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 1.1.1m and 1.1.1n [15 Mar 2022] 11 12 *) Fixed a bug in the BN_mod_sqrt() function that can cause it to loop forever 13 for non-prime moduli. 14 15 Internally this function is used when parsing certificates that contain 16 elliptic curve public keys in compressed form or explicit elliptic curve 17 parameters with a base point encoded in compressed form. 18 19 It is possible to trigger the infinite loop by crafting a certificate that 20 has invalid explicit curve parameters. 21 22 Since certificate parsing happens prior to verification of the certificate 23 signature, any process that parses an externally supplied certificate may 24 thus be subject to a denial of service attack. The infinite loop can also 25 be reached when parsing crafted private keys as they can contain explicit 26 elliptic curve parameters. 27 28 Thus vulnerable situations include: 29 30 - TLS clients consuming server certificates 31 - TLS servers consuming client certificates 32 - Hosting providers taking certificates or private keys from customers 33 - Certificate authorities parsing certification requests from subscribers 34 - Anything else which parses ASN.1 elliptic curve parameters 35 36 Also any other applications that use the BN_mod_sqrt() where the attacker 37 can control the parameter values are vulnerable to this DoS issue. 38 (CVE-2022-0778) 39 [Tomáš Mráz] 40 41 *) Add ciphersuites based on DHE_PSK (RFC 4279) and ECDHE_PSK (RFC 5489) 42 to the list of ciphersuites providing Perfect Forward Secrecy as 43 required by SECLEVEL >= 3. 44 45 [Dmitry Belyavskiy, Nicola Tuveri] 46 47 Changes between 1.1.1l and 1.1.1m [14 Dec 2021] 48 49 *) Avoid loading of a dynamic engine twice. 50 51 [Bernd Edlinger] 52 53 *) Fixed building on Debian with kfreebsd kernels 54 55 [Mattias Ellert] 56 57 *) Prioritise DANE TLSA issuer certs over peer certs 58 59 [Viktor Dukhovni] 60 61 *) Fixed random API for MacOS prior to 10.12 62 63 These MacOS versions don't support the CommonCrypto APIs 64 65 [Lenny Primak] 66 67 Changes between 1.1.1k and 1.1.1l [24 Aug 2021] 68 69 *) Fixed an SM2 Decryption Buffer Overflow. 70 71 In order to decrypt SM2 encrypted data an application is expected to call the 72 API function EVP_PKEY_decrypt(). Typically an application will call this 73 function twice. The first time, on entry, the "out" parameter can be NULL and, 74 on exit, the "outlen" parameter is populated with the buffer size required to 75 hold the decrypted plaintext. The application can then allocate a sufficiently 76 sized buffer and call EVP_PKEY_decrypt() again, but this time passing a non-NULL 77 value for the "out" parameter. 78 79 A bug in the implementation of the SM2 decryption code means that the 80 calculation of the buffer size required to hold the plaintext returned by the 81 first call to EVP_PKEY_decrypt() can be smaller than the actual size required by 82 the second call. This can lead to a buffer overflow when EVP_PKEY_decrypt