465 3466 *) Enhance SSL/TLS certificate chain handling to support different 3467 chains for each certificate instead of one chain in the parent SSL_CTX. 3468 [Steve Henson] 3469 3470 *) Support for fixed DH ciphersuite client authentication: where both 3471 server and client use DH certificates with common parameters. 3472 [Steve Henson] 3473 3474 *) Support for fixed DH ciphersuites: those requiring DH server 3475 certificates. 3476 [Steve Henson] 3477 3478 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of 3479 the certificate. 3480 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 3481 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 3482 X509_CINF_get_signature were reverted post internal team review. 3483 3484 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 3485 3486 *) Build fixes for the Windows and OpenVMS platforms 3487 [Matt Caswell and Richard Levitte] 3488 3489 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 3490 3491 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 3492 message can cause a segmentation fault in OpenSSL due to a NULL pointer 3493 dereference. This could lead to a Denial Of Service attack. Thanks to 3494 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 3495 (CVE-2014-3571) 3496 [Steve Henson] 3497 3498 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 3499 dtls1_buffer_record function under certain conditions. In particular this 3500 could occur if an attacker sent repeated DTLS records with the same 3501 sequence number but for the next epoch. The memory leak could be exploited 3502 by an attacker in a Denial of Service attack through memory exhaustion. 3503 Thanks to Chris Mueller for reporting this issue. 3504 (CVE-2015-0206) 3505 [Matt Caswell] 3506 3507 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 3508 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 3509 method would be set to NULL which could later result in a NULL pointer 3510 dereference. Thanks to Frank Schmirler for reporting this issue. 3511 (CVE-2014-3569) 3512 [Kurt Roeckx] 3513 3514 *) Abort handshake if server key exchange message is omitted for ephemeral 3515 ECDH ciphersuites. 3516 3517 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 3518 reporting this issue. 3519 (CVE-2014-3572) 3520 [Steve Henson] 3521 3522 *) Remove non-export ephemeral RSA code on client and server. This code 3523 violated the TLS standard by allowing the use of temporary RSA keys in 3524 non-export ciphersuites and could be used by a server to effectively 3525 downgrade the RSA key length used to a value smaller than the server 3526 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 3527 INRIA or reporting this issue. 3528 (CVE-2015-0204) 3529 [Steve Henson] 3530 3531 *) Fixed issue where DH client certificates are accepted without verification. 3532 An OpenSSL server will accept a DH certificate for client authentication 3533 without the certificate verify message. This effectively allows a client to 3534 authenticate without the use of a private key. This only affects servers 3535 which trust a client certificate authority which issues certificates 3536 containing DH keys: these are extremely rare and hardly ever encountered. 3537 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 3538 this issue. 3539 (CVE-2015-0205) 3540 [Steve Henson] 3541 3542 *) Ensure that the session ID context of an SSL is updated when its 3543 SSL_CTX is updated via SSL_set_SSL_CTX. 3544 3545 The session ID context is typically set from the parent SSL_CTX, 3546 and can vary with the CTX. 3547 [Adam Langley] 3548 3549 *) Fix various certificate fingerprint issues. 3550 3551 By using non-DER or invalid encodings outside the signed portion of a 3552 certificate the fingerprint can be changed without breaking the signature. 3553 Although no details of the signed portion of the certificate can be changed 3554 this can cause problems with some applications: e.g. those using the 3555 certificate fingerprint for blacklists. 3556 3557 1. Reject signatures with non zero unused bits. 3558 3559 If the BIT STRING containing the signature has non zero unused bits reject 3560 the signature. All current signature algorithms require zero unused bits. 3561 3562 2. Check certificate algorithm consistency. 3563 3564 Check the AlgorithmIdentifier inside TBS matches the one in the 3565 certificate signature. NB: this will result in signature failure 3566 errors for some broken certificates. 3567 3568 Thanks to Konrad Kraszewski from Google for reporting this issue. 3569 3570 3. Check DSA/ECDSA signatures use DER. 3571 3572 Re-encode DSA/ECDSA signatures and compare with the original received 3573 signature. Return an error if there is a mismatch. 3574 3575 This will reject various cases including garbage after signature 3576 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 3577 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 3578 (negative or with leading zeroes). 3579 3580 Further analysis was conducted and fixes were developed by Stephen Henson 3581 of the OpenSSL core team. 3582 3583 (CVE-2014-8275) 3584 [Steve Henson] 3585 3586 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 3587 results on some platforms, including x86_64. This bug occurs at random 3588 with a very low probability, and is not known to be exploitable in any 3589 way, though its exact impact is difficult to determine. Thanks to Pieter 3590 Wuille (Blockstream) who reported this issue and also suggested an initial 3591 fix. Further analysis was conducted by the OpenSSL development team and 3592 Adam Langley of Google. The final fix was developed by Andy Polyakov of 3593 the OpenSSL core team. 3594 (CVE-2014-3570) 3595 [Andy Polyakov] 3596 3597 *) Do not resume sessions on the server if the negotiated protocol 3598 version does not match the session's version. Resuming with a different 3599 version, while not strictly forbidden by the RFC, is of questionable 3600 sanity and breaks all known clients. 3601 [David Benjamin, Emilia Käsper] 3602 3603 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 3604 early CCS messages during renegotiation. (Note that because 3605 renegotiation is encrypted, this early CCS was not exploitable.) 3606 [Emilia Käsper] 3607 3608 *) Tighten client-side session ticket handling during renegotiation: 3609 ensure that the client only accepts a session ticket if the server sends 3610 the extension anew in the ServerHello. Previously, a TLS client would 3611 reuse the old extension state and thus accept a session ticket if one was 3612 announced in the initial ServerHello. 3613 3614 Similarly, ensure that the client requires a session ticket if one 3615 was advertised in the ServerHello. Previously, a TLS client would 3616 ignore a missing NewSessionTicket message. 3617 [Emilia Käsper] 3618 3619 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 3620 3621 *) SRTP Memory Leak. 3622 3623 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 3624 sends a carefully crafted handshake message, to cause OpenSSL to fail 3625 to free up to 64k of memory causing a memory leak. This could be 3626 exploited in a Denial Of Service attack. This issue affects OpenSSL 3627 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 3628 whether SRTP is used or configured. Implementations of OpenSSL that 3629 have been compiled with OPENSSL_NO_SRTP defined are not affected. 3630 3631 The fix was developed by the OpenSSL team. 3632 (CVE-2014-3513) 3633 [OpenSSL team] 3634 3635 *) Session Ticket Memory Leak. 3636 3637 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 3638 integrity of that ticket is first verified. In the event of a session 3639 ticket integrity check failing, OpenSSL will fail to free memory 3640 causing a memory leak. By sending a large number of invalid session 3641 tickets an attacker could exploit this issue in a Denial Of Service 3642 attack. 3643 (CVE-2014-3567) 3644 [Steve Henson] 3645 3646 *) Build option no-ssl3 is incomplete. 3647 3648 When OpenSSL is configured with "no-ssl3" as a build option, servers 3649 could accept and complete a SSL 3.0 handshake, and clients could be 3650 configured to send them. 3651 (CVE-2014-3568) 3652 [Akamai and the OpenSSL team] 3653 3654 *) Add support for TLS_FALLBACK_SCSV. 3655 Client applications doing fallback retries should call 3656 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 3657 (CVE-2014-3566) 3658 [Adam Langley, Bodo Moeller] 3659 3660 *) Add additional DigestInfo checks. 3661 3662 Re-encode DigestInto in DER and check against the original when 3663 verifying RSA signature: this will reject any improperly encoded 3664 DigestInfo structures. 3665 3666 Note: this is a precautionary measure and no attacks are currently known. 3667 3668 [Steve Henson] 3669 3670 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 3671 3672 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 3673 SRP code can be overrun an internal buffer. Add sanity check that 3674 g, A, B < N to SRP code. 3675 3676 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 3677 Group for discovering this issue. 3678 (CVE-2014-3512) 3679 [Steve Henson] 3680 3681 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 3682 TLS 1.0 instead of higher protocol versions when the ClientHello message 3683 is badly fragmented. This allows a man-in-the-middle attacker to force a 3684 downgrade to TLS 1.0 even if both the server and the client support a 3685 higher protocol version, by modifying the client's TLS records. 3686 3687 Thanks to David Benjamin and Adam Langley (Google) for discovering and 3688 researching this issue. 3689 (CVE-2014-3511) 3690 [David Benjamin] 3691 3692 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 3693 to a denial of service attack. A malicious server can crash the client 3694 with a null pointer dereference (read) by specifying an anonymous (EC)DH 3695 ciphersuite and sending carefully crafted handshake messages. 3696 3697 Thanks to Felix Gröbert (Google) for discovering and researching this 3698 issue. 3699 (CVE-2014-3510) 3700 [Emilia Käsper] 3701 3702 *) By sending carefully crafted DTLS packets an attacker could cause openssl 3703 to leak memory. This can be exploited through a Denial of Service attack. 3704 Thanks to Adam Langley for discovering and researching this issue. 3705 (CVE-2014-3507) 3706 [Adam Langley] 3707 3708 *) An attacker can force openssl to consume large amounts of memory whilst 3709 processing DTLS handshake messages. This can be exploited through a 3710 Denial of Service attack. 3711 Thanks to Adam Langley for discovering and researching this issue. 3712 (CVE-2014-3506) 3713 [Adam Langley] 3714 3715 *) An attacker can force an error condition which causes openssl to crash 3716 whilst processing DTLS packets due to memory being freed twice. This 3717 can be exploited through a Denial of Service attack. 3718 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 3719 this issue. 3720 (CVE-2014-3505) 3721 [Adam Langley] 3722 3723 *) If a multithreaded client connects to a malicious server using a resumed 3724 session and the server sends an ec point format extension it could write 3725 up to 255 bytes to freed memory. 3726 3727 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 3728 issue. 3729 (CVE-2014-3509) 3730 [Gabor Tyukasz] 3731 3732 *) A malicious server can crash an OpenSSL client with a null pointer 3733 dereference (read) by specifying an SRP ciphersuite even though it was not 3734 properly negotiated with the client. This can be exploited through a 3735 Denial of Service attack. 3736 3737 Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for 3738 discovering and researching this issue. 3739 (CVE-2014-5139) 3740 [Steve Henson] 3741 3742 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 3743 X509_name_oneline, X509_name_print_ex et al. to leak some information 3744 from the stack. Applications may be affected if they echo pretty printing 3745 output to the attacker. 3746 3747 Thanks to Ivan Fratric (Google) for discovering this issue. 3748 (CVE-2014-3508) 3749 [Emilia Käsper, and Steve Henson] 3750 3751 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 3752 for corner cases. (Certain input points at infinity could lead to 3753 bogus results, with non-infinity inputs mapped to infinity too.) 3754 [Bodo Moeller] 3755 3756 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 3757 3758 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 3759 handshake can force the use of weak keying material in OpenSSL 3760 SSL/TLS clients and servers. 3761 3762 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 3763 researching this issue. (CVE-2014-0224) 3764 [KIKUCHI Masashi, Steve Henson] 3765 3766 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 3767 OpenSSL DTLS client the code can be made to recurse eventually crashing 3768 in a DoS attack. 3769 3770 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 3771 (CVE-2014-0221) 3772 [Imre Rad, Steve Henson] 3773 3774 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 3775 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 3776 client or server. This is potentially exploitable to run arbitrary 3777 code on a vulnerable client or server. 3778 3779 Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195) 3780 [Jüri Aedla, Steve Henson] 3781 3782 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 3783 are subject to a denial of service attack. 3784 3785 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering 3786 this issue. (CVE-2014-3470) 3787 [Felix Gröbert, Ivan Fratric, Steve Henson] 3788 3789 *) Harmonize version and its documentation. -f flag is used to display 3790 compilation flags. 3791 [mancha <[email protected]>] 3792 3793 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 3794 in i2d_ECPrivateKey. 3795 [mancha <[email protected]>] 3796 3797 *) Fix some double frees. These are not thought to be exploitable. 3798 [mancha <[email protected]>] 3799 3800 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 3801 3802 *) A missing bounds check in the handling of the TLS heartbeat extension 3803 can be used to reveal up to 64k of memory to a connected client or 3804 server. 3805 3806 Thanks for Neel Mehta of Google Security for discovering this bug and to 3807 Adam Langley <[email protected]> and Bodo Moeller <[email protected]> for 3808 preparing the fix (CVE-2014-0160) 3809 [Adam Langley, Bodo Moeller] 3810 3811 *) Fix for the attack described in the paper "Recovering OpenSSL 3812 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 3813 by Yuval Yarom and Naomi Benger. Details can be obtained from: 3814 http://eprint.iacr.org/2014/140 3815 3816 Thanks to Yuval Yarom and Naomi Benger for discovering this 3817 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 3818 [Yuval Yarom and Naomi Benger] 3819 3820 *) TLS pad extension: draft-agl-tls-padding-03 3821 3822 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 3823 TLS client Hello record length value would otherwise be > 255 and 3824 less that 512 pad with a dummy extension containing zeroes so it 3825 is at least 512 bytes long. 3826 3827 [Adam Langley, Steve Henson] 3828 3829 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 3830 3831 *) Fix for TLS record tampering bug. A carefully crafted invalid 3832 handshake could crash OpenSSL with a NULL pointer exception. 3833 Thanks to Anton Johansson for reporting this issues. 3834 (CVE-2013-4353) 3835 3836 *) Keep original DTLS digest and encryption contexts in retransmission 3837 structures so we can use the previous session parameters if they need 3838 to be resent. (CVE-2013-6450) 3839 [Steve Henson] 3840 3841 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 3842 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 3843 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 3844 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 3845 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 3846 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 3847 [Rob Stradling, Adam Langley] 3848 3849 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 3850 3851 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 3852 supporting platforms or when small records were transferred. 3853 [Andy Polyakov, Steve Henson] 3854 3855 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 3856 3857 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 3858 3859 This addresses the flaw in CBC record processing discovered by 3860 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 3861 at: http://www.isg.rhul.ac.uk/tls/ 3862 3863 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 3864 Security Group at Royal Holloway, University of London 3865 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 3866 Emilia Käsper for the initial patch. 3867 (CVE-2013-0169) 3868 [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 3869 3870 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 3871 ciphersuites which can be exploited in a denial of service attack. 3872 Thanks go to and to Adam Langley <[email protected]> for discovering 3873 and detecting this bug and to Wolfgang Ettlinger 3874 <[email protected]> for independently discovering this issue. 3875 (CVE-2012-2686) 3876 [Adam Langley] 3877 3878 *) Return an error when checking OCSP signatures when key is NULL. 3879 This fixes a DoS attack. (CVE-2013-0166) 3880 [Steve Henson] 3881 3882 *) Make openssl verify return errors. 3883 [Chris Palmer <[email protected]> and Ben Laurie] 3884 3885 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 3886 the right response is stapled. Also change SSL_get_certificate() 3887 so it returns the certificate actually sent. 3888 See http://rt.openssl.org/Ticket/Display.html?id=2836. 3889 [Rob Stradling <[email protected]>] 3890 3891 *) Fix possible deadlock when decoding public keys. 3892 [Steve Henson] 3893 3894 *) Don't use TLS 1.0 record version number in initial client hello 3895 if renegotiating. 3896 [Steve Henson] 3897 3898 Changes between 1.0.1b and 1.0.1c [10 May 2012] 3899 3900 *) Sanity check record length before skipping explicit IV in TLS 3901 1.2, 1.1 and DTLS to fix DoS attack. 3902 3903 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 3904 fuzzing as a service testing platform. 3905 (CVE-2012-2333) 3906 [Steve Henson] 3907 3908 *) Initialise tkeylen properly when encrypting CMS messages. 3909 Thanks to Solar Designer of Openwall for reporting this issue. 3910 [Steve Henson] 3911 3912 *) In FIPS mode don't try to use composite ciphers as they are not 3913 approved. 3914 [Steve Henson] 3915 3916 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 3917 3918 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 3919 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 3920 mean any application compiled against OpenSSL 1.0.0 headers setting 3921 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling 3922 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 3923 0x10000000L Any application which was previously compiled against 3924 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 3925 will need to be recompiled as a result. Letting be results in 3926 inability to disable specifically TLS 1.1 and in client context, 3927 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 3928 [Steve Henson] 3929 3930 *) In order to ensure interoperability SSL_OP_NO_protocolX does not 3931 disable just protocol X, but all protocols above X *if* there are 3932 protocols *below* X still enabled. In more practical terms it means 3933 that if application wants to disable TLS1.0 in favor of TLS1.1 and 3934 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 3935 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 3936 client side. 3937 [Andy Polyakov] 3938 3939 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 3940 3941 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 3942 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 3943 in CRYPTO_realloc_clean. 3944 3945 Thanks to Tavis Ormandy, Google Security Team, for discovering this 3946 issue and to Adam Langley <[email protected]> for fixing it. 3947 (CVE-2012-2110) 3948 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 3949 3950 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 3951 [Adam Langley] 3952 3953 *) Workarounds for some broken servers that "hang" if a client hello 3954 record length exceeds 255 bytes. 3955 3956 1. Do not use record version number > TLS 1.0 in initial client 3957 hello: some (but not all) hanging servers will now work. 3958 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 3959 the number of ciphers sent in the client hello. This should be 3960 set to an even number, such as 50, for example by passing: 3961 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 3962 Most broken servers should now work. 3963 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 3964 TLS 1.2 client support entirely. 3965 [Steve Henson] 3966 3967 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 3968 [Andy Polyakov] 3969 3970 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 3971 3972 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 3973 STRING form instead of a DigestInfo. 3974 [Steve Henson] 3975 3976 *) The format used for MDC2 RSA signatures is inconsistent between EVP 3977 and the RSA_sign/RSA_verify functions. This was made more apparent when 3978 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 3979 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 3980 the correct format in RSA_verify so both forms transparently work. 3981 [Steve Henson] 3982 3983 *) Some servers which support TLS 1.0 can choke if we initially indicate 3984 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 3985 encrypted premaster secret. As a workaround use the maximum permitted 3986 client version in client hello, this should keep such servers happy 3987 and still work with previous versions of OpenSSL. 3988 [Steve Henson] 3989 3990 *) Add support for TLS/DTLS heartbeats. 3991 [Robin Seggelmann <[email protected]>] 3992 3993 *) Add support for SCTP. 3994 [Robin Seggelmann <[email protected]>] 3995 3996 *) Improved PRNG seeding for VOS. 3997 [Paul Green <[email protected]>] 3998 3999 *) Extensive assembler packs updates, most notably: 4000 4001 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 4002 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 4003 - x86_64: bit-sliced AES implementation; 4004 - ARM: NEON support, contemporary platforms optimizations; 4005 - s390x: z196 support; 4006 - *: GHASH and GF(2^m) multiplication implementations; 4007 4008 [Andy Polyakov] 4009 4010 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 4011 (removal of unnecessary code) 4012 [Peter Sylvester <[email protected]>] 4013 4014 *) Add TLS key material exporter from RFC 5705. 4015 [Eric Rescorla] 4016 4017 *) Add DTLS-SRTP negotiation from RFC 5764. 4018 [Eric Rescorla] 4019 4020 *) Add Next Protocol Negotiation, 4021 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 4022 disabled with a no-npn flag to config or Configure. Code donated 4023 by Google. 4024 [Adam Langley <[email protected]> and Ben Laurie] 4025 4026 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 4027 NIST-P256, NIST-P521, with constant-time single point multiplication on 4028 typical inputs. Compiler support for the nonstandard type __uint128_t is 4029 required to use this (present in gcc 4.4 and later, for 64-bit builds). 4030 Code made available under Apache License version 2.0. 4031 4032 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 4033 line to include this in your build of OpenSSL, and run "make depend" (or 4034 "make update"). This enables the following EC_METHODs: 4035 4036 EC_GFp_nistp224_method() 4037 EC_GFp_nistp256_method() 4038 EC_GFp_nistp521_method() 4039 4040 EC_GROUP_new_by_curve_name() will automatically use these (while 4041 EC_GROUP_new_curve_GFp() currently prefers the more flexible 4042 implementations). 4043 [Emilia Käsper, Adam Langley, Bodo Moeller (Google)] 4044 4045 *) Use type ossl_ssize_t instead of ssize_t which isn't available on 4046 all platforms. Move ssize_t definition from e_os.h to the public 4047 header file e_os2.h as it now appears in public header file cms.h 4048 [Steve Henson] 4049 4050 *) New -sigopt option to the ca, req and x509 utilities. Additional 4051 signature parameters can be passed using this option and in 4052 particular PSS. 4053 [Steve Henson] 4054 4055 *) Add RSA PSS signing function. This will generate and set the 4056 appropriate AlgorithmIdentifiers for PSS based on those in the 4057 corresponding EVP_MD_CTX structure. No application support yet. 4058 [Steve Henson] 4059 4060 *) Support for companion algorithm specific ASN1 signing routines. 4061 New function ASN1_item_sign_ctx() signs a pre-initialised 4062 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 4063 the appropriate parameters. 4064 [Steve Henson] 4065 4066 *) Add new algorithm specific ASN1 verification initialisation function 4067 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 4068 handling will be the same no matter what EVP_PKEY_METHOD is used. 4069 Add a PSS handler to support verification of PSS signatures: checked 4070 against a number of sample certificates. 4071 [Steve Henson] 4072 4073 *) Add signature printing for PSS. Add PSS OIDs. 4074 [Steve Henson, Martin Kaiser <[email protected]>] 4075 4076 *) Add algorithm specific signature printing. An individual ASN1 method 4077 can now print out signatures instead of the standard hex dump. 4078 4079 More complex signatures (e.g. PSS) can print out more meaningful 4080 information. Include DSA version that prints out the signature 4081 parameters r, s. 4082 [Steve Henson] 4083 4084 *) Password based recipient info support for CMS library: implementing 4085 RFC3211. 4086 [Steve Henson] 4087 4088 *) Split password based encryption into PBES2 and PBKDF2 functions. This 4089 neatly separates the code into cipher and PBE sections and is required 4090 for some algorithms that split PBES2 into separate pieces (such as 4091 password based CMS). 4092 [Steve Henson] 4093 4094 *) Session-handling fixes: 4095 - Fix handling of connections that are resuming with a session ID, 4096 but also support Session Tickets. 4097 - Fix a bug that suppressed issuing of a new ticket if the client 4098 presented a ticket with an expired session. 4099 - Try to set the ticket lifetime hint to something reasonable. 4100 - Make tickets shorter by excluding irrelevant information. 4101 - On the client side, don't ignore renewed tickets. 4102 [Adam Langley, Bodo Moeller (Google)] 4103 4104 *) Fix PSK session representation. 4105 [Bodo Moeller] 4106 4107 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 4108 4109 This work was sponsored by Intel. 4110 [Andy Polyakov] 4111 4112 *) Add GCM support to TLS library. Some custom code is needed to split 4113 the IV between the fixed (from PRF) and explicit (from TLS record) 4114 portions. This adds all GCM ciphersuites supported by RFC5288 and 4115 RFC5289. Generalise some AES* cipherstrings to include GCM and 4116 add a special AESGCM string for GCM only. 4117 [Steve Henson] 4118 4119 *) Expand range of ctrls for AES GCM. Permit setting invocation 4120 field on decrypt and retrieval of invocation field only on encrypt. 4121 [Steve Henson] 4122 4123 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 4124 As required by RFC5289 these ciphersuites cannot be used if for 4125 versions of TLS earlier than 1.2. 4126 [Steve Henson] 4127 4128 *) For FIPS capable OpenSSL interpret a NULL default public key method 4129 as unset and return the appropriate default but do *not* set the default. 4130 This means we can return the appropriate method in applications that 4131 switch between FIPS and non-FIPS modes. 4132 [Steve Henson] 4133 4134 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 4135 ENGINE is used then we cannot handle that in the FIPS module so we 4136 keep original code iff non-FIPS operations are allowed. 4137 [Steve Henson] 4138 4139 *) Add -attime option to openssl utilities. 4140 [Peter Eckersley <[email protected]>, Ben Laurie and Steve Henson] 4141 4142 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 4143 [Steve Henson] 4144 4145 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 4146 FIPS EC methods unconditionally for now. 4147 [Steve Henson] 4148 4149 *) New build option no-ec2m to disable characteristic 2 code. 4150 [Steve Henson] 4151 4152 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 4153 all cases can be covered as some introduce binary incompatibilities. 4154 [Steve Henson] 4155 4156 *) Redirect RSA operations to FIPS module including keygen, 4157 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 4158 [Steve Henson] 4159 4160 *) Add similar low level API blocking to ciphers. 4161 [Steve Henson] 4162 4163 *) Low level digest APIs are not approved in FIPS mode: any attempt 4164 to use these will cause a fatal error. Applications that *really* want 4165 to use them can use the private_* version instead. 4166 [Steve Henson] 4167 4168 *) Redirect cipher operations to FIPS module for FIPS builds. 4169 [Steve Henson] 4170 4171 *) Redirect digest operations to FIPS module for FIPS builds. 4172 [Steve Henson] 4173 4174 *) Update build system to add "fips" flag which will link in fipscanister.o 4175 for static and shared library builds embedding a signature if needed. 4176 [Steve Henson] 4177 4178 *) Output TLS supported curves in preference order instead of numerical 4179 order. This is currently hardcoded for the highest order curves first. 4180 This should be configurable so applications can judge speed vs strength. 4181 [Steve Henson] 4182 4183 *) Add TLS v1.2 server support for client authentication. 4184 [Steve Henson] 4185 4186 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 4187 and enable MD5. 4188 [Steve Henson] 4189 4190 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 4191 FIPS modules versions. 4192 [Steve Henson] 4193 4194 *) Add TLS v1.2 client side support for client authentication. Keep cache 4195 of handshake records longer as we don't know the hash algorithm to use 4196 until after the certificate request message is received. 4197 [Steve Henson] 4198 4199 *) Initial TLS v1.2 client support. Add a default signature algorithms 4200 extension including all the algorithms we support. Parse new signature 4201 format in client key exchange. Relax some ECC signing restrictions for 4202 TLS v1.2 as indicated in RFC5246. 4203 [Steve Henson] 4204 4205 *) Add server support for TLS v1.2 signature algorithms extension. Switch 4206 to new signature format when needed using client digest preference. 4207 All server ciphersuites should now work correctly in TLS v1.2. No client 4208 support yet and no support for client certificates. 4209 [Steve Henson] 4210 4211 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 4212 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 4213 ciphersuites. At present only RSA key exchange ciphersuites work with 4214 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 4215 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 4216 and version checking. 4217 [Steve Henson] 4218 4219 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 4220 with this defined it will not be affected by any changes to ssl internal 4221 structures. Add several utility functions to allow openssl application 4222 to work with OPENSSL_NO_SSL_INTERN defined. 4223 [Steve Henson] 4224 4225 *) A long standing patch to add support for SRP from EdelWeb (Peter 4226 Sylvester and Christophe Renou) was integrated. 4227 [Christophe Renou <[email protected]>, Peter Sylvester 4228 <[email protected]>, Tom Wu <[email protected]>, and 4229 Ben Laurie] 4230 4231 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 4232 [Steve Henson] 4233 4234 *) Permit abbreviated handshakes when renegotiating using the function 4235 SSL_renegotiate_abbreviated(). 4236 [Robin Seggelmann <[email protected]>] 4237 4238 *) Add call to ENGINE_register_all_complete() to 4239 ENGINE_load_builtin_engines(), so some implementations get used 4240 automatically instead of needing explicit application support. 4241 [Steve Henson] 4242 4243 *) Add support for TLS key exporter as described in RFC5705. 4244 [Robin Seggelmann <[email protected]>, Steve Henson] 4245 4246 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 4247 a few changes are required: 4248 4249 Add SSL_OP_NO_TLSv1_1 flag. 4250 Add TLSv1_1 methods. 4251 Update version checking logic to handle version 1.1. 4252 Add explicit IV handling (ported from DTLS code). 4253 Add command line options to s_client/s_server. 4254 [Steve Henson] 4255 4256 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 4257 4258 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 4259 in CMS and PKCS7 code. When RSA decryption fails use a random key for 4260 content decryption and always return the same error. Note: this attack 4261 needs on average 2^20 messages so it only affects automated senders. The 4262 old behaviour can be re-enabled in the CMS code by setting the 4263 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 4264 an MMA defence is not necessary. 4265 Thanks to Ivan Nestlerode <[email protected]> for discovering 4266 this issue. (CVE-2012-0884) 4267 [Steve Henson] 4268 4269 *) Fix CVE-2011-4619: make sure we really are receiving a 4270 client hello before rejecting multiple SGC restarts. Thanks to 4271 Ivan Nestlerode <[email protected]> for discovering this bug. 4272 [Steve Henson] 4273 4274 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 4275 4276 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 4277 Thanks to Antonio Martin, Enterprise Secure Access Research and 4278 Development, Cisco Systems, Inc. for discovering this bug and 4279 preparing a fix. (CVE-2012-0050) 4280 [Antonio Martin] 4281 4282 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 4283 4284 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 4285 of the Vaudenay padding oracle attack on CBC mode encryption 4286 which enables an efficient plaintext recovery attack against 4287 the OpenSSL implementation of DTLS. Their attack exploits timing 4288 differences arising during decryption processing. A research 4289 paper describing this attack can be found at: 4290 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 4291 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 4292 Security Group at Royal Holloway, University of London 4293 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 4294 <[email protected]> and Michael Tuexen <[email protected]> 4295 for preparing the fix. (CVE-2011-4108) 4296 [Robin Seggelmann, Michael Tuexen] 4297 4298 *) Clear bytes used for block padding of SSL 3.0 records. 4299 (CVE-2011-4576) 4300 [Adam Langley (Google)] 4301 4302 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 4303 Kadianakis <[email protected]> for discovering this issue and 4304 Adam Langley for preparing the fix. (CVE-2011-4619) 4305 [Adam Langley (Google)] 4306 4307 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 4308 [Andrey Kulikov <[email protected]>] 4309 4310 *) Prevent malformed RFC3779 data triggering an assertion failure. 4311 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 4312 and Rob Austein <[email protected]> for fixing it. (CVE-2011-4577) 4313 [Rob Austein <[email protected]>] 4314 4315 *) Improved PRNG seeding for VOS. 4316 [Paul Green <[email protected]>] 4317 4318 *) Fix ssl_ciph.c set-up race. 4319 [Adam Langley (Google)] 4320 4321 *) Fix spurious failures in ecdsatest.c. 4322 [Emilia Käsper (Google)] 4323 4324 *) Fix the BIO_f_buffer() implementation (which was mixing different 4325 interpretations of the '..._len' fields). 4326 [Adam Langley (Google)] 4327 4328 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 4329 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 4330 threads won't reuse the same blinding coefficients. 4331 4332 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 4333 lock to call BN_BLINDING_invert_ex, and avoids one use of 4334 BN_BLINDING_update for each BN_BLINDING structure (previously, 4335 the last update always remained unused). 4336 [Emilia Käsper (Google)] 4337 4338 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 4339 [Bob Buckholz (Google)] 4340 4341 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 4342 4343 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 4344 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 4345 [Kaspar Brand <[email protected]>] 4346 4347 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 4348 for multi-threaded use of ECDH. (CVE-2011-3210) 4349 [Adam Langley (Google)] 4350 4351 *) Fix x509_name_ex_d2i memory leak on bad inputs. 4352 [Bodo Moeller] 4353 4354 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 4355 signature public key algorithm by using OID xref utilities instead. 4356 Before this you could only use some ECC ciphersuites with SHA1 only. 4357 [Steve Henson] 4358 4359 *) Add protection against ECDSA timing attacks as mentioned in the paper 4360 by Billy Bob Brumley and Nicola Tuveri, see: 4361 4362 http://eprint.iacr.org/2011/232.pdf 4363 4364 [Billy Bob Brumley and Nicola Tuveri] 4365 4366 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 4367 4368 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 4369 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 4370 4371 *) Fix bug in string printing code: if *any* escaping is enabled we must 4372 escape the escape character (backslash) or the resulting string is 4373 ambiguous. 4374 [Steve Henson] 4375 4376 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 4377 4378 *) Disable code workaround for ancient and obsolete Netscape browsers 4379 and servers: an attacker can use it in a ciphersuite downgrade attack. 4380 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 4381 [Steve Henson] 4382 4383 *) Fixed J-PAKE implementation error, originally discovered by 4384 Sebastien Martini, further info and confirmation from Stefan 4385 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 4386 [Ben Laurie] 4387 4388 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 4389 4390 *) Fix extension code to avoid race conditions which can result in a buffer 4391 overrun vulnerability: resumed sessions must not be modified as they can 4392 be shared by multiple threads. CVE-2010-3864 4393 [Steve Henson] 4394 4395 *) Fix WIN32 build system to correctly link an ENGINE directory into 4396 a DLL. 4397 [Steve Henson] 4398 4399 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 4400 4401 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 4402 (CVE-2010-1633) 4403 [Steve Henson, Peter-Michael Hager <[email protected]>] 4404 4405 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 4406 4407 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 4408 context. The operation can be customised via the ctrl mechanism in 4409 case ENGINEs want to include additional functionality. 4410 [Steve Henson] 4411 4412 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 4413 [Steve Henson] 4414 4415 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 4416 output hashes compatible with older versions of OpenSSL. 4417 [Willy Weisz <[email protected]>] 4418 4419 *) Fix compression algorithm handling: if resuming a session use the 4420 compression algorithm of the resumed session instead of determining 4421 it from client hello again. Don't allow server to change algorithm. 4422 [Steve Henson] 4423 4424 *) Add load_crls() function to apps tidying load_certs() too. Add option 4425 to verify utility to allow additional CRLs to be included. 4426 [Steve Henson] 4427 4428 *) Update OCSP request code to permit adding custom headers to the request: 4429 some responders need this. 4430 [Steve Henson] 4431 4432 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 4433 correctly. 4434 [Julia Lawall <[email protected]>] 4435 4436 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 4437 needlessly dereferenced structures, used obsolete functions and 4438 didn't handle all updated verify codes correctly. 4439 [Steve Henson] 4440 4441 *) Disable MD2 in the default configuration. 4442 [Steve Henson] 4443 4444 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 4445 indicate the initial BIO being pushed or popped. This makes it possible 4446 to determine whether the BIO is the one explicitly called or as a result 4447 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 4448 it handles reference counts correctly and doesn't zero out the I/O bio 4449 when it is not being explicitly popped. WARNING: applications which 4450 included workarounds for the old buggy behaviour will need to be modified 4451 or they could free up already freed BIOs. 4452 [Steve Henson] 4453 4454 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 4455 renaming to all platforms (within the 0.9.8 branch, this was 4456 done conditionally on Netware platforms to avoid a name clash). 4457 [Guenter <[email protected]>] 4458 4459 *) Add ECDHE and PSK support to DTLS. 4460 [Michael Tuexen <[email protected]>] 4461 4462 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 4463 be used on C++. 4464 [Steve Henson] 4465 4466 *) Add "missing" function EVP_MD_flags() (without this the only way to 4467 retrieve a digest flags is by accessing the structure directly. Update 4468 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 4469 or cipher is registered as in the "from" argument. Print out all 4470 registered digests in the dgst usage message instead of manually 4471 attempting to work them out. 4472 [Steve Henson] 4473 4474 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 4475 this allows the use of compression and extensions. Change default cipher 4476 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 4477 by default unless an application cipher string requests it. 4478 [Steve Henson] 4479 4480 *) Alter match criteria in PKCS12_parse(). It used to try to use local 4481 key ids to find matching certificates and keys but some PKCS#12 files 4482 don't follow the (somewhat unwritten) rules and this strategy fails. 4483 Now just gather all certificates together and the first private key 4484 then look for the first certificate that matches the key. 4485 [Steve Henson] 4486 4487 *) Support use of registered digest and cipher names for dgst and cipher 4488 commands instead of having to add each one as a special case. So now 4489 you can do: 4490 4491 openssl sha256 foo 4492 4493 as well as: 4494 4495 openssl dgst -sha256 foo 4496 4497 and this works for ENGINE based algorithms too. 4498 4499 [Steve Henson] 4500 4501 *) Update Gost ENGINE to support parameter files. 4502 [Victor B. Wagner <[email protected]>] 4503 4504 *) Support GeneralizedTime in ca utility. 4505 [Oliver Martin <[email protected]>, Steve Henson] 4506 4507 *) Enhance the hash format used for certificate directory links. The new 4508 form uses the canonical encoding (meaning equivalent names will work 4509 even if they aren't identical) and uses SHA1 instead of MD5. This form 4510 is incompatible with the older format and as a result c_rehash should 4511 be used to rebuild symbolic links. 4512 [Steve Henson] 4513 4514 *) Make PKCS#8 the default write format for private keys, replacing the 4515 traditional format. This form is standardised, more secure and doesn't 4516 include an implicit MD5 dependency. 4517 [Steve Henson] 4518 4519 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 4520 committed to OpenSSL should pass this lot as a minimum. 4521 [Steve Henson] 4522 4523 *) Add session ticket override functionality for use by EAP-FAST. 4524 [Jouni Malinen <[email protected]>] 4525 4526 *) Modify HMAC functions to return a value. Since these can be implemented 4527 in an ENGINE errors can occur. 4528 [Steve Henson] 4529 4530 *) Type-checked OBJ_bsearch_ex. 4531 [Ben Laurie] 4532 4533 *) Type-checked OBJ_bsearch. Also some constification necessitated 4534 by type-checking. Still to come: TXT_DB, bsearch(?), 4535 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 4536 CONF_VALUE. 4537 [Ben Laurie] 4538 4539 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 4540 seconds to a tm structure directly, instead of going through OS 4541 specific date routines. This avoids any issues with OS routines such 4542 as the year 2038 bug. New *_adj() functions for ASN1 time structures 4543 and X509_time_adj_ex() to cover the extended range. The existing 4544 X509_time_adj() is still usable and will no longer have any date issues. 4545 [Steve Henson] 4546 4547 *) Delta CRL support. New use deltas option which will attempt to locate 4548 and search any appropriate delta CRLs available. 4549 4550 This work was sponsored by Google. 4551 [Steve Henson] 4552 4553 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 4554 code and add additional score elements. Validate alternate CRL paths 4555 as part of the CRL checking and indicate a new error "CRL path validation 4556 error" in this case. Applications wanting additional details can use 4557 the verify callback and check the new "parent" field. If this is not 4558 NULL CRL path validation is taking place. Existing applications won't 4559 see this because it requires extended CRL support which is off by 4560 default. 4561 4562 This work was sponsored by Google. 4563 [Steve Henson] 4564 4565 *) Support for freshest CRL extension. 4566 4567 This work was sponsored by Google. 4568 [Steve Henson] 4569 4570 *) Initial indirect CRL support. Currently only supported in the CRLs 4571 passed directly and not via lookup. Process certificate issuer 4572 CRL entry extension and lookup CRL entries by bother issuer name 4573 and serial number. Check and process CRL issuer entry in IDP extension. 4574 4575 This work was sponsored by Google. 4576 [Steve Henson] 4577 4578 *) Add support for distinct certificate and CRL paths. The CRL issuer 4579 certificate is validated separately in this case. Only enabled if 4580 an extended CRL support flag is set: this flag will enable additional 4581 CRL functionality in future. 4582 4583 This work was sponsored by Google. 4584 [Steve Henson] 4585 4586 *) Add support for policy mappings extension. 4587 4588 This work was sponsored by Google. 4589 [Steve Henson] 4590 4591 *) Fixes to pathlength constraint, self issued certificate handling, 4592 policy processing to align with RFC3280 and PKITS tests. 4593 4594 This work was sponsored by Google. 4595 [Steve Henson] 4596 4597 *) Support for name constraints certificate extension. DN, email, DNS 4598 and URI types are currently supported. 4599 4600 This work was sponsored by Google. 4601 [Steve Henson] 4602 4603 *) To cater for systems that provide a pointer-based thread ID rather 4604 than numeric, deprecate the current numeric thread ID mechanism and 4605 replace it with a structure and associated callback type. This 4606 mechanism allows a numeric "hash" to be extracted from a thread ID in 4607 either case, and on platforms where pointers are larger than 'long', 4608 mixing is done to help ensure the numeric 'hash' is usable even if it 4609 can't be guaranteed unique. The default mechanism is to use "&errno" 4610 as a pointer-based thread ID to distinguish between threads. 4611 4612 Applications that want to provide their own thread IDs should now use 4613 CRYPTO_THREADID_set_callback() to register a callback that will call 4614 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 4615 4616 Note that ERR_remove_state() is now deprecated, because it is tied 4617 to the assumption that thread IDs are numeric. ERR_remove_state(0) 4618 to free the current thread's error state should be replaced by 4619 ERR_remove_thread_state(NULL). 4620 4621 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 4622 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 4623 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 4624 application was previously providing a numeric thread callback that 4625 was inappropriate for distinguishing threads, then uniqueness might 4626 have been obtained with &errno that happened immediately in the 4627 intermediate development versions of OpenSSL; this is no longer the 4628 case, the numeric thread callback will now override the automatic use 4629 of &errno.) 4630 [Geoff Thorpe, with help from Bodo Moeller] 4631 4632 *) Initial support for different CRL issuing certificates. This covers a 4633 simple case where the self issued certificates in the chain exist and 4634 the real CRL issuer is higher in the existing chain. 4635 4636 This work was sponsored by Google. 4637 [Steve Henson] 4638 4639 *) Removed effectively defunct crypto/store from the build. 4640 [Ben Laurie] 4641 4642 *) Revamp of STACK to provide stronger type-checking. Still to come: 4643 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 4644 ASN1_STRING, CONF_VALUE. 4645 [Ben Laurie] 4646 4647 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 4648 RAM on SSL connections. This option can save about 34k per idle SSL. 4649 [Nick Mathewson] 4650 4651 *) Revamp of LHASH to provide stronger type-checking. Still to come: 4652 STACK, TXT_DB, bsearch, qsort. 4653 [Ben Laurie] 4654 4655 *) Initial support for Cryptographic Message Syntax (aka CMS) based 4656 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 4657 support for data, signedData, compressedData, digestedData and 4658 encryptedData, envelopedData types included. Scripts to check against 4659 RFC4134 examples draft and interop and consistency checks of many 4660 content types and variants. 4661 [Steve Henson] 4662 4663 *) Add options to enc utility to support use of zlib compression BIO. 4664 [Steve Henson] 4665 4666 *) Extend mk1mf to support importing of options and assembly language 4667 files from Configure script, currently only included in VC-WIN32. 4668 The assembly language rules can now optionally generate the source 4669 files from the associated perl scripts. 4670 [Steve Henson] 4671 4672 *) Implement remaining functionality needed to support GOST ciphersuites. 4673 Interop testing has been performed using CryptoPro implementations. 4674 [Victor B. Wagner <[email protected]>] 4675 4676 *) s390x assembler pack. 4677 [Andy Polyakov] 4678 4679 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 4680 "family." 4681 [Andy Polyakov] 4682 4683 *) Implement Opaque PRF Input TLS extension as specified in 4684 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 4685 official specification yet and no extension type assignment by 4686 IANA exists, this extension (for now) will have to be explicitly 4687 enabled when building OpenSSL by providing the extension number 4688 to use. For example, specify an option 4689 4690 -DTLSEXT_TYPE_opaque_prf_input=0x9527 4691 4692 to the "config" or "Configure" script to enable the extension, 4693 assuming extension number 0x9527 (which is a completely arbitrary 4694 and unofficial assignment based on the MD5 hash of the Internet 4695 Draft). Note that by doing so, you potentially lose 4696 interoperability with other TLS implementations since these might 4697 be using the same extension number for other purposes. 4698 4699 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 4700 opaque PRF input value to use in the handshake. This will create 4701 an internal copy of the length-'len' string at 'src', and will 4702 return non-zero for success. 4703 4704 To get more control and flexibility, provide a callback function 4705 by using 4706 4707 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 4708 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 4709 4710 where 4711 4712 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 4713 void *arg; 4714 4715 Callback function 'cb' will be called in handshakes, and is 4716 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 4717 Argument 'arg' is for application purposes (the value as given to 4718 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 4719 be provided to the callback function). The callback function 4720 has to return non-zero to report success: usually 1 to use opaque 4721 PRF input just if possible, or 2 to enforce use of the opaque PRF 4722 input. In the latter case, the library will abort the handshake 4723 if opaque PRF input is not successfully negotiated. 4724 4725 Arguments 'peerinput' and 'len' given to the callback function 4726 will always be NULL and 0 in the case of a client. A server will 4727 see the client's opaque PRF input through these variables if 4728 available (NULL and 0 otherwise). Note that if the server 4729 provides an opaque PRF input, the length must be the same as the 4730 length of the client's opaque PRF input. 4731 4732 Note that the callback function will only be called when creating 4733 a new session (session resumption can resume whatever was 4734 previously negotiated), and will not be called in SSL 2.0 4735 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 4736 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 4737 for applications that need to enforce opaque PRF input. 4738 4739 [Bodo Moeller] 4740 4741 *) Update ssl code to support digests other than SHA1+MD5 for handshake 4742 MAC. 4743 4744 [Victor B. Wagner <[email protected]>] 4745 4746 *) Add RFC4507 support to OpenSSL. This includes the corrections in 4747 RFC4507bis. The encrypted ticket format is an encrypted encoded 4748 SSL_SESSION structure, that way new session features are automatically 4749 supported. 4750 4751 If a client application caches session in an SSL_SESSION structure 4752 support is transparent because tickets are now stored in the encoded 4753 SSL_SESSION. 4754 4755 The SSL_CTX structure automatically generates keys for ticket 4756 protection in servers so again support should be possible 4757 with no application modification. 4758 4759 If a client or server wishes to disable RFC4507 support then the option 4760 SSL_OP_NO_TICKET can be set. 4761 4762 Add a TLS extension debugging callback to allow the contents of any client 4763 or server extensions to be examined. 4764 4765 This work was sponsored by Google. 4766 [Steve Henson] 4767 4768 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 4769 OpenSSL should now compile cleanly on gcc 4.2 4770 [Peter Hartley <[email protected]>, Steve Henson] 4771 4772 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 4773 support including streaming MAC support: this is required for GOST 4774 ciphersuite support. 4775 [Victor B. Wagner <[email protected]>, Steve Henson] 4776 4777 *) Add option -stream to use PKCS#7 streaming in smime utility. New 4778 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 4779 to output in BER and PEM format. 4780 [Steve Henson] 4781 4782 *) Experimental support for use of HMAC via EVP_PKEY interface. This 4783 allows HMAC to be handled via the EVP_DigestSign*() interface. The 4784 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 4785 ENGINE support for HMAC keys which are unextractable. New -mac and 4786 -macopt options to dgst utility. 4787 [Steve Henson] 4788 4789 *) New option -sigopt to dgst utility. Update dgst to use 4790 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 4791 alternative signing parameters such as X9.31 or PSS in the dgst 4792 utility. 4793 [Steve Henson] 4794 4795 *) Change ssl_cipher_apply_rule(), the internal function that does 4796 the work each time a ciphersuite string requests enabling 4797 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 4798 removing ("!foo+bar") a class of ciphersuites: Now it maintains 4799 the order of disabled ciphersuites such that those ciphersuites 4800 that most recently went from enabled to disabled not only stay 4801 in order with respect to each other, but also have higher priority 4802 than other disabled ciphersuites the next time ciphersuites are 4803 enabled again. 4804 4805 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 4806 the same ciphersuites as with "HIGH" alone, but in a specific 4807 order where the PSK ciphersuites come first (since they are the 4808 most recently disabled ciphersuites when "HIGH" is parsed). 4809 4810 Also, change ssl_create_cipher_list() (using this new 4811 functionality) such that between otherwise identical 4812 ciphersuites, ephemeral ECDH is preferred over ephemeral DH in 4813 the default order. 4814 [Bodo Moeller] 4815 4816 *) Change ssl_create_cipher_list() so that it automatically 4817 arranges the ciphersuites in reasonable order before starting 4818 to process the rule string. Thus, the definition for "DEFAULT" 4819 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 4820 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 4821 This makes it much easier to arrive at a reasonable default order 4822 in applications for which anonymous ciphers are OK (meaning 4823 that you can't actually use DEFAULT). 4824 [Bodo Moeller; suggested by Victor Duchovni] 4825 4826 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 4827 processing) into multiple integers instead of setting 4828 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 4829 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 4830 (These masks as well as the individual bit definitions are hidden 4831 away into the non-exported interface ssl/ssl_locl.h, so this 4832 change to the definition of the SSL_CIPHER structure shouldn't 4833 affect applications.) This give us more bits for each of these 4834 categories, so there is no longer a need to coagulate AES128 and 4835 AES256 into a single algorithm bit, and to coagulate Camellia128 4836 and Camellia256 into a single algorithm bit, which has led to all 4837 kinds of kludges. 4838 4839 Thus, among other things, the kludge introduced in 0.9.7m and 4840 0.9.8e for masking out AES256 independently of AES128 or masking 4841 out Camellia256 independently of AES256 is not needed here in 0.9.9. 4842 4843 With the change, we also introduce new ciphersuite aliases that 4844 so far were missing: "AES128", "AES256", "CAMELLIA128", and 4845 "CAMELLIA256". 4846 [Bodo Moeller] 4847 4848 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 4849 Use the leftmost N bytes of the signature input if the input is 4850 larger than the prime q (with N being the size in bytes of q). 4851 [Nils Larsch] 4852 4853 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 4854 it yet and it is largely untested. 4855 [Steve Henson] 4856 4857 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 4858 [Nils Larsch] 4859 4860 *) Initial incomplete changes to avoid need for function casts in OpenSSL 4861 some compilers (gcc 4.2 and later) reject their use. Safestack is 4862 reimplemented. Update ASN1 to avoid use of legacy functions. 4863 [Steve Henson] 4864 4865 *) Win32/64 targets are linked with Winsock2. 4866 [Andy Polyakov] 4867 4868 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 4869 to external functions. This can be used to increase CRL handling 4870 efficiency especially when CRLs are very large by (for example) storing 4871 the CRL revoked certificates in a database. 4872 [Steve Henson] 4873 4874 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 4875 new CRLs added to a directory can be used. New command line option 4876 -verify_return_error to s_client and s_server. This causes real errors 4877 to be returned by the verify callback instead of carrying on no matter 4878 what. This reflects the way a "real world" verify callback would behave. 4879 [Steve Henson] 4880 4881 *) GOST engine, supporting several GOST algorithms and public key formats. 4882 Kindly donated by Cryptocom. 4883 [Cryptocom] 4884 4885 *) Partial support for Issuing Distribution Point CRL extension. CRLs 4886 partitioned by DP are handled but no indirect CRL or reason partitioning 4887 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 4888 selected via a scoring technique which handles IDP and AKID in CRLs. 4889 [Steve Henson] 4890 4891 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 4892 will ultimately be used for all verify operations: this will remove the 4893 X509_STORE dependency on certificate verification and allow alternative 4894 lookup methods. X509_STORE based implementations of these two callbacks. 4895 [Steve Henson] 4896 4897 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 4898 Modify get_crl() to find a valid (unexpired) CRL if possible. 4899 [Steve Henson] 4900 4901 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 4902 this would be called X509_CRL_cmp() but that name is already used by 4903 a function that just compares CRL issuer names. Cache several CRL 4904 extensions in X509_CRL structure and cache CRLDP in X509. 4905 [Steve Henson] 4906 4907 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 4908 this maps equivalent X509_NAME structures into a consistent structure. 4909 Name comparison can then be performed rapidly using memcmp(). 4910 [Steve Henson] 4911 4912 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 4913 utility. 4914 [Steve Henson] 4915 4916 *) Allow digests to supply their own micalg string for S/MIME type using 4917 the ctrl EVP_MD_CTRL_MICALG. 4918 [Steve Henson] 4919 4920 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 4921 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 4922 ctrl. It can then customise the structure before and/or after signing 4923 if necessary. 4924 [Steve Henson] 4925 4926 *) New function OBJ_add_sigid() to allow application defined signature OIDs 4927 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 4928 to free up any added signature OIDs. 4929 [Steve Henson] 4930 4931 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 4932 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 4933 digest and cipher tables. New options added to openssl utility: 4934 list-message-digest-algorithms and list-cipher-algorithms. 4935 [Steve Henson] 4936 4937 *) Change the array representation of binary polynomials: the list 4938 of degrees of non-zero coefficients is now terminated with -1. 4939 Previously it was terminated with 0, which was also part of the 4940 value; thus, the array representation was not applicable to 4941 polynomials where t^0 has coefficient zero. This change makes 4942 the array representation useful in a more general context. 4943 [Douglas Stebila] 4944 4945 *) Various modifications and fixes to SSL/TLS cipher string 4946 handling. For ECC, the code now distinguishes between fixed ECDH 4947 with RSA certificates on the one hand and with ECDSA certificates 4948 on the other hand, since these are separate ciphersuites. The 4949 unused code for Fortezza ciphersuites has been removed. 4950 4951 For consistency with EDH, ephemeral ECDH is now called "EECDH" 4952 (not "ECDHE"). For consistency with the code for DH 4953 certificates, use of ECDH certificates is now considered ECDH 4954 authentication, not RSA or ECDSA authentication (the latter is 4955 merely the CA's signing algorithm and not actively used in the 4956 protocol). 4957 4958 The temporary ciphersuite alias "ECCdraft" is no longer 4959 available, and ECC ciphersuites are no longer excluded from "ALL" 4960 and "DEFAULT". The following aliases now exist for RFC 4492 4961 ciphersuites, most of these by analogy with the DH case: 4962 4963 kECDHr - ECDH cert, signed with RSA 4964 kECDHe - ECDH cert, signed with ECDSA 4965 kECDH - ECDH cert (signed with either RSA or ECDSA) 4966 kEECDH - ephemeral ECDH 4967 ECDH - ECDH cert or ephemeral ECDH 4968 4969 aECDH - ECDH cert 4970 aECDSA - ECDSA cert 4971 ECDSA - ECDSA cert 4972 4973 AECDH - anonymous ECDH 4974 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 4975 4976 [Bodo Moeller] 4977 4978 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 4979 Use correct micalg parameters depending on digest(s) in signed message. 4980 [Steve Henson] 4981 4982 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 4983 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 4984 [Steve Henson] 4985 4986 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 4987 an engine to register a method. Add ENGINE lookups for methods and 4988 functional reference processing. 4989 [Steve Henson] 4990 4991 *) New functions EVP_Digest{Sign,Verify)*. These are enhanced versions of 4992 EVP_{Sign,Verify}* which allow an application to customise the signature 4993 process. 4994 [Steve Henson] 4995 4996 *) New -resign option to smime utility. This adds one or more signers 4997 to an existing PKCS#7 signedData structure. Also -md option to use an 4998 alternative message digest algorithm for signing. 4999 [Steve Henson] 5000 5001 *) Tidy up PKCS#7 routines and add new functions to make it easier to 5002 create PKCS7 structures containing multiple signers. Update smime 5003 application to support multiple signers. 5004 [Steve Henson] 5005 5006 *) New -macalg option to pkcs12 utility to allow setting of an alternative 5007 digest MAC. 5008 [Steve Henson] 5009 5010 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 5011 Reorganize PBE internals to lookup from a static table using NIDs, 5012 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 5013 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 5014 PRF which will be automatically used with PBES2. 5015 [Steve Henson] 5016 5017 *) Replace the algorithm specific calls to generate keys in "req" with the 5018 new API. 5019 [Steve Henson] 5020 5021 *) Update PKCS#7 enveloped data routines to use new API. This is now 5022 supported by any public key method supporting the encrypt operation. A 5023 ctrl is added to allow the public key algorithm to examine or modify 5024 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 5025 a no op. 5026 [Steve Henson] 5027 5028 *) Add a ctrl to asn1 method to allow a public key algorithm to express 5029 a default digest type to use. In most cases this will be SHA1 but some 5030 algorithms (such as GOST) need to specify an alternative digest. The 5031 return value indicates how strong the preference is 1 means optional and 5032 2 is mandatory (that is it is the only supported type). Modify 5033 ASN1_item_sign() to accept a NULL digest argument to indicate it should 5034 use the default md. Update openssl utilities to use the default digest 5035 type for signing if it is not explicitly indicated. 5036 [Steve Henson] 5037 5038 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 5039 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 5040 signing method from the key type. This effectively removes the link 5041 between digests and public key types. 5042 [Steve Henson] 5043 5044 *) Add an OID cross reference table and utility functions. Its purpose is to 5045 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 5046 rsaEncryption. This will allow some of the algorithm specific hackery 5047 needed to use the correct OID to be removed. 5048 [Steve Henson] 5049 5050 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 5051 structures for PKCS7_sign(). They are now set up by the relevant public 5052 key ASN1 method. 5053 [Steve Henson] 5054 5055 *) Add provisional EC pkey method with support for ECDSA and ECDH. 5056 [Steve Henson] 5057 5058 *) Add support for key derivation (agreement) in the API, DH method and 5059 pkeyutl. 5060 [Steve Henson] 5061 5062 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 5063 public and private key formats. As a side effect these add additional 5064 command line functionality not previously available: DSA signatures can be 5065 generated and verified using pkeyutl and DH key support and generation in 5066 pkey, genpkey. 5067 [Steve Henson] 5068 5069 *) BeOS support. 5070 [Oliver Tappe <[email protected]>] 5071 5072 *) New make target "install_html_docs" installs HTML renditions of the 5073 manual pages. 5074 [Oliver Tappe <[email protected]>] 5075 5076 *) New utility "genpkey" this is analogous to "genrsa" etc except it can 5077 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 5078 support key and parameter generation and add initial key generation 5079 functionality for RSA. 5080 [Steve Henson] 5081 5082 *) Add functions for main EVP_PKEY_method operations. The undocumented 5083 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 5084 EVP_PKEY_{encrypt,decrypt}_old. 5085 [Steve Henson] 5086 5087 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 5088 key API, doesn't do much yet. 5089 [Steve Henson] 5090 5091 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 5092 public key algorithms. New option to openssl utility: 5093 "list-public-key-algorithms" to print out info. 5094 [Steve Henson] 5095 5096 *) Implement the Supported Elliptic Curves Extension for 5097 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 5098 [Douglas Stebila] 5099 5100 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 5101 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 5102 [Steve Henson] 5103 5104 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 5105 utilities such as rsa, dsa, dsaparam etc except they process any key 5106 type. 5107 [Steve Henson] 5108 5109 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 5110 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 5111 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 5112 structure. 5113 [Steve Henson] 5114 5115 *) Initial support for pluggable public key ASN1. 5116 De-spaghettify the public key ASN1 handling. Move public and private 5117 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 5118 algorithm specific handling to a single module within the relevant 5119 algorithm directory. Add functions to allow (near) opaque processing 5120 of public and private key structures. 5121 [Steve Henson] 5122 5123 *) Implement the Supported Point Formats Extension for 5124 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 5125 [Douglas Stebila] 5126 5127 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 5128 for the psk identity [hint] and the psk callback functions to the 5129 SSL_SESSION, SSL and SSL_CTX structure. 5130 5131 New ciphersuites: 5132 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 5133 PSK-AES256-CBC-SHA 5134 5135 New functions: 5136 SSL_CTX_use_psk_identity_hint 5137 SSL_get_psk_identity_hint 5138 SSL_get_psk_identity 5139 SSL_use_psk_identity_hint 5140 5141 [Mika Kousa and Pasi Eronen of Nokia Corporation] 5142 5143 *) Add RFC 3161 compliant time stamp request creation, response generation 5144 and response verification functionality. 5145 [Zoltán Glózik <[email protected]>, The OpenTSA Project] 5146 5147 *) Add initial support for TLS extensions, specifically for the server_name 5148 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 5149 have new members for a host name. The SSL data structure has an 5150 additional member SSL_CTX *initial_ctx so that new sessions can be 5151 stored in that context to allow for session resumption, even after the 5152 SSL has been switched to a new SSL_CTX in reaction to a client's 5153 server_name extension. 5154 5155 New functions (subject to change): 5156 5157 SSL_get_servername() 5158 SSL_get_servername_type() 5159 SSL_set_SSL_CTX() 5160 5161 New CTRL codes and macros (subject to change): 5162 5163 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 5164 - SSL_CTX_set_tlsext_servername_callback() 5165 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 5166 - SSL_CTX_set_tlsext_servername_arg() 5167 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 5168 5169 openssl s_client has a new '-servername ...' option. 5170 5171 openssl s_server has new options '-servername_host ...', '-cert2 ...', 5172 '-key2 ...', '-servername_fatal' (subject to change). This allows 5173 testing the HostName extension for a specific single host name ('-cert' 5174 and '-key' remain fallbacks for handshakes without HostName 5175 negotiation). If the unrecognized_name alert has to be sent, this by 5176 default is a warning; it becomes fatal with the '-servername_fatal' 5177 option. 5178 5179 [Peter Sylvester, Remy Allais, Christophe Renou] 5180 5181 *) Whirlpool hash implementation is added. 5182 [Andy Polyakov] 5183 5184 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 5185 bn(64,32). Because of instruction set limitations it doesn't have 5186 any negative impact on performance. This was done mostly in order 5187 to make it possible to share assembler modules, such as bn_mul_mont 5188 implementations, between 32- and 64-bit builds without hassle. 5189 [Andy Polyakov] 5190 5191 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 5192 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 5193 macro. 5194 [Bodo Moeller] 5195 5196 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 5197 dedicated Montgomery multiplication procedure, is introduced. 5198 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 5199 "64-bit" performance on certain 32-bit targets. 5200 [Andy Polyakov] 5201 5202 *) New option SSL_OP_NO_COMP to disable use of compression selectively 5203 in SSL structures. New SSL ctrl to set maximum send fragment size. 5204 Save memory by setting the I/O buffer sizes dynamically instead of 5205 using the maximum available value. 5206 [Steve Henson] 5207 5208 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 5209 in addition to the text details. 5210 [Bodo Moeller] 5211 5212 *) Very, very preliminary EXPERIMENTAL support for printing of general 5213 ASN1 structures. This currently produces rather ugly output and doesn't 5214 handle several customised structures at all. 5215 [Steve Henson] 5216 5217 *) Integrated support for PVK file format and some related formats such 5218 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 5219 these in the 'rsa' and 'dsa' utilities. 5220 [Steve Henson] 5221 5222 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 5223 [Steve Henson] 5224 5225 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 5226 place for the (very old) "NETSCAPE" format certificates which are now 5227 handled using new ASN1 code equivalents. 5228 [Steve Henson] 5229 5230 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 5231 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 5232 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 5233 [Nils Larsch] 5234 5235 *) Modify CRL distribution points extension code to print out previously 5236 unsupported fields. Enhance extension setting code to allow setting of 5237 all fields. 5238 [Steve Henson] 5239 5240 *) Add print and set support for Issuing Distribution Point CRL extension. 5241 [Steve Henson] 5242 5243 *) Change 'Configure' script to enable Camellia by default. 5244 [NTT] 5245 5246 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 5247 5248 *) When rejecting SSL/TLS records due to an incorrect version number, never 5249 update s->server with a new major version number. As of 5250 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 5251 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 5252 the previous behavior could result in a read attempt at NULL when 5253 receiving specific incorrect SSL/TLS records once record payload 5254 protection is active. (CVE-2010-0740) 5255 [Bodo Moeller, Adam Langley <[email protected]>] 5256 5257 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 5258 could be crashed if the relevant tables were not present (e.g. chrooted). 5259 [Tomas Hoger <[email protected]>] 5260 5261 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 5262 5263 *) Always check bn_wexpand() return values for failure. (CVE-2009-3245) 5264 [Martin Olsson, Neel Mehta] 5265 5266 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 5267 accommodate for stack sorting, always a write lock!). 5268 [Bodo Moeller] 5269 5270 *) On some versions of WIN32 Heap32Next is very slow. This can cause 5271 excessive delays in the RAND_poll(): over a minute. As a workaround 5272 include a time check in the inner Heap32Next loop too. 5273 [Steve Henson] 5274 5275 *) The code that handled flushing of data in SSL/TLS originally used the 5276 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 5277 the problem outlined in PR#1949. The fix suggested there however can 5278 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 5279 of Apache). So instead simplify the code to flush unconditionally. 5280 This should be fine since flushing with no data to flush is a no op. 5281 [Steve Henson] 5282 5283 *) Handle TLS versions 2.0 and later properly and correctly use the 5284 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 5285 off ancient servers have a habit of sticking around for a while... 5286 [Steve Henson] 5287 5288 *) Modify compression code so it frees up structures without using the 5289 ex_data callbacks. This works around a problem where some applications 5290 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 5291 restarting) then use compression (e.g. SSL with compression) later. 5292 This results in significant per-connection memory leaks and 5293 has caused some security issues including CVE-2008-1678 and 5294 CVE-2009-4355. 5295 [Steve Henson] 5296 5297 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 5298 change when encrypting or decrypting. 5299 [Bodo Moeller] 5300 5301 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 5302 connect and renegotiate with servers which do not support RI. 5303 Until RI is more widely deployed this option is enabled by default. 5304 [Steve Henson] 5305 5306 *) Add "missing" ssl ctrls to clear options and mode. 5307 [Steve Henson] 5308 5309 *) If client attempts to renegotiate and doesn't support RI respond with 5310 a no_renegotiation alert as required by RFC5746. Some renegotiating 5311 TLS clients will continue a connection gracefully when they receive 5312 the alert. Unfortunately OpenSSL mishandled this alert and would hang 5313 waiting for a server hello which it will never receive. Now we treat a 5314 received no_renegotiation alert as a fatal error. This is because 5315 applications requesting a renegotiation might well expect it to succeed 5316 and would have no code in place to handle the server denying it so the 5317 only safe thing to do is to terminate the connection. 5318 [Steve Henson] 5319 5320 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 5321 peer supports secure renegotiation and 0 otherwise. Print out peer 5322 renegotiation support in s_client/s_server. 5323 [Steve Henson] 5324 5325 *) Replace the highly broken and deprecated SPKAC certification method with 5326 the updated NID creation version. This should correctly handle UTF8. 5327 [Steve Henson] 5328 5329 *) Implement RFC5746. Re-enable renegotiation but require the extension 5330 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 5331 turns out to be a bad idea. It has been replaced by 5332 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 5333 SSL_CTX_set_options(). This is really not recommended unless you 5334 know what you are doing. 5335 [Eric Rescorla <[email protected]>, Ben Laurie, Steve Henson] 5336 5337 *) Fixes to stateless session resumption handling. Use initial_ctx when 5338 issuing and attempting to decrypt tickets in case it has changed during 5339 servername handling. Use a non-zero length session ID when attempting 5340 stateless session resumption: this makes it possible to determine if 5341 a resumption has occurred immediately after receiving server hello 5342 (several places in OpenSSL subtly assume this) instead of later in 5343 the handshake. 5344 [Steve Henson] 5345 5346 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 5347 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 5348 fixes for a few places where the return code is not checked 5349 correctly. 5350 [Julia Lawall <[email protected]>] 5351 5352 *) Add --strict-warnings option to Configure script to include devteam 5353 warnings in other configurations. 5354 [Steve Henson] 5355 5356 *) Add support for --libdir option and LIBDIR variable in makefiles. This 5357 makes it possible to install openssl libraries in locations which 5358 have names other than "lib", for example "/usr/lib64" which some 5359 systems need. 5360 [Steve Henson, based on patch from Jeremy Utley] 5361 5362 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 5363 X690 8.9.12 and can produce some misleading textual output of OIDs. 5364 [Steve Henson, reported by Dan Kaminsky] 5365 5366 *) Delete MD2 from algorithm tables. This follows the recommendation in 5367 several standards that it is not used in new applications due to 5368 several cryptographic weaknesses. For binary compatibility reasons 5369 the MD2 API is still compiled in by default. 5370 [Steve Henson] 5371 5372 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 5373 and restored. 5374 [Steve Henson] 5375 5376 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 5377 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 5378 clash. 5379 [Guenter <[email protected]>] 5380 5381 *) Fix the server certificate chain building code to use X509_verify_cert(), 5382 it used to have an ad-hoc builder which was unable to cope with anything 5383 other than a simple chain. 5384 [David Woodhouse <[email protected]>, Steve Henson] 5385 5386 *) Don't check self signed certificate signatures in X509_verify_cert() 5387 by default (a flag can override this): it just wastes time without 5388 adding any security. As a useful side effect self signed root CAs 5389 with non-FIPS digests are now usable in FIPS mode. 5390 [Steve Henson] 5391 5392 *) In dtls1_process_out_of_seq_message() the check if the current message 5393 is already buffered was missing. For every new message was memory 5394 allocated, allowing an attacker to perform an denial of service attack 5395 with sending out of seq handshake messages until there is no memory 5396 left. Additionally every future message was buffered, even if the 5397 sequence number made no sense and would be part of another handshake. 5398 So only messages with sequence numbers less than 10 in advance will be 5399 buffered. (CVE-2009-1378) 5400 [Robin Seggelmann, discovered by Daniel Mentz] 5401 5402 *) Records are buffered if they arrive with a future epoch to be 5403 processed after finishing the corresponding handshake. There is 5404 currently no limitation to this buffer allowing an attacker to perform 5405 a DOS attack with sending records with future epochs until there is no 5406 memory left. This patch adds the pqueue_size() function to determine 5407 the size of a buffer and limits the record buffer to 100 entries. 5408 (CVE-2009-1377) 5409 [Robin Seggelmann, discovered by Daniel Mentz] 5410 5411 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 5412 parent structure is freed. (CVE-2009-1379) 5413 [Daniel Mentz] 5414 5415 *) Handle non-blocking I/O properly in SSL_shutdown() call. 5416 [Darryl Miles <[email protected]>] 5417 5418 *) Add 2.5.4.* OIDs 5419 [Ilya O. <[email protected]>] 5420 5421 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 5422 5423 *) Disable renegotiation completely - this fixes a severe security 5424 problem (CVE-2009-3555) at the cost of breaking all 5425 renegotiation. Renegotiation can be re-enabled by setting 5426 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 5427 run-time. This is really not recommended unless you know what 5428 you're doing. 5429 [Ben Laurie] 5430 5431 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 5432 5433 *) Don't set val to NULL when freeing up structures, it is freed up by 5434 underlying code. If sizeof(void *) > sizeof(long) this can result in 5435 zeroing past the valid field. (CVE-2009-0789) 5436 [Paolo Ganci <[email protected]>] 5437 5438 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 5439 checked correctly. This would allow some invalid signed attributes to 5440 appear to verify correctly. (CVE-2009-0591) 5441 [Ivan Nestlerode <[email protected]>] 5442 5443 *) Reject UniversalString and BMPString types with invalid lengths. This 5444 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 5445 a legal length. (CVE-2009-0590) 5446 [Steve Henson] 5447 5448 *) Set S/MIME signing as the default purpose rather than setting it 5449 unconditionally. This allows applications to override it at the store 5450 level. 5451 [Steve Henson] 5452 5453 *) Permit restricted recursion of ASN1 strings. This is needed in practice 5454 to handle some structures. 5455 [Steve Henson] 5456 5457 *) Improve efficiency of mem_gets: don't search whole buffer each time 5458 for a '\n' 5459 [Jeremy Shapiro <[email protected]>] 5460 5461 *) New -hex option for openssl rand. 5462 [Matthieu Herrb] 5463 5464 *) Print out UTF8String and NumericString when parsing ASN1. 5465 [Steve Henson] 5466 5467 *) Support NumericString type for name components. 5468 [Steve Henson] 5469 5470 *) Allow CC in the environment to override the automatically chosen 5471 compiler. Note that nothing is done to ensure flags work with the 5472 chosen compiler. 5473 [Ben Laurie] 5474 5475 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 5476 5477 *) Properly check EVP_VerifyFinal() and similar return values 5478 (CVE-2008-5077). 5479 [Ben Laurie, Bodo Moeller, Google Security Team] 5480 5481 *) Enable TLS extensions by default. 5482 [Ben Laurie] 5483 5484 *) Allow the CHIL engine to be loaded, whether the application is 5485 multithreaded or not. (This does not release the developer from the 5486 obligation to set up the dynamic locking callbacks.) 5487 [Sander Temme <[email protected]>] 5488 5489 *) Use correct exit code if there is an error in dgst command. 5490 [Steve Henson; problem pointed out by Roland Dirlewanger] 5491 5492 *) Tweak Configure so that you need to say "experimental-jpake" to enable 5493 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 5494 [Bodo Moeller] 5495 5496 *) Add experimental JPAKE support, including demo authentication in 5497 s_client and s_server. 5498 [Ben Laurie] 5499 5500 *) Set the comparison function in v3_addr_canonize(). 5501 [Rob Austein <[email protected]>] 5502 5503 *) Add support for XMPP STARTTLS in s_client. 5504 [Philip Paeps <[email protected]>] 5505 5506 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 5507 to ensure that even with this option, only ciphersuites in the 5508 server's preference list will be accepted. (Note that the option 5509 applies only when resuming a session, so the earlier behavior was 5510 just about the algorithm choice for symmetric cryptography.) 5511 [Bodo Moeller] 5512 5513 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 5514 5515 *) Fix NULL pointer dereference if a DTLS server received 5516 ChangeCipherSpec as first record (CVE-2009-1386). 5517 [PR #1679] 5518 5519 *) Fix a state transition in s3_srvr.c and d1_srvr.c 5520 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 5521 [Nagendra Modadugu] 5522 5523 *) The fix in 0.9.8c that supposedly got rid of unsafe 5524 double-checked locking was incomplete for RSA blinding, 5525 addressing just one layer of what turns out to have been 5526 doubly unsafe triple-checked locking. 5527 5528 So now fix this for real by retiring the MONT_HELPER macro 5529 in crypto/rsa/rsa_eay.c. 5530 5531 [Bodo Moeller; problem pointed out by Marius Schilder] 5532 5533 *) Various precautionary measures: 5534 5535 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 5536 5537 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 5538 (NB: This would require knowledge of the secret session ticket key 5539 to exploit, in which case you'd be SOL either way.) 5540 5541 - Change bn_nist.c so that it will properly handle input BIGNUMs 5542 outside the expected range. 5543 5544 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 5545 builds. 5546 5547 [Neel Mehta, Bodo Moeller] 5548 5549 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 5550 the load fails. Useful for distros. 5551 [Ben Laurie and the FreeBSD team] 5552 5553 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 5554 [Steve Henson] 5555 5556 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 5557 [Huang Ying] 5558 5559 *) Expand ENGINE to support engine supplied SSL client certificate functions. 5560 5561 This work was sponsored by Logica. 5562 [Steve Henson] 5563 5564 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 5565 keystores. Support for SSL/TLS client authentication too. 5566 Not compiled unless enable-capieng specified to Configure. 5567 5568 This work was sponsored by Logica. 5569 [Steve Henson] 5570 5571 *) Fix bug in X509_ATTRIBUTE creation: don't set attribute using 5572 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 5573 attribute creation routines such as certificate requests and PKCS#12 5574 files. 5575 [Steve Henson] 5576 5577 Changes between 0.9.8g and 0.9.8h [28 May 2008] 5578 5579 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 5580 handshake which could lead to a client crash as found using the 5581 Codenomicon TLS test suite (CVE-2008-1672) 5582 [Steve Henson, Mark Cox] 5583 5584 *) Fix double free in TLS server name extensions which could lead to 5585 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 5586 [Joe Orton] 5587 5588 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 5589 5590 Clear the error queue to ensure that error entries left from 5591 older function calls do not interfere with the correct operation. 5592 [Lutz Jaenicke, Erik de Castro Lopo] 5593 5594 *) Remove root CA certificates of commercial CAs: 5595 5596 The OpenSSL project does not recommend any specific CA and does not 5597 have any policy with respect to including or excluding any CA. 5598 Therefore it does not make any sense to ship an arbitrary selection 5599 of root CA certificates with the OpenSSL software. 5600 [Lutz Jaenicke] 5601 5602 *) RSA OAEP patches to fix two separate invalid memory reads. 5603 The first one involves inputs when 'lzero' is greater than 5604 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 5605 before the beginning of from). The second one involves inputs where 5606 the 'db' section contains nothing but zeroes (there is a one-byte 5607 invalid read after the end of 'db'). 5608 [Ivan Nestlerode <[email protected]>] 5609 5610 *) Partial backport from 0.9.9-dev: 5611 5612 Introduce bn_mul_mont (dedicated Montgomery multiplication 5613 procedure) as a candidate for BIGNUM assembler implementation. 5614 While 0.9.9-dev uses assembler for various architectures, only 5615 x86_64 is available by default here in the 0.9.8 branch, and 5616 32-bit x86 is available through a compile-time setting. 5617 5618 To try the 32-bit x86 assembler implementation, use Configure 5619 option "enable-montasm" (which exists only for this backport). 5620 5621 As "enable-montasm" for 32-bit x86 disclaims code stability 5622 anyway, in this constellation we activate additional code 5623 backported from 0.9.9-dev for further performance improvements, 5624 namely BN_from_montgomery_word. (To enable this otherwise, 5625 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 5626 5627 [Andy Polyakov (backport partially by Bodo Moeller)] 5628 5629 *) Add TLS session ticket callback. This allows an application to set 5630 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 5631 values. This is useful for key rollover for example where several key 5632 sets may exist with different names. 5633 [Steve Henson] 5634 5635 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 5636 This was broken until now in 0.9.8 releases, such that the only way 5637 a registered ENGINE could be used (assuming it initialises 5638 successfully on the host) was to explicitly set it as the default 5639 for the relevant algorithms. This is in contradiction with 0.9.7 5640 behaviour and the documentation. With this fix, when an ENGINE is 5641 registered into a given algorithm's table of implementations, the 5642 'uptodate' flag is reset so that auto-discovery will be used next 5643 time a new context for that algorithm attempts to select an 5644 implementation. 5645 [Ian Lister (tweaked by Geoff Thorpe)] 5646 5647 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 5648 implementation in the following ways: 5649 5650 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 5651 hard coded. 5652 5653 Lack of BER streaming support means one pass streaming processing is 5654 only supported if data is detached: setting the streaming flag is 5655 ignored for embedded content. 5656 5657 CMS support is disabled by default and must be explicitly enabled 5658 with the enable-cms configuration option. 5659 [Steve Henson] 5660 5661 *) Update the GMP engine glue to do direct copies between BIGNUM and 5662 mpz_t when openssl and GMP use the same limb size. Otherwise the 5663 existing "conversion via a text string export" trick is still used. 5664 [Paul Sheer <[email protected]>] 5665 5666 *) Zlib compression BIO. This is a filter BIO which compressed and 5667 uncompresses any data passed through it. 5668 [Steve Henson] 5669 5670 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 5671 RFC3394 compatible AES key wrapping. 5672 [Steve Henson] 5673 5674 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 5675 sets string data without copying. X509_ALGOR_set0() and 5676 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 5677 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 5678 from an X509_ATTRIBUTE structure optionally checking it occurs only 5679 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 5680 data. 5681 [Steve Henson] 5682 5683 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 5684 to get the expected BN_FLG_CONSTTIME behavior. 5685 [Bodo Moeller (Google)] 5686 5687 *) Netware support: 5688 5689 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 5690 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 5691 - added some more tests to do_tests.pl 5692 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 5693 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 5694 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 5695 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 5696 - various changes to netware.pl to enable gcc-cross builds on Win32 5697 platform 5698 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 5699 - various changes to fix missing prototype warnings 5700 - fixed x86nasm.pl to create correct asm files for NASM COFF output 5701 - added AES, WHIRLPOOL and CPUID assembler code to build files 5702 - added missing AES assembler make rules to mk1mf.pl 5703 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 5704 [Guenter Knauf <[email protected]>] 5705 5706 *) Implement certificate status request TLS extension defined in RFC3546. 5707 A client can set the appropriate parameters and receive the encoded 5708 OCSP response via a callback. A server can query the supplied parameters 5709 and set the encoded OCSP response in the callback. Add simplified examples 5710 to s_client and s_server. 5711 [Steve Henson] 5712 5713 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 5714 5715 *) Fix various bugs: 5716 + Binary incompatibility of ssl_ctx_st structure 5717 + DTLS interoperation with non-compliant servers 5718 + Don't call get_session_cb() without proposed session 5719 + Fix ia64 assembler code 5720 [Andy Polyakov, Steve Henson] 5721 5722 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 5723 5724 *) DTLS Handshake overhaul. There were longstanding issues with 5725 OpenSSL DTLS implementation, which were making it impossible for 5726 RFC 4347 compliant client to communicate with OpenSSL server. 5727 Unfortunately just fixing these incompatibilities would "cut off" 5728 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 5729 server keeps tolerating non RFC compliant syntax. The opposite is 5730 not true, 0.9.8f client can not communicate with earlier server. 5731 This update even addresses CVE-2007-4995. 5732 [Andy Polyakov] 5733 5734 *) Changes to avoid need for function casts in OpenSSL: some compilers 5735 (gcc 4.2 and later) reject their use. 5736 [Kurt Roeckx <[email protected]>, Peter Hartley <[email protected]>, 5737 Steve Henson] 5738 5739 *) Add RFC4507 support to OpenSSL. This includes the corrections in 5740 RFC4507bis. The encrypted ticket format is an encrypted encoded 5741 SSL_SESSION structure, that way new session features are automatically 5742 supported. 5743 5744 If a client application caches session in an SSL_SESSION structure 5745 support is transparent because tickets are now stored in the encoded 5746 SSL_SESSION. 5747 5748 The SSL_CTX structure automatically generates keys for ticket 5749 protection in servers so again support should be possible 5750 with no application modification. 5751 5752 If a client or server wishes to disable RFC4507 support then the option 5753 SSL_OP_NO_TICKET can be set. 5754 5755 Add a TLS extension debugging callback to allow the contents of any client 5756 or server extensions to be examined. 5757 5758 This work was sponsored by Google. 5759 [Steve Henson] 5760 5761 *) Add initial support for TLS extensions, specifically for the server_name 5762 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 5763 have new members for a host name. The SSL data structure has an 5764 additional member SSL_CTX *initial_ctx so that new sessions can be 5765 stored in that context to allow for session resumption, even after the 5766 SSL has been switched to a new SSL_CTX in reaction to a client's 5767 server_name extension. 5768 5769 New functions (subject to change): 5770 5771 SSL_get_servername() 5772 SSL_get_servername_type() 5773 SSL_set_SSL_CTX() 5774 5775 New CTRL codes and macros (subject to change): 5776 5777 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 5778 - SSL_CTX_set_tlsext_servername_callback() 5779 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 5780 - SSL_CTX_set_tlsext_servername_arg() 5781 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 5782 5783 openssl s_client has a new '-servername ...' option. 5784 5785 openssl s_server has new options '-servername_host ...', '-cert2 ...', 5786 '-key2 ...', '-servername_fatal' (subject to change). This allows 5787 testing the HostName extension for a specific single host name ('-cert' 5788 and '-key' remain fallbacks for handshakes without HostName 5789 negotiation). If the unrecognized_name alert has to be sent, this by 5790 default is a warning; it becomes fatal with the '-servername_fatal' 5791 option. 5792 5793 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 5794 5795 *) Add AES and SSE2 assembly language support to VC++ build. 5796 [Steve Henson] 5797 5798 *) Mitigate attack on final subtraction in Montgomery reduction. 5799 [Andy Polyakov] 5800 5801 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 5802 (which previously caused an internal error). 5803 [Bodo Moeller] 5804 5805 *) Squeeze another 10% out of IGE mode when in != out. 5806 [Ben Laurie] 5807 5808 *) AES IGE mode speedup. 5809 [Dean Gaudet (Google)] 5810 5811 *) Add the Korean symmetric 128-bit cipher SEED (see 5812 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 5813 add SEED ciphersuites from RFC 4162: 5814 5815 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 5816 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 5817 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 5818 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 5819 5820 To minimize changes between patchlevels in the OpenSSL 0.9.8 5821 series, SEED remains excluded from compilation unless OpenSSL 5822 is configured with 'enable-seed'. 5823 [KISA, Bodo Moeller] 5824 5825 *) Mitigate branch prediction attacks, which can be practical if a 5826 single processor is shared, allowing a spy process to extract 5827 information. For detailed background information, see 5828 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 5829 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 5830 and Necessary Software Countermeasures"). The core of the change 5831 are new versions BN_div_no_branch() and 5832 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 5833 respectively, which are slower, but avoid the security-relevant 5834 conditional branches. These are automatically called by BN_div() 5835 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 5836 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 5837 remove a conditional branch. 5838 5839 BN_FLG_CONSTTIME is the new name for the previous 5840 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 5841 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 5842 in the exponent causes BN_mod_exp_mont() to use the alternative 5843 implementation in BN_mod_exp_mont_consttime().) The old name 5844 remains as a deprecated alias. 5845 5846 Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 5847 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 5848 constant-time implementations for more than just exponentiation. 5849 Here too the old name is kept as a deprecated alias. 5850 5851 BN_BLINDING_new() will now use BN_dup() for the modulus so that 5852 the BN_BLINDING structure gets an independent copy of the 5853 modulus. This means that the previous "BIGNUM *m" argument to 5854 BN_BLINDING_new() and to BN_BLINDING_create_param() now 5855 essentially becomes "const BIGNUM *m", although we can't actually 5856 change this in the header file before 0.9.9. It allows 5857 RSA_setup_blinding() to use BN_with_flags() on the modulus to 5858 enable BN_FLG_CONSTTIME. 5859 5860 [Matthew D Wood (Intel Corp)] 5861 5862 *) In the SSL/TLS server implementation, be strict about session ID 5863 context matching (which matters if an application uses a single 5864 external cache for different purposes). Previously, 5865 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 5866 set. This did ensure strict client verification, but meant that, 5867 with applications using a single external cache for quite 5868 different requirements, clients could circumvent ciphersuite 5869 restrictions for a given session ID context by starting a session 5870 in a different context. 5871 [Bodo Moeller] 5872 5873 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 5874 a ciphersuite string such as "DEFAULT:RSA" cannot enable 5875 authentication-only ciphersuites. 5876 [Bodo Moeller] 5877 5878 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 5879 not complete and could lead to a possible single byte overflow 5880 (CVE-2007-5135) [Ben Laurie] 5881 5882 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 5883 5884 *) Since AES128 and AES256 (and similarly Camellia128 and 5885 Camellia256) share a single mask bit in the logic of 5886 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 5887 kludge to work properly if AES128 is available and AES256 isn't 5888 (or if Camellia128 is available and Camellia256 isn't). 5889 [Victor Duchovni] 5890 5891 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 5892 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 5893 When a point or a seed is encoded in a BIT STRING, we need to 5894 prevent the removal of trailing zero bits to get the proper DER 5895 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 5896 of a NamedBitList, for which trailing 0 bits need to be removed.) 5897 [Bodo Moeller] 5898 5899 *) Have SSL/TLS server implementation tolerate "mismatched" record 5900 protocol version while receiving ClientHello even if the 5901 ClientHello is fragmented. (The server can't insist on the 5902 particular protocol version it has chosen before the ServerHello 5903 message has informed the client about his choice.) 5904 [Bodo Moeller] 5905 5906 *) Add RFC 3779 support. 5907 [Rob Austein for ARIN, Ben Laurie] 5908 5909 *) Load error codes if they are not already present instead of using a 5910 static variable. This allows them to be cleanly unloaded and reloaded. 5911 Improve header file function name parsing. 5912 [Steve Henson] 5913 5914 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 5915 or CAPABILITY handshake as required by RFCs. 5916 [Goetz Babin-Ebell] 5917 5918 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 5919 5920 *) Introduce limits to prevent malicious keys being able to 5921 cause a denial of service. (CVE-2006-2940) 5922 [Steve Henson, Bodo Moeller] 5923 5924 *) Fix ASN.1 parsing of certain invalid structures that can result 5925 in a denial of service. (CVE-2006-2937) [Steve Henson] 5926 5927 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 5928 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 5929 5930 *) Fix SSL client code which could crash if connecting to a 5931 malicious SSLv2 server. (CVE-2006-4343) 5932 [Tavis Ormandy and Will Drewry, Google Security Team] 5933 5934 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 5935 match only those. Before that, "AES256-SHA" would be interpreted 5936 as a pattern and match "AES128-SHA" too (since AES128-SHA got 5937 the same strength classification in 0.9.7h) as we currently only 5938 have a single AES bit in the ciphersuite description bitmap. 5939 That change, however, also applied to ciphersuite strings such as 5940 "RC4-MD5" that intentionally matched multiple ciphersuites -- 5941 namely, SSL 2.0 ciphersuites in addition to the more common ones 5942 from SSL 3.0/TLS 1.0. 5943 5944 So we change the selection algorithm again: Naming an explicit 5945 ciphersuite selects this one ciphersuite, and any other similar 5946 ciphersuite (same bitmap) from *other* protocol versions. 5947 Thus, "RC4-MD5" again will properly select both the SSL 2.0 5948 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 5949 5950 Since SSL 2.0 does not have any ciphersuites for which the 5951 128/256 bit distinction would be relevant, this works for now. 5952 The proper fix will be to use different bits for AES128 and 5953 AES256, which would have avoided the problems from the beginning; 5954 however, bits are scarce, so we can only do this in a new release 5955 (not just a patchlevel) when we can change the SSL_CIPHER 5956 definition to split the single 'unsigned long mask' bitmap into 5957 multiple values to extend the available space. 5958 5959 [Bodo Moeller] 5960 5961 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 5962 5963 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 5964 (CVE-2006-4339) [Ben Laurie and Google Security Team] 5965 5966 *) Add AES IGE and biIGE modes. 5967 [Ben Laurie] 5968 5969 *) Change the Unix randomness entropy gathering to use poll() when 5970 possible instead of select(), since the latter has some 5971 undesirable limitations. 5972 [Darryl Miles via Richard Levitte and Bodo Moeller] 5973 5974 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 5975 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 5976 cannot be implicitly activated as part of, e.g., the "AES" alias. 5977 However, please upgrade to OpenSSL 0.9.9[-dev] for 5978 non-experimental use of the ECC ciphersuites to get TLS extension 5979 support, which is required for curve and point format negotiation 5980 to avoid potential handshake problems. 5981 [Bodo Moeller] 5982 5983 *) Disable rogue ciphersuites: 5984 5985 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 5986 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 5987 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 5988 5989 The latter two were purportedly from 5990 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 5991 appear there. 5992 5993 Also deactivate the remaining ciphersuites from 5994 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 5995 unofficial, and the ID has long expired. 5996 [Bodo Moeller] 5997 5998 *) Fix RSA blinding Heisenbug (problems sometimes occurred on 5999 dual-core machines) and other potential thread-safety issues. 6000 [Bodo Moeller] 6001 6002 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 6003 versions), which is now available for royalty-free use 6004 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 6005 Also, add Camellia TLS ciphersuites from RFC 4132. 6006 6007 To minimize changes between patchlevels in the OpenSSL 0.9.8 6008 series, Camellia remains excluded from compilation unless OpenSSL 6009 is configured with 'enable-camellia'. 6010 [NTT] 6011 6012 *) Disable the padding bug check when compression is in use. The padding 6013 bug check assumes the first packet is of even length, this is not 6014 necessarily true if compression is enabled and can result in false 6015 positives causing handshake failure. The actual bug test is ancient 6016 code so it is hoped that implementations will either have fixed it by 6017 now or any which still have the bug do not support compression. 6018 [Steve Henson] 6019 6020 Changes between 0.9.8a and 0.9.8b [04 May 2006] 6021 6022 *) When applying a cipher rule check to see if string match is an explicit 6023 cipher suite and only match that one cipher suite if it is. 6024 [Steve Henson] 6025 6026 *) Link in manifests for VC++ if needed. 6027 [Austin Ziegler <[email protected]>] 6028 6029 *) Update support for ECC-based TLS ciphersuites according to 6030 draft-ietf-tls-ecc-12.txt with proposed changes (but without 6031 TLS extensions, which are supported starting with the 0.9.9 6032 branch, not in the OpenSSL 0.9.8 branch). 6033 [Douglas Stebila] 6034 6035 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 6036 opaque EVP_CIPHER_CTX handling. 6037 [Steve Henson] 6038 6039 *) Fixes and enhancements to zlib compression code. We now only use 6040 "zlib1.dll" and use the default __cdecl calling convention on Win32 6041 to conform with the standards mentioned here: 6042 http://www.zlib.net/DLL_FAQ.txt 6043 Static zlib linking now works on Windows and the new --with-zlib-include 6044 --with-zlib-lib options to Configure can be used to supply the location 6045 of the headers and library. Gracefully handle case where zlib library 6046 can't be loaded. 6047 [Steve Henson] 6048 6049 *) Several fixes and enhancements to the OID generation code. The old code 6050 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 6051 handle numbers larger than ULONG_MAX, truncated printing and had a 6052 non standard OBJ_obj2txt() behaviour. 6053 [Steve Henson] 6054 6055 *) Add support for building of engines under engine/ as shared libraries 6056 under VC++ build system. 6057 [Steve Henson] 6058 6059 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 6060 Hopefully, we will not see any false combination of paths any more. 6061 [Richard Levitte] 6062 6063 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 6064 6065 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 6066 (part of SSL_OP_ALL). This option used to disable the 6067 countermeasure against man-in-the-middle protocol-version 6068 rollback in the SSL 2.0 server implementation, which is a bad 6069 idea. (CVE-2005-2969) 6070 6071 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 6072 for Information Security, National Institute of Advanced Industrial 6073 Science and Technology [AIST], Japan)] 6074 6075 *) Add two function to clear and return the verify parameter flags. 6076 [Steve Henson] 6077 6078 *) Keep cipherlists sorted in the source instead of sorting them at 6079 runtime, thus removing the need for a lock. 6080 [Nils Larsch] 6081 6082 *) Avoid some small subgroup attacks in Diffie-Hellman. 6083 [Nick Mathewson and Ben Laurie] 6084 6085 *) Add functions for well-known primes. 6086 [Nick Mathewson] 6087 6088 *) Extended Windows CE support. 6089 [Satoshi Nakamura and Andy Polyakov] 6090 6091 *) Initialize SSL_METHOD structures at compile time instead of during 6092 runtime, thus removing the need for a lock. 6093 [Steve Henson] 6094 6095 *) Make PKCS7_decrypt() work even if no certificate is supplied by 6096 attempting to decrypt each encrypted key in turn. Add support to 6097 smime utility. 6098 [Steve Henson] 6099 6100 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 6101 6102 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 6103 OpenSSL 0.9.8.] 6104 6105 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 6106 [Richard Levitte] 6107 6108 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 6109 key into the same file any more. 6110 [Richard Levitte] 6111 6112 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 6113 [Andy Polyakov] 6114 6115 *) Add -utf8 command line and config file option to 'ca'. 6116 [Stefan <[email protected]] 6117 6118 *) Removed the macro des_crypt(), as it seems to conflict with some 6119 libraries. Use DES_crypt(). 6120 [Richard Levitte] 6121 6122 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 6123 involves renaming the source and generated shared-libs for 6124 both. The engines will accept the corrected or legacy ids 6125 ('ncipher' and '4758_cca' respectively) when binding. NB, 6126 this only applies when building 'shared'. 6127 [Corinna Vinschen <[email protected]> and Geoff Thorpe] 6128 6129 *) Add attribute functions to EVP_PKEY structure. Modify 6130 PKCS12_create() to recognize a CSP name attribute and 6131 use it. Make -CSP option work again in pkcs12 utility. 6132 [Steve Henson] 6133 6134 *) Add new functionality to the bn blinding code: 6135 - automatic re-creation of the BN_BLINDING parameters after 6136 a fixed number of uses (currently 32) 6137 - add new function for parameter creation 6138 - introduce flags to control the update behaviour of the 6139 BN_BLINDING parameters 6140 - hide BN_BLINDING structure 6141 Add a second BN_BLINDING slot to the RSA structure to improve 6142 performance when a single RSA object is shared among several 6143 threads. 6144 [Nils Larsch] 6145 6146 *) Add support for DTLS. 6147 [Nagendra Modadugu <[email protected]> and Ben Laurie] 6148 6149 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 6150 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 6151 [Walter Goulet] 6152 6153 *) Remove buggy and incomplete DH cert support from 6154 ssl/ssl_rsa.c and ssl/s3_both.c 6155 [Nils Larsch] 6156 6157 *) Use SHA-1 instead of MD5 as the default digest algorithm for 6158 the apps/openssl applications. 6159 [Nils Larsch] 6160 6161 *) Compile clean with "-Wall -Wmissing-prototypes 6162 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 6163 DEBUG_SAFESTACK must also be set. 6164 [Ben Laurie] 6165 6166 *) Change ./Configure so that certain algorithms can be disabled by default. 6167 The new counterpiece to "no-xxx" is "enable-xxx". 6168 6169 The patented RC5 and MDC2 algorithms will now be disabled unless 6170 "enable-rc5" and "enable-mdc2", respectively, are specified. 6171 6172 (IDEA remains enabled despite being patented. This is because IDEA 6173 is frequently required for interoperability, and there is no license 6174 fee for non-commercial use. As before, "no-idea" can be used to 6175 avoid this algorithm.) 6176 6177 [Bodo Moeller] 6178 6179 *) Add processing of proxy certificates (see RFC 3820). This work was 6180 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 6181 EGEE (Enabling Grids for E-science in Europe). 6182 [Richard Levitte] 6183 6184 *) RC4 performance overhaul on modern architectures/implementations, such 6185 as Intel P4, IA-64 and AMD64. 6186 [Andy Polyakov] 6187 6188 *) New utility extract-section.pl. This can be used specify an alternative 6189 section number in a pod file instead of having to treat each file as 6190 a separate case in Makefile. This can be done by adding two lines to the 6191 pod file: 6192 6193 =for comment openssl_section:XXX 6194 6195 The blank line is mandatory. 6196 6197 [Steve Henson] 6198 6199 *) New arguments -certform, -keyform and -pass for s_client and s_server 6200 to allow alternative format key and certificate files and passphrase 6201 sources. 6202 [Steve Henson] 6203 6204 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 6205 update associated structures and add various utility functions. 6206 6207 Add new policy related verify parameters, include policy checking in 6208 standard verify code. Enhance 'smime' application with extra parameters 6209 to support policy checking and print out. 6210 [Steve Henson] 6211 6212 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 6213 Nehemiah processors. These extensions support AES encryption in hardware 6214 as well as RNG (though RNG support is currently disabled). 6215 [Michal Ludvig <[email protected]>, with help from Andy Polyakov] 6216 6217 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 6218 [Geoff Thorpe] 6219 6220 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 6221 [Andy Polyakov and a number of other people] 6222 6223 *) Improved PowerPC platform support. Most notably BIGNUM assembler 6224 implementation contributed by IBM. 6225 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 6226 6227 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 6228 exponent rather than 'unsigned long'. There is a corresponding change to 6229 the new 'rsa_keygen' element of the RSA_METHOD structure. 6230 [Jelte Jansen, Geoff Thorpe] 6231 6232 *) Functionality for creating the initial serial number file is now 6233 moved from CA.pl to the 'ca' utility with a new option -create_serial. 6234 6235 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 6236 number file to 1, which is bound to cause problems. To avoid 6237 the problems while respecting compatibility between different 0.9.7 6238 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 6239 CA.pl for serial number initialization. With the new release 0.9.8, 6240 we can fix the problem directly in the 'ca' utility.) 6241 [Steve Henson] 6242 6243 *) Reduced header interdependencies by declaring more opaque objects in 6244 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 6245 give fewer recursive includes, which could break lazy source code - so 6246 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 6247 developers should define this symbol when building and using openssl to 6248 ensure they track the recommended behaviour, interfaces, [etc], but 6249 backwards-compatible behaviour prevails when this isn't defined. 6250 [Geoff Thorpe] 6251 6252 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 6253 [Steve Henson] 6254 6255 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 6256 This will generate a random key of the appropriate length based on the 6257 cipher context. The EVP_CIPHER can provide its own random key generation 6258 routine to support keys of a specific form. This is used in the des and 6259 3des routines to generate a key of the correct parity. Update S/MIME 6260 code to use new functions and hence generate correct parity DES keys. 6261 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 6262 valid (weak or incorrect parity). 6263 [Steve Henson] 6264 6265 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 6266 as looking them up. This is useful when the verified structure may contain 6267 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 6268 present unless the new PKCS7_NO_CRL flag is asserted. 6269 [Steve Henson] 6270 6271 *) Extend ASN1 oid configuration module. It now additionally accepts the 6272 syntax: 6273 6274 shortName = some long name, 1.2.3.4 6275 [Steve Henson] 6276 6277 *) Reimplemented the BN_CTX implementation. There is now no more static 6278 limitation on the number of variables it can handle nor the depth of the 6279 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 6280 information can now expand as required, and rather than having a single 6281 static array of bignums, BN_CTX now uses a linked-list of such arrays 6282 allowing it to expand on demand whilst maintaining the usefulness of 6283 BN_CTX's "bundling". 6284 [Geoff Thorpe] 6285 6286 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 6287 to allow all RSA operations to function using a single BN_CTX. 6288 [Geoff Thorpe] 6289 6290 *) Preliminary support for certificate policy evaluation and checking. This 6291 is initially intended to pass the tests outlined in "Conformance Testing 6292 of Relying Party Client Certificate Path Processing Logic" v1.07. 6293 [Steve Henson] 6294 6295 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 6296 remained unused and not that useful. A variety of other little bignum 6297 tweaks and fixes have also been made continuing on from the audit (see 6298 below). 6299 [Geoff Thorpe] 6300 6301 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 6302 associated ASN1, EVP and SSL functions and old ASN1 macros. 6303 [Richard Levitte] 6304 6305 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 6306 and this should never fail. So the return value from the use of 6307 BN_set_word() (which can fail due to needless expansion) is now deprecated; 6308 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 6309 [Geoff Thorpe] 6310 6311 *) BN_CTX_get() should return zero-valued bignums, providing the same 6312 initialised value as BN_new(). 6313 [Geoff Thorpe, suggested by Ulf Möller] 6314 6315 *) Support for inhibitAnyPolicy certificate extension. 6316 [Steve Henson] 6317 6318 *) An audit of the BIGNUM code is underway, for which debugging code is 6319 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 6320 is considered valid when processing BIGNUMs, and causes execution to 6321 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 6322 further steps are taken to deliberately pollute unused data in BIGNUM 6323 structures to try and expose faulty code further on. For now, openssl will 6324 (in its default mode of operation) continue to tolerate the inconsistent 6325 forms that it has tolerated in the past, but authors and packagers should 6326 consider trying openssl and their own applications when compiled with 6327 these debugging symbols defined. It will help highlight potential bugs in 6328 their own code, and will improve the test coverage for OpenSSL itself. At 6329 some point, these tighter rules will become openssl's default to improve 6330 maintainability, though the assert()s and other overheads will remain only 6331 in debugging configurations. See bn.h for more details. 6332 [Geoff Thorpe, Nils Larsch, Ulf Möller] 6333 6334 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 6335 that can only be obtained through BN_CTX_new() (which implicitly 6336 initialises it). The presence of this function only made it possible 6337 to overwrite an existing structure (and cause memory leaks). 6338 [Geoff Thorpe] 6339 6340 *) Because of the callback-based approach for implementing LHASH as a 6341 template type, lh_insert() adds opaque objects to hash-tables and 6342 lh_doall() or lh_doall_arg() are typically used with a destructor callback 6343 to clean up those corresponding objects before destroying the hash table 6344 (and losing the object pointers). So some over-zealous constifications in 6345 LHASH have been relaxed so that lh_insert() does not take (nor store) the 6346 objects as "const" and the lh_doall[_arg] callback wrappers are not 6347 prototyped to have "const" restrictions on the object pointers they are 6348 given (and so aren't required to cast them away any more). 6349 [Geoff Thorpe] 6350 6351 *) The tmdiff.h API was so ugly and minimal that our own timing utility 6352 (speed) prefers to use its own implementation. The two implementations 6353 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 6354 its object type properly exposed (MS_TM) instead of casting to/from "char 6355 *". This may still change yet if someone realises MS_TM and "ms_time_***" 6356 aren't necessarily the greatest nomenclatures - but this is what was used 6357 internally to the implementation so I've used that for now. 6358 [Geoff Thorpe] 6359 6360 *) Ensure that deprecated functions do not get compiled when 6361 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 6362 the self-tests were still using deprecated key-generation functions so 6363 these have been updated also. 6364 [Geoff Thorpe] 6365 6366 *) Reorganise PKCS#7 code to separate the digest location functionality 6367 into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest(). 6368 New function PKCS7_set_digest() to set the digest type for PKCS#7 6369 digestedData type. Add additional code to correctly generate the 6370 digestedData type and add support for this type in PKCS7 initialization 6371 functions. 6372 [Steve Henson] 6373 6374 *) New function PKCS7_set0_type_other() this initializes a PKCS7 6375 structure of type "other". 6376 [Steve Henson] 6377 6378 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 6379 sure the loop does correctly stop and breaking ("division by zero") 6380 modulus operations are not performed. The (pre-generated) prime 6381 table crypto/bn/bn_prime.h was already correct, but it could not be 6382 re-generated on some platforms because of the "division by zero" 6383 situation in the script. 6384 [Ralf S. Engelschall] 6385 6386 *) Update support for ECC-based TLS ciphersuites according to 6387 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 6388 SHA-1 now is only used for "small" curves (where the 6389 representation of a field element takes up to 24 bytes); for 6390 larger curves, the field element resulting from ECDH is directly 6391 used as premaster secret. 6392 [Douglas Stebila (Sun Microsystems Laboratories)] 6393 6394 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 6395 curve secp160r1 to the tests. 6396 [Douglas Stebila (Sun Microsystems Laboratories)] 6397 6398 *) Add the possibility to load symbols globally with DSO. 6399 [Götz Babin-Ebell <[email protected]> via Richard Levitte] 6400 6401 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 6402 control of the error stack. 6403 [Richard Levitte] 6404 6405 *) Add support for STORE in ENGINE. 6406 [Richard Levitte] 6407 6408 *) Add the STORE type. The intention is to provide a common interface 6409 to certificate and key stores, be they simple file-based stores, or 6410 HSM-type store, or LDAP stores, or... 6411 NOTE: The code is currently UNTESTED and isn't really used anywhere. 6412 [Richard Levitte] 6413 6414 *) Add a generic structure called OPENSSL_ITEM. This can be used to 6415 pass a list of arguments to any function as well as provide a way 6416 for a function to pass data back to the caller. 6417 [Richard Levitte] 6418 6419 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 6420 works like BUF_strdup() but can be used to duplicate a portion of 6421 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 6422 a memory area. 6423 [Richard Levitte] 6424 6425 *) Add the function sk_find_ex() which works like sk_find(), but will 6426 return an index to an element even if an exact match couldn't be 6427 found. The index is guaranteed to point at the element where the 6428 searched-for key would be inserted to preserve sorting order. 6429 [Richard Levitte] 6430 6431 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 6432 takes an extra flags argument for optional functionality. Currently, 6433 the following flags are defined: 6434 6435 OBJ_BSEARCH_VALUE_ON_NOMATCH 6436 This one gets OBJ_bsearch_ex() to return a pointer to the first 6437 element where the comparing function returns a negative or zero 6438 number. 6439 6440 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 6441 This one gets OBJ_bsearch_ex() to return a pointer to the first 6442 element where the comparing function returns zero. This is useful 6443 if there are more than one element where the comparing function 6444 returns zero. 6445 [Richard Levitte] 6446 6447 *) Make it possible to create self-signed certificates with 'openssl ca' 6448 in such a way that the self-signed certificate becomes part of the 6449 CA database and uses the same mechanisms for serial number generation 6450 as all other certificate signing. The new flag '-selfsign' enables 6451 this functionality. Adapt CA.sh and CA.pl.in. 6452 [Richard Levitte] 6453 6454 *) Add functionality to check the public key of a certificate request 6455 against a given private. This is useful to check that a certificate 6456 request can be signed by that key (self-signing). 6457 [Richard Levitte] 6458 6459 *) Make it possible to have multiple active certificates with the same 6460 subject in the CA index file. This is done only if the keyword 6461 'unique_subject' is set to 'no' in the main CA section (default 6462 if 'CA_default') of the configuration file. The value is saved 6463 with the database itself in a separate index attribute file, 6464 named like the index file with '.attr' appended to the name. 6465 [Richard Levitte] 6466 6467 *) Generate multi-valued AVAs using '+' notation in config files for 6468 req and dirName. 6469 [Steve Henson] 6470 6471 *) Support for nameConstraints certificate extension. 6472 [Steve Henson] 6473 6474 *) Support for policyConstraints certificate extension. 6475 [Steve Henson] 6476 6477 *) Support for policyMappings certificate extension. 6478 [Steve Henson] 6479 6480 *) Make sure the default DSA_METHOD implementation only uses its 6481 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 6482 and change its own handlers to be NULL so as to remove unnecessary 6483 indirection. This lets alternative implementations fallback to the 6484 default implementation more easily. 6485 [Geoff Thorpe] 6486 6487 *) Support for directoryName in GeneralName related extensions 6488 in config files. 6489 [Steve Henson] 6490 6491 *) Make it possible to link applications using Makefile.shared. 6492 Make that possible even when linking against static libraries! 6493 [Richard Levitte] 6494 6495 *) Support for single pass processing for S/MIME signing. This now 6496 means that S/MIME signing can be done from a pipe, in addition 6497 cleartext signing (multipart/signed type) is effectively streaming 6498 and the signed data does not need to be all held in memory. 6499 6500 This is done with a new flag PKCS7_STREAM. When this flag is set 6501 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 6502 is done after the data is output (and digests calculated) in 6503 SMIME_write_PKCS7(). 6504 [Steve Henson] 6505 6506 *) Add full support for -rpath/-R, both in shared libraries and 6507 applications, at least on the platforms where it's known how 6508 to do it. 6509 [Richard Levitte] 6510 6511 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 6512 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 6513 will now compute a table of multiples of the generator that 6514 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 6515 faster (notably in the case of a single point multiplication, 6516 scalar * generator). 6517 [Nils Larsch, Bodo Moeller] 6518 6519 *) IPv6 support for certificate extensions. The various extensions 6520 which use the IP:a.b.c.d can now take IPv6 addresses using the 6521 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 6522 correctly. 6523 [Steve Henson] 6524 6525 *) Added an ENGINE that implements RSA by performing private key 6526 exponentiations with the GMP library. The conversions to and from 6527 GMP's mpz_t format aren't optimised nor are any montgomery forms 6528 cached, and on x86 it appears OpenSSL's own performance has caught up. 6529 However there are likely to be other architectures where GMP could 6530 provide a boost. This ENGINE is not built in by default, but it can be 6531 specified at Configure time and should be accompanied by the necessary 6532 linker additions, eg; 6533 ./config -DOPENSSL_USE_GMP -lgmp 6534 [Geoff Thorpe] 6535 6536 *) "openssl engine" will not display ENGINE/DSO load failure errors when 6537 testing availability of engines with "-t" - the old behaviour is 6538 produced by increasing the feature's verbosity with "-tt". 6539 [Geoff Thorpe] 6540 6541 *) ECDSA routines: under certain error conditions uninitialized BN objects 6542 could be freed. Solution: make sure initialization is performed early 6543 enough. (Reported and fix supplied by Nils Larsch <[email protected]> 6544 via PR#459) 6545 [Lutz Jaenicke] 6546 6547 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 6548 and DH_METHOD (eg. by ENGINE implementations) to override the normal 6549 software implementations. For DSA and DH, parameter generation can 6550 also be overridden by providing the appropriate method callbacks. 6551 [Geoff Thorpe] 6552 6553 *) Change the "progress" mechanism used in key-generation and 6554 primality testing to functions that take a new BN_GENCB pointer in 6555 place of callback/argument pairs. The new API functions have "_ex" 6556 postfixes and the older functions are reimplemented as wrappers for 6557 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 6558 declarations of the old functions to help (graceful) attempts to 6559 migrate to the new functions. Also, the new key-generation API 6560 functions operate on a caller-supplied key-structure and return 6561 success/failure rather than returning a key or NULL - this is to 6562 help make "keygen" another member function of RSA_METHOD etc. 6563 6564 Example for using the new callback interface: 6565 6566 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 6567 void *my_arg = ...; 6568 BN_GENCB my_cb; 6569 6570 BN_GENCB_set(&my_cb, my_callback, my_arg); 6571 6572 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 6573 /* For the meaning of a, b in calls to my_callback(), see the 6574 * documentation of the function that calls the callback. 6575 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 6576 * my_callback should return 1 if it wants BN_is_prime_ex() 6577 * to continue, or 0 to stop. 6578 */ 6579 6580 [Geoff Thorpe] 6581 6582 *) Change the ZLIB compression method to be stateful, and make it 6583 available to TLS with the number defined in 6584 draft-ietf-tls-compression-04.txt. 6585 [Richard Levitte] 6586 6587 *) Add the ASN.1 structures and functions for CertificatePair, which 6588 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 6589 6590 CertificatePair ::= SEQUENCE { 6591 forward [0] Certificate OPTIONAL, 6592 reverse [1] Certificate OPTIONAL, 6593 -- at least one of the pair shall be present -- } 6594 6595 Also implement the PEM functions to read and write certificate 6596 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 6597 6598 This needed to be defined, mostly for the sake of the LDAP 6599 attribute crossCertificatePair, but may prove useful elsewhere as 6600 well. 6601 [Richard Levitte] 6602 6603 *) Make it possible to inhibit symlinking of shared libraries in 6604 Makefile.shared, for Cygwin's sake. 6605 [Richard Levitte] 6606 6607 *) Extend the BIGNUM API by creating a function 6608 void BN_set_negative(BIGNUM *a, int neg); 6609 and a macro that behave like 6610 int BN_is_negative(const BIGNUM *a); 6611 6612 to avoid the need to access 'a->neg' directly in applications. 6613 [Nils Larsch] 6614 6615 *) Implement fast modular reduction for pseudo-Mersenne primes 6616 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 6617 EC_GROUP_new_curve_GFp() will now automatically use this 6618 if applicable. 6619 [Nils Larsch <[email protected]>] 6620 6621 *) Add new lock type (CRYPTO_LOCK_BN). 6622 [Bodo Moeller] 6623 6624 *) Change the ENGINE framework to automatically load engines 6625 dynamically from specific directories unless they could be 6626 found to already be built in or loaded. Move all the 6627 current engines except for the cryptodev one to a new 6628 directory engines/. 6629 The engines in engines/ are built as shared libraries if 6630 the "shared" options was given to ./Configure or ./config. 6631 Otherwise, they are inserted in libcrypto.a. 6632 /usr/local/ssl/engines is the default directory for dynamic 6633 engines, but that can be overridden at configure time through 6634 the usual use of --prefix and/or --openssldir, and at run 6635 time with the environment variable OPENSSL_ENGINES. 6636 [Geoff Thorpe and Richard Levitte] 6637 6638 *) Add Makefile.shared, a helper makefile to build shared 6639 libraries. Adapt Makefile.org. 6640 [Richard Levitte] 6641 6642 *) Add version info to Win32 DLLs. 6643 [Peter 'Luna' Runestig" <[email protected]>] 6644 6645 *) Add new 'medium level' PKCS#12 API. Certificates and keys 6646 can be added using this API to created arbitrary PKCS#12 6647 files while avoiding the low level API. 6648 6649 New options to PKCS12_create(), key or cert can be NULL and 6650 will then be omitted from the output file. The encryption 6651 algorithm NIDs can be set to -1 for no encryption, the mac 6652 iteration count can be set to 0 to omit the mac. 6653 6654 Enhance pkcs12 utility by making the -nokeys and -nocerts 6655 options work when creating a PKCS#12 file. New option -nomac 6656 to omit the mac, NONE can be set for an encryption algorithm. 6657 New code is modified to use the enhanced PKCS12_create() 6658 instead of the low level API. 6659 [Steve Henson] 6660 6661 *) Extend ASN1 encoder to support indefinite length constructed 6662 encoding. This can output sequences tags and octet strings in 6663 this form. Modify pk7_asn1.c to support indefinite length 6664 encoding. This is experimental and needs additional code to 6665 be useful, such as an ASN1 bio and some enhanced streaming 6666 PKCS#7 code. 6667 6668 Extend template encode functionality so that tagging is passed 6669 down to the template encoder. 6670 [Steve Henson] 6671 6672 *) Let 'openssl req' fail if an argument to '-newkey' is not 6673 recognized instead of using RSA as a default. 6674 [Bodo Moeller] 6675 6676 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 6677 As these are not official, they are not included in "ALL"; 6678 the "ECCdraft" ciphersuite group alias can be used to select them. 6679 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 6680 6681 *) Add ECDH engine support. 6682 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 6683 6684 *) Add ECDH in new directory crypto/ecdh/. 6685 [Douglas Stebila (Sun Microsystems Laboratories)] 6686 6687 *) Let BN_rand_range() abort with an error after 100 iterations 6688 without success (which indicates a broken PRNG). 6689 [Bodo Moeller] 6690 6691 *) Change BN_mod_sqrt() so that it verifies that the input value 6692 is really the square of the return value. (Previously, 6693 BN_mod_sqrt would show GIGO behaviour.) 6694 [Bodo Moeller] 6695 6696 *) Add named elliptic curves over binary fields from X9.62, SECG, 6697 and WAP/WTLS; add OIDs that were still missing. 6698 6699 [Sheueling Chang Shantz and Douglas Stebila 6700 (Sun Microsystems Laboratories)] 6701 6702 *) Extend the EC library for elliptic curves over binary fields 6703 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 6704 New EC_METHOD: 6705 6706 EC_GF2m_simple_method 6707 6708 New API functions: 6709 6710 EC_GROUP_new_curve_GF2m 6711 EC_GROUP_set_curve_GF2m 6712 EC_GROUP_get_curve_GF2m 6713 EC_POINT_set_affine_coordinates_GF2m 6714 EC_POINT_get_affine_coordinates_GF2m 6715 EC_POINT_set_compressed_coordinates_GF2m 6716 6717 Point compression for binary fields is disabled by default for 6718 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 6719 enable it). 6720 6721 As binary polynomials are represented as BIGNUMs, various members 6722 of the EC_GROUP and EC_POINT data structures can be shared 6723 between the implementations for prime fields and binary fields; 6724 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 6725 are essentially identical to their ..._GFp counterparts. 6726 (For simplicity, the '..._GFp' prefix has been dropped from 6727 various internal method names.) 6728 6729 An internal 'field_div' method (similar to 'field_mul' and 6730 'field_sqr') has been added; this is used only for binary fields. 6731 6732 [Sheueling Chang Shantz and Douglas Stebila 6733 (Sun Microsystems Laboratories)] 6734 6735 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 6736 through methods ('mul', 'precompute_mult'). 6737 6738 The generic implementations (now internally called 'ec_wNAF_mul' 6739 and 'ec_wNAF_precomputed_mult') remain the default if these 6740 methods are undefined. 6741 6742 [Sheueling Chang Shantz and Douglas Stebila 6743 (Sun Microsystems Laboratories)] 6744 6745 *) New function EC_GROUP_get_degree, which is defined through 6746 EC_METHOD. For curves over prime fields, this returns the bit 6747 length of the modulus. 6748 6749 [Sheueling Chang Shantz and Douglas Stebila 6750 (Sun Microsystems Laboratories)] 6751 6752 *) New functions EC_GROUP_dup, EC_POINT_dup. 6753 (These simply call ..._new and ..._copy). 6754 6755 [Sheueling Chang Shantz and Douglas Stebila 6756 (Sun Microsystems Laboratories)] 6757 6758 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 6759 Polynomials are represented as BIGNUMs (where the sign bit is not 6760 used) in the following functions [macros]: 6761 6762 BN_GF2m_add 6763 BN_GF2m_sub [= BN_GF2m_add] 6764 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 6765 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 6766 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 6767 BN_GF2m_mod_inv 6768 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 6769 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 6770 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 6771 BN_GF2m_cmp [= BN_ucmp] 6772 6773 (Note that only the 'mod' functions are actually for fields GF(2^m). 6774 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 6775 6776 For some functions, an the irreducible polynomial defining a 6777 field can be given as an 'unsigned int[]' with strictly 6778 decreasing elements giving the indices of those bits that are set; 6779 i.e., p[] represents the polynomial 6780 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 6781 where 6782 p[0] > p[1] > ... > p[k] = 0. 6783 This applies to the following functions: 6784 6785 BN_GF2m_mod_arr 6786 BN_GF2m_mod_mul_arr 6787 BN_GF2m_mod_sqr_arr 6788 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 6789 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 6790 BN_GF2m_mod_exp_arr 6791 BN_GF2m_mod_sqrt_arr 6792 BN_GF2m_mod_solve_quad_arr 6793 BN_GF2m_poly2arr 6794 BN_GF2m_arr2poly 6795 6796 Conversion can be performed by the following functions: 6797 6798 BN_GF2m_poly2arr 6799 BN_GF2m_arr2poly 6800 6801 bntest.c has additional tests for binary polynomial arithmetic. 6802 6803 Two implementations for BN_GF2m_mod_div() are available. 6804 The default algorithm simply uses BN_GF2m_mod_inv() and 6805 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 6806 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 6807 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 6808 6809 [Sheueling Chang Shantz and Douglas Stebila 6810 (Sun Microsystems Laboratories)] 6811 6812 *) Add new error code 'ERR_R_DISABLED' that can be used when some 6813 functionality is disabled at compile-time. 6814 [Douglas Stebila <[email protected]>] 6815 6816 *) Change default behaviour of 'openssl asn1parse' so that more 6817 information is visible when viewing, e.g., a certificate: 6818 6819 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 6820 mode the content of non-printable OCTET STRINGs is output in a 6821 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 6822 avoid the appearance of a printable string. 6823 [Nils Larsch <[email protected]>] 6824 6825 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 6826 functions 6827 EC_GROUP_set_asn1_flag() 6828 EC_GROUP_get_asn1_flag() 6829 EC_GROUP_set_point_conversion_form() 6830 EC_GROUP_get_point_conversion_form() 6831 These control ASN1 encoding details: 6832 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 6833 has been set to OPENSSL_EC_NAMED_CURVE. 6834 - Points are encoded in uncompressed form by default; options for 6835 asn1_for are as for point2oct, namely 6836 POINT_CONVERSION_COMPRESSED 6837 POINT_CONVERSION_UNCOMPRESSED 6838 POINT_CONVERSION_HYBRID 6839 6840 Also add 'seed' and 'seed_len' members to EC_GROUP with access 6841 functions 6842 EC_GROUP_set_seed() 6843 EC_GROUP_get0_seed() 6844 EC_GROUP_get_seed_len() 6845 This is used only for ASN1 purposes (so far). 6846 [Nils Larsch <[email protected]>] 6847 6848 *) Add 'field_type' member to EC_METHOD, which holds the NID 6849 of the appropriate field type OID. The new function 6850 EC_METHOD_get_field_type() returns this value. 6851 [Nils Larsch <[email protected]>] 6852 6853 *) Add functions 6854 EC_POINT_point2bn() 6855 EC_POINT_bn2point() 6856 EC_POINT_point2hex() 6857 EC_POINT_hex2point() 6858 providing useful interfaces to EC_POINT_point2oct() and 6859 EC_POINT_oct2point(). 6860 [Nils Larsch <[email protected]>] 6861 6862 *) Change internals of the EC library so that the functions 6863 EC_GROUP_set_generator() 6864 EC_GROUP_get_generator() 6865 EC_GROUP_get_order() 6866 EC_GROUP_get_cofactor() 6867 are implemented directly in crypto/ec/ec_lib.c and not dispatched 6868 to methods, which would lead to unnecessary code duplication when 6869 adding different types of curves. 6870 [Nils Larsch <[email protected]> with input by Bodo Moeller] 6871 6872 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 6873 arithmetic, and such that modified wNAFs are generated 6874 (which avoid length expansion in many cases). 6875 [Bodo Moeller] 6876 6877 *) Add a function EC_GROUP_check_discriminant() (defined via 6878 EC_METHOD) that verifies that the curve discriminant is non-zero. 6879 6880 Add a function EC_GROUP_check() that makes some sanity tests 6881 on a EC_GROUP, its generator and order. This includes 6882 EC_GROUP_check_discriminant(). 6883 [Nils Larsch <[email protected]>] 6884 6885 *) Add ECDSA in new directory crypto/ecdsa/. 6886 6887 Add applications 'openssl ecparam' and 'openssl ecdsa' 6888 (these are based on 'openssl dsaparam' and 'openssl dsa'). 6889 6890 ECDSA support is also included in various other files across the 6891 library. Most notably, 6892 - 'openssl req' now has a '-newkey ecdsa:file' option; 6893 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 6894 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 6895 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 6896 them suitable for ECDSA where domain parameters must be 6897 extracted before the specific public key; 6898 - ECDSA engine support has been added. 6899 [Nils Larsch <[email protected]>] 6900 6901 *) Include some named elliptic curves, and add OIDs from X9.62, 6902 SECG, and WAP/WTLS. Each curve can be obtained from the new 6903 function 6904 EC_GROUP_new_by_curve_name(), 6905 and the list of available named curves can be obtained with 6906 EC_get_builtin_curves(). 6907 Also add a 'curve_name' member to EC_GROUP objects, which can be 6908 accessed via 6909 EC_GROUP_set_curve_name() 6910 EC_GROUP_get_curve_name() 6911 [Nils Larsch <[email protected], Bodo Moeller] 6912 6913 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 6914 was actually never needed) and in BN_mul(). The removal in BN_mul() 6915 required a small change in bn_mul_part_recursive() and the addition 6916 of the functions bn_cmp_part_words(), bn_sub_part_words() and 6917 bn_add_part_words(), which do the same thing as bn_cmp_words(), 6918 bn_sub_words() and bn_add_words() except they take arrays with 6919 differing sizes. 6920 [Richard Levitte] 6921 6922 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 6923 6924 *) Cleanse PEM buffers before freeing them since they may contain 6925 sensitive data. 6926 [Benjamin Bennett <[email protected]>] 6927 6928 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 6929 a ciphersuite string such as "DEFAULT:RSA" cannot enable 6930 authentication-only ciphersuites. 6931 [Bodo Moeller] 6932 6933 *) Since AES128 and AES256 share a single mask bit in the logic of 6934 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 6935 kludge to work properly if AES128 is available and AES256 isn't. 6936 [Victor Duchovni] 6937 6938 *) Expand security boundary to match 1.1.1 module. 6939 [Steve Henson] 6940 6941 *) Remove redundant features: hash file source, editing of test vectors 6942 modify fipsld to use external fips_premain.c signature. 6943 [Steve Henson] 6944 6945 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 6946 run algorithm test programs. 6947 [Steve Henson] 6948 6949 *) Make algorithm test programs more tolerant of whitespace. 6950 [Steve Henson] 6951 6952 *) Have SSL/TLS server implementation tolerate "mismatched" record 6953 protocol version while receiving ClientHello even if the 6954 ClientHello is fragmented. (The server can't insist on the 6955 particular protocol version it has chosen before the ServerHello 6956 message has informed the client about his choice.) 6957 [Bodo Moeller] 6958 6959 *) Load error codes if they are not already present instead of using a 6960 static variable. This allows them to be cleanly unloaded and reloaded. 6961 [Steve Henson] 6962 6963 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 6964 6965 *) Introduce limits to prevent malicious keys being able to 6966 cause a denial of service. (CVE-2006-2940) 6967 [Steve Henson, Bodo Moeller] 6968 6969 *) Fix ASN.1 parsing of certain invalid structures that can result 6970 in a denial of service. (CVE-2006-2937) [Steve Henson] 6971 6972 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 6973 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 6974 6975 *) Fix SSL client code which could crash if connecting to a 6976 malicious SSLv2 server. (CVE-2006-4343) 6977 [Tavis Ormandy and Will Drewry, Google Security Team] 6978 6979 *) Change ciphersuite string processing so that an explicit 6980 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 6981 will no longer include "AES128-SHA"), and any other similar 6982 ciphersuite (same bitmap) from *other* protocol versions (so that 6983 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 6984 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 6985 changes from 0.9.8b and 0.9.8d. 6986 [Bodo Moeller] 6987 6988 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 6989 6990 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 6991 (CVE-2006-4339) [Ben Laurie and Google Security Team] 6992 6993 *) Change the Unix randomness entropy gathering to use poll() when 6994 possible instead of select(), since the latter has some 6995 undesirable limitations. 6996 [Darryl Miles via Richard Levitte and Bodo Moeller] 6997 6998 *) Disable rogue ciphersuites: 6999 7000 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 7001 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 7002 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 7003 7004 The latter two were purportedly from 7005 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 7006 appear there. 7007 7008 Also deactivate the remaining ciphersuites from 7009 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 7010 unofficial, and the ID has long expired. 7011 [Bodo Moeller] 7012 7013 *) Fix RSA blinding Heisenbug (problems sometimes occurred on 7014 dual-core machines) and other potential thread-safety issues. 7015 [Bodo Moeller] 7016 7017 Changes between 0.9.7i and 0.9.7j [04 May 2006] 7018 7019 *) Adapt fipsld and the build system to link against the validated FIPS 7020 module in FIPS mode. 7021 [Steve Henson] 7022 7023 *) Fixes for VC++ 2005 build under Windows. 7024 [Steve Henson] 7025 7026 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 7027 from a Windows bash shell such as MSYS. It is autodetected from the 7028 "config" script when run from a VC++ environment. Modify standard VC++ 7029 build to use fipscanister.o from the GNU make build. 7030 [Steve Henson] 7031 7032 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 7033 7034 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 7035 The value now differs depending on if you build for FIPS or not. 7036 BEWARE! A program linked with a shared FIPSed libcrypto can't be 7037 safely run with a non-FIPSed libcrypto, as it may crash because of 7038 the difference induced by this change. 7039 [Andy Polyakov] 7040 7041 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 7042 7043 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 7044 (part of SSL_OP_ALL). This option used to disable the 7045 countermeasure against man-in-the-middle protocol-version 7046 rollback in the SSL 2.0 server implementation, which is a bad 7047 idea. (CVE-2005-2969) 7048 7049 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 7050 for Information Security, National Institute of Advanced Industrial 7051 Science and Technology [AIST], Japan)] 7052 7053 *) Minimal support for X9.31 signatures and PSS padding modes. This is 7054 mainly for FIPS compliance and not fully integrated at this stage. 7055 [Steve Henson] 7056 7057 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 7058 the exponentiation using a fixed-length exponent. (Otherwise, 7059 the information leaked through timing could expose the secret key 7060 after many signatures; cf. Bleichenbacher's attack on DSA with 7061 biased k.) 7062 [Bodo Moeller] 7063 7064 *) Make a new fixed-window mod_exp implementation the default for 7065 RSA, DSA, and DH private-key operations so that the sequence of 7066 squares and multiplies and the memory access pattern are 7067 independent of the particular secret key. This will mitigate 7068 cache-timing and potential related attacks. 7069 7070 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 7071 and this is automatically used by BN_mod_exp_mont() if the new flag 7072 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 7073 will use this BN flag for private exponents unless the flag 7074 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 7075 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 7076 7077 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 7078 7079 *) Change the client implementation for SSLv23_method() and 7080 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 7081 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 7082 (Previously, the SSL 2.0 backwards compatible Client Hello 7083 message format would be used even with SSL_OP_NO_SSLv2.) 7084 [Bodo Moeller] 7085 7086 *) Add support for smime-type MIME parameter in S/MIME messages which some 7087 clients need. 7088 [Steve Henson] 7089 7090 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 7091 a threadsafe manner. Modify rsa code to use new function and add calls 7092 to dsa and dh code (which had race conditions before). 7093 [Steve Henson] 7094 7095 *) Include the fixed error library code in the C error file definitions 7096 instead of fixing them up at runtime. This keeps the error code 7097 structures constant. 7098 [Steve Henson] 7099 7100 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 7101 7102 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 7103 OpenSSL 0.9.8.] 7104 7105 *) Fixes for newer kerberos headers. NB: the casts are needed because 7106 the 'length' field is signed on one version and unsigned on another 7107 with no (?) obvious way to tell the difference, without these VC++ 7108 complains. Also the "definition" of FAR (blank) is no longer included 7109 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 7110 some needed definitions. 7111 [Steve Henson] 7112 7113 *) Undo Cygwin change. 7114 [Ulf Möller] 7115 7116 *) Added support for proxy certificates according to RFC 3820. 7117 Because they may be a security thread to unaware applications, 7118 they must be explicitly allowed in run-time. See 7119 docs/HOWTO/proxy_certificates.txt for further information. 7120 [Richard Levitte] 7121 7122 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 7123 7124 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 7125 server and client random values. Previously 7126 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 7127 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 7128 7129 This change has negligible security impact because: 7130 7131 1. Server and client random values still have 24 bytes of pseudo random 7132 data. 7133 7134 2. Server and client random values are sent in the clear in the initial 7135 handshake. 7136 7137 3. The master secret is derived using the premaster secret (48 bytes in 7138 size for static RSA ciphersuites) as well as client server and random 7139 values. 7140 7141 The OpenSSL team would like to thank the UK NISCC for bringing this issue 7142 to our attention. 7143 7144 [Stephen Henson, reported by UK NISCC] 7145 7146 *) Use Windows randomness collection on Cygwin. 7147 [Ulf Möller] 7148 7149 *) Fix hang in EGD/PRNGD query when communication socket is closed 7150 prematurely by EGD/PRNGD. 7151 [Darren Tucker <[email protected]> via Lutz Jänicke, resolves #1014] 7152 7153 *) Prompt for pass phrases when appropriate for PKCS12 input format. 7154 [Steve Henson] 7155 7156 *) Back-port of selected performance improvements from development 7157 branch, as well as improved support for PowerPC platforms. 7158 [Andy Polyakov] 7159 7160 *) Add lots of checks for memory allocation failure, error codes to indicate 7161 failure and freeing up memory if a failure occurs. 7162 [Nauticus Networks SSL Team <[email protected]>, Steve Henson] 7163 7164 *) Add new -passin argument to dgst. 7165 [Steve Henson] 7166 7167 *) Perform some character comparisons of different types in X509_NAME_cmp: 7168 this is needed for some certificates that re-encode DNs into UTF8Strings 7169 (in violation of RFC3280) and can't or won't issue name rollover 7170 certificates. 7171 [Steve Henson] 7172 7173 *) Make an explicit check during certificate validation to see that 7174 the CA setting in each certificate on the chain is correct. As a 7175 side effect always do the following basic checks on extensions, 7176 not just when there's an associated purpose to the check: 7177 7178 - if there is an unhandled critical extension (unless the user 7179 has chosen to ignore this fault) 7180 - if the path length has been exceeded (if one is set at all) 7181 - that certain extensions fit the associated purpose (if one has 7182 been given) 7183 [Richard Levitte] 7184 7185 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 7186 7187 *) Avoid a race condition when CRLs are checked in a multi threaded 7188 environment. This would happen due to the reordering of the revoked 7189 entries during signature checking and serial number lookup. Now the 7190 encoding is cached and the serial number sort performed under a lock. 7191 Add new STACK function sk_is_sorted(). 7192 [Steve Henson] 7193 7194 *) Add Delta CRL to the extension code. 7195 [Steve Henson] 7196 7197 *) Various fixes to s3_pkt.c so alerts are sent properly. 7198 [David Holmes <[email protected]>] 7199 7200 *) Reduce the chances of duplicate issuer name and serial numbers (in 7201 violation of RFC3280) using the OpenSSL certificate creation utilities. 7202 This is done by creating a random 64 bit value for the initial serial 7203 number when a serial number file is created or when a self signed 7204 certificate is created using 'openssl req -x509'. The initial serial 7205 number file is created using 'openssl x509 -next_serial' in CA.pl 7206 rather than being initialized to 1. 7207 [Steve Henson] 7208 7209 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 7210 7211 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 7212 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 7213 [Joe Orton, Steve Henson] 7214 7215 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 7216 (CVE-2004-0112) 7217 [Joe Orton, Steve Henson] 7218 7219 *) Make it possible to have multiple active certificates with the same 7220 subject in the CA index file. This is done only if the keyword 7221 'unique_subject' is set to 'no' in the main CA section (default 7222 if 'CA_default') of the configuration file. The value is saved 7223 with the database itself in a separate index attribute file, 7224 named like the index file with '.attr' appended to the name. 7225 [Richard Levitte] 7226 7227 *) X509 verify fixes. Disable broken certificate workarounds when 7228 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 7229 keyUsage extension present. Don't accept CRLs with unhandled critical 7230 extensions: since verify currently doesn't process CRL extensions this 7231 rejects a CRL with *any* critical extensions. Add new verify error codes 7232 for these cases. 7233 [Steve Henson] 7234 7235 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 7236 A clarification of RFC2560 will require the use of OCTET STRINGs and 7237 some implementations cannot handle the current raw format. Since OpenSSL 7238 copies and compares OCSP nonces as opaque blobs without any attempt at 7239 parsing them this should not create any compatibility issues. 7240 [Steve Henson] 7241 7242 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 7243 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 7244 this HMAC (and other) operations are several times slower than OpenSSL 7245 < 0.9.7. 7246 [Steve Henson] 7247 7248 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 7249 [Peter Sylvester <[email protected]>] 7250 7251 *) Use the correct content when signing type "other". 7252 [Steve Henson] 7253 7254 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 7255 7256 *) Fix various bugs revealed by running the NISCC test suite: 7257 7258 Stop out of bounds reads in the ASN1 code when presented with 7259 invalid tags (CVE-2003-0543 and CVE-2003-0544). 7260 7261 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 7262 7263 If verify callback ignores invalid public key errors don't try to check 7264 certificate signature with the NULL public key. 7265 7266 [Steve Henson] 7267 7268 *) New -ignore_err option in ocsp application to stop the server 7269 exiting on the first error in a request. 7270 [Steve Henson] 7271 7272 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 7273 if the server requested one: as stated in TLS 1.0 and SSL 3.0 7274 specifications. 7275 [Steve Henson] 7276 7277 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 7278 extra data after the compression methods not only for TLS 1.0 7279 but also for SSL 3.0 (as required by the specification). 7280 [Bodo Moeller; problem pointed out by Matthias Loepfe] 7281 7282 *) Change X509_certificate_type() to mark the key as exported/exportable 7283 when it's 512 *bits* long, not 512 bytes. 7284 [Richard Levitte] 7285 7286 *) Change AES_cbc_encrypt() so it outputs exact multiple of 7287 blocks during encryption. 7288 [Richard Levitte] 7289 7290 *) Various fixes to base64 BIO and non blocking I/O. On write 7291 flushes were not handled properly if the BIO retried. On read 7292 data was not being buffered properly and had various logic bugs. 7293 This also affects blocking I/O when the data being decoded is a 7294 certain size. 7295 [Steve Henson] 7296 7297 *) Various S/MIME bugfixes and compatibility changes: 7298 output correct application/pkcs7 MIME type if 7299 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 7300 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 7301 of files as .eml work). Correctly handle very long lines in MIME 7302 parser. 7303 [Steve Henson] 7304 7305 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 7306 7307 *) Countermeasure against the Klima-Pokorny-Rosa extension of 7308 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 7309 a protocol version number mismatch like a decryption error 7310 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 7311 [Bodo Moeller] 7312 7313 *) Turn on RSA blinding by default in the default implementation 7314 to avoid a timing attack. Applications that don't want it can call 7315 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 7316 They would be ill-advised to do so in most cases. 7317 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 7318 7319 *) Change RSA blinding code so that it works when the PRNG is not 7320 seeded (in this case, the secret RSA exponent is abused as 7321 an unpredictable seed -- if it is not unpredictable, there 7322 is no point in blinding anyway). Make RSA blinding thread-safe 7323 by remembering the creator's thread ID in rsa->blinding and 7324 having all other threads use local one-time blinding factors 7325 (this requires more computation than sharing rsa->blinding, but 7326 avoids excessive locking; and if an RSA object is not shared 7327 between threads, blinding will still be very fast). 7328 [Bodo Moeller] 7329 7330 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 7331 ENGINE as defaults for all supported algorithms irrespective of 7332 the 'flags' parameter. 'flags' is now honoured, so applications 7333 should make sure they are passing it correctly. 7334 [Geoff Thorpe] 7335 7336 *) Target "mingw" now allows native Windows code to be generated in 7337 the Cygwin environment as well as with the MinGW compiler. 7338 [Ulf Moeller] 7339 7340 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 7341 7342 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 7343 via timing by performing a MAC computation even if incorrect 7344 block cipher padding has been found. This is a countermeasure 7345 against active attacks where the attacker has to distinguish 7346 between bad padding and a MAC verification error. (CVE-2003-0078) 7347 7348 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 7349 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 7350 Martin Vuagnoux (EPFL, Ilion)] 7351 7352 *) Make the no-err option work as intended. The intention with no-err 7353 is not to have the whole error stack handling routines removed from 7354 libcrypto, it's only intended to remove all the function name and 7355 reason texts, thereby removing some of the footprint that may not 7356 be interesting if those errors aren't displayed anyway. 7357 7358 NOTE: it's still possible for any application or module to have its 7359 own set of error texts inserted. The routines are there, just not 7360 used by default when no-err is given. 7361 [Richard Levitte] 7362 7363 *) Add support for FreeBSD on IA64. 7364 [[email protected] via Richard Levitte, resolves #454] 7365 7366 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 7367 Kerberos function mit_des_cbc_cksum(). Before this change, 7368 the value returned by DES_cbc_cksum() was like the one from 7369 mit_des_cbc_cksum(), except the bytes were swapped. 7370 [Kevin Greaney <[email protected]> and Richard Levitte] 7371 7372 *) Allow an application to disable the automatic SSL chain building. 7373 Before this a rather primitive chain build was always performed in 7374 ssl3_output_cert_chain(): an application had no way to send the 7375 correct chain if the automatic operation produced an incorrect result. 7376 7377 Now the chain builder is disabled if either: 7378 7379 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 7380 7381 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 7382 7383 The reasoning behind this is that an application would not want the 7384 auto chain building to take place if extra chain certificates are 7385 present and it might also want a means of sending no additional 7386 certificates (for example the chain has two certificates and the 7387 root is omitted). 7388 [Steve Henson] 7389 7390 *) Add the possibility to build without the ENGINE framework. 7391 [Steven Reddie <[email protected]> via Richard Levitte] 7392 7393 *) Under Win32 gmtime() can return NULL: check return value in 7394 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 7395 [Steve Henson] 7396 7397 *) DSA routines: under certain error conditions uninitialized BN objects 7398 could be freed. Solution: make sure initialization is performed early 7399 enough. (Reported and fix supplied by Ivan D Nestlerode <[email protected]>, 7400 Nils Larsch <[email protected]> via PR#459) 7401 [Lutz Jaenicke] 7402 7403 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 7404 checked on reconnect on the client side, therefore session resumption 7405 could still fail with a "ssl session id is different" error. This 7406 behaviour is masked when SSL_OP_ALL is used due to 7407 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 7408 Behaviour observed by Crispin Flowerday <[email protected]> as 7409 followup to PR #377. 7410 [Lutz Jaenicke] 7411 7412 *) IA-32 assembler support enhancements: unified ELF targets, support 7413 for SCO/Caldera platforms, fix for Cygwin shared build. 7414 [Andy Polyakov] 7415 7416 *) Add support for FreeBSD on sparc64. As a consequence, support for 7417 FreeBSD on non-x86 processors is separate from x86 processors on 7418 the config script, much like the NetBSD support. 7419 [Richard Levitte & Kris Kennaway <[email protected]>] 7420 7421 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 7422 7423 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 7424 OpenSSL 0.9.7.] 7425 7426 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 7427 code (06) was taken as the first octet of the session ID and the last 7428 octet was ignored consequently. As a result SSLv2 client side session 7429 caching could not have worked due to the session ID mismatch between 7430 client and server. 7431 Behaviour observed by Crispin Flowerday <[email protected]> as 7432 PR #377. 7433 [Lutz Jaenicke] 7434 7435 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 7436 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 7437 removed entirely. 7438 [Richard Levitte] 7439 7440 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 7441 seems that in spite of existing for more than a year, many application 7442 author have done nothing to provide the necessary callbacks, which 7443 means that this particular engine will not work properly anywhere. 7444 This is a very unfortunate situation which forces us, in the name 7445 of usability, to give the hw_ncipher.c a static lock, which is part 7446 of libcrypto. 7447 NOTE: This is for the 0.9.7 series ONLY. This hack will never 7448 appear in 0.9.8 or later. We EXPECT application authors to have 7449 dealt properly with this when 0.9.8 is released (unless we actually 7450 make such changes in the libcrypto locking code that changes will 7451 have to be made anyway). 7452 [Richard Levitte] 7453 7454 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 7455 octets have been read, EOF or an error occurs. Without this change 7456 some truncated ASN1 structures will not produce an error. 7457 [Steve Henson] 7458 7459 *) Disable Heimdal support, since it hasn't been fully implemented. 7460 Still give the possibility to force the use of Heimdal, but with 7461 warnings and a request that patches get sent to openssl-dev. 7462 [Richard Levitte] 7463 7464 *) Add the VC-CE target, introduce the WINCE sysname, and add 7465 INSTALL.WCE and appropriate conditionals to make it build. 7466 [Steven Reddie <[email protected]> via Richard Levitte] 7467 7468 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 7469 cygssl-x.y.z.dll, where x, y and z are the major, minor and 7470 edit numbers of the version. 7471 [Corinna Vinschen <[email protected]> and Richard Levitte] 7472 7473 *) Introduce safe string copy and catenation functions 7474 (BUF_strlcpy() and BUF_strlcat()). 7475 [Ben Laurie (CHATS) and Richard Levitte] 7476 7477 *) Avoid using fixed-size buffers for one-line DNs. 7478 [Ben Laurie (CHATS)] 7479 7480 *) Add BUF_MEM_grow_clean() to avoid information leakage when 7481 resizing buffers containing secrets, and use where appropriate. 7482 [Ben Laurie (CHATS)] 7483 7484 *) Avoid using fixed size buffers for configuration file location. 7485 [Ben Laurie (CHATS)] 7486 7487 *) Avoid filename truncation for various CA files. 7488 [Ben Laurie (CHATS)] 7489 7490 *) Use sizeof in preference to magic numbers. 7491 [Ben Laurie (CHATS)] 7492 7493 *) Avoid filename truncation in cert requests. 7494 [Ben Laurie (CHATS)] 7495 7496 *) Add assertions to check for (supposedly impossible) buffer 7497 overflows. 7498 [Ben Laurie (CHATS)] 7499 7500 *) Don't cache truncated DNS entries in the local cache (this could 7501 potentially lead to a spoofing attack). 7502 [Ben Laurie (CHATS)] 7503 7504 *) Fix various buffers to be large enough for hex/decimal 7505 representations in a platform independent manner. 7506 [Ben Laurie (CHATS)] 7507 7508 *) Add CRYPTO_realloc_clean() to avoid information leakage when 7509 resizing buffers containing secrets, and use where appropriate. 7510 [Ben Laurie (CHATS)] 7511 7512 *) Add BIO_indent() to avoid much slightly worrying code to do 7513 indents. 7514 [Ben Laurie (CHATS)] 7515 7516 *) Convert sprintf()/BIO_puts() to BIO_printf(). 7517 [Ben Laurie (CHATS)] 7518 7519 *) buffer_gets() could terminate with the buffer only half 7520 full. Fixed. 7521 [Ben Laurie (CHATS)] 7522 7523 *) Add assertions to prevent user-supplied crypto functions from 7524 overflowing internal buffers by having large block sizes, etc. 7525 [Ben Laurie (CHATS)] 7526 7527 *) New OPENSSL_assert() macro (similar to assert(), but enabled 7528 unconditionally). 7529 [Ben Laurie (CHATS)] 7530 7531 *) Eliminate unused copy of key in RC4. 7532 [Ben Laurie (CHATS)] 7533 7534 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 7535 [Ben Laurie (CHATS)] 7536 7537 *) Fix off-by-one error in EGD path. 7538 [Ben Laurie (CHATS)] 7539 7540 *) If RANDFILE path is too long, ignore instead of truncating. 7541 [Ben Laurie (CHATS)] 7542 7543 *) Eliminate unused and incorrectly sized X.509 structure 7544 CBCParameter. 7545 [Ben Laurie (CHATS)] 7546 7547 *) Eliminate unused and dangerous function knumber(). 7548 [Ben Laurie (CHATS)] 7549 7550 *) Eliminate unused and dangerous structure, KSSL_ERR. 7551 [Ben Laurie (CHATS)] 7552 7553 *) Protect against overlong session ID context length in an encoded 7554 session object. Since these are local, this does not appear to be 7555 exploitable. 7556 [Ben Laurie (CHATS)] 7557 7558 *) Change from security patch (see 0.9.6e below) that did not affect 7559 the 0.9.6 release series: 7560 7561 Remote buffer overflow in SSL3 protocol - an attacker could 7562 supply an oversized master key in Kerberos-enabled versions. 7563 (CVE-2002-0657) 7564 [Ben Laurie (CHATS)] 7565 7566 *) Change the SSL kerb5 codes to match RFC 2712. 7567 [Richard Levitte] 7568 7569 *) Make -nameopt work fully for req and add -reqopt switch. 7570 [Michael Bell <[email protected]>, Steve Henson] 7571 7572 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 7573 [Steve Henson, reported by Yngve Nysaeter Pettersen <[email protected]>] 7574 7575 *) Make sure tests can be performed even if the corresponding algorithms 7576 have been removed entirely. This was also the last step to make 7577 OpenSSL compilable with DJGPP under all reasonable conditions. 7578 [Richard Levitte, Doug Kaufman <[email protected]>] 7579 7580 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 7581 to allow version independent disabling of normally unselected ciphers, 7582 which may be activated as a side-effect of selecting a single cipher. 7583 7584 (E.g., cipher list string "RSA" enables ciphersuites that are left 7585 out of "ALL" because they do not provide symmetric encryption. 7586 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 7587 [Lutz Jaenicke, Bodo Moeller] 7588 7589 *) Add appropriate support for separate platform-dependent build 7590 directories. The recommended way to make a platform-dependent 7591 build directory is the following (tested on Linux), maybe with 7592 some local tweaks: 7593 7594 # Place yourself outside of the OpenSSL source tree. In 7595 # this example, the environment variable OPENSSL_SOURCE 7596 # is assumed to contain the absolute OpenSSL source directory. 7597 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 7598 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 7599 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 7600 mkdir -p `dirname $F` 7601 ln -s $OPENSSL_SOURCE/$F $F 7602 done 7603 7604 To be absolutely sure not to disturb the source tree, a "make clean" 7605 is a good thing. If it isn't successful, don't worry about it, 7606 it probably means the source directory is very clean. 7607 [Richard Levitte] 7608 7609 *) Make sure any ENGINE control commands make local copies of string 7610 pointers passed to them whenever necessary. Otherwise it is possible 7611 the caller may have overwritten (or deallocated) the original string 7612 data when a later ENGINE operation tries to use the stored values. 7613 [Götz Babin-Ebell <[email protected]>] 7614 7615 *) Improve diagnostics in file reading and command-line digests. 7616 [Ben Laurie aided and abetted by Solar Designer <[email protected]>] 7617 7618 *) Add AES modes CFB and OFB to the object database. Correct an 7619 error in AES-CFB decryption. 7620 [Richard Levitte] 7621 7622 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 7623 allows existing EVP_CIPHER_CTX structures to be reused after 7624 calling EVP_*Final(). This behaviour is used by encryption 7625 BIOs and some applications. This has the side effect that 7626 applications must explicitly clean up cipher contexts with 7627 EVP_CIPHER_CTX_cleanup() or they will leak memory. 7628 [Steve Henson] 7629 7630 *) Check the values of dna and dnb in bn_mul_recursive before calling 7631 bn_mul_comba (a non zero value means the a or b arrays do not contain 7632 n2 elements) and fallback to bn_mul_normal if either is not zero. 7633 [Steve Henson] 7634 7635 *) Fix escaping of non-ASCII characters when using the -subj option 7636 of the "openssl req" command line tool. (Robert Joop <[email protected]>) 7637 [Lutz Jaenicke] 7638 7639 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 7640 form for "surname", serialNumber has no short form. 7641 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 7642 therefore remove "mail" short name for "internet 7". 7643 The OID for unique identifiers in X509 certificates is 7644 x500UniqueIdentifier, not uniqueIdentifier. 7645 Some more OID additions. (Michael Bell <[email protected]>) 7646 [Lutz Jaenicke] 7647 7648 *) Add an "init" command to the ENGINE config module and auto initialize 7649 ENGINEs. Without any "init" command the ENGINE will be initialized 7650 after all ctrl commands have been executed on it. If init=1 the 7651 ENGINE is initialized at that point (ctrls before that point are run 7652 on the uninitialized ENGINE and after on the initialized one). If 7653 init=0 then the ENGINE will not be initialized at all. 7654 [Steve Henson] 7655 7656 *) Fix the 'app_verify_callback' interface so that the user-defined 7657 argument is actually passed to the callback: In the 7658 SSL_CTX_set_cert_verify_callback() prototype, the callback 7659 declaration has been changed from 7660 int (*cb)() 7661 into 7662 int (*cb)(X509_STORE_CTX *,void *); 7663 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 7664 i=s->ctx->app_verify_callback(&ctx) 7665 has been changed into 7666 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 7667 7668 To update applications using SSL_CTX_set_cert_verify_callback(), 7669 a dummy argument can be added to their callback functions. 7670 [D. K. Smetters <[email protected]>] 7671 7672 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 7673 [Maurice Gittens <[email protected]>, touchups by Geoff Thorpe] 7674 7675 *) Add and OPENSSL_LOAD_CONF define which will cause 7676 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 7677 This allows older applications to transparently support certain 7678 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 7679 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 7680 load the config file and OPENSSL_add_all_algorithms_conf() which will 7681 always load it have also been added. 7682 [Steve Henson] 7683 7684 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 7685 Adjust NIDs and EVP layer. 7686 [Stephen Sprunk <[email protected]> and Richard Levitte] 7687 7688 *) Config modules support in openssl utility. 7689 7690 Most commands now load modules from the config file, 7691 though in a few (such as version) this isn't done 7692 because it couldn't be used for anything. 7693 7694 In the case of ca and req the config file used is 7695 the same as the utility itself: that is the -config 7696 command line option can be used to specify an 7697 alternative file. 7698 [Steve Henson] 7699 7700 *) Move default behaviour from OPENSSL_config(). If appname is NULL 7701 use "openssl_conf" if filename is NULL use default openssl config file. 7702 [Steve Henson] 7703 7704 *) Add an argument to OPENSSL_config() to allow the use of an alternative 7705 config section name. Add a new flag to tolerate a missing config file 7706 and move code to CONF_modules_load_file(). 7707 [Steve Henson] 7708 7709 *) Support for crypto accelerator cards from Accelerated Encryption 7710 Processing, www.aep.ie. (Use engine 'aep') 7711 The support was copied from 0.9.6c [engine] and adapted/corrected 7712 to work with the new engine framework. 7713 [AEP Inc. and Richard Levitte] 7714 7715 *) Support for SureWare crypto accelerator cards from Baltimore 7716 Technologies. (Use engine 'sureware') 7717 The support was copied from 0.9.6c [engine] and adapted 7718 to work with the new engine framework. 7719 [Richard Levitte] 7720 7721 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 7722 make the newer ENGINE framework commands for the CHIL engine work. 7723 [Toomas Kiisk <[email protected]> and Richard Levitte] 7724 7725 *) Make it possible to produce shared libraries on ReliantUNIX. 7726 [Robert Dahlem <[email protected]> via Richard Levitte] 7727 7728 *) Add the configuration target debug-linux-ppro. 7729 Make 'openssl rsa' use the general key loading routines 7730 implemented in apps.c, and make those routines able to 7731 handle the key format FORMAT_NETSCAPE and the variant 7732 FORMAT_IISSGC. 7733 [Toomas Kiisk <[email protected]> via Richard Levitte] 7734 7735 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 7736 [Toomas Kiisk <[email protected]> via Richard Levitte] 7737 7738 *) Add -keyform to rsautl, and document -engine. 7739 [Richard Levitte, inspired by Toomas Kiisk <[email protected]>] 7740 7741 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 7742 BIO_R_NO_SUCH_FILE error code rather than the generic 7743 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 7744 [Ben Laurie] 7745 7746 *) Add new functions 7747 ERR_peek_last_error 7748 ERR_peek_last_error_line 7749 ERR_peek_last_error_line_data. 7750 These are similar to 7751 ERR_peek_error 7752 ERR_peek_error_line 7753 ERR_peek_error_line_data, 7754 but report on the latest error recorded rather than the first one 7755 still in the error queue. 7756 [Ben Laurie, Bodo Moeller] 7757 7758 *) default_algorithms option in ENGINE config module. This allows things 7759 like: 7760 default_algorithms = ALL 7761 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 7762 [Steve Henson] 7763 7764 *) Preliminary ENGINE config module. 7765 [Steve Henson] 7766 7767 *) New experimental application configuration code. 7768 [Steve Henson] 7769 7770 *) Change the AES code to follow the same name structure as all other 7771 symmetric ciphers, and behave the same way. Move everything to 7772 the directory crypto/aes, thereby obsoleting crypto/rijndael. 7773 [Stephen Sprunk <[email protected]> and Richard Levitte] 7774 7775 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 7776 [Ben Laurie and Theo de Raadt] 7777 7778 *) Add option to output public keys in req command. 7779 [Massimiliano Pala [email protected]] 7780 7781 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 7782 (up to about 10% better than before for P-192 and P-224). 7783 [Bodo Moeller] 7784 7785 *) New functions/macros 7786 7787 SSL_CTX_set_msg_callback(ctx, cb) 7788 SSL_CTX_set_msg_callback_arg(ctx, arg) 7789 SSL_set_msg_callback(ssl, cb) 7790 SSL_set_msg_callback_arg(ssl, arg) 7791 7792 to request calling a callback function 7793 7794 void cb(int write_p, int version, int content_type, 7795 const void *buf, size_t len, SSL *ssl, void *arg) 7796 7797 whenever a protocol message has been completely received 7798 (write_p == 0) or sent (write_p == 1). Here 'version' is the 7799 protocol version according to which the SSL library interprets 7800 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 7801 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 7802 the content type as defined in the SSL 3.0/TLS 1.0 protocol 7803 specification (change_cipher_spec(20), alert(21), handshake(22)). 7804 'buf' and 'len' point to the actual message, 'ssl' to the 7805 SSL object, and 'arg' is the application-defined value set by 7806 SSL[_CTX]_set_msg_callback_arg(). 7807 7808 'openssl s_client' and 'openssl s_server' have new '-msg' options 7809 to enable a callback that displays all protocol messages. 7810 [Bodo Moeller] 7811 7812 *) Change the shared library support so shared libraries are built as 7813 soon as the corresponding static library is finished, and thereby get 7814 openssl and the test programs linked against the shared library. 7815 This still only happens when the keyword "shard" has been given to 7816 the configuration scripts. 7817 7818 NOTE: shared library support is still an experimental thing, and 7819 backward binary compatibility is still not guaranteed. 7820 ["Maciej W. Rozycki" <[email protected]> and Richard Levitte] 7821 7822 *) Add support for Subject Information Access extension. 7823 [Peter Sylvester <[email protected]>] 7824 7825 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 7826 additional bytes when new memory had to be allocated, not just 7827 when reusing an existing buffer. 7828 [Bodo Moeller] 7829 7830 *) New command line and configuration option 'utf8' for the req command. 7831 This allows field values to be specified as UTF8 strings. 7832 [Steve Henson] 7833 7834 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 7835 runs for the former and machine-readable output for the latter. 7836 [Ben Laurie] 7837 7838 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 7839 of the e-mail address in the DN (i.e., it will go into a certificate 7840 extension only). The new configuration file option 'email_in_dn = no' 7841 has the same effect. 7842 [Massimiliano Pala [email protected]] 7843 7844 *) Change all functions with names starting with des_ to be starting 7845 with DES_ instead. Add wrappers that are compatible with libdes, 7846 but are named _ossl_old_des_*. Finally, add macros that map the 7847 des_* symbols to the corresponding _ossl_old_des_* if libdes 7848 compatibility is desired. If OpenSSL 0.9.6c compatibility is 7849 desired, the des_* symbols will be mapped to DES_*, with one 7850 exception. 7851 7852 Since we provide two compatibility mappings, the user needs to 7853 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 7854 compatibility is desired. The default (i.e., when that macro 7855 isn't defined) is OpenSSL 0.9.6c compatibility. 7856 7857 There are also macros that enable and disable the support of old 7858 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 7859 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 7860 are defined, the default will apply: to support the old des routines. 7861 7862 In either case, one must include openssl/des.h to get the correct 7863 definitions. Do not try to just include openssl/des_old.h, that 7864 won't work. 7865 7866 NOTE: This is a major break of an old API into a new one. Software 7867 authors are encouraged to switch to the DES_ style functions. Some 7868 time in the future, des_old.h and the libdes compatibility functions 7869 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 7870 default), and then completely removed. 7871 [Richard Levitte] 7872 7873 *) Test for certificates which contain unsupported critical extensions. 7874 If such a certificate is found during a verify operation it is 7875 rejected by default: this behaviour can be overridden by either 7876 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 7877 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 7878 X509_supported_extension() has also been added which returns 1 if a 7879 particular extension is supported. 7880 [Steve Henson] 7881 7882 *) Modify the behaviour of EVP cipher functions in similar way to digests 7883 to retain compatibility with existing code. 7884 [Steve Henson] 7885 7886 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 7887 compatibility with existing code. In particular the 'ctx' parameter does 7888 not have to be to be initialized before the call to EVP_DigestInit() and 7889 it is tidied up after a call to EVP_DigestFinal(). New function 7890 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 7891 EVP_MD_CTX_copy() changed to not require the destination to be 7892 initialized valid and new function EVP_MD_CTX_copy_ex() added which 7893 requires the destination to be valid. 7894 7895 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 7896 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 7897 [Steve Henson] 7898 7899 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 7900 so that complete 'Handshake' protocol structures are kept in memory 7901 instead of overwriting 'msg_type' and 'length' with 'body' data. 7902 [Bodo Moeller] 7903 7904 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 7905 [Massimo Santin via Richard Levitte] 7906 7907 *) Major restructuring to the underlying ENGINE code. This includes 7908 reduction of linker bloat, separation of pure "ENGINE" manipulation 7909 (initialisation, etc) from functionality dealing with implementations 7910 of specific crypto interfaces. This change also introduces integrated 7911 support for symmetric ciphers and digest implementations - so ENGINEs 7912 can now accelerate these by providing EVP_CIPHER and EVP_MD 7913 implementations of their own. This is detailed in crypto/engine/README 7914 as it couldn't be adequately described here. However, there are a few 7915 API changes worth noting - some RSA, DSA, DH, and RAND functions that 7916 were changed in the original introduction of ENGINE code have now 7917 reverted back - the hooking from this code to ENGINE is now a good 7918 deal more passive and at run-time, operations deal directly with 7919 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 7920 dereferencing through an ENGINE pointer any more. Also, the ENGINE 7921 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 7922 they were not being used by the framework as there is no concept of a 7923 BIGNUM_METHOD and they could not be generalised to the new 7924 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 7925 ENGINE_cpy() has been removed as it cannot be consistently defined in 7926 the new code. 7927 [Geoff Thorpe] 7928 7929 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 7930 [Steve Henson] 7931 7932 *) Change mkdef.pl to sort symbols that get the same entry number, 7933 and make sure the automatically generated functions ERR_load_* 7934 become part of libeay.num as well. 7935 [Richard Levitte] 7936 7937 *) New function SSL_renegotiate_pending(). This returns true once 7938 renegotiation has been requested (either SSL_renegotiate() call 7939 or HelloRequest/ClientHello received from the peer) and becomes 7940 false once a handshake has been completed. 7941 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 7942 sends a HelloRequest, but does not ensure that a handshake takes 7943 place. SSL_renegotiate_pending() is useful for checking if the 7944 client has followed the request.) 7945 [Bodo Moeller] 7946 7947 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 7948 By default, clients may request session resumption even during 7949 renegotiation (if session ID contexts permit); with this option, 7950 session resumption is possible only in the first handshake. 7951 7952 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 7953 more bits available for options that should not be part of 7954 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 7955 [Bodo Moeller] 7956 7957 *) Add some demos for certificate and certificate request creation. 7958 [Steve Henson] 7959 7960 *) Make maximum certificate chain size accepted from the peer application 7961 settable (SSL*_get/set_max_cert_list()), as proposed by 7962 "Douglas E. Engert" <[email protected]>. 7963 [Lutz Jaenicke] 7964 7965 *) Add support for shared libraries for Unixware-7 7966 (Boyd Lynn Gerber <[email protected]>). 7967 [Lutz Jaenicke] 7968 7969 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 7970 be done prior to destruction. Use this to unload error strings from 7971 ENGINEs that load their own error strings. NB: This adds two new API 7972 functions to "get" and "set" this destroy handler in an ENGINE. 7973 [Geoff Thorpe] 7974 7975 *) Alter all existing ENGINE implementations (except "openssl" and 7976 "openbsd") to dynamically instantiate their own error strings. This 7977 makes them more flexible to be built both as statically-linked ENGINEs 7978 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 7979 Also, add stub code to each that makes building them as self-contained 7980 shared-libraries easier (see README.ENGINE). 7981 [Geoff Thorpe] 7982 7983 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 7984 implementations into applications that are completely implemented in 7985 self-contained shared-libraries. The "dynamic" ENGINE exposes control 7986 commands that can be used to configure what shared-library to load and 7987 to control aspects of the way it is handled. Also, made an update to 7988 the README.ENGINE file that brings its information up-to-date and 7989 provides some information and instructions on the "dynamic" ENGINE 7990 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 7991 [Geoff Thorpe] 7992 7993 *) Make it possible to unload ranges of ERR strings with a new 7994 "ERR_unload_strings" function. 7995 [Geoff Thorpe] 7996 7997 *) Add a copy() function to EVP_MD. 7998 [Ben Laurie] 7999 8000 *) Make EVP_MD routines take a context pointer instead of just the 8001 md_data void pointer. 8002 [Ben Laurie] 8003 8004 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 8005 that the digest can only process a single chunk of data 8006 (typically because it is provided by a piece of 8007 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 8008 is only going to provide a single chunk of data, and hence the 8009 framework needn't accumulate the data for oneshot drivers. 8010 [Ben Laurie] 8011 8012 *) As with "ERR", make it possible to replace the underlying "ex_data" 8013 functions. This change also alters the storage and management of global 8014 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 8015 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 8016 index counters. The API functions that use this state have been changed 8017 to take a "class_index" rather than pointers to the class's local STACK 8018 and counter, and there is now an API function to dynamically create new 8019 classes. This centralisation allows us to (a) plug a lot of the 8020 thread-safety problems that existed, and (b) makes it possible to clean 8021 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 8022 such data would previously have always leaked in application code and 8023 workarounds were in place to make the memory debugging turn a blind eye 8024 to it. Application code that doesn't use this new function will still 8025 leak as before, but their memory debugging output will announce it now 8026 rather than letting it slide. 8027 8028 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 8029 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 8030 has a return value to indicate success or failure. 8031 [Geoff Thorpe] 8032 8033 *) Make it possible to replace the underlying "ERR" functions such that the 8034 global state (2 LHASH tables and 2 locks) is only used by the "default" 8035 implementation. This change also adds two functions to "get" and "set" 8036 the implementation prior to it being automatically set the first time 8037 any other ERR function takes place. Ie. an application can call "get", 8038 pass the return value to a module it has just loaded, and that module 8039 can call its own "set" function using that value. This means the 8040 module's "ERR" operations will use (and modify) the error state in the 8041 application and not in its own statically linked copy of OpenSSL code. 8042 [Geoff Thorpe] 8043 8044 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 8045 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 8046 the operation, and provides a more encapsulated way for external code 8047 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 8048 to use these functions rather than manually incrementing the counts. 8049 8050 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 8051 [Geoff Thorpe] 8052 8053 *) Add EVP test program. 8054 [Ben Laurie] 8055 8056 *) Add symmetric cipher support to ENGINE. Expect the API to change! 8057 [Ben Laurie] 8058 8059 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 8060 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 8061 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 8062 These allow a CRL to be built without having to access X509_CRL fields 8063 directly. Modify 'ca' application to use new functions. 8064 [Steve Henson] 8065 8066 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 8067 bug workarounds. Rollback attack detection is a security feature. 8068 The problem will only arise on OpenSSL servers when TLSv1 is not 8069 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 8070 Software authors not wanting to support TLSv1 will have special reasons 8071 for their choice and can explicitly enable this option. 8072 [Bodo Moeller, Lutz Jaenicke] 8073 8074 *) Rationalise EVP so it can be extended: don't include a union of 8075 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 8076 (similar to those existing for EVP_CIPHER_CTX). 8077 Usage example: 8078 8079 EVP_MD_CTX md; 8080 8081 EVP_MD_CTX_init(&md); /* new function call */ 8082 EVP_DigestInit(&md, EVP_sha1()); 8083 EVP_DigestUpdate(&md, in, len); 8084 EVP_DigestFinal(&md, out, NULL); 8085 EVP_MD_CTX_cleanup(&md); /* new function call */ 8086 8087 [Ben Laurie] 8088 8089 *) Make DES key schedule conform to the usual scheme, as well as 8090 correcting its structure. This means that calls to DES functions 8091 now have to pass a pointer to a des_key_schedule instead of a 8092 plain des_key_schedule (which was actually always a pointer 8093 anyway): E.g., 8094 8095 des_key_schedule ks; 8096 8097 des_set_key_checked(..., &ks); 8098 des_ncbc_encrypt(..., &ks, ...); 8099 8100 (Note that a later change renames 'des_...' into 'DES_...'.) 8101 [Ben Laurie] 8102 8103 *) Initial reduction of linker bloat: the use of some functions, such as 8104 PEM causes large amounts of unused functions to be linked in due to 8105 poor organisation. For example pem_all.c contains every PEM function 8106 which has a knock on effect of linking in large amounts of (unused) 8107 ASN1 code. Grouping together similar functions and splitting unrelated 8108 functions prevents this. 8109 [Steve Henson] 8110 8111 *) Cleanup of EVP macros. 8112 [Ben Laurie] 8113 8114 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 8115 correct _ecb suffix. 8116 [Ben Laurie] 8117 8118 *) Add initial OCSP responder support to ocsp application. The 8119 revocation information is handled using the text based index 8120 use by the ca application. The responder can either handle 8121 requests generated internally, supplied in files (for example 8122 via a CGI script) or using an internal minimal server. 8123 [Steve Henson] 8124 8125 *) Add configuration choices to get zlib compression for TLS. 8126 [Richard Levitte] 8127 8128 *) Changes to Kerberos SSL for RFC 2712 compliance: 8129 1. Implemented real KerberosWrapper, instead of just using 8130 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <[email protected]>] 8131 2. Implemented optional authenticator field of KerberosWrapper. 8132 8133 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 8134 and authenticator structs; see crypto/krb5/. 8135 8136 Generalized Kerberos calls to support multiple Kerberos libraries. 8137 [Vern Staats <[email protected]>, 8138 Jeffrey Altman <[email protected]> 8139 via Richard Levitte] 8140 8141 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 8142 already does with RSA. testdsa.h now has 'priv_key/pub_key' 8143 values for each of the key sizes rather than having just 8144 parameters (and 'speed' generating keys each time). 8145 [Geoff Thorpe] 8146 8147 *) Speed up EVP routines. 8148 Before: 8149encrypt 8150type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 8151des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 8152des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 8153des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 8154decrypt 8155des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 8156des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 8157des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 8158 After: 8159encrypt 8160des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 8161decrypt 8162des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 8163 [Ben Laurie] 8164 8165 *) Added the OS2-EMX target. 8166 ["Brian Havard" <[email protected]> and Richard Levitte] 8167 8168 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 8169 to support NCONF routines in extension code. New function CONF_set_nconf() 8170 to allow functions which take an NCONF to also handle the old LHASH 8171 structure: this means that the old CONF compatible routines can be 8172 retained (in particular wrt extensions) without having to duplicate the 8173 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 8174 [Steve Henson] 8175 8176 *) Enhance the general user interface with mechanisms for inner control 8177 and with possibilities to have yes/no kind of prompts. 8178 [Richard Levitte] 8179 8180 *) Change all calls to low level digest routines in the library and 8181 applications to use EVP. Add missing calls to HMAC_cleanup() and 8182 don't assume HMAC_CTX can be copied using memcpy(). 8183 [Verdon Walker <[email protected]>, Steve Henson] 8184 8185 *) Add the possibility to control engines through control names but with 8186 arbitrary arguments instead of just a string. 8187 Change the key loaders to take a UI_METHOD instead of a callback 8188 function pointer. NOTE: this breaks binary compatibility with earlier 8189 versions of OpenSSL [engine]. 8190 Adapt the nCipher code for these new conditions and add a card insertion 8191 callback. 8192 [Richard Levitte] 8193 8194 *) Enhance the general user interface with mechanisms to better support 8195 dialog box interfaces, application-defined prompts, the possibility 8196 to use defaults (for example default passwords from somewhere else) 8197 and interrupts/cancellations. 8198 [Richard Levitte] 8199 8200 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 8201 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 8202 [Steve Henson] 8203 8204 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 8205 tidy up some unnecessarily weird code in 'sk_new()'). 8206 [Geoff, reported by Diego Tartara <[email protected]>] 8207 8208 *) Change the key loading routines for ENGINEs to use the same kind 8209 callback (pem_password_cb) as all other routines that need this 8210 kind of callback. 8211 [Richard Levitte] 8212 8213 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 8214 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 8215 than this minimum value is recommended. 8216 [Lutz Jaenicke] 8217 8218 *) New random seeder for OpenVMS, using the system process statistics 8219 that are easily reachable. 8220 [Richard Levitte] 8221 8222 *) Windows apparently can't transparently handle global 8223 variables defined in DLLs. Initialisations such as: 8224 8225 const ASN1_ITEM *it = &ASN1_INTEGER_it; 8226 8227 won't compile. This is used by the any applications that need to 8228 declare their own ASN1 modules. This was fixed by adding the option 8229 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 8230 needed for static libraries under Win32. 8231 [Steve Henson] 8232 8233 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 8234 setting of purpose and trust fields. New X509_STORE trust and 8235 purpose functions and tidy up setting in other SSL functions. 8236 [Steve Henson] 8237 8238 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 8239 structure. These are inherited by X509_STORE_CTX when it is 8240 initialised. This allows various defaults to be set in the 8241 X509_STORE structure (such as flags for CRL checking and custom 8242 purpose or trust settings) for functions which only use X509_STORE_CTX 8243 internally such as S/MIME. 8244 8245 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 8246 trust settings if they are not set in X509_STORE. This allows X509_STORE 8247 purposes and trust (in S/MIME for example) to override any set by default. 8248 8249 Add command line options for CRL checking to smime, s_client and s_server 8250 applications. 8251 [Steve Henson] 8252 8253 *) Initial CRL based revocation checking. If the CRL checking flag(s) 8254 are set then the CRL is looked up in the X509_STORE structure and 8255 its validity and signature checked, then if the certificate is found 8256 in the CRL the verify fails with a revoked error. 8257 8258 Various new CRL related callbacks added to X509_STORE_CTX structure. 8259 8260 Command line options added to 'verify' application to support this. 8261 8262 This needs some additional work, such as being able to handle multiple 8263 CRLs with different times, extension based lookup (rather than just 8264 by subject name) and ultimately more complete V2 CRL extension 8265 handling. 8266 [Steve Henson] 8267 8268 *) Add a general user interface API (crypto/ui/). This is designed 8269 to replace things like des_read_password and friends (backward 8270 compatibility functions using this new API are provided). 8271 The purpose is to remove prompting functions from the DES code 8272 section as well as provide for prompting through dialog boxes in 8273 a window system and the like. 8274 [Richard Levitte] 8275 8276 *) Add "ex_data" support to ENGINE so implementations can add state at a 8277 per-structure level rather than having to store it globally. 8278 [Geoff] 8279 8280 *) Make it possible for ENGINE structures to be copied when retrieved by 8281 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 8282 This causes the "original" ENGINE structure to act like a template, 8283 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 8284 operational state can be localised to each ENGINE structure, despite the 8285 fact they all share the same "methods". New ENGINE structures returned in 8286 this case have no functional references and the return value is the single 8287 structural reference. This matches the single structural reference returned 8288 by ENGINE_by_id() normally, when it is incremented on the pre-existing 8289 ENGINE structure. 8290 [Geoff] 8291 8292 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 8293 needs to match any other type at all we need to manually clear the 8294 tag cache. 8295 [Steve Henson] 8296 8297 *) Changes to the "openssl engine" utility to include; 8298 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 8299 about an ENGINE's available control commands. 8300 - executing control commands from command line arguments using the 8301 '-pre' and '-post' switches. '-post' is only used if '-t' is 8302 specified and the ENGINE is successfully initialised. The syntax for 8303 the individual commands are colon-separated, for example; 8304 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 8305 [Geoff] 8306 8307 *) New dynamic control command support for ENGINEs. ENGINEs can now 8308 declare their own commands (numbers), names (strings), descriptions, 8309 and input types for run-time discovery by calling applications. A 8310 subset of these commands are implicitly classed as "executable" 8311 depending on their input type, and only these can be invoked through 8312 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 8313 can be based on user input, config files, etc). The distinction is 8314 that "executable" commands cannot return anything other than a boolean 8315 result and can only support numeric or string input, whereas some 8316 discoverable commands may only be for direct use through 8317 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 8318 pointers, or other custom uses. The "executable" commands are to 8319 support parameterisations of ENGINE behaviour that can be 8320 unambiguously defined by ENGINEs and used consistently across any 8321 OpenSSL-based application. Commands have been added to all the 8322 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 8323 control over shared-library paths without source code alterations. 8324 [Geoff] 8325 8326 *) Changed all ENGINE implementations to dynamically allocate their 8327 ENGINEs rather than declaring them statically. Apart from this being 8328 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 8329 this also allows the implementations to compile without using the 8330 internal engine_int.h header. 8331 [Geoff] 8332 8333 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 8334 'const' value. Any code that should be able to modify a RAND_METHOD 8335 should already have non-const pointers to it (ie. they should only 8336 modify their own ones). 8337 [Geoff] 8338 8339 *) Made a variety of little tweaks to the ENGINE code. 8340 - "atalla" and "ubsec" string definitions were moved from header files 8341 to C code. "nuron" string definitions were placed in variables 8342 rather than hard-coded - allowing parameterisation of these values 8343 later on via ctrl() commands. 8344 - Removed unused "#if 0"'d code. 8345 - Fixed engine list iteration code so it uses ENGINE_free() to release 8346 structural references. 8347 - Constified the RAND_METHOD element of ENGINE structures. 8348 - Constified various get/set functions as appropriate and added 8349 missing functions (including a catch-all ENGINE_cpy that duplicates 8350 all ENGINE values onto a new ENGINE except reference counts/state). 8351 - Removed NULL parameter checks in get/set functions. Setting a method 8352 or function to NULL is a way of cancelling out a previously set 8353 value. Passing a NULL ENGINE parameter is just plain stupid anyway 8354 and doesn't justify the extra error symbols and code. 8355 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 8356 flags from engine_int.h to engine.h. 8357 - Changed prototypes for ENGINE handler functions (init(), finish(), 8358 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 8359 [Geoff] 8360 8361 *) Implement binary inversion algorithm for BN_mod_inverse in addition 8362 to the algorithm using long division. The binary algorithm can be 8363 used only if the modulus is odd. On 32-bit systems, it is faster 8364 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 8365 roughly 5-15% for 256-bit moduli), so we use it only for moduli 8366 up to 450 bits. In 64-bit environments, the binary algorithm 8367 appears to be advantageous for much longer moduli; here we use it 8368 for moduli up to 2048 bits. 8369 [Bodo Moeller] 8370 8371 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 8372 could not support the combine flag in choice fields. 8373 [Steve Henson] 8374 8375 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 8376 extensions from a certificate request to the certificate. 8377 [Steve Henson] 8378 8379 *) Allow multiple 'certopt' and 'nameopt' options to be separated 8380 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 8381 file: this allows the display of the certificate about to be 8382 signed to be customised, to allow certain fields to be included 8383 or excluded and extension details. The old system didn't display 8384 multicharacter strings properly, omitted fields not in the policy 8385 and couldn't display additional details such as extensions. 8386 [Steve Henson] 8387 8388 *) Function EC_POINTs_mul for multiple scalar multiplication 8389 of an arbitrary number of elliptic curve points 8390 \sum scalars[i]*points[i], 8391 optionally including the generator defined for the EC_GROUP: 8392 scalar*generator + \sum scalars[i]*points[i]. 8393 8394 EC_POINT_mul is a simple wrapper function for the typical case 8395 that the point list has just one item (besides the optional 8396 generator). 8397 [Bodo Moeller] 8398 8399 *) First EC_METHODs for curves over GF(p): 8400 8401 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 8402 operations and provides various method functions that can also 8403 operate with faster implementations of modular arithmetic. 8404 8405 EC_GFp_mont_method() reuses most functions that are part of 8406 EC_GFp_simple_method, but uses Montgomery arithmetic. 8407 8408 [Bodo Moeller; point addition and point doubling 8409 implementation directly derived from source code provided by 8410 Lenka Fibikova <[email protected]>] 8411 8412 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 8413 crypto/ec/ec_lib.c): 8414 8415 Curves are EC_GROUP objects (with an optional group generator) 8416 based on EC_METHODs that are built into the library. 8417 8418 Points are EC_POINT objects based on EC_GROUP objects. 8419 8420 Most of the framework would be able to handle curves over arbitrary 8421 finite fields, but as there are no obvious types for fields other 8422 than GF(p), some functions are limited to that for now. 8423 [Bodo Moeller] 8424 8425 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 8426 that the file contains a complete HTTP response. 8427 [Richard Levitte] 8428 8429 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 8430 change the def and num file printf format specifier from "%-40sXXX" 8431 to "%-39s XXX". The latter will always guarantee a space after the 8432 field while the former will cause them to run together if the field 8433 is 40 of more characters long. 8434 [Steve Henson] 8435 8436 *) Constify the cipher and digest 'method' functions and structures 8437 and modify related functions to take constant EVP_MD and EVP_CIPHER 8438 pointers. 8439 [Steve Henson] 8440 8441 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 8442 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 8443 [Bodo Moeller] 8444 8445 *) Modify EVP_Digest*() routines so they now return values. Although the 8446 internal software routines can never fail additional hardware versions 8447 might. 8448 [Steve Henson] 8449 8450 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 8451 8452 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 8453 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 8454 8455 ASN1 error codes 8456 ERR_R_NESTED_ASN1_ERROR 8457 ... 8458 ERR_R_MISSING_ASN1_EOS 8459 were 4 .. 9, conflicting with 8460 ERR_LIB_RSA (= ERR_R_RSA_LIB) 8461 ... 8462 ERR_LIB_PEM (= ERR_R_PEM_LIB). 8463 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 8464 8465 Add new error code 'ERR_R_INTERNAL_ERROR'. 8466 [Bodo Moeller] 8467 8468 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 8469 suffices. 8470 [Bodo Moeller] 8471 8472 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 8473 sets the subject name for a new request or supersedes the 8474 subject name in a given request. Formats that can be parsed are 8475 'CN=Some Name, OU=myOU, C=IT' 8476 and 8477 'CN=Some Name/OU=myOU/C=IT'. 8478 8479 Add options '-batch' and '-verbose' to 'openssl req'. 8480 [Massimiliano Pala <[email protected]>] 8481 8482 *) Introduce the possibility to access global variables through 8483 functions on platform were that's the best way to handle exporting 8484 global variables in shared libraries. To enable this functionality, 8485 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 8486 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 8487 is normally done by Configure or something similar). 8488 8489 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 8490 in the source file (foo.c) like this: 8491 8492 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 8493 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 8494 8495 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 8496 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 8497 8498 OPENSSL_DECLARE_GLOBAL(int,foo); 8499 #define foo OPENSSL_GLOBAL_REF(foo) 8500 OPENSSL_DECLARE_GLOBAL(double,bar); 8501 #define bar OPENSSL_GLOBAL_REF(bar) 8502 8503 The #defines are very important, and therefore so is including the 8504 header file everywhere where the defined globals are used. 8505 8506 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 8507 of ASN.1 items, but that structure is a bit different. 8508 8509 The largest change is in util/mkdef.pl which has been enhanced with 8510 better and easier to understand logic to choose which symbols should 8511 go into the Windows .def files as well as a number of fixes and code 8512 cleanup (among others, algorithm keywords are now sorted 8513 lexicographically to avoid constant rewrites). 8514 [Richard Levitte] 8515 8516 *) In BN_div() keep a copy of the sign of 'num' before writing the 8517 result to 'rm' because if rm==num the value will be overwritten 8518 and produce the wrong result if 'num' is negative: this caused 8519 problems with BN_mod() and BN_nnmod(). 8520 [Steve Henson] 8521 8522 *) Function OCSP_request_verify(). This checks the signature on an 8523 OCSP request and verifies the signer certificate. The signer 8524 certificate is just checked for a generic purpose and OCSP request 8525 trust settings. 8526 [Steve Henson] 8527 8528 *) Add OCSP_check_validity() function to check the validity of OCSP 8529 responses. OCSP responses are prepared in real time and may only 8530 be a few seconds old. Simply checking that the current time lies 8531 between thisUpdate and nextUpdate max reject otherwise valid responses 8532 caused by either OCSP responder or client clock inaccuracy. Instead 8533 we allow thisUpdate and nextUpdate to fall within a certain period of 8534 the current time. The age of the response can also optionally be 8535 checked. Two new options -validity_period and -status_age added to 8536 ocsp utility. 8537 [Steve Henson] 8538 8539 *) If signature or public key algorithm is unrecognized print out its 8540 OID rather that just UNKNOWN. 8541 [Steve Henson] 8542 8543 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 8544 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 8545 ID to be generated from the issuer certificate alone which can then be 8546 passed to OCSP_id_issuer_cmp(). 8547 [Steve Henson] 8548 8549 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 8550 ASN1 modules to export functions returning ASN1_ITEM pointers 8551 instead of the ASN1_ITEM structures themselves. This adds several 8552 new macros which allow the underlying ASN1 function/structure to 8553 be accessed transparently. As a result code should not use ASN1_ITEM 8554 references directly (such as &X509_it) but instead use the relevant 8555 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 8556 use of the new ASN1 code on platforms where exporting structures 8557 is problematical (for example in shared libraries) but exporting 8558 functions returning pointers to structures is not. 8559 [Steve Henson] 8560 8561 *) Add support for overriding the generation of SSL/TLS session IDs. 8562 These callbacks can be registered either in an SSL_CTX or per SSL. 8563 The purpose of this is to allow applications to control, if they wish, 8564 the arbitrary values chosen for use as session IDs, particularly as it 8565 can be useful for session caching in multiple-server environments. A 8566 command-line switch for testing this (and any client code that wishes 8567 to use such a feature) has been added to "s_server". 8568 [Geoff Thorpe, Lutz Jaenicke] 8569 8570 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 8571 of the form '#if defined(...) || defined(...) || ...' and 8572 '#if !defined(...) && !defined(...) && ...'. This also avoids 8573 the growing number of special cases it was previously handling. 8574 [Richard Levitte] 8575 8576 *) Make all configuration macros available for application by making 8577 sure they are available in opensslconf.h, by giving them names starting 8578 with "OPENSSL_" to avoid conflicts with other packages and by making 8579 sure e_os2.h will cover all platform-specific cases together with 8580 opensslconf.h. 8581 Additionally, it is now possible to define configuration/platform- 8582 specific names (called "system identities"). In the C code, these 8583 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 8584 macro with the name beginning with "OPENSSL_SYS_", which is determined 8585 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 8586 what is available. 8587 [Richard Levitte] 8588 8589 *) New option -set_serial to 'req' and 'x509' this allows the serial 8590 number to use to be specified on the command line. Previously self 8591 signed certificates were hard coded with serial number 0 and the 8592 CA options of 'x509' had to use a serial number in a file which was 8593 auto incremented. 8594 [Steve Henson] 8595 8596 *) New options to 'ca' utility to support V2 CRL entry extensions. 8597 Currently CRL reason, invalidity date and hold instruction are 8598 supported. Add new CRL extensions to V3 code and some new objects. 8599 [Steve Henson] 8600 8601 *) New function EVP_CIPHER_CTX_set_padding() this is used to 8602 disable standard block padding (aka PKCS#5 padding) in the EVP 8603 API, which was previously mandatory. This means that the data is 8604 not padded in any way and so the total length much be a multiple 8605 of the block size, otherwise an error occurs. 8606 [Steve Henson] 8607 8608 *) Initial (incomplete) OCSP SSL support. 8609 [Steve Henson] 8610 8611 *) New function OCSP_parse_url(). This splits up a URL into its host, 8612 port and path components: primarily to parse OCSP URLs. New -url 8613 option to ocsp utility. 8614 [Steve Henson] 8615 8616 *) New nonce behavior. The return value of OCSP_check_nonce() now 8617 reflects the various checks performed. Applications can decide 8618 whether to tolerate certain situations such as an absent nonce 8619 in a response when one was present in a request: the ocsp application 8620 just prints out a warning. New function OCSP_add1_basic_nonce() 8621 this is to allow responders to include a nonce in a response even if 8622 the request is nonce-less. 8623 [Steve Henson] 8624 8625 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 8626 skipped when using openssl x509 multiple times on a single input file, 8627 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 8628 [Bodo Moeller] 8629 8630 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 8631 set string type: to handle setting ASN1_TIME structures. Fix ca 8632 utility to correctly initialize revocation date of CRLs. 8633 [Steve Henson] 8634 8635 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 8636 the clients preferred ciphersuites and rather use its own preferences. 8637 Should help to work around M$ SGC (Server Gated Cryptography) bug in 8638 Internet Explorer by ensuring unchanged hash method during stepup. 8639 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 8640 [Lutz Jaenicke] 8641 8642 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 8643 to aes and add a new 'exist' option to print out symbols that don't 8644 appear to exist. 8645 [Steve Henson] 8646 8647 *) Additional options to ocsp utility to allow flags to be set and 8648 additional certificates supplied. 8649 [Steve Henson] 8650 8651 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 8652 OCSP client a number of certificate to only verify the response 8653 signature against. 8654 [Richard Levitte] 8655 8656 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 8657 handle the new API. Currently only ECB, CBC modes supported. Add new 8658 AES OIDs. 8659 8660 Add TLS AES ciphersuites as described in RFC3268, "Advanced 8661 Encryption Standard (AES) Ciphersuites for Transport Layer 8662 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 8663 not enabled by default and were not part of the "ALL" ciphersuite 8664 alias because they were not yet official; they could be 8665 explicitly requested by specifying the "AESdraft" ciphersuite 8666 group alias. In the final release of OpenSSL 0.9.7, the group 8667 alias is called "AES" and is part of "ALL".) 8668 [Ben Laurie, Steve Henson, Bodo Moeller] 8669 8670 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 8671 request to response. 8672 [Steve Henson] 8673 8674 *) Functions for OCSP responders. OCSP_request_onereq_count(), 8675 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 8676 extract information from a certificate request. OCSP_response_create() 8677 creates a response and optionally adds a basic response structure. 8678 OCSP_basic_add1_status() adds a complete single response to a basic 8679 response and returns the OCSP_SINGLERESP structure just added (to allow 8680 extensions to be included for example). OCSP_basic_add1_cert() adds a 8681 certificate to a basic response and OCSP_basic_sign() signs a basic 8682 response with various flags. New helper functions ASN1_TIME_check() 8683 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 8684 (converts ASN1_TIME to GeneralizedTime). 8685 [Steve Henson] 8686 8687 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 8688 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 8689 structure from a certificate. X509_pubkey_digest() digests the public_key 8690 contents: this is used in various key identifiers. 8691 [Steve Henson] 8692 8693 *) Make sk_sort() tolerate a NULL argument. 8694 [Steve Henson reported by Massimiliano Pala <[email protected]>] 8695 8696 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 8697 passed by the function are trusted implicitly. If any of them signed the 8698 response then it is assumed to be valid and is not verified. 8699 [Steve Henson] 8700 8701 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 8702 to data. This was previously part of the PKCS7 ASN1 code. This 8703 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 8704 [Steve Henson, reported by Kenneth R. Robinette 8705 <[email protected]>] 8706 8707 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 8708 routines: without these tracing memory leaks is very painful. 8709 Fix leaks in PKCS12 and PKCS7 routines. 8710 [Steve Henson] 8711 8712 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 8713 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 8714 effectively meant GeneralizedTime would never be used. Now it 8715 is initialised to -1 but X509_time_adj() now has to check the value 8716 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 8717 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 8718 [Steve Henson, reported by Kenneth R. Robinette 8719 <[email protected]>] 8720 8721 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 8722 result in a zero length in the ASN1_INTEGER structure which was 8723 not consistent with the structure when d2i_ASN1_INTEGER() was used 8724 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 8725 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 8726 where it did not print out a minus for negative ASN1_INTEGER. 8727 [Steve Henson] 8728 8729 *) Add summary printout to ocsp utility. The various functions which 8730 convert status values to strings have been renamed to: 8731 OCSP_response_status_str(), OCSP_cert_status_str() and 8732 OCSP_crl_reason_str() and are no longer static. New options 8733 to verify nonce values and to disable verification. OCSP response 8734 printout format cleaned up. 8735 [Steve Henson] 8736 8737 *) Add additional OCSP certificate checks. These are those specified 8738 in RFC2560. This consists of two separate checks: the CA of the 8739 certificate being checked must either be the OCSP signer certificate 8740 or the issuer of the OCSP signer certificate. In the latter case the 8741 OCSP signer certificate must contain the OCSP signing extended key 8742 usage. This check is performed by attempting to match the OCSP 8743 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 8744 in the OCSP_CERTID structures of the response. 8745 [Steve Henson] 8746 8747 *) Initial OCSP certificate verification added to OCSP_basic_verify() 8748 and related routines. This uses the standard OpenSSL certificate 8749 verify routines to perform initial checks (just CA validity) and 8750 to obtain the certificate chain. Then additional checks will be 8751 performed on the chain. Currently the root CA is checked to see 8752 if it is explicitly trusted for OCSP signing. This is used to set 8753 a root CA as a global signing root: that is any certificate that 8754 chains to that CA is an acceptable OCSP signing certificate. 8755 [Steve Henson] 8756 8757 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 8758 extensions from a separate configuration file. 8759 As when reading extensions from the main configuration file, 8760 the '-extensions ...' option may be used for specifying the 8761 section to use. 8762 [Massimiliano Pala <[email protected]>] 8763 8764 *) New OCSP utility. Allows OCSP requests to be generated or 8765 read. The request can be sent to a responder and the output 8766 parsed, outputted or printed in text form. Not complete yet: 8767 still needs to check the OCSP response validity. 8768 [Steve Henson] 8769 8770 *) New subcommands for 'openssl ca': 8771 'openssl ca -status <serial>' prints the status of the cert with 8772 the given serial number (according to the index file). 8773 'openssl ca -updatedb' updates the expiry status of certificates 8774 in the index file. 8775 [Massimiliano Pala <[email protected]>] 8776 8777 *) New '-newreq-nodes' command option to CA.pl. This is like 8778 '-newreq', but calls 'openssl req' with the '-nodes' option 8779 so that the resulting key is not encrypted. 8780 [Damien Miller <[email protected]>] 8781 8782 *) New configuration for the GNU Hurd. 8783 [Jonathan Bartlett <[email protected]> via Richard Levitte] 8784 8785 *) Initial code to implement OCSP basic response verify. This 8786 is currently incomplete. Currently just finds the signer's 8787 certificate and verifies the signature on the response. 8788 [Steve Henson] 8789 8790 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 8791 value of OPENSSLDIR. This is available via the new '-d' option 8792 to 'openssl version', and is also included in 'openssl version -a'. 8793 [Bodo Moeller] 8794 8795 *) Allowing defining memory allocation callbacks that will be given 8796 file name and line number information in additional arguments 8797 (a const char* and an int). The basic functionality remains, as 8798 well as the original possibility to just replace malloc(), 8799 realloc() and free() by functions that do not know about these 8800 additional arguments. To register and find out the current 8801 settings for extended allocation functions, the following 8802 functions are provided: 8803 8804 CRYPTO_set_mem_ex_functions 8805 CRYPTO_set_locked_mem_ex_functions 8806 CRYPTO_get_mem_ex_functions 8807 CRYPTO_get_locked_mem_ex_functions 8808 8809 These work the same way as CRYPTO_set_mem_functions and friends. 8810 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 8811 extended allocation function is enabled. 8812 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 8813 a conventional allocation function is enabled. 8814 [Richard Levitte, Bodo Moeller] 8815 8816 *) Finish off removing the remaining LHASH function pointer casts. 8817 There should no longer be any prototype-casting required when using 8818 the LHASH abstraction, and any casts that remain are "bugs". See 8819 the callback types and macros at the head of lhash.h for details 8820 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 8821 [Geoff Thorpe] 8822 8823 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 8824 If /dev/[u]random devices are not available or do not return enough 8825 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 8826 be queried. 8827 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 8828 /etc/entropy will be queried once each in this sequence, querying stops 8829 when enough entropy was collected without querying more sockets. 8830 [Lutz Jaenicke] 8831 8832 *) Change the Unix RAND_poll() variant to be able to poll several 8833 random devices, as specified by DEVRANDOM, until a sufficient amount 8834 of data has been collected. We spend at most 10 ms on each file 8835 (select timeout) and read in non-blocking mode. DEVRANDOM now 8836 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 8837 (previously it was just the string "/dev/urandom"), so on typical 8838 platforms the 10 ms delay will never occur. 8839 Also separate out the Unix variant to its own file, rand_unix.c. 8840 For VMS, there's a currently-empty rand_vms.c. 8841 [Richard Levitte] 8842 8843 *) Move OCSP client related routines to ocsp_cl.c. These 8844 provide utility functions which an application needing 8845 to issue a request to an OCSP responder and analyse the 8846 response will typically need: as opposed to those which an 8847 OCSP responder itself would need which will be added later. 8848 8849 OCSP_request_sign() signs an OCSP request with an API similar 8850 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 8851 response. OCSP_response_get1_basic() extracts basic response 8852 from response. OCSP_resp_find_status(): finds and extracts status 8853 information from an OCSP_CERTID structure (which will be created 8854 when the request structure is built). These are built from lower 8855 level functions which work on OCSP_SINGLERESP structures but 8856 won't normally be used unless the application wishes to examine 8857 extensions in the OCSP response for example. 8858 8859 Replace nonce routines with a pair of functions. 8860 OCSP_request_add1_nonce() adds a nonce value and optionally 8861 generates a random value. OCSP_check_nonce() checks the 8862 validity of the nonce in an OCSP response. 8863 [Steve Henson] 8864 8865 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 8866 This doesn't copy the supplied OCSP_CERTID and avoids the 8867 need to free up the newly created id. Change return type 8868 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 8869 This can then be used to add extensions to the request. 8870 Deleted OCSP_request_new(), since most of its functionality 8871 is now in OCSP_REQUEST_new() (and the case insensitive name 8872 clash) apart from the ability to set the request name which 8873 will be added elsewhere. 8874 [Steve Henson] 8875 8876 *) Update OCSP API. Remove obsolete extensions argument from 8877 various functions. Extensions are now handled using the new 8878 OCSP extension code. New simple OCSP HTTP function which 8879 can be used to send requests and parse the response. 8880 [Steve Henson] 8881 8882 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 8883 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 8884 uses the special reorder version of SET OF to sort the attributes 8885 and reorder them to match the encoded order. This resolves a long 8886 standing problem: a verify on a PKCS7 structure just after signing 8887 it used to fail because the attribute order did not match the 8888 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 8889 it uses the received order. This is necessary to tolerate some broken 8890 software that does not order SET OF. This is handled by encoding 8891 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 8892 to produce the required SET OF. 8893 [Steve Henson] 8894 8895 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 8896 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 8897 files to get correct declarations of the ASN.1 item variables. 8898 [Richard Levitte] 8899 8900 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 8901 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 8902 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 8903 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 8904 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 8905 ASN1_ITEM and no wrapper functions. 8906 [Steve Henson] 8907 8908 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 8909 replace the old function pointer based I/O routines. Change most of 8910 the *_d2i_bio() and *_d2i_fp() functions to use these. 8911 [Steve Henson] 8912 8913 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 8914 lines, recognize more "algorithms" that can be deselected, and make 8915 it complain about algorithm deselection that isn't recognised. 8916 [Richard Levitte] 8917 8918 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 8919 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 8920 to use new functions. Add NO_ASN1_OLD which can be set to remove 8921 some old style ASN1 functions: this can be used to determine if old 8922 code will still work when these eventually go away. 8923 [Steve Henson] 8924 8925 *) New extension functions for OCSP structures, these follow the 8926 same conventions as certificates and CRLs. 8927 [Steve Henson] 8928 8929 *) New function X509V3_add1_i2d(). This automatically encodes and 8930 adds an extension. Its behaviour can be customised with various 8931 flags to append, replace or delete. Various wrappers added for 8932 certificates and CRLs. 8933 [Steve Henson] 8934 8935 *) Fix to avoid calling the underlying ASN1 print routine when 8936 an extension cannot be parsed. Correct a typo in the 8937 OCSP_SERVICELOC extension. Tidy up print OCSP format. 8938 [Steve Henson] 8939 8940 *) Make mkdef.pl parse some of the ASN1 macros and add appropriate 8941 entries for variables. 8942 [Steve Henson] 8943 8944 *) Add functionality to apps/openssl.c for detecting locking 8945 problems: As the program is single-threaded, all we have 8946 to do is register a locking callback using an array for 8947 storing which locks are currently held by the program. 8948 [Bodo Moeller] 8949 8950 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 8951 SSL_get_ex_data_X509_STORE_idx(), which is used in 8952 ssl_verify_cert_chain() and thus can be called at any time 8953 during TLS/SSL handshakes so that thread-safety is essential. 8954 Unfortunately, the ex_data design is not at all suited 8955 for multi-threaded use, so it probably should be abolished. 8956 [Bodo Moeller] 8957 8958 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 8959 [Broadcom, tweaked and integrated by Geoff Thorpe] 8960 8961 *) Move common extension printing code to new function 8962 X509V3_print_extensions(). Reorganise OCSP print routines and 8963 implement some needed OCSP ASN1 functions. Add OCSP extensions. 8964 [Steve Henson] 8965 8966 *) New function X509_signature_print() to remove duplication in some 8967 print routines. 8968 [Steve Henson] 8969 8970 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 8971 set (this was treated exactly the same as SET OF previously). This 8972 is used to reorder the STACK representing the structure to match the 8973 encoding. This will be used to get round a problem where a PKCS7 8974 structure which was signed could not be verified because the STACK 8975 order did not reflect the encoded order. 8976 [Steve Henson] 8977 8978 *) Reimplement the OCSP ASN1 module using the new code. 8979 [Steve Henson] 8980 8981 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 8982 for its ASN1 operations. The old style function pointers still exist 8983 for now but they will eventually go away. 8984 [Steve Henson] 8985 8986 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 8987 completely replaces the old ASN1 functionality with a table driven 8988 encoder and decoder which interprets an ASN1_ITEM structure describing 8989 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 8990 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 8991 has also been converted to the new form. 8992 [Steve Henson] 8993 8994 *) Change BN_mod_exp_recp so that negative moduli are tolerated 8995 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 8996 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 8997 for negative moduli. 8998 [Bodo Moeller] 8999 9000 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 9001 of not touching the result's sign bit. 9002 [Bodo Moeller] 9003 9004 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 9005 set. 9006 [Bodo Moeller] 9007 9008 *) Changed the LHASH code to use prototypes for callbacks, and created 9009 macros to declare and implement thin (optionally static) functions 9010 that provide type-safety and avoid function pointer casting for the 9011 type-specific callbacks. 9012 [Geoff Thorpe] 9013 9014 *) Added Kerberos Cipher Suites to be used with TLS, as written in 9015 RFC 2712. 9016 [Veers Staats <[email protected]>, 9017 Jeffrey Altman <[email protected]>, via Richard Levitte] 9018 9019 *) Reformat the FAQ so the different questions and answers can be divided 9020 in sections depending on the subject. 9021 [Richard Levitte] 9022 9023 *) Have the zlib compression code load ZLIB.DLL dynamically under 9024 Windows. 9025 [Richard Levitte] 9026 9027 *) New function BN_mod_sqrt for computing square roots modulo a prime 9028 (using the probabilistic Tonelli-Shanks algorithm unless 9029 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 9030 be handled deterministically). 9031 [Lenka Fibikova <[email protected]>, Bodo Moeller] 9032 9033 *) Make BN_mod_inverse faster by explicitly handling small quotients 9034 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 9035 512 bits], about 30% for larger ones [1024 or 2048 bits].) 9036 [Bodo Moeller] 9037 9038 *) New function BN_kronecker. 9039 [Bodo Moeller] 9040 9041 *) Fix BN_gcd so that it works on negative inputs; the result is 9042 positive unless both parameters are zero. 9043 Previously something reasonably close to an infinite loop was 9044 possible because numbers could be growing instead of shrinking 9045 in the implementation of Euclid's algorithm. 9046 [Bodo Moeller] 9047 9048 *) Fix BN_is_word() and BN_is_one() macros to take into account the 9049 sign of the number in question. 9050 9051 Fix BN_is_word(a,w) to work correctly for w == 0. 9052 9053 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 9054 because its test if the absolute value of 'a' equals 'w'. 9055 Note that BN_abs_is_word does *not* handle w == 0 reliably; 9056 it exists mostly for use in the implementations of BN_is_zero(), 9057 BN_is_one(), and BN_is_word(). 9058 [Bodo Moeller] 9059 9060 *) New function BN_swap. 9061 [Bodo Moeller] 9062 9063 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 9064 the exponentiation functions are more likely to produce reasonable 9065 results on negative inputs. 9066 [Bodo Moeller] 9067 9068 *) Change BN_mod_mul so that the result is always non-negative. 9069 Previously, it could be negative if one of the factors was negative; 9070 I don't think anyone really wanted that behaviour. 9071 [Bodo Moeller] 9072 9073 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 9074 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 9075 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 9076 and add new functions: 9077 9078 BN_nnmod 9079 BN_mod_sqr 9080 BN_mod_add 9081 BN_mod_add_quick 9082 BN_mod_sub 9083 BN_mod_sub_quick 9084 BN_mod_lshift1 9085 BN_mod_lshift1_quick 9086 BN_mod_lshift 9087 BN_mod_lshift_quick 9088 9089 These functions always generate non-negative results. 9090 9091 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 9092 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 9093 9094 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 9095 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 9096 be reduced modulo m. 9097 [Lenka Fibikova <[email protected]>, Bodo Moeller] 9098 9099#if 0 9100 The following entry accidentally appeared in the CHANGES file 9101 distributed with OpenSSL 0.9.7. The modifications described in 9102 it do *not* apply to OpenSSL 0.9.7. 9103 9104 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 9105 was actually never needed) and in BN_mul(). The removal in BN_mul() 9106 required a small change in bn_mul_part_recursive() and the addition 9107 of the functions bn_cmp_part_words(), bn_sub_part_words() and 9108 bn_add_part_words(), which do the same thing as bn_cmp_words(), 9109 bn_sub_words() and bn_add_words() except they take arrays with 9110 differing sizes. 9111 [Richard Levitte] 9112#endif 9113 9114 *) In 'openssl passwd', verify passwords read from the terminal 9115 unless the '-salt' option is used (which usually means that 9116 verification would just waste user's time since the resulting 9117 hash is going to be compared with some given password hash) 9118 or the new '-noverify' option is used. 9119 9120 This is an incompatible change, but it does not affect 9121 non-interactive use of 'openssl passwd' (passwords on the command 9122 line, '-stdin' option, '-in ...' option) and thus should not 9123 cause any problems. 9124 [Bodo Moeller] 9125 9126 *) Remove all references to RSAref, since there's no more need for it. 9127 [Richard Levitte] 9128 9129 *) Make DSO load along a path given through an environment variable 9130 (SHLIB_PATH) with shl_load(). 9131 [Richard Levitte] 9132 9133 *) Constify the ENGINE code as a result of BIGNUM constification. 9134 Also constify the RSA code and most things related to it. In a 9135 few places, most notable in the depth of the ASN.1 code, ugly 9136 casts back to non-const were required (to be solved at a later 9137 time) 9138 [Richard Levitte] 9139 9140 *) Make it so the openssl application has all engines loaded by default. 9141 [Richard Levitte] 9142 9143 *) Constify the BIGNUM routines a little more. 9144 [Richard Levitte] 9145 9146 *) Add the following functions: 9147 9148 ENGINE_load_cswift() 9149 ENGINE_load_chil() 9150 ENGINE_load_atalla() 9151 ENGINE_load_nuron() 9152 ENGINE_load_builtin_engines() 9153 9154 That way, an application can itself choose if external engines that 9155 are built-in in OpenSSL shall ever be used or not. The benefit is 9156 that applications won't have to be linked with libdl or other dso 9157 libraries unless it's really needed. 9158 9159 Changed 'openssl engine' to load all engines on demand. 9160 Changed the engine header files to avoid the duplication of some 9161 declarations (they differed!). 9162 [Richard Levitte] 9163 9164 *) 'openssl engine' can now list capabilities. 9165 [Richard Levitte] 9166 9167 *) Better error reporting in 'openssl engine'. 9168 [Richard Levitte] 9169 9170 *) Never call load_dh_param(NULL) in s_server. 9171 [Bodo Moeller] 9172 9173 *) Add engine application. It can currently list engines by name and 9174 identity, and test if they are actually available. 9175 [Richard Levitte] 9176 9177 *) Improve RPM specification file by forcing symbolic linking and making 9178 sure the installed documentation is also owned by root.root. 9179 [Damien Miller <[email protected]>] 9180 9181 *) Give the OpenSSL applications more possibilities to make use of 9182 keys (public as well as private) handled by engines. 9183 [Richard Levitte] 9184 9185 *) Add OCSP code that comes from CertCo. 9186 [Richard Levitte] 9187 9188 *) Add VMS support for the Rijndael code. 9189 [Richard Levitte] 9190 9191 *) Added untested support for Nuron crypto accelerator. 9192 [Ben Laurie] 9193 9194 *) Add support for external cryptographic devices. This code was 9195 previously distributed separately as the "engine" branch. 9196 [Geoff Thorpe, Richard Levitte] 9197 9198 *) Rework the filename-translation in the DSO code. It is now possible to 9199 have far greater control over how a "name" is turned into a filename 9200 depending on the operating environment and any oddities about the 9201 different shared library filenames on each system. 9202 [Geoff Thorpe] 9203 9204 *) Support threads on FreeBSD-elf in Configure. 9205 [Richard Levitte] 9206 9207 *) Fix for SHA1 assembly problem with MASM: it produces 9208 warnings about corrupt line number information when assembling 9209 with debugging information. This is caused by the overlapping 9210 of two sections. 9211 [Bernd Matthes <[email protected]>, Steve Henson] 9212 9213 *) NCONF changes. 9214 NCONF_get_number() has no error checking at all. As a replacement, 9215 NCONF_get_number_e() is defined (_e for "error checking") and is 9216 promoted strongly. The old NCONF_get_number is kept around for 9217 binary backward compatibility. 9218 Make it possible for methods to load from something other than a BIO, 9219 by providing a function pointer that is given a name instead of a BIO. 9220 For example, this could be used to load configuration data from an 9221 LDAP server. 9222 [Richard Levitte] 9223 9224 *) Fix for non blocking accept BIOs. Added new I/O special reason 9225 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 9226 with non blocking I/O was not possible because no retry code was 9227 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 9228 this case. 9229 [Steve Henson] 9230 9231 *) Added the beginnings of Rijndael support. 9232 [Ben Laurie] 9233 9234 *) Fix for bug in DirectoryString mask setting. Add support for 9235 X509_NAME_print_ex() in 'req' and X509_print_ex() function 9236 to allow certificate printing to more controllable, additional 9237 'certopt' option to 'x509' to allow new printing options to be 9238 set. 9239 [Steve Henson] 9240 9241 *) Clean old EAY MD5 hack from e_os.h. 9242 [Richard Levitte] 9243 9244 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 9245 9246 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 9247 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 9248 [Joe Orton, Steve Henson] 9249 9250 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 9251 9252 *) Fix additional bug revealed by the NISCC test suite: 9253 9254 Stop bug triggering large recursion when presented with 9255 certain ASN.1 tags (CVE-2003-0851) 9256 [Steve Henson] 9257 9258 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 9259 9260 *) Fix various bugs revealed by running the NISCC test suite: 9261 9262 Stop out of bounds reads in the ASN1 code when presented with 9263 invalid tags (CVE-2003-0543 and CVE-2003-0544). 9264 9265 If verify callback ignores invalid public key errors don't try to check 9266 certificate signature with the NULL public key. 9267 9268 [Steve Henson] 9269 9270 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 9271 if the server requested one: as stated in TLS 1.0 and SSL 3.0 9272 specifications. 9273 [Steve Henson] 9274 9275 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 9276 extra data after the compression methods not only for TLS 1.0 9277 but also for SSL 3.0 (as required by the specification). 9278 [Bodo Moeller; problem pointed out by Matthias Loepfe] 9279 9280 *) Change X509_certificate_type() to mark the key as exported/exportable 9281 when it's 512 *bits* long, not 512 bytes. 9282 [Richard Levitte] 9283 9284 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 9285 9286 *) Countermeasure against the Klima-Pokorny-Rosa extension of 9287 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 9288 a protocol version number mismatch like a decryption error 9289 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 9290 [Bodo Moeller] 9291 9292 *) Turn on RSA blinding by default in the default implementation 9293 to avoid a timing attack. Applications that don't want it can call 9294 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 9295 They would be ill-advised to do so in most cases. 9296 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 9297 9298 *) Change RSA blinding code so that it works when the PRNG is not 9299 seeded (in this case, the secret RSA exponent is abused as 9300 an unpredictable seed -- if it is not unpredictable, there 9301 is no point in blinding anyway). Make RSA blinding thread-safe 9302 by remembering the creator's thread ID in rsa->blinding and 9303 having all other threads use local one-time blinding factors 9304 (this requires more computation than sharing rsa->blinding, but 9305 avoids excessive locking; and if an RSA object is not shared 9306 between threads, blinding will still be very fast). 9307 [Bodo Moeller] 9308 9309 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 9310 9311 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 9312 via timing by performing a MAC computation even if incorrect 9313 block cipher padding has been found. This is a countermeasure 9314 against active attacks where the attacker has to distinguish 9315 between bad padding and a MAC verification error. (CVE-2003-0078) 9316 9317 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 9318 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 9319 Martin Vuagnoux (EPFL, Ilion)] 9320 9321 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 9322 9323 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 9324 memory from its contents. This is done with a counter that will 9325 place alternating values in each byte. This can be used to solve 9326 two issues: 1) the removal of calls to memset() by highly optimizing 9327 compilers, and 2) cleansing with other values than 0, since those can 9328 be read through on certain media, for example a swap space on disk. 9329 [Geoff Thorpe] 9330 9331 *) Bugfix: client side session caching did not work with external caching, 9332 because the session->cipher setting was not restored when reloading 9333 from the external cache. This problem was masked, when 9334 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 9335 (Found by Steve Haslam <[email protected]>.) 9336 [Lutz Jaenicke] 9337 9338 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 9339 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 9340 [Zeev Lieber <[email protected]>] 9341 9342 *) Undo an undocumented change introduced in 0.9.6e which caused 9343 repeated calls to OpenSSL_add_all_ciphers() and 9344 OpenSSL_add_all_digests() to be ignored, even after calling 9345 EVP_cleanup(). 9346 [Richard Levitte] 9347 9348 *) Change the default configuration reader to deal with last line not 9349 being properly terminated. 9350 [Richard Levitte] 9351 9352 *) Change X509_NAME_cmp() so it applies the special rules on handling 9353 DN values that are of type PrintableString, as well as RDNs of type 9354 emailAddress where the value has the type ia5String. 9355 [[email protected] via Richard Levitte] 9356 9357 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 9358 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 9359 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 9360 the bitwise-OR of the two for use by the majority of applications 9361 wanting this behaviour, and update the docs. The documented 9362 behaviour and actual behaviour were inconsistent and had been 9363 changing anyway, so this is more a bug-fix than a behavioural 9364 change. 9365 [Geoff Thorpe, diagnosed by Nadav Har'El] 9366 9367 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 9368 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 9369 [Bodo Moeller] 9370 9371 *) Fix initialization code race conditions in 9372 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 9373 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 9374 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 9375 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 9376 ssl2_get_cipher_by_char(), 9377 ssl3_get_cipher_by_char(). 9378 [Patrick McCormick <[email protected]>, Bodo Moeller] 9379 9380 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 9381 the cached sessions are flushed, as the remove_cb() might use ex_data 9382 contents. Bug found by Sam Varshavchik <[email protected]> 9383 (see [openssl.org #212]). 9384 [Geoff Thorpe, Lutz Jaenicke] 9385 9386 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 9387 length, instead of the encoding length to d2i_ASN1_OBJECT. 9388 [Steve Henson] 9389 9390 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 9391 9392 *) [In 0.9.6g-engine release:] 9393 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 9394 [Lynn Gazis <[email protected]>] 9395 9396 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 9397 9398 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 9399 and get fix the header length calculation. 9400 [Florian Weimer <[email protected]>, 9401 Alon Kantor <[email protected]> (and others), 9402 Steve Henson] 9403 9404 *) Use proper error handling instead of 'assertions' in buffer 9405 overflow checks added in 0.9.6e. This prevents DoS (the 9406 assertions could call abort()). 9407 [Arne Ansper <[email protected]>, Bodo Moeller] 9408 9409 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 9410 9411 *) Add various sanity checks to asn1_get_length() to reject 9412 the ASN1 length bytes if they exceed sizeof(long), will appear 9413 negative or the content length exceeds the length of the 9414 supplied buffer. 9415 [Steve Henson, Adi Stav <[email protected]>, James Yonan <[email protected]>] 9416 9417 *) Fix cipher selection routines: ciphers without encryption had no flags 9418 for the cipher strength set and where therefore not handled correctly 9419 by the selection routines (PR #130). 9420 [Lutz Jaenicke] 9421 9422 *) Fix EVP_dsa_sha macro. 9423 [Nils Larsch] 9424 9425 *) New option 9426 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 9427 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 9428 that was added in OpenSSL 0.9.6d. 9429 9430 As the countermeasure turned out to be incompatible with some 9431 broken SSL implementations, the new option is part of SSL_OP_ALL. 9432 SSL_OP_ALL is usually employed when compatibility with weird SSL 9433 implementations is desired (e.g. '-bugs' option to 's_client' and 9434 's_server'), so the new option is automatically set in many 9435 applications. 9436 [Bodo Moeller] 9437 9438 *) Changes in security patch: 9439 9440 Changes marked "(CHATS)" were sponsored by the Defense Advanced 9441 Research Projects Agency (DARPA) and Air Force Research Laboratory, 9442 Air Force Materiel Command, USAF, under agreement number 9443 F30602-01-2-0537. 9444 9445 *) Add various sanity checks to asn1_get_length() to reject 9446 the ASN1 length bytes if they exceed sizeof(long), will appear 9447 negative or the content length exceeds the length of the 9448 supplied buffer. (CVE-2002-0659) 9449 [Steve Henson, Adi Stav <[email protected]>, James Yonan <[email protected]>] 9450 9451 *) Assertions for various potential buffer overflows, not known to 9452 happen in practice. 9453 [Ben Laurie (CHATS)] 9454 9455 *) Various temporary buffers to hold ASCII versions of integers were 9456 too small for 64 bit platforms. (CVE-2002-0655) 9457 [Matthew Byng-Maddick <[email protected]> and Ben Laurie (CHATS)> 9458 9459 *) Remote buffer overflow in SSL3 protocol - an attacker could 9460 supply an oversized session ID to a client. (CVE-2002-0656) 9461 [Ben Laurie (CHATS)] 9462 9463 *) Remote buffer overflow in SSL2 protocol - an attacker could 9464 supply an oversized client master key. (CVE-2002-0656) 9465 [Ben Laurie (CHATS)] 9466 9467 Changes between 0.9.6c and 0.9.6d [9 May 2002] 9468 9469 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 9470 encoded as NULL) with id-dsa-with-sha1. 9471 [Nils Larsch <[email protected]>; problem pointed out by Bodo Moeller] 9472 9473 *) Check various X509_...() return values in apps/req.c. 9474 [Nils Larsch <[email protected]>] 9475 9476 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 9477 an end-of-file condition would erroneously be flagged, when the CRLF 9478 was just at the end of a processed block. The bug was discovered when 9479 processing data through a buffering memory BIO handing the data to a 9480 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 9481 <[email protected]> and Nedelcho Stanev. 9482 [Lutz Jaenicke] 9483 9484 *) Implement a countermeasure against a vulnerability recently found 9485 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 9486 before application data chunks to avoid the use of known IVs 9487 with data potentially chosen by the attacker. 9488 [Bodo Moeller] 9489 9490 *) Fix length checks in ssl3_get_client_hello(). 9491 [Bodo Moeller] 9492 9493 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 9494 to prevent ssl3_read_internal() from incorrectly assuming that 9495 ssl3_read_bytes() found application data while handshake 9496 processing was enabled when in fact s->s3->in_read_app_data was 9497 merely automatically cleared during the initial handshake. 9498 [Bodo Moeller; problem pointed out by Arne Ansper <[email protected]>] 9499 9500 *) Fix object definitions for Private and Enterprise: they were not 9501 recognized in their shortname (=lowercase) representation. Extend 9502 obj_dat.pl to issue an error when using undefined keywords instead 9503 of silently ignoring the problem (Svenning Sorensen 9504 <[email protected]>). 9505 [Lutz Jaenicke] 9506 9507 *) Fix DH_generate_parameters() so that it works for 'non-standard' 9508 generators, i.e. generators other than 2 and 5. (Previously, the 9509 code did not properly initialise the 'add' and 'rem' values to 9510 BN_generate_prime().) 9511 9512 In the new general case, we do not insist that 'generator' is 9513 actually a primitive root: This requirement is rather pointless; 9514 a generator of the order-q subgroup is just as good, if not 9515 better. 9516 [Bodo Moeller] 9517 9518 *) Map new X509 verification errors to alerts. Discovered and submitted by 9519 Tom Wu <[email protected]>. 9520 [Lutz Jaenicke] 9521 9522 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 9523 returning non-zero before the data has been completely received 9524 when using non-blocking I/O. 9525 [Bodo Moeller; problem pointed out by John Hughes] 9526 9527 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 9528 [Ben Laurie, Lutz Jaenicke] 9529 9530 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 9531 Yoram Zahavi <[email protected]>). 9532 [Lutz Jaenicke] 9533 9534 *) Add information about CygWin 1.3 and on, and preserve proper 9535 configuration for the versions before that. 9536 [Corinna Vinschen <[email protected]> and Richard Levitte] 9537 9538 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 9539 check whether we deal with a copy of a session and do not delete from 9540 the cache in this case. Problem reported by "Izhar Shoshani Levi" 9541 <[email protected]>. 9542 [Lutz Jaenicke] 9543 9544 *) Do not store session data into the internal session cache, if it 9545 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 9546 flag is set). Proposed by Aslam <[email protected]>. 9547 [Lutz Jaenicke] 9548 9549 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 9550 value is 0. 9551 [Richard Levitte] 9552 9553 *) [In 0.9.6d-engine release:] 9554 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 9555 [Toomas Kiisk <[email protected]> via Richard Levitte] 9556 9557 *) Add the configuration target linux-s390x. 9558 [Neale Ferguson <[email protected]> via Richard Levitte] 9559 9560 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 9561 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 9562 variable as an indication that a ClientHello message has been 9563 received. As the flag value will be lost between multiple 9564 invocations of ssl3_accept when using non-blocking I/O, the 9565 function may not be aware that a handshake has actually taken 9566 place, thus preventing a new session from being added to the 9567 session cache. 9568 9569 To avoid this problem, we now set s->new_session to 2 instead of 9570 using a local variable. 9571 [Lutz Jaenicke, Bodo Moeller] 9572 9573 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 9574 if the SSL_R_LENGTH_MISMATCH error is detected. 9575 [Geoff Thorpe, Bodo Moeller] 9576 9577 *) New 'shared_ldflag' column in Configure platform table. 9578 [Richard Levitte] 9579 9580 *) Fix EVP_CIPHER_mode macro. 9581 ["Dan S. Camper" <[email protected]>] 9582 9583 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 9584 type, we must throw them away by setting rr->length to 0. 9585 [D P Chang <[email protected]>] 9586 9587 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 9588 9589 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 9590 <[email protected]>. (The previous implementation 9591 worked incorrectly for those cases where range = 10..._2 and 9592 3*range is two bits longer than range.) 9593 [Bodo Moeller] 9594 9595 *) Only add signing time to PKCS7 structures if it is not already 9596 present. 9597 [Steve Henson] 9598 9599 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 9600 OBJ_ld_ce should be OBJ_id_ce. 9601 Also some ip-pda OIDs in crypto/objects/objects.txt were 9602 incorrect (cf. RFC 3039). 9603 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 9604 9605 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 9606 returns early because it has nothing to do. 9607 [Andy Schneider <[email protected]>] 9608 9609 *) [In 0.9.6c-engine release:] 9610 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 9611 [Andy Schneider <[email protected]>] 9612 9613 *) [In 0.9.6c-engine release:] 9614 Add support for Cryptographic Appliance's keyserver technology. 9615 (Use engine 'keyclient') 9616 [Cryptographic Appliances and Geoff Thorpe] 9617 9618 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 9619 is called via tools/c89.sh because arguments have to be 9620 rearranged (all '-L' options must appear before the first object 9621 modules). 9622 [Richard Shapiro <[email protected]>] 9623 9624 *) [In 0.9.6c-engine release:] 9625 Add support for Broadcom crypto accelerator cards, backported 9626 from 0.9.7. 9627 [Broadcom, Nalin Dahyabhai <[email protected]>, Mark Cox] 9628 9629 *) [In 0.9.6c-engine release:] 9630 Add support for SureWare crypto accelerator cards from 9631 Baltimore Technologies. (Use engine 'sureware') 9632 [Baltimore Technologies and Mark Cox] 9633 9634 *) [In 0.9.6c-engine release:] 9635 Add support for crypto accelerator cards from Accelerated 9636 Encryption Processing, www.aep.ie. (Use engine 'aep') 9637 [AEP Inc. and Mark Cox] 9638 9639 *) Add a configuration entry for gcc on UnixWare. 9640 [Gary Benson <[email protected]>] 9641 9642 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 9643 messages are stored in a single piece (fixed-length part and 9644 variable-length part combined) and fix various bugs found on the way. 9645 [Bodo Moeller] 9646 9647 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 9648 instead. BIO_gethostbyname() does not know what timeouts are 9649 appropriate, so entries would stay in cache even when they have 9650 become invalid. 9651 [Bodo Moeller; problem pointed out by Rich Salz <[email protected]> 9652 9653 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 9654 faced with a pathologically small ClientHello fragment that does 9655 not contain client_version: Instead of aborting with an error, 9656 simply choose the highest available protocol version (i.e., 9657 TLS 1.0 unless it is disabled). In practice, ClientHello 9658 messages are never sent like this, but this change gives us 9659 strictly correct behaviour at least for TLS. 9660 [Bodo Moeller] 9661 9662 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 9663 never resets s->method to s->ctx->method when called from within 9664 one of the SSL handshake functions. 9665 [Bodo Moeller; problem pointed out by Niko Baric] 9666 9667 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 9668 (sent using the client's version number) if client_version is 9669 smaller than the protocol version in use. Also change 9670 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 9671 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 9672 the client will at least see that alert. 9673 [Bodo Moeller] 9674 9675 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 9676 correctly. 9677 [Bodo Moeller] 9678 9679 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 9680 client receives HelloRequest while in a handshake. 9681 [Bodo Moeller; bug noticed by Andy Schneider <[email protected]>] 9682 9683 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 9684 should end in 'break', not 'goto end' which circumvents various 9685 cleanups done in state SSL_ST_OK. But session related stuff 9686 must be disabled for SSL_ST_OK in the case that we just sent a 9687 HelloRequest. 9688 9689 Also avoid some overhead by not calling ssl_init_wbio_buffer() 9690 before just sending a HelloRequest. 9691 [Bodo Moeller, Eric Rescorla <[email protected]>] 9692 9693 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 9694 reveal whether illegal block cipher padding was found or a MAC 9695 verification error occurred. (Neither SSLerr() codes nor alerts 9696 are directly visible to potential attackers, but the information 9697 may leak via logfiles.) 9698 9699 Similar changes are not required for the SSL 2.0 implementation 9700 because the number of padding bytes is sent in clear for SSL 2.0, 9701 and the extra bytes are just ignored. However ssl/s2_pkt.c 9702 failed to verify that the purported number of padding bytes is in 9703 the legal range. 9704 [Bodo Moeller] 9705 9706 *) Add OpenUNIX-8 support including shared libraries 9707 (Boyd Lynn Gerber <[email protected]>). 9708 [Lutz Jaenicke] 9709 9710 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 9711 'wristwatch attack' using huge encoding parameters (cf. 9712 James H. Manger's CRYPTO 2001 paper). Note that the 9713 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 9714 encoding parameters and hence was not vulnerable. 9715 [Bodo Moeller] 9716 9717 *) BN_sqr() bug fix. 9718 [Ulf Möller, reported by Jim Ellis <[email protected]>] 9719 9720 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 9721 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 9722 followed by modular reduction. 9723 [Bodo Moeller; pointed out by Adam Young <[email protected]>] 9724 9725 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 9726 equivalent based on BN_pseudo_rand() instead of BN_rand(). 9727 [Bodo Moeller] 9728 9729 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 9730 This function was broken, as the check for a new client hello message 9731 to handle SGC did not allow these large messages. 9732 (Tracked down by "Douglas E. Engert" <[email protected]>.) 9733 [Lutz Jaenicke] 9734 9735 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 9736 [Lutz Jaenicke] 9737 9738 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 9739 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <[email protected]>). 9740 [Lutz Jaenicke] 9741 9742 *) Rework the configuration and shared library support for Tru64 Unix. 9743 The configuration part makes use of modern compiler features and 9744 still retains old compiler behavior for those that run older versions 9745 of the OS. The shared library support part includes a variant that 9746 uses the RPATH feature, and is available through the special 9747 configuration target "alpha-cc-rpath", which will never be selected 9748 automatically. 9749 [Tim Mooney <[email protected]> via Richard Levitte] 9750 9751 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 9752 with the same message size as in ssl3_get_certificate_request(). 9753 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 9754 messages might inadvertently be reject as too long. 9755 [Petr Lampa <[email protected]>] 9756 9757 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 9758 [Andy Polyakov] 9759 9760 *) Modified SSL library such that the verify_callback that has been set 9761 specifically for an SSL object with SSL_set_verify() is actually being 9762 used. Before the change, a verify_callback set with this function was 9763 ignored and the verify_callback() set in the SSL_CTX at the time of 9764 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 9765 to allow the necessary settings. 9766 [Lutz Jaenicke] 9767 9768 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 9769 explicitly to NULL, as at least on Solaris 8 this seems not always to be 9770 done automatically (in contradiction to the requirements of the C 9771 standard). This made problems when used from OpenSSH. 9772 [Lutz Jaenicke] 9773 9774 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 9775 dh->length and always used 9776 9777 BN_rand_range(priv_key, dh->p). 9778 9779 BN_rand_range() is not necessary for Diffie-Hellman, and this 9780 specific range makes Diffie-Hellman unnecessarily inefficient if 9781 dh->length (recommended exponent length) is much smaller than the 9782 length of dh->p. We could use BN_rand_range() if the order of 9783 the subgroup was stored in the DH structure, but we only have 9784 dh->length. 9785 9786 So switch back to 9787 9788 BN_rand(priv_key, l, ...) 9789 9790 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 9791 otherwise. 9792 [Bodo Moeller] 9793 9794 *) In 9795 9796 RSA_eay_public_encrypt 9797 RSA_eay_private_decrypt 9798 RSA_eay_private_encrypt (signing) 9799 RSA_eay_public_decrypt (signature verification) 9800 9801 (default implementations for RSA_public_encrypt, 9802 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 9803 always reject numbers >= n. 9804 [Bodo Moeller] 9805 9806 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 9807 to synchronize access to 'locking_thread'. This is necessary on 9808 systems where access to 'locking_thread' (an 'unsigned long' 9809 variable) is not atomic. 9810 [Bodo Moeller] 9811 9812 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 9813 *before* setting the 'crypto_lock_rand' flag. The previous code had 9814 a race condition if 0 is a valid thread ID. 9815 [Travis Vitek <[email protected]>] 9816 9817 *) Add support for shared libraries under Irix. 9818 [Albert Chin-A-Young <[email protected]>] 9819 9820 *) Add configuration option to build on Linux on both big-endian and 9821 little-endian MIPS. 9822 [Ralf Baechle <[email protected]>] 9823 9824 *) Add the possibility to create shared libraries on HP-UX. 9825 [Richard Levitte] 9826 9827 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 9828 9829 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 9830 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 9831 Markku-Juhani O. Saarinen <[email protected]>: 9832 PRNG state recovery was possible based on the output of 9833 one PRNG request appropriately sized to gain knowledge on 9834 'md' followed by enough consecutive 1-byte PRNG requests 9835 to traverse all of 'state'. 9836 9837 1. When updating 'md_local' (the current thread's copy of 'md') 9838 during PRNG output generation, hash all of the previous 9839 'md_local' value, not just the half used for PRNG output. 9840 9841 2. Make the number of bytes from 'state' included into the hash 9842 independent from the number of PRNG bytes requested. 9843 9844 The first measure alone would be sufficient to avoid 9845 Markku-Juhani's attack. (Actually it had never occurred 9846 to me that the half of 'md_local' used for chaining was the 9847 half from which PRNG output bytes were taken -- I had always 9848 assumed that the secret half would be used.) The second 9849 measure makes sure that additional data from 'state' is never 9850 mixed into 'md_local' in small portions; this heuristically 9851 further strengthens the PRNG. 9852 [Bodo Moeller] 9853 9854 *) Fix crypto/bn/asm/mips3.s. 9855 [Andy Polyakov] 9856 9857 *) When only the key is given to "enc", the IV is undefined. Print out 9858 an error message in this case. 9859 [Lutz Jaenicke] 9860 9861 *) Handle special case when X509_NAME is empty in X509 printing routines. 9862 [Steve Henson] 9863 9864 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 9865 positive and less than q. 9866 [Bodo Moeller] 9867 9868 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 9869 used: it isn't thread safe and the add_lock_callback should handle 9870 that itself. 9871 [Paul Rose <[email protected]>] 9872 9873 *) Verify that incoming data obeys the block size in 9874 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 9875 [Bodo Moeller] 9876 9877 *) Fix OAEP check. 9878 [Ulf Möller, Bodo Möller] 9879 9880 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 9881 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 9882 when fixing the server behaviour for backwards-compatible 'client 9883 hello' messages. (Note that the attack is impractical against 9884 SSL 3.0 and TLS 1.0 anyway because length and version checking 9885 means that the probability of guessing a valid ciphertext is 9886 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 9887 paper.) 9888 9889 Before 0.9.5, the countermeasure (hide the error by generating a 9890 random 'decryption result') did not work properly because 9891 ERR_clear_error() was missing, meaning that SSL_get_error() would 9892 detect the supposedly ignored error. 9893 9894 Both problems are now fixed. 9895 [Bodo Moeller] 9896 9897 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 9898 (previously it was 1024). 9899 [Bodo Moeller] 9900 9901 *) Fix for compatibility mode trust settings: ignore trust settings 9902 unless some valid trust or reject settings are present. 9903 [Steve Henson] 9904 9905 *) Fix for blowfish EVP: its a variable length cipher. 9906 [Steve Henson] 9907 9908 *) Fix various bugs related to DSA S/MIME verification. Handle missing 9909 parameters in DSA public key structures and return an error in the 9910 DSA routines if parameters are absent. 9911 [Steve Henson] 9912 9913 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 9914 in the current directory if neither $RANDFILE nor $HOME was set. 9915 RAND_file_name() in 0.9.6a returned NULL in this case. This has 9916 caused some confusion to Windows users who haven't defined $HOME. 9917 Thus RAND_file_name() is changed again: e_os.h can define a 9918 DEFAULT_HOME, which will be used if $HOME is not set. 9919 For Windows, we use "C:"; on other platforms, we still require 9920 environment variables. 9921 9922 *) Move 'if (!initialized) RAND_poll()' into regions protected by 9923 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 9924 having multiple threads call RAND_poll() concurrently. 9925 [Bodo Moeller] 9926 9927 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 9928 combination of a flag and a thread ID variable. 9929 Otherwise while one thread is in ssleay_rand_bytes (which sets the 9930 flag), *other* threads can enter ssleay_add_bytes without obeying 9931 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 9932 that they do not hold after the first thread unsets add_do_not_lock). 9933 [Bodo Moeller] 9934 9935 *) Change bctest again: '-x' expressions are not available in all 9936 versions of 'test'. 9937 [Bodo Moeller] 9938 9939 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 9940 9941 *) Fix a couple of memory leaks in PKCS7_dataDecode() 9942 [Steve Henson, reported by Heyun Zheng <[email protected]>] 9943 9944 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 9945 the default extension for executables, if any. Also, make the perl 9946 scripts that use symlink() to test if it really exists and use "cp" 9947 if it doesn't. All this made OpenSSL compilable and installable in 9948 CygWin. 9949 [Richard Levitte] 9950 9951 *) Fix for asn1_GetSequence() for indefinite length constructed data. 9952 If SEQUENCE is length is indefinite just set c->slen to the total 9953 amount of data available. 9954 [Steve Henson, reported by [email protected]] 9955 [This change does not apply to 0.9.7.] 9956 9957 *) Change bctest to avoid here-documents inside command substitution 9958 (workaround for FreeBSD /bin/sh bug). 9959 For compatibility with Ultrix, avoid shell functions (introduced 9960 in the bctest version that searches along $PATH). 9961 [Bodo Moeller] 9962 9963 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 9964 with des_encrypt() defined on some operating systems, like Solaris 9965 and UnixWare. 9966 [Richard Levitte] 9967 9968 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 9969 On the Importance of Eliminating Errors in Cryptographic 9970 Computations, J. Cryptology 14 (2001) 2, 101-119, 9971 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 9972 [Ulf Moeller] 9973 9974 *) MIPS assembler BIGNUM division bug fix. 9975 [Andy Polyakov] 9976 9977 *) Disabled incorrect Alpha assembler code. 9978 [Richard Levitte] 9979 9980 *) Fix PKCS#7 decode routines so they correctly update the length 9981 after reading an EOC for the EXPLICIT tag. 9982 [Steve Henson] 9983 [This change does not apply to 0.9.7.] 9984 9985 *) Fix bug in PKCS#12 key generation routines. This was triggered 9986 if a 3DES key was generated with a 0 initial byte. Include 9987 PKCS12_BROKEN_KEYGEN compilation option to retain the old 9988 (but broken) behaviour. 9989 [Steve Henson] 9990 9991 *) Enhance bctest to search for a working bc along $PATH and print 9992 it when found. 9993 [Tim Rice <[email protected]> via Richard Levitte] 9994 9995 *) Fix memory leaks in err.c: free err_data string if necessary; 9996 don't write to the wrong index in ERR_set_error_data. 9997 [Bodo Moeller] 9998 9999 *) Implement ssl23_peek (analogous to ssl23_read), which previously 10000 did not exist. 10001 [Bodo Moeller] 10002 10003 *) Replace rdtsc with _emit statements for VC++ version 5. 10004 [Jeremy Cooper <[email protected]>] 10005 10006 *) Make it possible to reuse SSLv2 sessions. 10007 [Richard Levitte] 10008 10009 *) In copy_email() check for >= 0 as a return value for 10010 X509_NAME_get_index_by_NID() since 0 is a valid index. 10011 [Steve Henson reported by Massimiliano Pala <[email protected]>] 10012 10013 *) Avoid coredump with unsupported or invalid public keys by checking if 10014 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 10015 PKCS7_verify() fails with non detached data. 10016 [Steve Henson] 10017 10018 *) Don't use getenv in library functions when run as setuid/setgid. 10019 New function OPENSSL_issetugid(). 10020 [Ulf Moeller] 10021 10022 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 10023 due to incorrect handling of multi-threading: 10024 10025 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 10026 10027 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 10028 10029 3. Count how many times MemCheck_off() has been called so that 10030 nested use can be treated correctly. This also avoids 10031 inband-signalling in the previous code (which relied on the 10032 assumption that thread ID 0 is impossible). 10033 [Bodo Moeller] 10034 10035 *) Add "-rand" option also to s_client and s_server. 10036 [Lutz Jaenicke] 10037 10038 *) Fix CPU detection on Irix 6.x. 10039 [Kurt Hockenbury <[email protected]> and 10040 "Bruce W. Forsberg" <[email protected]>] 10041 10042 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 10043 was empty. 10044 [Steve Henson] 10045 [This change does not apply to 0.9.7.] 10046 10047 *) Use the cached encoding of an X509_NAME structure rather than 10048 copying it. This is apparently the reason for the libsafe "errors" 10049 but the code is actually correct. 10050 [Steve Henson] 10051 10052 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 10053 Bleichenbacher's DSA attack. 10054 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 10055 to be set and top=0 forces the highest bit to be set; top=-1 is new 10056 and leaves the highest bit random. 10057 [Ulf Moeller, Bodo Moeller] 10058 10059 *) In the NCONF_...-based implementations for CONF_... queries 10060 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 10061 a temporary CONF structure with the data component set to NULL 10062 (which gives segmentation faults in lh_retrieve). 10063 Instead, use NULL for the CONF pointer in CONF_get_string and 10064 CONF_get_number (which may use environment variables) and directly 10065 return NULL from CONF_get_section. 10066 [Bodo Moeller] 10067 10068 *) Fix potential buffer overrun for EBCDIC. 10069 [Ulf Moeller] 10070 10071 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 10072 keyUsage if basicConstraints absent for a CA. 10073 [Steve Henson] 10074 10075 *) Make SMIME_write_PKCS7() write mail header values with a format that 10076 is more generally accepted (no spaces before the semicolon), since 10077 some programs can't parse those values properly otherwise. Also make 10078 sure BIO's that break lines after each write do not create invalid 10079 headers. 10080 [Richard Levitte] 10081 10082 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 10083 macros previously used would not encode an empty SEQUENCE OF 10084 and break the signature. 10085 [Steve Henson] 10086 [This change does not apply to 0.9.7.] 10087 10088 *) Zero the premaster secret after deriving the master secret in 10089 DH ciphersuites. 10090 [Steve Henson] 10091 10092 *) Add some EVP_add_digest_alias registrations (as found in 10093 OpenSSL_add_all_digests()) to SSL_library_init() 10094 aka OpenSSL_add_ssl_algorithms(). This provides improved 10095 compatibility with peers using X.509 certificates 10096 with unconventional AlgorithmIdentifier OIDs. 10097 [Bodo Moeller] 10098 10099 *) Fix for Irix with NO_ASM. 10100 ["Bruce W. Forsberg" <[email protected]>] 10101 10102 *) ./config script fixes. 10103 [Ulf Moeller, Richard Levitte] 10104 10105 *) Fix 'openssl passwd -1'. 10106 [Bodo Moeller] 10107 10108 *) Change PKCS12_key_gen_asc() so it can cope with non null 10109 terminated strings whose length is passed in the passlen 10110 parameter, for example from PEM callbacks. This was done 10111 by adding an extra length parameter to asc2uni(). 10112 [Steve Henson, reported by <[email protected]>] 10113 10114 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 10115 call failed, free the DSA structure. 10116 [Bodo Moeller] 10117 10118 *) Fix to uni2asc() to cope with zero length Unicode strings. 10119 These are present in some PKCS#12 files. 10120 [Steve Henson] 10121 10122 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 10123 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 10124 when writing a 32767 byte record. 10125 [Bodo Moeller; problem reported by Eric Day <[email protected]>] 10126 10127 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 10128 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 10129 10130 (RSA objects have a reference count access to which is protected 10131 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 10132 so they are meant to be shared between threads.) 10133 [Bodo Moeller, Geoff Thorpe; original patch submitted by 10134 "Reddie, Steven" <[email protected]>] 10135 10136 *) Fix a deadlock in CRYPTO_mem_leaks(). 10137 [Bodo Moeller] 10138 10139 *) Use better test patterns in bntest. 10140 [Ulf Möller] 10141 10142 *) rand_win.c fix for Borland C. 10143 [Ulf Möller] 10144 10145 *) BN_rshift bugfix for n == 0. 10146 [Bodo Moeller] 10147 10148 *) Add a 'bctest' script that checks for some known 'bc' bugs 10149 so that 'make test' does not abort just because 'bc' is broken. 10150 [Bodo Moeller] 10151 10152 *) Store verify_result within SSL_SESSION also for client side to 10153 avoid potential security hole. (Re-used sessions on the client side 10154 always resulted in verify_result==X509_V_OK, not using the original 10155 result of the server certificate verification.) 10156 [Lutz Jaenicke] 10157 10158 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 10159 SSL3_RT_APPLICATION_DATA, return 0. 10160 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 10161 [Bodo Moeller] 10162 10163 *) Fix SSL_peek: 10164 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 10165 releases, have been re-implemented by renaming the previous 10166 implementations of ssl2_read and ssl3_read to ssl2_read_internal 10167 and ssl3_read_internal, respectively, and adding 'peek' parameters 10168 to them. The new ssl[23]_{read,peek} functions are calls to 10169 ssl[23]_read_internal with the 'peek' flag set appropriately. 10170 A 'peek' parameter has also been added to ssl3_read_bytes, which 10171 does the actual work for ssl3_read_internal. 10172 [Bodo Moeller] 10173 10174 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 10175 the method-specific "init()" handler. Also clean up ex_data after 10176 calling the method-specific "finish()" handler. Previously, this was 10177 happening the other way round. 10178 [Geoff Thorpe] 10179 10180 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 10181 The previous value, 12, was not always sufficient for BN_mod_exp(). 10182 [Bodo Moeller] 10183 10184 *) Make sure that shared libraries get the internal name engine with 10185 the full version number and not just 0. This should mark the 10186 shared libraries as not backward compatible. Of course, this should 10187 be changed again when we can guarantee backward binary compatibility. 10188 [Richard Levitte] 10189 10190 *) Fix typo in get_cert_by_subject() in by_dir.c 10191 [Jean-Marc Desperrier <[email protected]>] 10192 10193 *) Rework the system to generate shared libraries: 10194 10195 - Make note of the expected extension for the shared libraries and 10196 if there is a need for symbolic links from for example libcrypto.so.0 10197 to libcrypto.so.0.9.7. There is extended info in Configure for 10198 that. 10199 10200 - Make as few rebuilds of the shared libraries as possible. 10201 10202 - Still avoid linking the OpenSSL programs with the shared libraries. 10203 10204 - When installing, install the shared libraries separately from the 10205 static ones. 10206 [Richard Levitte] 10207 10208 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 10209 10210 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 10211 and not in SSL_clear because the latter is also used by the 10212 accept/connect functions; previously, the settings made by 10213 SSL_set_read_ahead would be lost during the handshake. 10214 [Bodo Moeller; problems reported by Anders Gertz <[email protected]>] 10215 10216 *) Correct util/mkdef.pl to be selective about disabled algorithms. 10217 Previously, it would create entries for disabled algorithms no 10218 matter what. 10219 [Richard Levitte] 10220 10221 *) Added several new manual pages for SSL_* function. 10222 [Lutz Jaenicke] 10223 10224 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 10225 10226 *) In ssl23_get_client_hello, generate an error message when faced 10227 with an initial SSL 3.0/TLS record that is too small to contain the 10228 first two bytes of the ClientHello message, i.e. client_version. 10229 (Note that this is a pathologic case that probably has never happened 10230 in real life.) The previous approach was to use the version number 10231 from the record header as a substitute; but our protocol choice 10232 should not depend on that one because it is not authenticated 10233 by the Finished messages. 10234 [Bodo Moeller] 10235 10236 *) More robust randomness gathering functions for Windows. 10237 [Jeffrey Altman <[email protected]>] 10238 10239 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 10240 not set then we don't setup the error code for issuer check errors 10241 to avoid possibly overwriting other errors which the callback does 10242 handle. If an application does set the flag then we assume it knows 10243 what it is doing and can handle the new informational codes 10244 appropriately. 10245 [Steve Henson] 10246 10247 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 10248 a general "ANY" type, as such it should be able to decode anything 10249 including tagged types. However it didn't check the class so it would 10250 wrongly interpret tagged types in the same way as their universal 10251 counterpart and unknown types were just rejected. Changed so that the 10252 tagged and unknown types are handled in the same way as a SEQUENCE: 10253 that is the encoding is stored intact. There is also a new type 10254 "V_ASN1_OTHER" which is used when the class is not universal, in this 10255 case we have no idea what the actual type is so we just lump them all 10256 together. 10257 [Steve Henson] 10258 10259 *) On VMS, stdout may very well lead to a file that is written to 10260 in a record-oriented fashion. That means that every write() will 10261 write a separate record, which will be read separately by the 10262 programs trying to read from it. This can be very confusing. 10263 10264 The solution is to put a BIO filter in the way that will buffer 10265 text until a linefeed is reached, and then write everything a 10266 line at a time, so every record written will be an actual line, 10267 not chunks of lines and not (usually doesn't happen, but I've 10268 seen it once) several lines in one record. BIO_f_linebuffer() is 10269 the answer. 10270 10271 Currently, it's a VMS-only method, because that's where it has 10272 been tested well enough. 10273 [Richard Levitte] 10274 10275 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 10276 it can return incorrect results. 10277 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 10278 but it was in 0.9.6-beta[12].) 10279 [Bodo Moeller] 10280 10281 *) Disable the check for content being present when verifying detached 10282 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 10283 include zero length content when signing messages. 10284 [Steve Henson] 10285 10286 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 10287 BIO_ctrl (for BIO pairs). 10288 [Bodo Möller] 10289 10290 *) Add DSO method for VMS. 10291 [Richard Levitte] 10292 10293 *) Bug fix: Montgomery multiplication could produce results with the 10294 wrong sign. 10295 [Ulf Möller] 10296 10297 *) Add RPM specification openssl.spec and modify it to build three 10298 packages. The default package contains applications, application 10299 documentation and run-time libraries. The devel package contains 10300 include files, static libraries and function documentation. The 10301 doc package contains the contents of the doc directory. The original 10302 openssl.spec was provided by Damien Miller <[email protected]>. 10303 [Richard Levitte] 10304 10305 *) Add a large number of documentation files for many SSL routines. 10306 [Lutz Jaenicke <[email protected]>] 10307 10308 *) Add a configuration entry for Sony News 4. 10309 [NAKAJI Hiroyuki <[email protected]>] 10310 10311 *) Don't set the two most significant bits to one when generating a 10312 random number < q in the DSA library. 10313 [Ulf Möller] 10314 10315 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 10316 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 10317 the underlying transport is blocking) if a handshake took place. 10318 (The default behaviour is needed by applications such as s_client 10319 and s_server that use select() to determine when to use SSL_read; 10320 but for applications that know in advance when to expect data, it 10321 just makes things more complicated.) 10322 [Bodo Moeller] 10323 10324 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 10325 from EGD. 10326 [Ben Laurie] 10327 10328 *) Add a few more EBCDIC conditionals that make `req' and `x509' 10329 work better on such systems. 10330 [Martin Kraemer <[email protected]>] 10331 10332 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 10333 Update PKCS12_parse() so it copies the friendlyName and the 10334 keyid to the certificates aux info. 10335 [Steve Henson] 10336 10337 *) Fix bug in PKCS7_verify() which caused an infinite loop 10338 if there was more than one signature. 10339 [Sven Uszpelkat <[email protected]>] 10340 10341 *) Major change in util/mkdef.pl to include extra information 10342 about each symbol, as well as presenting variables as well 10343 as functions. This change means that there's n more need 10344 to rebuild the .num files when some algorithms are excluded. 10345 [Richard Levitte] 10346 10347 *) Allow the verify time to be set by an application, 10348 rather than always using the current time. 10349 [Steve Henson] 10350 10351 *) Phase 2 verify code reorganisation. The certificate 10352 verify code now looks up an issuer certificate by a 10353 number of criteria: subject name, authority key id 10354 and key usage. It also verifies self signed certificates 10355 by the same criteria. The main comparison function is 10356 X509_check_issued() which performs these checks. 10357 10358 Lot of changes were necessary in order to support this 10359 without completely rewriting the lookup code. 10360 10361 Authority and subject key identifier are now cached. 10362 10363 The LHASH 'certs' is X509_STORE has now been replaced 10364 by a STACK_OF(X509_OBJECT). This is mainly because an 10365 LHASH can't store or retrieve multiple objects with 10366 the same hash value. 10367 10368 As a result various functions (which were all internal 10369 use only) have changed to handle the new X509_STORE 10370 structure. This will break anything that messed round 10371 with X509_STORE internally. 10372 10373 The functions X509_STORE_add_cert() now checks for an 10374 exact match, rather than just subject name. 10375 10376 The X509_STORE API doesn't directly support the retrieval 10377 of multiple certificates matching a given criteria, however 10378 this can be worked round by performing a lookup first 10379 (which will fill the cache with candidate certificates) 10380 and then examining the cache for matches. This is probably 10381 the best we can do without throwing out X509_LOOKUP 10382 entirely (maybe later...). 10383 10384 The X509_VERIFY_CTX structure has been enhanced considerably. 10385 10386 All certificate lookup operations now go via a get_issuer() 10387 callback. Although this currently uses an X509_STORE it 10388 can be replaced by custom lookups. This is a simple way 10389 to bypass the X509_STORE hackery necessary to make this 10390 work and makes it possible to use more efficient techniques 10391 in future. A very simple version which uses a simple 10392 STACK for its trusted certificate store is also provided 10393 using X509_STORE_CTX_trusted_stack(). 10394 10395 The verify_cb() and verify() callbacks now have equivalents 10396 in the X509_STORE_CTX structure. 10397 10398 X509_STORE_CTX also has a 'flags' field which can be used 10399 to customise the verify behaviour. 10400 [Steve Henson] 10401 10402 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 10403 excludes S/MIME capabilities. 10404 [Steve Henson] 10405 10406 *) When a certificate request is read in keep a copy of the 10407 original encoding of the signed data and use it when outputting 10408 again. Signatures then use the original encoding rather than 10409 a decoded, encoded version which may cause problems if the 10410 request is improperly encoded. 10411 [Steve Henson] 10412 10413 *) For consistency with other BIO_puts implementations, call 10414 buffer_write(b, ...) directly in buffer_puts instead of calling 10415 BIO_write(b, ...). 10416 10417 In BIO_puts, increment b->num_write as in BIO_write. 10418 [[email protected]] 10419 10420 *) Fix BN_mul_word for the case where the word is 0. (We have to use 10421 BN_zero, we may not return a BIGNUM with an array consisting of 10422 words set to zero.) 10423 [Bodo Moeller] 10424 10425 *) Avoid calling abort() from within the library when problems are 10426 detected, except if preprocessor symbols have been defined 10427 (such as REF_CHECK, BN_DEBUG etc.). 10428 [Bodo Moeller] 10429 10430 *) New openssl application 'rsautl'. This utility can be 10431 used for low level RSA operations. DER public key 10432 BIO/fp routines also added. 10433 [Steve Henson] 10434 10435 *) New Configure entry and patches for compiling on QNX 4. 10436 [Andreas Schneider <[email protected]>] 10437 10438 *) A demo state-machine implementation was sponsored by 10439 Nuron (http://www.nuron.com/) and is now available in 10440 demos/state_machine. 10441 [Ben Laurie] 10442 10443 *) New options added to the 'dgst' utility for signature 10444 generation and verification. 10445 [Steve Henson] 10446 10447 *) Unrecognized PKCS#7 content types are now handled via a 10448 catch all ASN1_TYPE structure. This allows unsupported 10449 types to be stored as a "blob" and an application can 10450 encode and decode it manually. 10451 [Steve Henson] 10452 10453 *) Fix various signed/unsigned issues to make a_strex.c 10454 compile under VC++. 10455 [Oscar Jacobsson <[email protected]>] 10456 10457 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 10458 length if passed a buffer. ASN1_INTEGER_to_BN failed 10459 if passed a NULL BN and its argument was negative. 10460 [Steve Henson, pointed out by Sven Heiberg <[email protected]>] 10461 10462 *) Modification to PKCS#7 encoding routines to output definite 10463 length encoding. Since currently the whole structures are in 10464 memory there's not real point in using indefinite length 10465 constructed encoding. However if OpenSSL is compiled with 10466 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 10467 [Steve Henson] 10468 10469 *) Added BIO_vprintf() and BIO_vsnprintf(). 10470 [Richard Levitte] 10471 10472 *) Added more prefixes to parse for in the strings written 10473 through a logging bio, to cover all the levels that are available 10474 through syslog. The prefixes are now: 10475 10476 PANIC, EMERG, EMR => LOG_EMERG 10477 ALERT, ALR => LOG_ALERT 10478 CRIT, CRI => LOG_CRIT 10479 ERROR, ERR => LOG_ERR 10480 WARNING, WARN, WAR => LOG_WARNING 10481 NOTICE, NOTE, NOT => LOG_NOTICE 10482 INFO, INF => LOG_INFO 10483 DEBUG, DBG => LOG_DEBUG 10484 10485 and as before, if none of those prefixes are present at the 10486 beginning of the string, LOG_ERR is chosen. 10487 10488 On Win32, the LOG_* levels are mapped according to this: 10489 10490 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 10491 LOG_WARNING => EVENTLOG_WARNING_TYPE 10492 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 10493 10494 [Richard Levitte] 10495 10496 *) Made it possible to reconfigure with just the configuration 10497 argument "reconf" or "reconfigure". The command line arguments 10498 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 10499 and are retrieved from there when reconfiguring. 10500 [Richard Levitte] 10501 10502 *) MD4 implemented. 10503 [Assar Westerlund <[email protected]>, Richard Levitte] 10504 10505 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 10506 [Richard Levitte] 10507 10508 *) The obj_dat.pl script was messing up the sorting of object 10509 names. The reason was that it compared the quoted version 10510 of strings as a result "OCSP" > "OCSP Signing" because 10511 " > SPACE. Changed script to store unquoted versions of 10512 names and add quotes on output. It was also omitting some 10513 names from the lookup table if they were given a default 10514 value (that is if SN is missing it is given the same 10515 value as LN and vice versa), these are now added on the 10516 grounds that if an object has a name we should be able to 10517 look it up. Finally added warning output when duplicate 10518 short or long names are found. 10519 [Steve Henson] 10520 10521 *) Changes needed for Tandem NSK. 10522 [Scott Uroff <[email protected]>] 10523 10524 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 10525 RSA_padding_check_SSLv23(), special padding was never detected 10526 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 10527 version rollback attacks was not effective. 10528 10529 In s23_clnt.c, don't use special rollback-attack detection padding 10530 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 10531 client; similarly, in s23_srvr.c, don't do the rollback check if 10532 SSL 2.0 is the only protocol enabled in the server. 10533 [Bodo Moeller] 10534 10535 *) Make it possible to get hexdumps of unprintable data with 'openssl 10536 asn1parse'. By implication, the functions ASN1_parse_dump() and 10537 BIO_dump_indent() are added. 10538 [Richard Levitte] 10539 10540 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 10541 these print out strings and name structures based on various 10542 flags including RFC2253 support and proper handling of 10543 multibyte characters. Added options to the 'x509' utility 10544 to allow the various flags to be set. 10545 [Steve Henson] 10546 10547 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 10548 Also change the functions X509_cmp_current_time() and 10549 X509_gmtime_adj() work with an ASN1_TIME structure, 10550 this will enable certificates using GeneralizedTime in validity 10551 dates to be checked. 10552 [Steve Henson] 10553 10554 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 10555 negative public key encodings) on by default, 10556 NO_NEG_PUBKEY_BUG can be set to disable it. 10557 [Steve Henson] 10558 10559 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 10560 content octets. An i2c_ASN1_OBJECT is unnecessary because 10561 the encoding can be trivially obtained from the structure. 10562 [Steve Henson] 10563 10564 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 10565 not read locks (CRYPTO_r_[un]lock). 10566 [Bodo Moeller] 10567 10568 *) A first attempt at creating official support for shared 10569 libraries through configuration. I've kept it so the 10570 default is static libraries only, and the OpenSSL programs 10571 are always statically linked for now, but there are 10572 preparations for dynamic linking in place. 10573 This has been tested on Linux and Tru64. 10574 [Richard Levitte] 10575 10576 *) Randomness polling function for Win9x, as described in: 10577 Peter Gutmann, Software Generation of Practically Strong 10578 Random Numbers. 10579 [Ulf Möller] 10580 10581 *) Fix so PRNG is seeded in req if using an already existing 10582 DSA key. 10583 [Steve Henson] 10584 10585 *) New options to smime application. -inform and -outform 10586 allow alternative formats for the S/MIME message including 10587 PEM and DER. The -content option allows the content to be 10588 specified separately. This should allow things like Netscape 10589 form signing output easier to verify. 10590 [Steve Henson] 10591 10592 *) Fix the ASN1 encoding of tags using the 'long form'. 10593 [Steve Henson] 10594 10595 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 10596 STRING types. These convert content octets to and from the 10597 underlying type. The actual tag and length octets are 10598 already assumed to have been read in and checked. These 10599 are needed because all other string types have virtually 10600 identical handling apart from the tag. By having versions 10601 of the ASN1 functions that just operate on content octets 10602 IMPLICIT tagging can be handled properly. It also allows 10603 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 10604 and ASN1_INTEGER are identical apart from the tag. 10605 [Steve Henson] 10606 10607 *) Change the handling of OID objects as follows: 10608 10609 - New object identifiers are inserted in objects.txt, following 10610 the syntax given in objects.README. 10611 - objects.pl is used to process obj_mac.num and create a new 10612 obj_mac.h. 10613 - obj_dat.pl is used to create a new obj_dat.h, using the data in 10614 obj_mac.h. 10615 10616 This is currently kind of a hack, and the perl code in objects.pl 10617 isn't very elegant, but it works as I intended. The simplest way 10618 to check that it worked correctly is to look in obj_dat.h and 10619 check the array nid_objs and make sure the objects haven't moved 10620 around (this is important!). Additions are OK, as well as 10621 consistent name changes. 10622 [Richard Levitte] 10623 10624 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 10625 [Bodo Moeller] 10626 10627 *) Addition of the command line parameter '-rand file' to 'openssl req'. 10628 The given file adds to whatever has already been seeded into the 10629 random pool through the RANDFILE configuration file option or 10630 environment variable, or the default random state file. 10631 [Richard Levitte] 10632 10633 *) mkstack.pl now sorts each macro group into lexical order. 10634 Previously the output order depended on the order the files 10635 appeared in the directory, resulting in needless rewriting 10636 of safestack.h . 10637 [Steve Henson] 10638 10639 *) Patches to make OpenSSL compile under Win32 again. Mostly 10640 work arounds for the VC++ problem that it treats func() as 10641 func(void). Also stripped out the parts of mkdef.pl that 10642 added extra typesafe functions: these no longer exist. 10643 [Steve Henson] 10644 10645 *) Reorganisation of the stack code. The macros are now all 10646 collected in safestack.h . Each macro is defined in terms of 10647 a "stack macro" of the form SKM_<name>(type, a, b). The 10648 DEBUG_SAFESTACK is now handled in terms of function casts, 10649 this has the advantage of retaining type safety without the 10650 use of additional functions. If DEBUG_SAFESTACK is not defined 10651 then the non typesafe macros are used instead. Also modified the 10652 mkstack.pl script to handle the new form. Needs testing to see 10653 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 10654 the default if no major problems. Similar behaviour for ASN1_SET_OF 10655 and PKCS12_STACK_OF. 10656 [Steve Henson] 10657 10658 *) When some versions of IIS use the 'NET' form of private key the 10659 key derivation algorithm is different. Normally MD5(password) is 10660 used as a 128 bit RC4 key. In the modified case 10661 MD5(MD5(password) + "SGCKEYSALT") is used instead. Added some 10662 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 10663 as the old Netscape_RSA functions except they have an additional 10664 'sgckey' parameter which uses the modified algorithm. Also added 10665 an -sgckey command line option to the rsa utility. Thanks to 10666 Adrian Peck <[email protected]> for posting details of the modified 10667 algorithm to openssl-dev. 10668 [Steve Henson] 10669 10670 *) The evp_local.h macros were using 'c.##kname' which resulted in 10671 invalid expansion on some systems (SCO 5.0.5 for example). 10672 Corrected to 'c.kname'. 10673 [Phillip Porch <[email protected]>] 10674 10675 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 10676 a STACK of email addresses from a certificate or request, these look 10677 in the subject name and the subject alternative name extensions and 10678 omit any duplicate addresses. 10679 [Steve Henson] 10680 10681 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 10682 This makes DSA verification about 2 % faster. 10683 [Bodo Moeller] 10684 10685 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 10686 (meaning that now 2^5 values will be precomputed, which is only 4 KB 10687 plus overhead for 1024 bit moduli). 10688 This makes exponentiations about 0.5 % faster for 1024 bit 10689 exponents (as measured by "openssl speed rsa2048"). 10690 [Bodo Moeller] 10691 10692 *) Rename memory handling macros to avoid conflicts with other 10693 software: 10694 Malloc => OPENSSL_malloc 10695 Malloc_locked => OPENSSL_malloc_locked 10696 Realloc => OPENSSL_realloc 10697 Free => OPENSSL_free 10698 [Richard Levitte] 10699 10700 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 10701 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 10702 [Bodo Moeller] 10703 10704 *) CygWin32 support. 10705 [John Jarvie <[email protected]>] 10706 10707 *) The type-safe stack code has been rejigged. It is now only compiled 10708 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 10709 by default all type-specific stack functions are "#define"d back to 10710 standard stack functions. This results in more streamlined output 10711 but retains the type-safety checking possibilities of the original 10712 approach. 10713 [Geoff Thorpe] 10714 10715 *) The STACK code has been cleaned up, and certain type declarations 10716 that didn't make a lot of sense have been brought in line. This has 10717 also involved a cleanup of sorts in safestack.h to more correctly 10718 map type-safe stack functions onto their plain stack counterparts. 10719 This work has also resulted in a variety of "const"ifications of 10720 lots of the code, especially "_cmp" operations which should normally 10721 be prototyped with "const" parameters anyway. 10722 [Geoff Thorpe] 10723 10724 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 10725 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 10726 (The PRNG state consists of two parts, the large pool 'state' and 'md', 10727 where all of 'md' is used each time the PRNG is used, but 'state' 10728 is used only indexed by a cyclic counter. As entropy may not be 10729 well distributed from the beginning, 'md' is important as a 10730 chaining variable. However, the output function chains only half 10731 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 10732 all of 'md', and seeding with STATE_SIZE dummy bytes will result 10733 in all of 'state' being rewritten, with the new values depending 10734 on virtually all of 'md'. This overcomes the 80 bit limitation.) 10735 [Bodo Moeller] 10736 10737 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 10738 the handshake is continued after ssl_verify_cert_chain(); 10739 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 10740 can lead to 'unexplainable' connection aborts later. 10741 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 10742 10743 *) Major EVP API cipher revision. 10744 Add hooks for extra EVP features. This allows various cipher 10745 parameters to be set in the EVP interface. Support added for variable 10746 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 10747 setting of RC2 and RC5 parameters. 10748 10749 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 10750 ciphers. 10751 10752 Remove lots of duplicated code from the EVP library. For example *every* 10753 cipher init() function handles the 'iv' in the same way according to the 10754 cipher mode. They also all do nothing if the 'key' parameter is NULL and 10755 for CFB and OFB modes they zero ctx->num. 10756 10757 New functionality allows removal of S/MIME code RC2 hack. 10758 10759 Most of the routines have the same form and so can be declared in terms 10760 of macros. 10761 10762 By shifting this to the top level EVP_CipherInit() it can be removed from 10763 all individual ciphers. If the cipher wants to handle IVs or keys 10764 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 10765 flags. 10766 10767 Change lots of functions like EVP_EncryptUpdate() to now return a 10768 value: although software versions of the algorithms cannot fail 10769 any installed hardware versions can. 10770 [Steve Henson] 10771 10772 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 10773 this option is set, tolerate broken clients that send the negotiated 10774 protocol version number instead of the requested protocol version 10775 number. 10776 [Bodo Moeller] 10777 10778 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 10779 i.e. non-zero for export ciphersuites, zero otherwise. 10780 Previous versions had this flag inverted, inconsistent with 10781 rsa_tmp_cb (..._TMP_RSA_CB). 10782 [Bodo Moeller; problem reported by Amit Chopra] 10783 10784 *) Add missing DSA library text string. Work around for some IIS 10785 key files with invalid SEQUENCE encoding. 10786 [Steve Henson] 10787 10788 *) Add a document (doc/standards.txt) that list all kinds of standards 10789 and so on that are implemented in OpenSSL. 10790 [Richard Levitte] 10791 10792 *) Enhance c_rehash script. Old version would mishandle certificates 10793 with the same subject name hash and wouldn't handle CRLs at all. 10794 Added -fingerprint option to crl utility, to support new c_rehash 10795 features. 10796 [Steve Henson] 10797 10798 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 10799 [Ulf Möller] 10800 10801 *) Fix for SSL server purpose checking. Server checking was 10802 rejecting certificates which had extended key usage present 10803 but no ssl client purpose. 10804 [Steve Henson, reported by Rene Grosser <[email protected]>] 10805 10806 *) Make PKCS#12 code work with no password. The PKCS#12 spec 10807 is a little unclear about how a blank password is handled. 10808 Since the password in encoded as a BMPString with terminating 10809 double NULL a zero length password would end up as just the 10810 double NULL. However no password at all is different and is 10811 handled differently in the PKCS#12 key generation code. NS 10812 treats a blank password as zero length. MSIE treats it as no 10813 password on export: but it will try both on import. We now do 10814 the same: PKCS12_parse() tries zero length and no password if 10815 the password is set to "" or NULL (NULL is now a valid password: 10816 it wasn't before) as does the pkcs12 application. 10817 [Steve Henson] 10818 10819 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 10820 perror when PEM_read_bio_X509_REQ fails, the error message must 10821 be obtained from the error queue. 10822 [Bodo Moeller] 10823 10824 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 10825 it in ERR_remove_state if appropriate, and change ERR_get_state 10826 accordingly to avoid race conditions (this is necessary because 10827 thread_hash is no longer constant once set). 10828 [Bodo Moeller] 10829 10830 *) Bugfix for linux-elf makefile.one. 10831 [Ulf Möller] 10832 10833 *) RSA_get_default_method() will now cause a default 10834 RSA_METHOD to be chosen if one doesn't exist already. 10835 Previously this was only set during a call to RSA_new() 10836 or RSA_new_method(NULL) meaning it was possible for 10837 RSA_get_default_method() to return NULL. 10838 [Geoff Thorpe] 10839 10840 *) Added native name translation to the existing DSO code 10841 that will convert (if the flag to do so is set) filenames 10842 that are sufficiently small and have no path information 10843 into a canonical native form. Eg. "blah" converted to 10844 "libblah.so" or "blah.dll" etc. 10845 [Geoff Thorpe] 10846 10847 *) New function ERR_error_string_n(e, buf, len) which is like 10848 ERR_error_string(e, buf), but writes at most 'len' bytes 10849 including the 0 terminator. For ERR_error_string_n, 'buf' 10850 may not be NULL. 10851 [Damien Miller <[email protected]>, Bodo Moeller] 10852 10853 *) CONF library reworked to become more general. A new CONF 10854 configuration file reader "class" is implemented as well as a 10855 new functions (NCONF_*, for "New CONF") to handle it. The now 10856 old CONF_* functions are still there, but are reimplemented to 10857 work in terms of the new functions. Also, a set of functions 10858 to handle the internal storage of the configuration data is 10859 provided to make it easier to write new configuration file 10860 reader "classes" (I can definitely see something reading a 10861 configuration file in XML format, for example), called _CONF_*, 10862 or "the configuration storage API"... 10863 10864 The new configuration file reading functions are: 10865 10866 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 10867 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 10868 10869 NCONF_default, NCONF_WIN32 10870 10871 NCONF_dump_fp, NCONF_dump_bio 10872 10873 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 10874 NCONF_new creates a new CONF object. This works in the same way 10875 as other interfaces in OpenSSL, like the BIO interface. 10876 NCONF_dump_* dump the internal storage of the configuration file, 10877 which is useful for debugging. All other functions take the same 10878 arguments as the old CONF_* functions with the exception of the 10879 first that must be a `CONF *' instead of a `LHASH *'. 10880 10881 To make it easier to use the new classes with the old CONF_* functions, 10882 the function CONF_set_default_method is provided. 10883 [Richard Levitte] 10884 10885 *) Add '-tls1' option to 'openssl ciphers', which was already 10886 mentioned in the documentation but had not been implemented. 10887 (This option is not yet really useful because even the additional 10888 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 10889 [Bodo Moeller] 10890 10891 *) Initial DSO code added into libcrypto for letting OpenSSL (and 10892 OpenSSL-based applications) load shared libraries and bind to 10893 them in a portable way. 10894 [Geoff Thorpe, with contributions from Richard Levitte] 10895 10896 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 10897 10898 *) Make sure _lrotl and _lrotr are only used with MSVC. 10899 10900 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 10901 (the default implementation of RAND_status). 10902 10903 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 10904 to '-clrext' (= clear extensions), as intended and documented. 10905 [Bodo Moeller; inconsistency pointed out by Michael Attili 10906 <[email protected]>] 10907 10908 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 10909 was larger than the MD block size. 10910 [Steve Henson, pointed out by Yost William <[email protected]>] 10911 10912 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 10913 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 10914 using the passed key: if the passed key was a private key the result 10915 of X509_print(), for example, would be to print out all the private key 10916 components. 10917 [Steve Henson] 10918 10919 *) des_quad_cksum() byte order bug fix. 10920 [Ulf Möller, using the problem description in krb4-0.9.7, where 10921 the solution is attributed to Derrick J Brashear <[email protected]>] 10922 10923 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 10924 discouraged. 10925 [Steve Henson, pointed out by Brian Korver <[email protected]>] 10926 10927 *) For easily testing in shell scripts whether some command 10928 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 10929 returns with exit code 0 iff no command of the given name is available. 10930 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 10931 the output goes to stdout and nothing is printed to stderr. 10932 Additional arguments are always ignored. 10933 10934 Since for each cipher there is a command of the same name, 10935 the 'no-cipher' compilation switches can be tested this way. 10936 10937 ('openssl no-XXX' is not able to detect pseudo-commands such 10938 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 10939 [Bodo Moeller] 10940 10941 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 10942 [Bodo Moeller] 10943 10944 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 10945 is set; it will be thrown away anyway because each handshake creates 10946 its own key. 10947 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 10948 to parameters -- in previous versions (since OpenSSL 0.9.3) the 10949 'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning 10950 you effectively got SSL_OP_SINGLE_DH_USE when using this macro. 10951 [Bodo Moeller] 10952 10953 *) New s_client option -ign_eof: EOF at stdin is ignored, and 10954 'Q' and 'R' lose their special meanings (quit/renegotiate). 10955 This is part of what -quiet does; unlike -quiet, -ign_eof 10956 does not suppress any output. 10957 [Richard Levitte] 10958 10959 *) Add compatibility options to the purpose and trust code. The 10960 purpose X509_PURPOSE_ANY is "any purpose" which automatically 10961 accepts a certificate or CA, this was the previous behaviour, 10962 with all the associated security issues. 10963 10964 X509_TRUST_COMPAT is the old trust behaviour: only and 10965 automatically trust self signed roots in certificate store. A 10966 new trust setting X509_TRUST_DEFAULT is used to specify that 10967 a purpose has no associated trust setting and it should instead 10968 use the value in the default purpose. 10969 [Steve Henson] 10970 10971 *) Fix the PKCS#8 DSA private key code so it decodes keys again 10972 and fix a memory leak. 10973 [Steve Henson] 10974 10975 *) In util/mkerr.pl (which implements 'make errors'), preserve 10976 reason strings from the previous version of the .c file, as 10977 the default to have only downcase letters (and digits) in 10978 automatically generated reasons codes is not always appropriate. 10979 [Bodo Moeller] 10980 10981 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 10982 using strerror. Previously, ERR_reason_error_string() returned 10983 library names as reason strings for SYSerr; but SYSerr is a special 10984 case where small numbers are errno values, not library numbers. 10985 [Bodo Moeller] 10986 10987 *) Add '-dsaparam' option to 'openssl dhparam' application. This 10988 converts DSA parameters into DH parameters. (When creating parameters, 10989 DSA_generate_parameters is used.) 10990 [Bodo Moeller] 10991 10992 *) Include 'length' (recommended exponent length) in C code generated 10993 by 'openssl dhparam -C'. 10994 [Bodo Moeller] 10995 10996 *) The second argument to set_label in perlasm was already being used 10997 so couldn't be used as a "file scope" flag. Moved to third argument 10998 which was free. 10999 [Steve Henson] 11000 11001 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 11002 instead of RAND_bytes for encryption IVs and salts. 11003 [Bodo Moeller] 11004 11005 *) Include RAND_status() into RAND_METHOD instead of implementing 11006 it only for md_rand.c Otherwise replacing the PRNG by calling 11007 RAND_set_rand_method would be impossible. 11008 [Bodo Moeller] 11009 11010 *) Don't let DSA_generate_key() enter an infinite loop if the random 11011 number generation fails. 11012 [Bodo Moeller] 11013 11014 *) New 'rand' application for creating pseudo-random output. 11015 [Bodo Moeller] 11016 11017 *) Added configuration support for Linux/IA64 11018 [Rolf Haberrecker <[email protected]>] 11019 11020 *) Assembler module support for Mingw32. 11021 [Ulf Möller] 11022 11023 *) Shared library support for HPUX (in shlib/). 11024 [Lutz Jaenicke <[email protected]> and Anonymous] 11025 11026 *) Shared library support for Solaris gcc. 11027 [Lutz Behnke <[email protected]>] 11028 11029 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 11030 11031 *) PKCS7_encrypt() was adding text MIME headers twice because they 11032 were added manually and by SMIME_crlf_copy(). 11033 [Steve Henson] 11034 11035 *) In bntest.c don't call BN_rand with zero bits argument. 11036 [Steve Henson, pointed out by Andrew W. Gray <[email protected]>] 11037 11038 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 11039 case was implemented. This caused BN_div_recp() to fail occasionally. 11040 [Ulf Möller] 11041 11042 *) Add an optional second argument to the set_label() in the perl 11043 assembly language builder. If this argument exists and is set 11044 to 1 it signals that the assembler should use a symbol whose 11045 scope is the entire file, not just the current function. This 11046 is needed with MASM which uses the format label:: for this scope. 11047 [Steve Henson, pointed out by Peter Runestig <[email protected]>] 11048 11049 *) Change the ASN1 types so they are typedefs by default. Before 11050 almost all types were #define'd to ASN1_STRING which was causing 11051 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 11052 for example. 11053 [Steve Henson] 11054 11055 *) Change names of new functions to the new get1/get0 naming 11056 convention: After 'get1', the caller owns a reference count 11057 and has to call ..._free; 'get0' returns a pointer to some 11058 data structure without incrementing reference counters. 11059 (Some of the existing 'get' functions increment a reference 11060 counter, some don't.) 11061 Similarly, 'set1' and 'add1' functions increase reference 11062 counters or duplicate objects. 11063 [Steve Henson] 11064 11065 *) Allow for the possibility of temp RSA key generation failure: 11066 the code used to assume it always worked and crashed on failure. 11067 [Steve Henson] 11068 11069 *) Fix potential buffer overrun problem in BIO_printf(). 11070 [Ulf Möller, using public domain code by Patrick Powell; problem 11071 pointed out by David Sacerdote <[email protected]>] 11072 11073 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 11074 RAND_egd() and RAND_status(). In the command line application, 11075 the EGD socket can be specified like a seed file using RANDFILE 11076 or -rand. 11077 [Ulf Möller] 11078 11079 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 11080 Some CAs (e.g. Verisign) distribute certificates in this form. 11081 [Steve Henson] 11082 11083 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 11084 list to exclude them. This means that no special compilation option 11085 is needed to use anonymous DH: it just needs to be included in the 11086 cipher list. 11087 [Steve Henson] 11088 11089 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 11090 EVP_MD_type. The old functionality is available in a new macro called 11091 EVP_MD_md(). Change code that uses it and update docs. 11092 [Steve Henson] 11093 11094 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 11095 where the 'void *' argument is replaced by a function pointer argument. 11096 Previously 'void *' was abused to point to functions, which works on 11097 many platforms, but is not correct. As these functions are usually 11098 called by macros defined in OpenSSL header files, most source code 11099 should work without changes. 11100 [Richard Levitte] 11101 11102 *) <openssl/opensslconf.h> (which is created by Configure) now contains 11103 sections with information on -D... compiler switches used for 11104 compiling the library so that applications can see them. To enable 11105 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 11106 must be defined. E.g., 11107 #define OPENSSL_ALGORITHM_DEFINES 11108 #include <openssl/opensslconf.h> 11109 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 11110 [Richard Levitte, Ulf and Bodo Möller] 11111 11112 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 11113 record layer. 11114 [Bodo Moeller] 11115 11116 *) Change the 'other' type in certificate aux info to a STACK_OF 11117 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 11118 the required ASN1 format: arbitrary types determined by an OID. 11119 [Steve Henson] 11120 11121 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 11122 argument to 'req'. This is not because the function is newer or 11123 better than others it just uses the work 'NEW' in the certificate 11124 request header lines. Some software needs this. 11125 [Steve Henson] 11126 11127 *) Reorganise password command line arguments: now passwords can be 11128 obtained from various sources. Delete the PEM_cb function and make 11129 it the default behaviour: i.e. if the callback is NULL and the 11130 usrdata argument is not NULL interpret it as a null terminated pass 11131 phrase. If usrdata and the callback are NULL then the pass phrase 11132 is prompted for as usual. 11133 [Steve Henson] 11134 11135 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 11136 the support is automatically enabled. The resulting binaries will 11137 autodetect the card and use it if present. 11138 [Ben Laurie and Compaq Inc.] 11139 11140 *) Work around for Netscape hang bug. This sends certificate request 11141 and server done in one record. Since this is perfectly legal in the 11142 SSL/TLS protocol it isn't a "bug" option and is on by default. See 11143 the bugs/SSLv3 entry for more info. 11144 [Steve Henson] 11145 11146 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 11147 [Andy Polyakov] 11148 11149 *) Add -rand argument to smime and pkcs12 applications and read/write 11150 of seed file. 11151 [Steve Henson] 11152 11153 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 11154 [Bodo Moeller] 11155 11156 *) Add command line password options to the remaining applications. 11157 [Steve Henson] 11158 11159 *) Bug fix for BN_div_recp() for numerators with an even number of 11160 bits. 11161 [Ulf Möller] 11162 11163 *) More tests in bntest.c, and changed test_bn output. 11164 [Ulf Möller] 11165 11166 *) ./config recognizes MacOS X now. 11167 [Andy Polyakov] 11168 11169 *) Bug fix for BN_div() when the first words of num and divisor are 11170 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 11171 [Ulf Möller] 11172 11173 *) Add support for various broken PKCS#8 formats, and command line 11174 options to produce them. 11175 [Steve Henson] 11176 11177 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 11178 get temporary BIGNUMs from a BN_CTX. 11179 [Ulf Möller] 11180 11181 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 11182 for p == 0. 11183 [Ulf Möller] 11184 11185 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 11186 include a #define from the old name to the new. The original intent 11187 was that statically linked binaries could for example just call 11188 SSLeay_add_all_ciphers() to just add ciphers to the table and not 11189 link with digests. This never worked because SSLeay_add_all_digests() 11190 and SSLeay_add_all_ciphers() were in the same source file so calling 11191 one would link with the other. They are now in separate source files. 11192 [Steve Henson] 11193 11194 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 11195 [Steve Henson] 11196 11197 *) Use a less unusual form of the Miller-Rabin primality test (it used 11198 a binary algorithm for exponentiation integrated into the Miller-Rabin 11199 loop, our standard modexp algorithms are faster). 11200 [Bodo Moeller] 11201 11202 *) Support for the EBCDIC character set completed. 11203 [Martin Kraemer <[email protected]>] 11204 11205 *) Source code cleanups: use const where appropriate, eliminate casts, 11206 use void * instead of char * in lhash. 11207 [Ulf Möller] 11208 11209 *) Bugfix: ssl3_send_server_key_exchange was not restartable 11210 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 11211 this the server could overwrite ephemeral keys that the client 11212 has already seen). 11213 [Bodo Moeller] 11214 11215 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 11216 using 50 iterations of the Rabin-Miller test. 11217 11218 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 11219 iterations of the Rabin-Miller test as required by the appendix 11220 to FIPS PUB 186[-1]) instead of DSA_is_prime. 11221 As BN_is_prime_fasttest includes trial division, DSA parameter 11222 generation becomes much faster. 11223 11224 This implies a change for the callback functions in DSA_is_prime 11225 and DSA_generate_parameters: The callback function is called once 11226 for each positive witness in the Rabin-Miller test, not just 11227 occasionally in the inner loop; and the parameters to the 11228 callback function now provide an iteration count for the outer 11229 loop rather than for the current invocation of the inner loop. 11230 DSA_generate_parameters additionally can call the callback 11231 function with an 'iteration count' of -1, meaning that a 11232 candidate has passed the trial division test (when q is generated 11233 from an application-provided seed, trial division is skipped). 11234 [Bodo Moeller] 11235 11236 *) New function BN_is_prime_fasttest that optionally does trial 11237 division before starting the Rabin-Miller test and has 11238 an additional BN_CTX * argument (whereas BN_is_prime always 11239 has to allocate at least one BN_CTX). 11240 'callback(1, -1, cb_arg)' is called when a number has passed the 11241 trial division stage. 11242 [Bodo Moeller] 11243 11244 *) Fix for bug in CRL encoding. The validity dates weren't being handled 11245 as ASN1_TIME. 11246 [Steve Henson] 11247 11248 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 11249 [Steve Henson] 11250 11251 *) New function BN_pseudo_rand(). 11252 [Ulf Möller] 11253 11254 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 11255 bignum version of BN_from_montgomery() with the working code from 11256 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 11257 the comments. 11258 [Ulf Möller] 11259 11260 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 11261 made it impossible to use the same SSL_SESSION data structure in 11262 SSL2 clients in multiple threads. 11263 [Bodo Moeller] 11264 11265 *) The return value of RAND_load_file() no longer counts bytes obtained 11266 by stat(). RAND_load_file(..., -1) is new and uses the complete file 11267 to seed the PRNG (previously an explicit byte count was required). 11268 [Ulf Möller, Bodo Möller] 11269 11270 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 11271 used (char *) instead of (void *) and had casts all over the place. 11272 [Steve Henson] 11273 11274 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 11275 [Ulf Möller] 11276 11277 *) Retain source code compatibility for BN_prime_checks macro: 11278 BN_is_prime(..., BN_prime_checks, ...) now uses 11279 BN_prime_checks_for_size to determine the appropriate number of 11280 Rabin-Miller iterations. 11281 [Ulf Möller] 11282 11283 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 11284 DH_CHECK_P_NOT_SAFE_PRIME. 11285 (Check if this is true? OpenPGP calls them "strong".) 11286 [Ulf Möller] 11287 11288 *) Merge the functionality of "dh" and "gendh" programs into a new program 11289 "dhparam". The old programs are retained for now but will handle DH keys 11290 (instead of parameters) in future. 11291 [Steve Henson] 11292 11293 *) Make the ciphers, s_server and s_client programs check the return values 11294 when a new cipher list is set. 11295 [Steve Henson] 11296 11297 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 11298 ciphers. Before when the 56bit ciphers were enabled the sorting was 11299 wrong. 11300 11301 The syntax for the cipher sorting has been extended to support sorting by 11302 cipher-strength (using the strength_bits hard coded in the tables). 11303 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 11304 11305 Fix a bug in the cipher-command parser: when supplying a cipher command 11306 string with an "undefined" symbol (neither command nor alphanumeric 11307 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 11308 an error is flagged. 11309 11310 Due to the strength-sorting extension, the code of the 11311 ssl_create_cipher_list() function was completely rearranged. I hope that 11312 the readability was also increased :-) 11313 [Lutz Jaenicke <[email protected]>] 11314 11315 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 11316 for the first serial number and places 2 in the serial number file. This 11317 avoids problems when the root CA is created with serial number zero and 11318 the first user certificate has the same issuer name and serial number 11319 as the root CA. 11320 [Steve Henson] 11321 11322 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 11323 the new code. Add documentation for this stuff. 11324 [Steve Henson] 11325 11326 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 11327 X509_*() to X509at_*() on the grounds that they don't handle X509 11328 structures and behave in an analogous way to the X509v3 functions: 11329 they shouldn't be called directly but wrapper functions should be used 11330 instead. 11331 11332 So we also now have some wrapper functions that call the X509at functions 11333 when passed certificate requests. (TO DO: similar things can be done with 11334 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 11335 things. Some of these need some d2i or i2d and print functionality 11336 because they handle more complex structures.) 11337 [Steve Henson] 11338 11339 *) Add missing #ifndefs that caused missing symbols when building libssl 11340 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 11341 NO_RSA in ssl/s2*.c. 11342 [Kris Kennaway <[email protected]>, modified by Ulf Möller] 11343 11344 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 11345 has a return value which indicates the quality of the random data 11346 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 11347 error queue. New function RAND_pseudo_bytes() generates output that is 11348 guaranteed to be unique but not unpredictable. RAND_add is like 11349 RAND_seed, but takes an extra argument for an entropy estimate 11350 (RAND_seed always assumes full entropy). 11351 [Ulf Möller] 11352 11353 *) Do more iterations of Rabin-Miller probable prime test (specifically, 11354 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 11355 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 11356 in crypto/bn/bn_prime.c for the complete table). This guarantees a 11357 false-positive rate of at most 2^-80 for random input. 11358 [Bodo Moeller] 11359 11360 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 11361 [Bodo Moeller] 11362 11363 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 11364 in the 0.9.5 release), this returns the chain 11365 from an X509_CTX structure with a dup of the stack and all 11366 the X509 reference counts upped: so the stack will exist 11367 after X509_CTX_cleanup() has been called. Modify pkcs12.c 11368 to use this. 11369 11370 Also make SSL_SESSION_print() print out the verify return 11371 code. 11372 [Steve Henson] 11373 11374 *) Add manpage for the pkcs12 command. Also change the default 11375 behaviour so MAC iteration counts are used unless the new 11376 -nomaciter option is used. This improves file security and 11377 only older versions of MSIE (4.0 for example) need it. 11378 [Steve Henson] 11379 11380 *) Honor the no-xxx Configure options when creating .DEF files. 11381 [Ulf Möller] 11382 11383 *) Add PKCS#10 attributes to field table: challengePassword, 11384 unstructuredName and unstructuredAddress. These are taken from 11385 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 11386 international characters are used. 11387 11388 More changes to X509_ATTRIBUTE code: allow the setting of types 11389 based on strings. Remove the 'loc' parameter when adding 11390 attributes because these will be a SET OF encoding which is sorted 11391 in ASN1 order. 11392 [Steve Henson] 11393 11394 *) Initial changes to the 'req' utility to allow request generation 11395 automation. This will allow an application to just generate a template 11396 file containing all the field values and have req construct the 11397 request. 11398 11399 Initial support for X509_ATTRIBUTE handling. Stacks of these are 11400 used all over the place including certificate requests and PKCS#7 11401 structures. They are currently handled manually where necessary with 11402 some primitive wrappers for PKCS#7. The new functions behave in a 11403 manner analogous to the X509 extension functions: they allow 11404 attributes to be looked up by NID and added. 11405 11406 Later something similar to the X509V3 code would be desirable to 11407 automatically handle the encoding, decoding and printing of the 11408 more complex types. The string types like challengePassword can 11409 be handled by the string table functions. 11410 11411 Also modified the multi byte string table handling. Now there is 11412 a 'global mask' which masks out certain types. The table itself 11413 can use the flag STABLE_NO_MASK to ignore the mask setting: this 11414 is useful when for example there is only one permissible type 11415 (as in countryName) and using the mask might result in no valid 11416 types at all. 11417 [Steve Henson] 11418 11419 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 11420 SSL_get_peer_finished to allow applications to obtain the latest 11421 Finished messages sent to the peer or expected from the peer, 11422 respectively. (SSL_get_peer_finished is usually the Finished message 11423 actually received from the peer, otherwise the protocol will be aborted.) 11424 11425 As the Finished message are message digests of the complete handshake 11426 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 11427 be used for external authentication procedures when the authentication 11428 provided by SSL/TLS is not desired or is not enough. 11429 [Bodo Moeller] 11430 11431 *) Enhanced support for Alpha Linux is added. Now ./config checks if 11432 the host supports BWX extension and if Compaq C is present on the 11433 $PATH. Just exploiting of the BWX extension results in 20-30% 11434 performance kick for some algorithms, e.g. DES and RC4 to mention 11435 a couple. Compaq C in turn generates ~20% faster code for MD5 and 11436 SHA1. 11437 [Andy Polyakov] 11438 11439 *) Add support for MS "fast SGC". This is arguably a violation of the 11440 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 11441 weak crypto and after checking the certificate is SGC a second one 11442 with strong crypto. MS SGC stops the first handshake after receiving 11443 the server certificate message and sends a second client hello. Since 11444 a server will typically do all the time consuming operations before 11445 expecting any further messages from the client (server key exchange 11446 is the most expensive) there is little difference between the two. 11447 11448 To get OpenSSL to support MS SGC we have to permit a second client 11449 hello message after we have sent server done. In addition we have to 11450 reset the MAC if we do get this second client hello. 11451 [Steve Henson] 11452 11453 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 11454 if a DER encoded private key is RSA or DSA traditional format. Changed 11455 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 11456 format DER encoded private key. Newer code should use PKCS#8 format which 11457 has the key type encoded in the ASN1 structure. Added DER private key 11458 support to pkcs8 application. 11459 [Steve Henson] 11460 11461 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 11462 ciphersuites has been selected (as required by the SSL 3/TLS 1 11463 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 11464 is set, we interpret this as a request to violate the specification 11465 (the worst that can happen is a handshake failure, and 'correct' 11466 behaviour would result in a handshake failure anyway). 11467 [Bodo Moeller] 11468 11469 *) In SSL_CTX_add_session, take into account that there might be multiple 11470 SSL_SESSION structures with the same session ID (e.g. when two threads 11471 concurrently obtain them from an external cache). 11472 The internal cache can handle only one SSL_SESSION with a given ID, 11473 so if there's a conflict, we now throw out the old one to achieve 11474 consistency. 11475 [Bodo Moeller] 11476 11477 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 11478 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 11479 some routines that use cipher OIDs: some ciphers do not have OIDs 11480 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 11481 example. 11482 [Steve Henson] 11483 11484 *) Simplify the trust setting structure and code. Now we just have 11485 two sequences of OIDs for trusted and rejected settings. These will 11486 typically have values the same as the extended key usage extension 11487 and any application specific purposes. 11488 11489 The trust checking code now has a default behaviour: it will just 11490 check for an object with the same NID as the passed id. Functions can 11491 be provided to override either the default behaviour or the behaviour 11492 for a given id. SSL client, server and email already have functions 11493 in place for compatibility: they check the NID and also return "trusted" 11494 if the certificate is self signed. 11495 [Steve Henson] 11496 11497 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 11498 traditional format into an EVP_PKEY structure. 11499 [Steve Henson] 11500 11501 *) Add a password callback function PEM_cb() which either prompts for 11502 a password if usr_data is NULL or otherwise assumes it is a null 11503 terminated password. Allow passwords to be passed on command line 11504 environment or config files in a few more utilities. 11505 [Steve Henson] 11506 11507 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 11508 keys. Add some short names for PKCS#8 PBE algorithms and allow them 11509 to be specified on the command line for the pkcs8 and pkcs12 utilities. 11510 Update documentation. 11511 [Steve Henson] 11512 11513 *) Support for ASN1 "NULL" type. This could be handled before by using 11514 ASN1_TYPE but there wasn't any function that would try to read a NULL 11515 and produce an error if it couldn't. For compatibility we also have 11516 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 11517 don't allocate anything because they don't need to. 11518 [Steve Henson] 11519 11520 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 11521 for details. 11522 [Andy Polyakov, Roy Woods <[email protected]>] 11523 11524 *) Rebuild of the memory allocation routines used by OpenSSL code and 11525 possibly others as well. The purpose is to make an interface that 11526 provide hooks so anyone can build a separate set of allocation and 11527 deallocation routines to be used by OpenSSL, for example memory 11528 pool implementations, or something else, which was previously hard 11529 since Malloc(), Realloc() and Free() were defined as macros having 11530 the values malloc, realloc and free, respectively (except for Win32 11531 compilations). The same is provided for memory debugging code. 11532 OpenSSL already comes with functionality to find memory leaks, but 11533 this gives people a chance to debug other memory problems. 11534 11535 With these changes, a new set of functions and macros have appeared: 11536 11537 CRYPTO_set_mem_debug_functions() [F] 11538 CRYPTO_get_mem_debug_functions() [F] 11539 CRYPTO_dbg_set_options() [F] 11540 CRYPTO_dbg_get_options() [F] 11541 CRYPTO_malloc_debug_init() [M] 11542 11543 The memory debug functions are NULL by default, unless the library 11544 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 11545 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 11546 gives the standard debugging functions that come with OpenSSL) or 11547 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 11548 provided by the library user) must be used. When the standard 11549 debugging functions are used, CRYPTO_dbg_set_options can be used to 11550 request additional information: 11551 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 11552 the CRYPTO_MDEBUG_xxx macro when compiling the library. 11553 11554 Also, things like CRYPTO_set_mem_functions will always give the 11555 expected result (the new set of functions is used for allocation 11556 and deallocation) at all times, regardless of platform and compiler 11557 options. 11558 11559 To finish it up, some functions that were never use in any other 11560 way than through macros have a new API and new semantic: 11561 11562 CRYPTO_dbg_malloc() 11563 CRYPTO_dbg_realloc() 11564 CRYPTO_dbg_free() 11565 11566 All macros of value have retained their old syntax. 11567 [Richard Levitte and Bodo Moeller] 11568 11569 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 11570 ordering of SMIMECapabilities wasn't in "strength order" and there 11571 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 11572 algorithm. 11573 [Steve Henson] 11574 11575 *) Some ASN1 types with illegal zero length encoding (INTEGER, 11576 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 11577 [Frans Heymans <[email protected]>, modified by Steve Henson] 11578 11579 *) Merge in my S/MIME library for OpenSSL. This provides a simple 11580 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 11581 functionality to handle multipart/signed properly) and a utility 11582 called 'smime' to call all this stuff. This is based on code I 11583 originally wrote for Celo who have kindly allowed it to be 11584 included in OpenSSL. 11585 [Steve Henson] 11586 11587 *) Add variants des_set_key_checked and des_set_key_unchecked of 11588 des_set_key (aka des_key_sched). Global variable des_check_key 11589 decides which of these is called by des_set_key; this way 11590 des_check_key behaves as it always did, but applications and 11591 the library itself, which was buggy for des_check_key == 1, 11592 have a cleaner way to pick the version they need. 11593 [Bodo Moeller] 11594 11595 *) New function PKCS12_newpass() which changes the password of a 11596 PKCS12 structure. 11597 [Steve Henson] 11598 11599 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 11600 dynamic mix. In both cases the ids can be used as an index into the 11601 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 11602 functions so they accept a list of the field values and the 11603 application doesn't need to directly manipulate the X509_TRUST 11604 structure. 11605 [Steve Henson] 11606 11607 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 11608 need initialising. 11609 [Steve Henson] 11610 11611 *) Modify the way the V3 extension code looks up extensions. This now 11612 works in a similar way to the object code: we have some "standard" 11613 extensions in a static table which is searched with OBJ_bsearch() 11614 and the application can add dynamic ones if needed. The file 11615 crypto/x509v3/ext_dat.h now has the info: this file needs to be 11616 updated whenever a new extension is added to the core code and kept 11617 in ext_nid order. There is a simple program 'tabtest.c' which checks 11618 this. New extensions are not added too often so this file can readily 11619 be maintained manually. 11620 11621 There are two big advantages in doing things this way. The extensions 11622 can be looked up immediately and no longer need to be "added" using 11623 X509V3_add_standard_extensions(): this function now does nothing. 11624 [Side note: I get *lots* of email saying the extension code doesn't 11625 work because people forget to call this function] 11626 Also no dynamic allocation is done unless new extensions are added: 11627 so if we don't add custom extensions there is no need to call 11628 X509V3_EXT_cleanup(). 11629 [Steve Henson] 11630 11631 *) Modify enc utility's salting as follows: make salting the default. Add a 11632 magic header, so unsalted files fail gracefully instead of just decrypting 11633 to garbage. This is because not salting is a big security hole, so people 11634 should be discouraged from doing it. 11635 [Ben Laurie] 11636 11637 *) Fixes and enhancements to the 'x509' utility. It allowed a message 11638 digest to be passed on the command line but it only used this 11639 parameter when signing a certificate. Modified so all relevant 11640 operations are affected by the digest parameter including the 11641 -fingerprint and -x509toreq options. Also -x509toreq choked if a 11642 DSA key was used because it didn't fix the digest. 11643 [Steve Henson] 11644 11645 *) Initial certificate chain verify code. Currently tests the untrusted 11646 certificates for consistency with the verify purpose (which is set 11647 when the X509_STORE_CTX structure is set up) and checks the pathlength. 11648 11649 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 11650 this is because it will reject chains with invalid extensions whereas 11651 every previous version of OpenSSL and SSLeay made no checks at all. 11652 11653 Trust code: checks the root CA for the relevant trust settings. Trust 11654 settings have an initial value consistent with the verify purpose: e.g. 11655 if the verify purpose is for SSL client use it expects the CA to be 11656 trusted for SSL client use. However the default value can be changed to 11657 permit custom trust settings: one example of this would be to only trust 11658 certificates from a specific "secure" set of CAs. 11659 11660 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 11661 which should be used for version portability: especially since the 11662 verify structure is likely to change more often now. 11663 11664 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 11665 to set them. If not set then assume SSL clients will verify SSL servers 11666 and vice versa. 11667 11668 Two new options to the verify program: -untrusted allows a set of 11669 untrusted certificates to be passed in and -purpose which sets the 11670 intended purpose of the certificate. If a purpose is set then the 11671 new chain verify code is used to check extension consistency. 11672 [Steve Henson] 11673 11674 *) Support for the authority information access extension. 11675 [Steve Henson] 11676 11677 *) Modify RSA and DSA PEM read routines to transparently handle 11678 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 11679 public keys in a format compatible with certificate 11680 SubjectPublicKeyInfo structures. Unfortunately there were already 11681 functions called *_PublicKey_* which used various odd formats so 11682 these are retained for compatibility: however the DSA variants were 11683 never in a public release so they have been deleted. Changed dsa/rsa 11684 utilities to handle the new format: note no releases ever handled public 11685 keys so we should be OK. 11686 11687 The primary motivation for this change is to avoid the same fiasco 11688 that dogs private keys: there are several incompatible private key 11689 formats some of which are standard and some OpenSSL specific and 11690 require various evil hacks to allow partial transparent handling and 11691 even then it doesn't work with DER formats. Given the option anything 11692 other than PKCS#8 should be dumped: but the other formats have to 11693 stay in the name of compatibility. 11694 11695 With public keys and the benefit of hindsight one standard format 11696 is used which works with EVP_PKEY, RSA or DSA structures: though 11697 it clearly returns an error if you try to read the wrong kind of key. 11698 11699 Added a -pubkey option to the 'x509' utility to output the public key. 11700 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 11701 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 11702 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 11703 that do the same as the EVP_PKEY_assign_*() except they up the 11704 reference count of the added key (they don't "swallow" the 11705 supplied key). 11706 [Steve Henson] 11707 11708 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 11709 CRLs would fail if the file contained no certificates or no CRLs: 11710 added a new function to read in both types and return the number 11711 read: this means that if none are read it will be an error. The 11712 DER versions of the certificate and CRL reader would always fail 11713 because it isn't possible to mix certificates and CRLs in DER format 11714 without choking one or the other routine. Changed this to just read 11715 a certificate: this is the best we can do. Also modified the code 11716 in apps/verify.c to take notice of return codes: it was previously 11717 attempting to read in certificates from NULL pointers and ignoring 11718 any errors: this is one reason why the cert and CRL reader seemed 11719 to work. It doesn't check return codes from the default certificate 11720 routines: these may well fail if the certificates aren't installed. 11721 [Steve Henson] 11722 11723 *) Code to support otherName option in GeneralName. 11724 [Steve Henson] 11725 11726 *) First update to verify code. Change the verify utility 11727 so it warns if it is passed a self signed certificate: 11728 for consistency with the normal behaviour. X509_verify 11729 has been modified to it will now verify a self signed 11730 certificate if *exactly* the same certificate appears 11731 in the store: it was previously impossible to trust a 11732 single self signed certificate. This means that: 11733 openssl verify ss.pem 11734 now gives a warning about a self signed certificate but 11735 openssl verify -CAfile ss.pem ss.pem 11736 is OK. 11737 [Steve Henson] 11738 11739 *) For servers, store verify_result in SSL_SESSION data structure 11740 (and add it to external session representation). 11741 This is needed when client certificate verifications fails, 11742 but an application-provided verification callback (set by 11743 SSL_CTX_set_cert_verify_callback) allows accepting the session 11744 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 11745 but returns 1): When the session is reused, we have to set 11746 ssl->verify_result to the appropriate error code to avoid 11747 security holes. 11748 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 11749 11750 *) Fix a bug in the new PKCS#7 code: it didn't consider the 11751 case in PKCS7_dataInit() where the signed PKCS7 structure 11752 didn't contain any existing data because it was being created. 11753 [Po-Cheng Chen <[email protected]>, slightly modified by Steve Henson] 11754 11755 *) Add a salt to the key derivation routines in enc.c. This 11756 forms the first 8 bytes of the encrypted file. Also add a 11757 -S option to allow a salt to be input on the command line. 11758 [Steve Henson] 11759 11760 *) New function X509_cmp(). Oddly enough there wasn't a function 11761 to compare two certificates. We do this by working out the SHA1 11762 hash and comparing that. X509_cmp() will be needed by the trust 11763 code. 11764 [Steve Henson] 11765 11766 *) SSL_get1_session() is like SSL_get_session(), but increments 11767 the reference count in the SSL_SESSION returned. 11768 [Geoff Thorpe <[email protected]>] 11769 11770 *) Fix for 'req': it was adding a null to request attributes. 11771 Also change the X509_LOOKUP and X509_INFO code to handle 11772 certificate auxiliary information. 11773 [Steve Henson] 11774 11775 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 11776 the 'enc' command. 11777 [Steve Henson] 11778 11779 *) Add the possibility to add extra information to the memory leak 11780 detecting output, to form tracebacks, showing from where each 11781 allocation was originated: CRYPTO_push_info("constant string") adds 11782 the string plus current file name and line number to a per-thread 11783 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 11784 is like calling CYRPTO_pop_info() until the stack is empty. 11785 Also updated memory leak detection code to be multi-thread-safe. 11786 [Richard Levitte] 11787 11788 *) Add options -text and -noout to pkcs7 utility and delete the 11789 encryption options which never did anything. Update docs. 11790 [Steve Henson] 11791 11792 *) Add options to some of the utilities to allow the pass phrase 11793 to be included on either the command line (not recommended on 11794 OSes like Unix) or read from the environment. Update the 11795 manpages and fix a few bugs. 11796 [Steve Henson] 11797 11798 *) Add a few manpages for some of the openssl commands. 11799 [Steve Henson] 11800 11801 *) Fix the -revoke option in ca. It was freeing up memory twice, 11802 leaking and not finding already revoked certificates. 11803 [Steve Henson] 11804 11805 *) Extensive changes to support certificate auxiliary information. 11806 This involves the use of X509_CERT_AUX structure and X509_AUX 11807 functions. An X509_AUX function such as PEM_read_X509_AUX() 11808 can still read in a certificate file in the usual way but it 11809 will also read in any additional "auxiliary information". By 11810 doing things this way a fair degree of compatibility can be 11811 retained: existing certificates can have this information added 11812 using the new 'x509' options. 11813 11814 Current auxiliary information includes an "alias" and some trust 11815 settings. The trust settings will ultimately be used in enhanced 11816 certificate chain verification routines: currently a certificate 11817 can only be trusted if it is self signed and then it is trusted 11818 for all purposes. 11819 [Steve Henson] 11820 11821 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 11822 The problem was that one of the replacement routines had not been working 11823 since SSLeay releases. For now the offending routine has been replaced 11824 with non-optimised assembler. Even so, this now gives around 95% 11825 performance improvement for 1024 bit RSA signs. 11826 [Mark Cox] 11827 11828 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 11829 handling. Most clients have the effective key size in bits equal to 11830 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 11831 A few however don't do this and instead use the size of the decrypted key 11832 to determine the RC2 key length and the AlgorithmIdentifier to determine 11833 the effective key length. In this case the effective key length can still 11834 be 40 bits but the key length can be 168 bits for example. This is fixed 11835 by manually forcing an RC2 key into the EVP_PKEY structure because the 11836 EVP code can't currently handle unusual RC2 key sizes: it always assumes 11837 the key length and effective key length are equal. 11838 [Steve Henson] 11839 11840 *) Add a bunch of functions that should simplify the creation of 11841 X509_NAME structures. Now you should be able to do: 11842 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 11843 and have it automatically work out the correct field type and fill in 11844 the structures. The more adventurous can try: 11845 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 11846 and it will (hopefully) work out the correct multibyte encoding. 11847 [Steve Henson] 11848 11849 *) Change the 'req' utility to use the new field handling and multibyte 11850 copy routines. Before the DN field creation was handled in an ad hoc 11851 way in req, ca, and x509 which was rather broken and didn't support 11852 BMPStrings or UTF8Strings. Since some software doesn't implement 11853 BMPStrings or UTF8Strings yet, they can be enabled using the config file 11854 using the dirstring_type option. See the new comment in the default 11855 openssl.cnf for more info. 11856 [Steve Henson] 11857 11858 *) Make crypto/rand/md_rand.c more robust: 11859 - Assure unique random numbers after fork(). 11860 - Make sure that concurrent threads access the global counter and 11861 md serializably so that we never lose entropy in them 11862 or use exactly the same state in multiple threads. 11863 Access to the large state is not always serializable because 11864 the additional locking could be a performance killer, and 11865 md should be large enough anyway. 11866 [Bodo Moeller] 11867 11868 *) New file apps/app_rand.c with commonly needed functionality 11869 for handling the random seed file. 11870 11871 Use the random seed file in some applications that previously did not: 11872 ca, 11873 dsaparam -genkey (which also ignored its '-rand' option), 11874 s_client, 11875 s_server, 11876 x509 (when signing). 11877 Except on systems with /dev/urandom, it is crucial to have a random 11878 seed file at least for key creation, DSA signing, and for DH exchanges; 11879 for RSA signatures we could do without one. 11880 11881 gendh and gendsa (unlike genrsa) used to read only the first byte 11882 of each file listed in the '-rand' option. The function as previously 11883 found in genrsa is now in app_rand.c and is used by all programs 11884 that support '-rand'. 11885 [Bodo Moeller] 11886 11887 *) In RAND_write_file, use mode 0600 for creating files; 11888 don't just chmod when it may be too late. 11889 [Bodo Moeller] 11890 11891 *) Report an error from X509_STORE_load_locations 11892 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 11893 [Bill Perry] 11894 11895 *) New function ASN1_mbstring_copy() this copies a string in either 11896 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 11897 into an ASN1_STRING type. A mask of permissible types is passed 11898 and it chooses the "minimal" type to use or an error if not type 11899 is suitable. 11900 [Steve Henson] 11901 11902 *) Add function equivalents to the various macros in asn1.h. The old 11903 macros are retained with an M_ prefix. Code inside the library can 11904 use the M_ macros. External code (including the openssl utility) 11905 should *NOT* in order to be "shared library friendly". 11906 [Steve Henson] 11907 11908 *) Add various functions that can check a certificate's extensions 11909 to see if it usable for various purposes such as SSL client, 11910 server or S/MIME and CAs of these types. This is currently 11911 VERY EXPERIMENTAL but will ultimately be used for certificate chain 11912 verification. Also added a -purpose flag to x509 utility to 11913 print out all the purposes. 11914 [Steve Henson] 11915 11916 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 11917 functions. 11918 [Steve Henson] 11919 11920 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 11921 for, obtain and decode and extension and obtain its critical flag. 11922 This allows all the necessary extension code to be handled in a 11923 single function call. 11924 [Steve Henson] 11925 11926 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 11927 platforms. See crypto/rc4/rc4_enc.c for further details. 11928 [Andy Polyakov] 11929 11930 *) New -noout option to asn1parse. This causes no output to be produced 11931 its main use is when combined with -strparse and -out to extract data 11932 from a file (which may not be in ASN.1 format). 11933 [Steve Henson] 11934 11935 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 11936 when producing the local key id. 11937 [Richard Levitte <[email protected]>] 11938 11939 *) New option -dhparam in s_server. This allows a DH parameter file to be 11940 stated explicitly. If it is not stated then it tries the first server 11941 certificate file. The previous behaviour hard coded the filename 11942 "server.pem". 11943 [Steve Henson] 11944 11945 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 11946 a public key to be input or output. For example: 11947 openssl rsa -in key.pem -pubout -out pubkey.pem 11948 Also added necessary DSA public key functions to handle this. 11949 [Steve Henson] 11950 11951 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 11952 in the message. This was handled by allowing 11953 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 11954 [Steve Henson, reported by Sampo Kellomaki <[email protected]>] 11955 11956 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 11957 to the end of the strings whereas this didn't. This would cause problems 11958 if strings read with d2i_ASN1_bytes() were later modified. 11959 [Steve Henson, reported by Arne Ansper <[email protected]>] 11960 11961 *) Fix for base64 decode bug. When a base64 bio reads only one line of 11962 data and it contains EOF it will end up returning an error. This is 11963 caused by input 46 bytes long. The cause is due to the way base64 11964 BIOs find the start of base64 encoded data. They do this by trying a 11965 trial decode on each line until they find one that works. When they 11966 do a flag is set and it starts again knowing it can pass all the 11967 data directly through the decoder. Unfortunately it doesn't reset 11968 the context it uses. This means that if EOF is reached an attempt 11969 is made to pass two EOFs through the context and this causes the 11970 resulting error. This can also cause other problems as well. As is 11971 usual with these problems it takes *ages* to find and the fix is 11972 trivial: move one line. 11973 [Steve Henson, reported by [email protected] (Ivan Nejgebauer) ] 11974 11975 *) Ugly workaround to get s_client and s_server working under Windows. The 11976 old code wouldn't work because it needed to select() on sockets and the 11977 tty (for keypresses and to see if data could be written). Win32 only 11978 supports select() on sockets so we select() with a 1s timeout on the 11979 sockets and then see if any characters are waiting to be read, if none 11980 are present then we retry, we also assume we can always write data to 11981 the tty. This isn't nice because the code then blocks until we've 11982 received a complete line of data and it is effectively polling the 11983 keyboard at 1s intervals: however it's quite a bit better than not 11984 working at all :-) A dedicated Windows application might handle this 11985 with an event loop for example. 11986 [Steve Henson] 11987 11988 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 11989 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 11990 will be called when RSA_sign() and RSA_verify() are used. This is useful 11991 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 11992 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 11993 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 11994 This necessitated the support of an extra signature type NID_md5_sha1 11995 for SSL signatures and modifications to the SSL library to use it instead 11996 of calling RSA_public_decrypt() and RSA_private_encrypt(). 11997 [Steve Henson] 11998 11999 *) Add new -verify -CAfile and -CApath options to the crl program, these 12000 will lookup a CRL issuers certificate and verify the signature in a 12001 similar way to the verify program. Tidy up the crl program so it 12002 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 12003 less strict. It will now permit CRL extensions even if it is not 12004 a V2 CRL: this will allow it to tolerate some broken CRLs. 12005 [Steve Henson] 12006 12007 *) Initialize all non-automatic variables each time one of the openssl 12008 sub-programs is started (this is necessary as they may be started 12009 multiple times from the "OpenSSL>" prompt). 12010 [Lennart Bang, Bodo Moeller] 12011 12012 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 12013 removing all other RSA functionality (this is what NO_RSA does). This 12014 is so (for example) those in the US can disable those operations covered 12015 by the RSA patent while allowing storage and parsing of RSA keys and RSA 12016 key generation. 12017 [Steve Henson] 12018 12019 *) Non-copying interface to BIO pairs. 12020 (still largely untested) 12021 [Bodo Moeller] 12022 12023 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 12024 ASCII string. This was handled independently in various places before. 12025 [Steve Henson] 12026 12027 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 12028 UTF8 strings a character at a time. 12029 [Steve Henson] 12030 12031 *) Use client_version from client hello to select the protocol 12032 (s23_srvr.c) and for RSA client key exchange verification 12033 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 12034 [Bodo Moeller] 12035 12036 *) Add various utility functions to handle SPKACs, these were previously 12037 handled by poking round in the structure internals. Added new function 12038 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 12039 print, verify and generate SPKACs. Based on an original idea from 12040 Massimiliano Pala <[email protected]> but extensively modified. 12041 [Steve Henson] 12042 12043 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 12044 [Andy Polyakov] 12045 12046 *) Allow the config file extension section to be overwritten on the 12047 command line. Based on an original idea from Massimiliano Pala 12048 <[email protected]>. The new option is called -extensions 12049 and can be applied to ca, req and x509. Also -reqexts to override 12050 the request extensions in req and -crlexts to override the crl extensions 12051 in ca. 12052 [Steve Henson] 12053 12054 *) Add new feature to the SPKAC handling in ca. Now you can include 12055 the same field multiple times by preceding it by "XXXX." for example: 12056 1.OU="Unit name 1" 12057 2.OU="Unit name 2" 12058 this is the same syntax as used in the req config file. 12059 [Steve Henson] 12060 12061 *) Allow certificate extensions to be added to certificate requests. These 12062 are specified in a 'req_extensions' option of the req section of the 12063 config file. They can be printed out with the -text option to req but 12064 are otherwise ignored at present. 12065 [Steve Henson] 12066 12067 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 12068 data read consists of only the final block it would not decrypted because 12069 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 12070 A misplaced 'break' also meant the decrypted final block might not be 12071 copied until the next read. 12072 [Steve Henson] 12073 12074 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 12075 a few extra parameters to the DH structure: these will be useful if 12076 for example we want the value of 'q' or implement X9.42 DH. 12077 [Steve Henson] 12078 12079 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 12080 provides hooks that allow the default DSA functions or functions on a 12081 "per key" basis to be replaced. This allows hardware acceleration and 12082 hardware key storage to be handled without major modification to the 12083 library. Also added low level modexp hooks and CRYPTO_EX structure and 12084 associated functions. 12085 [Steve Henson] 12086 12087 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 12088 as "read only": it can't be written to and the buffer it points to will 12089 not be freed. Reading from a read only BIO is much more efficient than 12090 a normal memory BIO. This was added because there are several times when 12091 an area of memory needs to be read from a BIO. The previous method was 12092 to create a memory BIO and write the data to it, this results in two 12093 copies of the data and an O(n^2) reading algorithm. There is a new 12094 function BIO_new_mem_buf() which creates a read only memory BIO from 12095 an area of memory. Also modified the PKCS#7 routines to use read only 12096 memory BIOs. 12097 [Steve Henson] 12098 12099 *) Bugfix: ssl23_get_client_hello did not work properly when called in 12100 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 12101 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 12102 but a retry condition occurred while trying to read the rest. 12103 [Bodo Moeller] 12104 12105 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 12106 NID_pkcs7_encrypted by default: this was wrong since this should almost 12107 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 12108 the encrypted data type: this is a more sensible place to put it and it 12109 allows the PKCS#12 code to be tidied up that duplicated this 12110 functionality. 12111 [Steve Henson] 12112 12113 *) Changed obj_dat.pl script so it takes its input and output files on 12114 the command line. This should avoid shell escape redirection problems 12115 under Win32. 12116 [Steve Henson] 12117 12118 *) Initial support for certificate extension requests, these are included 12119 in things like Xenroll certificate requests. Included functions to allow 12120 extensions to be obtained and added. 12121 [Steve Henson] 12122 12123 *) -crlf option to s_client and s_server for sending newlines as 12124 CRLF (as required by many protocols). 12125 [Bodo Moeller] 12126 12127 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 12128 12129 *) Install libRSAglue.a when OpenSSL is built with RSAref. 12130 [Ralf S. Engelschall] 12131 12132 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 12133 [Andrija Antonijevic <[email protected]>] 12134 12135 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 12136 program. 12137 [Steve Henson] 12138 12139 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 12140 DH parameters/keys (q is lost during that conversion, but the resulting 12141 DH parameters contain its length). 12142 12143 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 12144 much faster than DH_generate_parameters (which creates parameters 12145 where p = 2*q + 1), and also the smaller q makes DH computations 12146 much more efficient (160-bit exponentiation instead of 1024-bit 12147 exponentiation); so this provides a convenient way to support DHE 12148 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 12149 utter importance to use 12150 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 12151 or 12152 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 12153 when such DH parameters are used, because otherwise small subgroup 12154 attacks may become possible! 12155 [Bodo Moeller] 12156 12157 *) Avoid memory leak in i2d_DHparams. 12158 [Bodo Moeller] 12159 12160 *) Allow the -k option to be used more than once in the enc program: 12161 this allows the same encrypted message to be read by multiple recipients. 12162 [Steve Henson] 12163 12164 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 12165 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 12166 it will always use the numerical form of the OID, even if it has a short 12167 or long name. 12168 [Steve Henson] 12169 12170 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 12171 method only got called if p,q,dmp1,dmq1,iqmp components were present, 12172 otherwise bn_mod_exp was called. In the case of hardware keys for example 12173 no private key components need be present and it might store extra data 12174 in the RSA structure, which cannot be accessed from bn_mod_exp. 12175 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 12176 private key operations. 12177 [Steve Henson] 12178 12179 *) Added support for SPARC Linux. 12180 [Andy Polyakov] 12181 12182 *) pem_password_cb function type incompatibly changed from 12183 typedef int pem_password_cb(char *buf, int size, int rwflag); 12184 to 12185 ....(char *buf, int size, int rwflag, void *userdata); 12186 so that applications can pass data to their callbacks: 12187 The PEM[_ASN1]_{read,write}... functions and macros now take an 12188 additional void * argument, which is just handed through whenever 12189 the password callback is called. 12190 [Damien Miller <[email protected]>; tiny changes by Bodo Moeller] 12191 12192 New function SSL_CTX_set_default_passwd_cb_userdata. 12193 12194 Compatibility note: As many C implementations push function arguments 12195 onto the stack in reverse order, the new library version is likely to 12196 interoperate with programs that have been compiled with the old 12197 pem_password_cb definition (PEM_whatever takes some data that 12198 happens to be on the stack as its last argument, and the callback 12199 just ignores this garbage); but there is no guarantee whatsoever that 12200 this will work. 12201 12202 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 12203 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 12204 problems not only on Windows, but also on some Unix platforms. 12205 To avoid problematic command lines, these definitions are now in an 12206 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 12207 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 12208 [Bodo Moeller] 12209 12210 *) MIPS III/IV assembler module is reimplemented. 12211 [Andy Polyakov] 12212 12213 *) More DES library cleanups: remove references to srand/rand and 12214 delete an unused file. 12215 [Ulf Möller] 12216 12217 *) Add support for the free Netwide assembler (NASM) under Win32, 12218 since not many people have MASM (ml) and it can be hard to obtain. 12219 This is currently experimental but it seems to work OK and pass all 12220 the tests. Check out INSTALL.W32 for info. 12221 [Steve Henson] 12222 12223 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 12224 without temporary keys kept an extra copy of the server key, 12225 and connections with temporary keys did not free everything in case 12226 of an error. 12227 [Bodo Moeller] 12228 12229 *) New function RSA_check_key and new openssl rsa option -check 12230 for verifying the consistency of RSA keys. 12231 [Ulf Moeller, Bodo Moeller] 12232 12233 *) Various changes to make Win32 compile work: 12234 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 12235 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 12236 comparison" warnings. 12237 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 12238 [Steve Henson] 12239 12240 *) Add a debugging option to PKCS#5 v2 key generation function: when 12241 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 12242 derived keys are printed to stderr. 12243 [Steve Henson] 12244 12245 *) Copy the flags in ASN1_STRING_dup(). 12246 [Roman E. Pavlov <[email protected]>] 12247 12248 *) The x509 application mishandled signing requests containing DSA 12249 keys when the signing key was also DSA and the parameters didn't match. 12250 12251 It was supposed to omit the parameters when they matched the signing key: 12252 the verifying software was then supposed to automatically use the CA's 12253 parameters if they were absent from the end user certificate. 12254 12255 Omitting parameters is no longer recommended. The test was also 12256 the wrong way round! This was probably due to unusual behaviour in 12257 EVP_cmp_parameters() which returns 1 if the parameters match. 12258 This meant that parameters were omitted when they *didn't* match and 12259 the certificate was useless. Certificates signed with 'ca' didn't have 12260 this bug. 12261 [Steve Henson, reported by Doug Erickson <[email protected]>] 12262 12263 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 12264 The interface is as follows: 12265 Applications can use 12266 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 12267 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 12268 "off" is now the default. 12269 The library internally uses 12270 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 12271 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 12272 to disable memory-checking temporarily. 12273 12274 Some inconsistent states that previously were possible (and were 12275 even the default) are now avoided. 12276 12277 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 12278 with each memory chunk allocated; this is occasionally more helpful 12279 than just having a counter. 12280 12281 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 12282 12283 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 12284 extensions. 12285 [Bodo Moeller] 12286 12287 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 12288 which largely parallels "options", but is for changing API behaviour, 12289 whereas "options" are about protocol behaviour. 12290 Initial "mode" flags are: 12291 12292 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 12293 a single record has been written. 12294 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 12295 retries use the same buffer location. 12296 (But all of the contents must be 12297 copied!) 12298 [Bodo Moeller] 12299 12300 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 12301 worked. 12302 12303 *) Fix problems with no-hmac etc. 12304 [Ulf Möller, pointed out by Brian Wellington <[email protected]>] 12305 12306 *) New functions RSA_get_default_method(), RSA_set_method() and 12307 RSA_get_method(). These allows replacement of RSA_METHODs without having 12308 to mess around with the internals of an RSA structure. 12309 [Steve Henson] 12310 12311 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 12312 Also really enable memory leak checks in openssl.c and in some 12313 test programs. 12314 [Chad C. Mulligan, Bodo Moeller] 12315 12316 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 12317 up the length of negative integers. This has now been simplified to just 12318 store the length when it is first determined and use it later, rather 12319 than trying to keep track of where data is copied and updating it to 12320 point to the end. 12321 [Steve Henson, reported by Brien Wheeler 12322 <[email protected]>] 12323 12324 *) Add a new function PKCS7_signatureVerify. This allows the verification 12325 of a PKCS#7 signature but with the signing certificate passed to the 12326 function itself. This contrasts with PKCS7_dataVerify which assumes the 12327 certificate is present in the PKCS#7 structure. This isn't always the 12328 case: certificates can be omitted from a PKCS#7 structure and be 12329 distributed by "out of band" means (such as a certificate database). 12330 [Steve Henson] 12331 12332 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 12333 function prototypes in pem.h, also change util/mkdef.pl to add the 12334 necessary function names. 12335 [Steve Henson] 12336 12337 *) mk1mf.pl (used by Windows builds) did not properly read the 12338 options set by Configure in the top level Makefile, and Configure 12339 was not even able to write more than one option correctly. 12340 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 12341 [Bodo Moeller] 12342 12343 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 12344 file to be loaded from a BIO or FILE pointer. The BIO version will 12345 for example allow memory BIOs to contain config info. 12346 [Steve Henson] 12347 12348 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 12349 Whoever hopes to achieve shared-library compatibility across versions 12350 must use this, not the compile-time macro. 12351 (Exercise 0.9.4: Which is the minimum library version required by 12352 such programs?) 12353 Note: All this applies only to multi-threaded programs, others don't 12354 need locks. 12355 [Bodo Moeller] 12356 12357 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 12358 through a BIO pair triggered the default case, i.e. 12359 SSLerr(...,SSL_R_UNKNOWN_STATE). 12360 [Bodo Moeller] 12361 12362 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 12363 can use the SSL library even if none of the specific BIOs is 12364 appropriate. 12365 [Bodo Moeller] 12366 12367 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 12368 for the encoded length. 12369 [Jeon KyoungHo <[email protected]>] 12370 12371 *) Add initial documentation of the X509V3 functions. 12372 [Steve Henson] 12373 12374 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 12375 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 12376 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 12377 secure PKCS#8 private key format with a high iteration count. 12378 [Steve Henson] 12379 12380 *) Fix determination of Perl interpreter: A perl or perl5 12381 _directory_ in $PATH was also accepted as the interpreter. 12382 [Ralf S. Engelschall] 12383 12384 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 12385 wrong with it but it was very old and did things like calling 12386 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 12387 unusual formatting. 12388 [Steve Henson] 12389 12390 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 12391 to use the new extension code. 12392 [Steve Henson] 12393 12394 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 12395 with macros. This should make it easier to change their form, add extra 12396 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 12397 constant. 12398 [Steve Henson] 12399 12400 *) Add to configuration table a new entry that can specify an alternative 12401 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 12402 according to Mark Crispin <[email protected]>. 12403 [Bodo Moeller] 12404 12405#if 0 12406 *) DES CBC did not update the IV. Weird. 12407 [Ben Laurie] 12408#else 12409 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 12410 Changing the behaviour of the former might break existing programs -- 12411 where IV updating is needed, des_ncbc_encrypt can be used. 12412#endif 12413 12414 *) When bntest is run from "make test" it drives bc to check its 12415 calculations, as well as internally checking them. If an internal check 12416 fails, it needs to cause bc to give a non-zero result or make test carries 12417 on without noticing the failure. Fixed. 12418 [Ben Laurie] 12419 12420 *) DES library cleanups. 12421 [Ulf Möller] 12422 12423 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 12424 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 12425 ciphers. NOTE: although the key derivation function has been verified 12426 against some published test vectors it has not been extensively tested 12427 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 12428 of v2.0. 12429 [Steve Henson] 12430 12431 *) Instead of "mkdir -p", which is not fully portable, use new 12432 Perl script "util/mkdir-p.pl". 12433 [Bodo Moeller] 12434 12435 *) Rewrite the way password based encryption (PBE) is handled. It used to 12436 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 12437 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 12438 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 12439 the 'parameter' field of the AlgorithmIdentifier is passed to the 12440 underlying key generation function so it must do its own ASN1 parsing. 12441 This has also changed the EVP_PBE_CipherInit() function which now has a 12442 'parameter' argument instead of literal salt and iteration count values 12443 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 12444 [Steve Henson] 12445 12446 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 12447 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 12448 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 12449 KEY" because this clashed with PKCS#8 unencrypted string. Since this 12450 value was just used as a "magic string" and not used directly its 12451 value doesn't matter. 12452 [Steve Henson] 12453 12454 *) Introduce some semblance of const correctness to BN. Shame C doesn't 12455 support mutable. 12456 [Ben Laurie] 12457 12458 *) "linux-sparc64" configuration (ultrapenguin). 12459 [Ray Miller <[email protected]>] 12460 "linux-sparc" configuration. 12461 [Christian Forster <[email protected]>] 12462 12463 *) config now generates no-xxx options for missing ciphers. 12464 [Ulf Möller] 12465 12466 *) Support the EBCDIC character set (work in progress). 12467 File ebcdic.c not yet included because it has a different license. 12468 [Martin Kraemer <[email protected]>] 12469 12470 *) Support BS2000/OSD-POSIX. 12471 [Martin Kraemer <[email protected]>] 12472 12473 *) Make callbacks for key generation use void * instead of char *. 12474 [Ben Laurie] 12475 12476 *) Make S/MIME samples compile (not yet tested). 12477 [Ben Laurie] 12478 12479 *) Additional typesafe stacks. 12480 [Ben Laurie] 12481 12482 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 12483 [Bodo Moeller] 12484 12485 12486 Changes between 0.9.3 and 0.9.3a [29 May 1999] 12487 12488 *) New configuration variant "sco5-gcc". 12489 12490 *) Updated some demos. 12491 [Sean O Riordain, Wade Scholine] 12492 12493 *) Add missing BIO_free at exit of pkcs12 application. 12494 [Wu Zhigang] 12495 12496 *) Fix memory leak in conf.c. 12497 [Steve Henson] 12498 12499 *) Updates for Win32 to assembler version of MD5. 12500 [Steve Henson] 12501 12502 *) Set #! path to perl in apps/der_chop to where we found it 12503 instead of using a fixed path. 12504 [Bodo Moeller] 12505 12506 *) SHA library changes for irix64-mips4-cc. 12507 [Andy Polyakov] 12508 12509 *) Improvements for VMS support. 12510 [Richard Levitte] 12511 12512 12513 Changes between 0.9.2b and 0.9.3 [24 May 1999] 12514 12515 *) Bignum library bug fix. IRIX 6 passes "make test" now! 12516 This also avoids the problems with SC4.2 and unpatched SC5. 12517 [Andy Polyakov <[email protected]>] 12518 12519 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 12520 These are required because of the typesafe stack would otherwise break 12521 existing code. If old code used a structure member which used to be STACK 12522 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 12523 sk_num or sk_value it would produce an error because the num, data members 12524 are not present in STACK_OF. Now it just produces a warning. sk_set 12525 replaces the old method of assigning a value to sk_value 12526 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 12527 that does this will no longer work (and should use sk_set instead) but 12528 this could be regarded as a "questionable" behaviour anyway. 12529 [Steve Henson] 12530 12531 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 12532 correctly handle encrypted S/MIME data. 12533 [Steve Henson] 12534 12535 *) Change type of various DES function arguments from des_cblock 12536 (which means, in function argument declarations, pointer to char) 12537 to des_cblock * (meaning pointer to array with 8 char elements), 12538 which allows the compiler to do more typechecking; it was like 12539 that back in SSLeay, but with lots of ugly casts. 12540 12541 Introduce new type const_des_cblock. 12542 [Bodo Moeller] 12543 12544 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 12545 problems: find RecipientInfo structure that matches recipient certificate 12546 and initialise the ASN1 structures properly based on passed cipher. 12547 [Steve Henson] 12548 12549 *) Belatedly make the BN tests actually check the results. 12550 [Ben Laurie] 12551 12552 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 12553 to and from BNs: it was completely broken. New compilation option 12554 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 12555 key elements as negative integers. 12556 [Steve Henson] 12557 12558 *) Reorganize and speed up MD5. 12559 [Andy Polyakov <[email protected]>] 12560 12561 *) VMS support. 12562 [Richard Levitte <[email protected]>] 12563 12564 *) New option -out to asn1parse to allow the parsed structure to be 12565 output to a file. This is most useful when combined with the -strparse 12566 option to examine the output of things like OCTET STRINGS. 12567 [Steve Henson] 12568 12569 *) Make SSL library a little more fool-proof by not requiring any longer 12570 that SSL_set_{accept,connect}_state be called before 12571 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 12572 in many applications because usually everything *appeared* to work as 12573 intended anyway -- now it really works as intended). 12574 [Bodo Moeller] 12575 12576 *) Move openssl.cnf out of lib/. 12577 [Ulf Möller] 12578 12579 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 12580 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 12581 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 12582 [Ralf S. Engelschall] 12583 12584 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 12585 handle PKCS#7 enveloped data properly. 12586 [Sebastian Akerman <[email protected]>, modified by Steve] 12587 12588 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 12589 copying pointers. The cert_st handling is changed by this in 12590 various ways (and thus what used to be known as ctx->default_cert 12591 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 12592 any longer when s->cert does not give us what we need). 12593 ssl_cert_instantiate becomes obsolete by this change. 12594 As soon as we've got the new code right (possibly it already is?), 12595 we have solved a couple of bugs of the earlier code where s->cert 12596 was used as if it could not have been shared with other SSL structures. 12597 12598 Note that using the SSL API in certain dirty ways now will result 12599 in different behaviour than observed with earlier library versions: 12600 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 12601 does not influence s as it used to. 12602 12603 In order to clean up things more thoroughly, inside SSL_SESSION 12604 we don't use CERT any longer, but a new structure SESS_CERT 12605 that holds per-session data (if available); currently, this is 12606 the peer's certificate chain and, for clients, the server's certificate 12607 and temporary key. CERT holds only those values that can have 12608 meaningful defaults in an SSL_CTX. 12609 [Bodo Moeller] 12610 12611 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 12612 from the internal representation. Various PKCS#7 fixes: remove some 12613 evil casts and set the enc_dig_alg field properly based on the signing 12614 key type. 12615 [Steve Henson] 12616 12617 *) Allow PKCS#12 password to be set from the command line or the 12618 environment. Let 'ca' get its config file name from the environment 12619 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 12620 and 'x509'). 12621 [Steve Henson] 12622 12623 *) Allow certificate policies extension to use an IA5STRING for the 12624 organization field. This is contrary to the PKIX definition but 12625 VeriSign uses it and IE5 only recognises this form. Document 'x509' 12626 extension option. 12627 [Steve Henson] 12628 12629 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 12630 without disallowing inline assembler and the like for non-pedantic builds. 12631 [Ben Laurie] 12632 12633 *) Support Borland C++ builder. 12634 [Janez Jere <[email protected]>, modified by Ulf Möller] 12635 12636 *) Support Mingw32. 12637 [Ulf Möller] 12638 12639 *) SHA-1 cleanups and performance enhancements. 12640 [Andy Polyakov <[email protected]>] 12641 12642 *) Sparc v8plus assembler for the bignum library. 12643 [Andy Polyakov <[email protected]>] 12644 12645 *) Accept any -xxx and +xxx compiler options in Configure. 12646 [Ulf Möller] 12647 12648 *) Update HPUX configuration. 12649 [Anonymous] 12650 12651 *) Add missing sk_<type>_unshift() function to safestack.h 12652 [Ralf S. Engelschall] 12653 12654 *) New function SSL_CTX_use_certificate_chain_file that sets the 12655 "extra_cert"s in addition to the certificate. (This makes sense 12656 only for "PEM" format files, as chains as a whole are not 12657 DER-encoded.) 12658 [Bodo Moeller] 12659 12660 *) Support verify_depth from the SSL API. 12661 x509_vfy.c had what can be considered an off-by-one-error: 12662 Its depth (which was not part of the external interface) 12663 was actually counting the number of certificates in a chain; 12664 now it really counts the depth. 12665 [Bodo Moeller] 12666 12667 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 12668 instead of X509err, which often resulted in confusing error 12669 messages since the error codes are not globally unique 12670 (e.g. an alleged error in ssl3_accept when a certificate 12671 didn't match the private key). 12672 12673 *) New function SSL_CTX_set_session_id_context that allows to set a default 12674 value (so that you don't need SSL_set_session_id_context for each 12675 connection using the SSL_CTX). 12676 [Bodo Moeller] 12677 12678 *) OAEP decoding bug fix. 12679 [Ulf Möller] 12680 12681 *) Support INSTALL_PREFIX for package builders, as proposed by 12682 David Harris. 12683 [Bodo Moeller] 12684 12685 *) New Configure options "threads" and "no-threads". For systems 12686 where the proper compiler options are known (currently Solaris 12687 and Linux), "threads" is the default. 12688 [Bodo Moeller] 12689 12690 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 12691 [Bodo Moeller] 12692 12693 *) Install various scripts to $(OPENSSLDIR)/misc, not to 12694 $(INSTALLTOP)/bin -- they shouldn't clutter directories 12695 such as /usr/local/bin. 12696 [Bodo Moeller] 12697 12698 *) "make linux-shared" to build shared libraries. 12699 [Niels Poppe <[email protected]>] 12700 12701 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 12702 [Ulf Möller] 12703 12704 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 12705 extension adding in x509 utility. 12706 [Steve Henson] 12707 12708 *) Remove NOPROTO sections and error code comments. 12709 [Ulf Möller] 12710 12711 *) Partial rewrite of the DEF file generator to now parse the ANSI 12712 prototypes. 12713 [Steve Henson] 12714 12715 *) New Configure options --prefix=DIR and --openssldir=DIR. 12716 [Ulf Möller] 12717 12718 *) Complete rewrite of the error code script(s). It is all now handled 12719 by one script at the top level which handles error code gathering, 12720 header rewriting and C source file generation. It should be much better 12721 than the old method: it now uses a modified version of Ulf's parser to 12722 read the ANSI prototypes in all header files (thus the old K&R definitions 12723 aren't needed for error creation any more) and do a better job of 12724 translating function codes into names. The old 'ASN1 error code embedded 12725 in a comment' is no longer necessary and it doesn't use .err files which 12726 have now been deleted. Also the error code call doesn't have to appear all 12727 on one line (which resulted in some large lines...). 12728 [Steve Henson] 12729 12730 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 12731 [Bodo Moeller] 12732 12733 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 12734 0 (which usually indicates a closed connection), but continue reading. 12735 [Bodo Moeller] 12736 12737 *) Fix some race conditions. 12738 [Bodo Moeller] 12739 12740 *) Add support for CRL distribution points extension. Add Certificate 12741 Policies and CRL distribution points documentation. 12742 [Steve Henson] 12743 12744 *) Move the autogenerated header file parts to crypto/opensslconf.h. 12745 [Ulf Möller] 12746 12747 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 12748 8 of keying material. Merlin has also confirmed interop with this fix 12749 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 12750 [Merlin Hughes <[email protected]>] 12751 12752 *) Fix lots of warnings. 12753 [Richard Levitte <[email protected]>] 12754 12755 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 12756 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 12757 [Richard Levitte <[email protected]>] 12758 12759 *) Fix problems with sizeof(long) == 8. 12760 [Andy Polyakov <[email protected]>] 12761 12762 *) Change functions to ANSI C. 12763 [Ulf Möller] 12764 12765 *) Fix typos in error codes. 12766 [Martin Kraemer <[email protected]>, Ulf Möller] 12767 12768 *) Remove defunct assembler files from Configure. 12769 [Ulf Möller] 12770 12771 *) SPARC v8 assembler BIGNUM implementation. 12772 [Andy Polyakov <[email protected]>] 12773 12774 *) Support for Certificate Policies extension: both print and set. 12775 Various additions to support the r2i method this uses. 12776 [Steve Henson] 12777 12778 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 12779 return a const string when you are expecting an allocated buffer. 12780 [Ben Laurie] 12781 12782 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 12783 types DirectoryString and DisplayText. 12784 [Steve Henson] 12785 12786 *) Add code to allow r2i extensions to access the configuration database, 12787 add an LHASH database driver and add several ctx helper functions. 12788 [Steve Henson] 12789 12790 *) Fix an evil bug in bn_expand2() which caused various BN functions to 12791 fail when they extended the size of a BIGNUM. 12792 [Steve Henson] 12793 12794 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 12795 support typesafe stack. 12796 [Steve Henson] 12797 12798 *) Fix typo in SSL_[gs]et_options(). 12799 [Nils Frostberg <[email protected]>] 12800 12801 *) Delete various functions and files that belonged to the (now obsolete) 12802 old X509V3 handling code. 12803 [Steve Henson] 12804 12805 *) New Configure option "rsaref". 12806 [Ulf Möller] 12807 12808 *) Don't auto-generate pem.h. 12809 [Bodo Moeller] 12810 12811 *) Introduce type-safe ASN.1 SETs. 12812 [Ben Laurie] 12813 12814 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 12815 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 12816 12817 *) Introduce type-safe STACKs. This will almost certainly break lots of code 12818 that links with OpenSSL (well at least cause lots of warnings), but fear 12819 not: the conversion is trivial, and it eliminates loads of evil casts. A 12820 few STACKed things have been converted already. Feel free to convert more. 12821 In the fullness of time, I'll do away with the STACK type altogether. 12822 [Ben Laurie] 12823 12824 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 12825 specified in <certfile> by updating the entry in the index.txt file. 12826 This way one no longer has to edit the index.txt file manually for 12827 revoking a certificate. The -revoke option does the gory details now. 12828 [Massimiliano Pala <[email protected]>, Ralf S. Engelschall] 12829 12830 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 12831 `-text' option at all and this way the `-noout -text' combination was 12832 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 12833 [Ralf S. Engelschall] 12834 12835 *) Make sure a corresponding plain text error message exists for the 12836 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 12837 verify callback function determined that a certificate was revoked. 12838 [Ralf S. Engelschall] 12839 12840 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 12841 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 12842 all available ciphers including rc5, which was forgotten until now. 12843 In order to let the testing shell script know which algorithms 12844 are available, a new (up to now undocumented) command 12845 "openssl list-cipher-commands" is used. 12846 [Bodo Moeller] 12847 12848 *) Bugfix: s_client occasionally would sleep in select() when 12849 it should have checked SSL_pending() first. 12850 [Bodo Moeller] 12851 12852 *) New functions DSA_do_sign and DSA_do_verify to provide access to 12853 the raw DSA values prior to ASN.1 encoding. 12854 [Ulf Möller] 12855 12856 *) Tweaks to Configure 12857 [Niels Poppe <[email protected]>] 12858 12859 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 12860 yet... 12861 [Steve Henson] 12862 12863 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 12864 [Ulf Möller] 12865 12866 *) New config option to avoid instructions that are illegal on the 80386. 12867 The default code is faster, but requires at least a 486. 12868 [Ulf Möller] 12869 12870 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 12871 SSL2_SERVER_VERSION (not used at all) macros, which are now the 12872 same as SSL2_VERSION anyway. 12873 [Bodo Moeller] 12874 12875 *) New "-showcerts" option for s_client. 12876 [Bodo Moeller] 12877 12878 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 12879 application. Various cleanups and fixes. 12880 [Steve Henson] 12881 12882 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 12883 modify error routines to work internally. Add error codes and PBE init 12884 to library startup routines. 12885 [Steve Henson] 12886 12887 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 12888 packing functions to asn1 and evp. Changed function names and error 12889 codes along the way. 12890 [Steve Henson] 12891 12892 *) PKCS12 integration: and so it begins... First of several patches to 12893 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 12894 objects to objects.h 12895 [Steve Henson] 12896 12897 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 12898 and display support for Thawte strong extranet extension. 12899 [Steve Henson] 12900 12901 *) Add LinuxPPC support. 12902 [Jeff Dubrule <[email protected]>] 12903 12904 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 12905 bn_div_words in alpha.s. 12906 [Hannes Reinecke <[email protected]> and Ben Laurie] 12907 12908 *) Make sure the RSA OAEP test is skipped under -DRSAref because 12909 OAEP isn't supported when OpenSSL is built with RSAref. 12910 [Ulf Moeller <[email protected]>] 12911 12912 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 12913 so they no longer are missing under -DNOPROTO. 12914 [Soren S. Jorvang <[email protected]>] 12915 12916 12917 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 12918 12919 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 12920 doesn't work when the session is reused. Coming soon! 12921 [Ben Laurie] 12922 12923 *) Fix a security hole, that allows sessions to be reused in the wrong 12924 context thus bypassing client cert protection! All software that uses 12925 client certs and session caches in multiple contexts NEEDS PATCHING to 12926 allow session reuse! A fuller solution is in the works. 12927 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 12928 12929 *) Some more source tree cleanups (removed obsolete files 12930 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 12931 permission on "config" script to be executable) and a fix for the INSTALL 12932 document. 12933 [Ulf Moeller <[email protected]>] 12934 12935 *) Remove some legacy and erroneous uses of malloc, free instead of 12936 Malloc, Free. 12937 [Lennart Bang <[email protected]>, with minor changes by Steve] 12938 12939 *) Make rsa_oaep_test return non-zero on error. 12940 [Ulf Moeller <[email protected]>] 12941 12942 *) Add support for native Solaris shared libraries. Configure 12943 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 12944 if someone would make that last step automatic. 12945 [Matthias Loepfe <[email protected]>] 12946 12947 *) ctx_size was not built with the right compiler during "make links". Fixed. 12948 [Ben Laurie] 12949 12950 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 12951 except NULL ciphers". This means the default cipher list will no longer 12952 enable NULL ciphers. They need to be specifically enabled e.g. with 12953 the string "DEFAULT:eNULL". 12954 [Steve Henson] 12955 12956 *) Fix to RSA private encryption routines: if p < q then it would 12957 occasionally produce an invalid result. This will only happen with 12958 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 12959 [Steve Henson] 12960 12961 *) Be less restrictive and allow also `perl util/perlpath.pl 12962 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 12963 because this way one can also use an interpreter named `perl5' (which is 12964 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 12965 installed as `perl'). 12966 [Matthias Loepfe <[email protected]>] 12967 12968 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 12969 [Matthias Loepfe <[email protected]>] 12970 12971 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 12972 advapi32.lib to Win32 build and change the pem test comparison 12973 to fc.exe (thanks to Ulrich Kroener <[email protected]> for the 12974 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 12975 and crypto/des/ede_cbcm_enc.c. 12976 [Steve Henson] 12977 12978 *) DES quad checksum was broken on big-endian architectures. Fixed. 12979 [Ben Laurie] 12980 12981 *) Comment out two functions in bio.h that aren't implemented. Fix up the 12982 Win32 test batch file so it (might) work again. The Win32 test batch file 12983 is horrible: I feel ill.... 12984 [Steve Henson] 12985 12986 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 12987 in e_os.h. Audit of header files to check ANSI and non ANSI 12988 sections: 10 functions were absent from non ANSI section and not exported 12989 from Windows DLLs. Fixed up libeay.num for new functions. 12990 [Steve Henson] 12991 12992 *) Make `openssl version' output lines consistent. 12993 [Ralf S. Engelschall] 12994 12995 *) Fix Win32 symbol export lists for BIO functions: Added 12996 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 12997 to ms/libeay{16,32}.def. 12998 [Ralf S. Engelschall] 12999 13000 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 13001 fine under Unix and passes some trivial tests I've now added. But the 13002 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 13003 added to make sure no one expects that this stuff really works in the 13004 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 13005 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 13006 openssl_bio.xs. 13007 [Ralf S. Engelschall] 13008 13009 *) Fix the generation of two part addresses in perl. 13010 [Kenji Miyake <[email protected]>, integrated by Ben Laurie] 13011 13012 *) Add config entry for Linux on MIPS. 13013 [John Tobey <[email protected]>] 13014 13015 *) Make links whenever Configure is run, unless we are on Windoze. 13016 [Ben Laurie] 13017 13018 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 13019 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 13020 in CRLs. 13021 [Steve Henson] 13022 13023 *) Add a useful kludge to allow package maintainers to specify compiler and 13024 other platforms details on the command line without having to patch the 13025 Configure script every time: One now can use ``perl Configure 13026 <id>:<details>'', i.e. platform ids are allowed to have details appended 13027 to them (separated by colons). This is treated as there would be a static 13028 pre-configured entry in Configure's %table under key <id> with value 13029 <details> and ``perl Configure <id>'' is called. So, when you want to 13030 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 13031 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 13032 now, which overrides the FreeBSD-elf entry on-the-fly. 13033 [Ralf S. Engelschall] 13034 13035 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 13036 [Ben Laurie] 13037 13038 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 13039 on the `perl Configure ...' command line. This way one can compile 13040 OpenSSL libraries with Position Independent Code (PIC) which is needed 13041 for linking it into DSOs. 13042 [Ralf S. Engelschall] 13043 13044 *) Remarkably, export ciphers were totally broken and no-one had noticed! 13045 Fixed. 13046 [Ben Laurie] 13047 13048 *) Cleaned up the LICENSE document: The official contact for any license 13049 questions now is the OpenSSL core team under [email protected]. 13050 And add a paragraph about the dual-license situation to make sure people 13051 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 13052 to the OpenSSL toolkit. 13053 [Ralf S. Engelschall] 13054 13055 *) General source tree makefile cleanups: Made `making xxx in yyy...' 13056 display consistent in the source tree and replaced `/bin/rm' by `rm'. 13057 Additionally cleaned up the `make links' target: Remove unnecessary 13058 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 13059 to speed processing and no longer clutter the display with confusing 13060 stuff. Instead only the actually done links are displayed. 13061 [Ralf S. Engelschall] 13062 13063 *) Permit null encryption ciphersuites, used for authentication only. It used 13064 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 13065 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 13066 encryption. 13067 [Ben Laurie] 13068 13069 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 13070 signed attributes when verifying signatures (this would break them), 13071 the detached data encoding was wrong and public keys obtained using 13072 X509_get_pubkey() weren't freed. 13073 [Steve Henson] 13074 13075 *) Add text documentation for the BUFFER functions. Also added a work around 13076 to a Win95 console bug. This was triggered by the password read stuff: the 13077 last character typed gets carried over to the next fread(). If you were 13078 generating a new cert request using 'req' for example then the last 13079 character of the passphrase would be CR which would then enter the first 13080 field as blank. 13081 [Steve Henson] 13082 13083 *) Added the new `Includes OpenSSL Cryptography Software' button as 13084 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 13085 button and can be used by applications based on OpenSSL to show the 13086 relationship to the OpenSSL project. 13087 [Ralf S. Engelschall] 13088 13089 *) Remove confusing variables in function signatures in files 13090 ssl/ssl_lib.c and ssl/ssl.h. 13091 [Lennart Bong <[email protected]>] 13092 13093 *) Don't install bss_file.c under PREFIX/include/ 13094 [Lennart Bong <[email protected]>] 13095 13096 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 13097 functions that return function pointers and has support for NT specific 13098 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 13099 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 13100 unsigned to signed types: this was killing the Win32 compile. 13101 [Steve Henson] 13102 13103 *) Add new certificate file to stack functions, 13104 SSL_add_dir_cert_subjects_to_stack() and 13105 SSL_add_file_cert_subjects_to_stack(). These largely supplant 13106 SSL_load_client_CA_file(), and can be used to add multiple certs easily 13107 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 13108 This means that Apache-SSL and similar packages don't have to mess around 13109 to add as many CAs as they want to the preferred list. 13110 [Ben Laurie] 13111 13112 *) Experiment with doxygen documentation. Currently only partially applied to 13113 ssl/ssl_lib.c. 13114 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 13115 openssl.doxy as the configuration file. 13116 [Ben Laurie] 13117 13118 *) Get rid of remaining C++-style comments which strict C compilers hate. 13119 [Ralf S. Engelschall, pointed out by Carlos Amengual] 13120 13121 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 13122 compiled in by default: it has problems with large keys. 13123 [Steve Henson] 13124 13125 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 13126 DH private keys and/or callback functions which directly correspond to 13127 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 13128 is needed for applications which have to configure certificates on a 13129 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 13130 (e.g. s_server). 13131 For the RSA certificate situation is makes no difference, but 13132 for the DSA certificate situation this fixes the "no shared cipher" 13133 problem where the OpenSSL cipher selection procedure failed because the 13134 temporary keys were not overtaken from the context and the API provided 13135 no way to reconfigure them. 13136 The new functions now let applications reconfigure the stuff and they 13137 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 13138 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 13139 non-public-API function ssl_cert_instantiate() is used as a helper 13140 function and also to reduce code redundancy inside ssl_rsa.c. 13141 [Ralf S. Engelschall] 13142 13143 *) Move s_server -dcert and -dkey options out of the undocumented feature 13144 area because they are useful for the DSA situation and should be 13145 recognized by the users. 13146 [Ralf S. Engelschall] 13147 13148 *) Fix the cipher decision scheme for export ciphers: the export bits are 13149 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 13150 SSL_EXP_MASK. So, the original variable has to be used instead of the 13151 already masked variable. 13152 [Richard Levitte <[email protected]>] 13153 13154 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 13155 [Richard Levitte <[email protected]>] 13156 13157 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 13158 from `int' to `unsigned int' because it's a length and initialized by 13159 EVP_DigestFinal() which expects an `unsigned int *'. 13160 [Richard Levitte <[email protected]>] 13161 13162 *) Don't hard-code path to Perl interpreter on shebang line of Configure 13163 script. Instead use the usual Shell->Perl transition trick. 13164 [Ralf S. Engelschall] 13165 13166 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 13167 (in addition to RSA certificates) to match the behaviour of `openssl dsa 13168 -noout -modulus' as it's already the case for `openssl rsa -noout 13169 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 13170 currently the public key is printed (a decision which was already done by 13171 `openssl dsa -modulus' in the past) which serves a similar purpose. 13172 Additionally the NO_RSA no longer completely removes the whole -modulus 13173 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 13174 now, too. 13175 [Ralf S. Engelschall] 13176 13177 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 13178 BIO. See the source (crypto/evp/bio_ok.c) for more info. 13179 [Arne Ansper <[email protected]>] 13180 13181 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 13182 to be added. Now both 'req' and 'ca' can use new objects defined in the 13183 config file. 13184 [Steve Henson] 13185 13186 *) Add cool BIO that does syslog (or event log on NT). 13187 [Arne Ansper <[email protected]>, integrated by Ben Laurie] 13188 13189 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 13190 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 13191 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 13192 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 13193 [Ben Laurie] 13194 13195 *) Add preliminary config info for new extension code. 13196 [Steve Henson] 13197 13198 *) Make RSA_NO_PADDING really use no padding. 13199 [Ulf Moeller <[email protected]>] 13200 13201 *) Generate errors when private/public key check is done. 13202 [Ben Laurie] 13203 13204 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 13205 for some CRL extensions and new objects added. 13206 [Steve Henson] 13207 13208 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 13209 key usage extension and fuller support for authority key id. 13210 [Steve Henson] 13211 13212 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 13213 padding method for RSA, which is recommended for new applications in PKCS 13214 #1 v2.0 (RFC 2437, October 1998). 13215 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 13216 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 13217 against Bleichbacher's attack on RSA. 13218 [Ulf Moeller <[email protected]>, reformatted, corrected and integrated by 13219 Ben Laurie] 13220 13221 *) Updates to the new SSL compression code 13222 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13223 13224 *) Fix so that the version number in the master secret, when passed 13225 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 13226 (because the server will not accept higher), that the version number 13227 is 0x03,0x01, not 0x03,0x00 13228 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13229 13230 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 13231 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 13232 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 13233 [Steve Henson] 13234 13235 *) Support for RAW extensions where an arbitrary extension can be 13236 created by including its DER encoding. See apps/openssl.cnf for 13237 an example. 13238 [Steve Henson] 13239 13240 *) Make sure latest Perl versions don't interpret some generated C array 13241 code as Perl array code in the crypto/err/err_genc.pl script. 13242 [Lars Weber <[email protected]>] 13243 13244 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 13245 not many people have the assembler. Various Win32 compilation fixes and 13246 update to the INSTALL.W32 file with (hopefully) more accurate Win32 13247 build instructions. 13248 [Steve Henson] 13249 13250 *) Modify configure script 'Configure' to automatically create crypto/date.h 13251 file under Win32 and also build pem.h from pem.org. New script 13252 util/mkfiles.pl to create the MINFO file on environments that can't do a 13253 'make files': perl util/mkfiles.pl >MINFO should work. 13254 [Steve Henson] 13255 13256 *) Major rework of DES function declarations, in the pursuit of correctness 13257 and purity. As a result, many evil casts evaporated, and some weirdness, 13258 too. You may find this causes warnings in your code. Zapping your evil 13259 casts will probably fix them. Mostly. 13260 [Ben Laurie] 13261 13262 *) Fix for a typo in asn1.h. Bug fix to object creation script 13263 obj_dat.pl. It considered a zero in an object definition to mean 13264 "end of object": none of the objects in objects.h have any zeros 13265 so it wasn't spotted. 13266 [Steve Henson, reported by Erwann ABALEA <[email protected]>] 13267 13268 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 13269 Masking (CBCM). In the absence of test vectors, the best I have been able 13270 to do is check that the decrypt undoes the encrypt, so far. Send me test 13271 vectors if you have them. 13272 [Ben Laurie] 13273 13274 *) Correct calculation of key length for export ciphers (too much space was 13275 allocated for null ciphers). This has not been tested! 13276 [Ben Laurie] 13277 13278 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 13279 message is now correct (it understands "crypto" and "ssl" on its 13280 command line). There is also now an "update" option. This will update 13281 the util/ssleay.num and util/libeay.num files with any new functions. 13282 If you do a: 13283 perl util/mkdef.pl crypto ssl update 13284 it will update them. 13285 [Steve Henson] 13286 13287 *) Overhauled the Perl interface (perl/*): 13288 - ported BN stuff to OpenSSL's different BN library 13289 - made the perl/ source tree CVS-aware 13290 - renamed the package from SSLeay to OpenSSL (the files still contain 13291 their history because I've copied them in the repository) 13292 - removed obsolete files (the test scripts will be replaced 13293 by better Test::Harness variants in the future) 13294 [Ralf S. Engelschall] 13295 13296 *) First cut for a very conservative source tree cleanup: 13297 1. merge various obsolete readme texts into doc/ssleay.txt 13298 where we collect the old documents and readme texts. 13299 2. remove the first part of files where I'm already sure that we no 13300 longer need them because of three reasons: either they are just temporary 13301 files which were left by Eric or they are preserved original files where 13302 I've verified that the diff is also available in the CVS via "cvs diff 13303 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 13304 the crypto/md/ stuff). 13305 [Ralf S. Engelschall] 13306 13307 *) More extension code. Incomplete support for subject and issuer alt 13308 name, issuer and authority key id. Change the i2v function parameters 13309 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 13310 what that's for :-) Fix to ASN1 macro which messed up 13311 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 13312 [Steve Henson] 13313 13314 *) Preliminary support for ENUMERATED type. This is largely copied from the 13315 INTEGER code. 13316 [Steve Henson] 13317 13318 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 13319 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13320 13321 *) Make sure `make rehash' target really finds the `openssl' program. 13322 [Ralf S. Engelschall, Matthias Loepfe <[email protected]>] 13323 13324 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 13325 like to hear about it if this slows down other processors. 13326 [Ben Laurie] 13327 13328 *) Add CygWin32 platform information to Configure script. 13329 [Alan Batie <[email protected]>] 13330 13331 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 13332 [Rainer W. Gerling <[email protected]>] 13333 13334 *) New program nseq to manipulate netscape certificate sequences 13335 [Steve Henson] 13336 13337 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 13338 few typos. 13339 [Steve Henson] 13340 13341 *) Fixes to BN code. Previously the default was to define BN_RECURSION 13342 but the BN code had some problems that would cause failures when 13343 doing certificate verification and some other functions. 13344 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13345 13346 *) Add ASN1 and PEM code to support netscape certificate sequences. 13347 [Steve Henson] 13348 13349 *) Add ASN1 and PEM code to support netscape certificate sequences. 13350 [Steve Henson] 13351 13352 *) Add several PKIX and private extended key usage OIDs. 13353 [Steve Henson] 13354 13355 *) Modify the 'ca' program to handle the new extension code. Modify 13356 openssl.cnf for new extension format, add comments. 13357 [Steve Henson] 13358 13359 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 13360 and add a sample to openssl.cnf so req -x509 now adds appropriate 13361 CA extensions. 13362 [Steve Henson] 13363 13364 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 13365 error code, add initial support to X509_print() and x509 application. 13366 [Steve Henson] 13367 13368 *) Takes a deep breath and start adding X509 V3 extension support code. Add 13369 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 13370 stuff is currently isolated and isn't even compiled yet. 13371 [Steve Henson] 13372 13373 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 13374 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 13375 Removed the versions check from X509 routines when loading extensions: 13376 this allows certain broken certificates that don't set the version 13377 properly to be processed. 13378 [Steve Henson] 13379 13380 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 13381 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 13382 can still be regenerated with "make depend". 13383 [Ben Laurie] 13384 13385 *) Spelling mistake in C version of CAST-128. 13386 [Ben Laurie, reported by Jeremy Hylton <[email protected]>] 13387 13388 *) Changes to the error generation code. The perl script err-code.pl 13389 now reads in the old error codes and retains the old numbers, only 13390 adding new ones if necessary. It also only changes the .err files if new 13391 codes are added. The makefiles have been modified to only insert errors 13392 when needed (to avoid needlessly modifying header files). This is done 13393 by only inserting errors if the .err file is newer than the auto generated 13394 C file. To rebuild all the error codes from scratch (the old behaviour) 13395 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 13396 or delete all the .err files. 13397 [Steve Henson] 13398 13399 *) CAST-128 was incorrectly implemented for short keys. The C version has 13400 been fixed, but is untested. The assembler versions are also fixed, but 13401 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 13402 to regenerate it if needed. 13403 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 13404 Hagino <[email protected]>] 13405 13406 *) File was opened incorrectly in randfile.c. 13407 [Ulf Möller <[email protected]>] 13408 13409 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 13410 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 13411 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 13412 al: it's just almost always a UTCTime. Note this patch adds new error 13413 codes so do a "make errors" if there are problems. 13414 [Steve Henson] 13415 13416 *) Correct Linux 1 recognition in config. 13417 [Ulf Möller <[email protected]>] 13418 13419 *) Remove pointless MD5 hash when using DSA keys in ca. 13420 [Anonymous <[email protected]>] 13421 13422 *) Generate an error if given an empty string as a cert directory. Also 13423 generate an error if handed NULL (previously returned 0 to indicate an 13424 error, but didn't set one). 13425 [Ben Laurie, reported by Anonymous <[email protected]>] 13426 13427 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 13428 [Ben Laurie] 13429 13430 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 13431 parameters. This was causing a warning which killed off the Win32 compile. 13432 [Steve Henson] 13433 13434 *) Remove C++ style comments from crypto/bn/bn_local.h. 13435 [Neil Costigan <[email protected]>] 13436 13437 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 13438 based on a text string, looking up short and long names and finally 13439 "dot" format. The "dot" format stuff didn't work. Added new function 13440 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 13441 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 13442 OID is not part of the table. 13443 [Steve Henson] 13444 13445 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 13446 X509_LOOKUP_by_alias(). 13447 [Ben Laurie] 13448 13449 *) Sort openssl functions by name. 13450 [Ben Laurie] 13451 13452 *) Get the gendsa program working (hopefully) and add it to app list. Remove 13453 encryption from sample DSA keys (in case anyone is interested the password 13454 was "1234"). 13455 [Steve Henson] 13456 13457 *) Make _all_ *_free functions accept a NULL pointer. 13458 [Frans Heymans <[email protected]>] 13459 13460 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 13461 NULL pointers. 13462 [Anonymous <[email protected]>] 13463 13464 *) s_server should send the CAfile as acceptable CAs, not its own cert. 13465 [Bodo Moeller <[email protected]>] 13466 13467 *) Don't blow it for numeric -newkey arguments to apps/req. 13468 [Bodo Moeller <[email protected]>] 13469 13470 *) Temp key "for export" tests were wrong in s3_srvr.c. 13471 [Anonymous <[email protected]>] 13472 13473 *) Add prototype for temp key callback functions 13474 SSL_CTX_set_tmp_{rsa,dh}_callback(). 13475 [Ben Laurie] 13476 13477 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 13478 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 13479 [Steve Henson] 13480 13481 *) X509_name_add_entry() freed the wrong thing after an error. 13482 [Arne Ansper <[email protected]>] 13483 13484 *) rsa_eay.c would attempt to free a NULL context. 13485 [Arne Ansper <[email protected]>] 13486 13487 *) BIO_s_socket() had a broken should_retry() on Windoze. 13488 [Arne Ansper <[email protected]>] 13489 13490 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 13491 [Arne Ansper <[email protected]>] 13492 13493 *) Make sure the already existing X509_STORE->depth variable is initialized 13494 in X509_STORE_new(), but document the fact that this variable is still 13495 unused in the certificate verification process. 13496 [Ralf S. Engelschall] 13497 13498 *) Fix the various library and apps files to free up pkeys obtained from 13499 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 13500 [Steve Henson] 13501 13502 *) Fix reference counting in X509_PUBKEY_get(). This makes 13503 demos/maurice/example2.c work, amongst others, probably. 13504 [Steve Henson and Ben Laurie] 13505 13506 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 13507 `openssl' and second, the shortcut symlinks for the `openssl <command>' 13508 are no longer created. This way we have a single and consistent command 13509 line interface `openssl <command>', similar to `cvs <command>'. 13510 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 13511 13512 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 13513 BIT STRING wrapper always have zero unused bits. 13514 [Steve Henson] 13515 13516 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 13517 [Steve Henson] 13518 13519 *) Make the top-level INSTALL documentation easier to understand. 13520 [Paul Sutton] 13521 13522 *) Makefiles updated to exit if an error occurs in a sub-directory 13523 make (including if user presses ^C) [Paul Sutton] 13524 13525 *) Make Montgomery context stuff explicit in RSA data structure. 13526 [Ben Laurie] 13527 13528 *) Fix build order of pem and err to allow for generated pem.h. 13529 [Ben Laurie] 13530 13531 *) Fix renumbering bug in X509_NAME_delete_entry(). 13532 [Ben Laurie] 13533 13534 *) Enhanced the err-ins.pl script so it makes the error library number 13535 global and can add a library name. This is needed for external ASN1 and 13536 other error libraries. 13537 [Steve Henson] 13538 13539 *) Fixed sk_insert which never worked properly. 13540 [Steve Henson] 13541 13542 *) Fix ASN1 macros so they can handle indefinite length constructed 13543 EXPLICIT tags. Some non standard certificates use these: they can now 13544 be read in. 13545 [Steve Henson] 13546 13547 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 13548 into a single doc/ssleay.txt bundle. This way the information is still 13549 preserved but no longer messes up this directory. Now it's new room for 13550 the new set of documentation files. 13551 [Ralf S. Engelschall] 13552 13553 *) SETs were incorrectly DER encoded. This was a major pain, because they 13554 shared code with SEQUENCEs, which aren't coded the same. This means that 13555 almost everything to do with SETs or SEQUENCEs has either changed name or 13556 number of arguments. 13557 [Ben Laurie, based on a partial fix by GP Jayan <[email protected]>] 13558 13559 *) Fix test data to work with the above. 13560 [Ben Laurie] 13561 13562 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 13563 was already fixed by Eric for 0.9.1 it seems. 13564 [Ben Laurie - pointed out by Ulf Möller <[email protected]>] 13565 13566 *) Autodetect FreeBSD3. 13567 [Ben Laurie] 13568 13569 *) Fix various bugs in Configure. This affects the following platforms: 13570 nextstep 13571 ncr-scde 13572 unixware-2.0 13573 unixware-2.0-pentium 13574 sco5-cc. 13575 [Ben Laurie] 13576 13577 *) Eliminate generated files from CVS. Reorder tests to regenerate files 13578 before they are needed. 13579 [Ben Laurie] 13580 13581 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 13582 [Ben Laurie] 13583 13584 13585 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 13586 13587 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 13588 changed SSLeay to OpenSSL in version strings. 13589 [Ralf S. Engelschall] 13590 13591 *) Some fixups to the top-level documents. 13592 [Paul Sutton] 13593 13594 *) Fixed the nasty bug where rsaref.h was not found under compile-time 13595 because the symlink to include/ was missing. 13596 [Ralf S. Engelschall] 13597 13598 *) Incorporated the popular no-RSA/DSA-only patches 13599 which allow to compile a RSA-free SSLeay. 13600 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 13601 13602 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 13603 when "ssleay" is still not found. 13604 [Ralf S. Engelschall] 13605 13606 *) Added more platforms to Configure: Cray T3E, HPUX 11, 13607 [Ralf S. Engelschall, Beckmann <[email protected]>] 13608 13609 *) Updated the README file. 13610 [Ralf S. Engelschall] 13611 13612 *) Added various .cvsignore files in the CVS repository subdirs 13613 to make a "cvs update" really silent. 13614 [Ralf S. Engelschall] 13615 13616 *) Recompiled the error-definition header files and added 13617 missing symbols to the Win32 linker tables. 13618 [Ralf S. Engelschall] 13619 13620 *) Cleaned up the top-level documents; 13621 o new files: CHANGES and LICENSE 13622 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 13623 o merged COPYRIGHT into LICENSE 13624 o removed obsolete TODO file 13625 o renamed MICROSOFT to INSTALL.W32 13626 [Ralf S. Engelschall] 13627 13628 *) Removed dummy files from the 0.9.1b source tree: 13629 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 13630 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 13631 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 13632 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 13633 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 13634 [Ralf S. Engelschall] 13635 13636 *) Added various platform portability fixes. 13637 [Mark J. Cox] 13638 13639 *) The Genesis of the OpenSSL rpject: 13640 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 13641 Young and Tim J. Hudson created while they were working for C2Net until 13642 summer 1998. 13643 [The OpenSSL Project] 13644 13645 13646 Changes between 0.9.0b and 0.9.1b [not released] 13647 13648 *) Updated a few CA certificates under certs/ 13649 [Eric A. Young] 13650 13651 *) Changed some BIGNUM api stuff. 13652 [Eric A. Young] 13653 13654 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 13655 DGUX x86, Linux Alpha, etc. 13656 [Eric A. Young] 13657 13658 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 13659 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 13660 available). 13661 [Eric A. Young] 13662 13663 *) Add -strparse option to asn1pars program which parses nested 13664 binary structures 13665 [Dr Stephen Henson <[email protected]>] 13666 13667 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 13668 [Eric A. Young] 13669 13670 *) DSA fix for "ca" program. 13671 [Eric A. Young] 13672 13673 *) Added "-genkey" option to "dsaparam" program. 13674 [Eric A. Young] 13675 13676 *) Added RIPE MD160 (rmd160) message digest. 13677 [Eric A. Young] 13678 13679 *) Added -a (all) option to "ssleay version" command. 13680 [Eric A. Young] 13681 13682 *) Added PLATFORM define which is the id given to Configure. 13683 [Eric A. Young] 13684 13685 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 13686 [Eric A. Young] 13687 13688 *) Extended the ASN.1 parser routines. 13689 [Eric A. Young] 13690 13691 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 13692 [Eric A. Young] 13693 13694 *) Added a BN_CTX to the BN library. 13695 [Eric A. Young] 13696 13697 *) Fixed the weak key values in DES library 13698 [Eric A. Young] 13699 13700 *) Changed API in EVP library for cipher aliases. 13701 [Eric A. Young] 13702 13703 *) Added support for RC2/64bit cipher. 13704 [Eric A. Young] 13705 13706 *) Converted the lhash library to the crypto/mem.c functions. 13707 [Eric A. Young] 13708 13709 *) Added more recognized ASN.1 object ids. 13710 [Eric A. Young] 13711 13712 *) Added more RSA padding checks for SSL/TLS. 13713 [Eric A. Young] 13714 13715 *) Added BIO proxy/filter functionality. 13716 [Eric A. Young] 13717 13718 *) Added extra_certs to SSL_CTX which can be used 13719 send extra CA certificates to the client in the CA cert chain sending 13720 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 13721 [Eric A. Young] 13722 13723 *) Now Fortezza is denied in the authentication phase because 13724 this is key exchange mechanism is not supported by SSLeay at all. 13725 [Eric A. Young] 13726 13727 *) Additional PKCS1 checks. 13728 [Eric A. Young] 13729 13730 *) Support the string "TLSv1" for all TLS v1 ciphers. 13731 [Eric A. Young] 13732 13733 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 13734 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 13735 [Eric A. Young] 13736 13737 *) Fixed a few memory leaks. 13738 [Eric A. Young] 13739 13740 *) Fixed various code and comment typos. 13741 [Eric A. Young] 13742 13743 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 13744 bytes sent in the client random. 13745 [Edward Bishop <[email protected]>] 13746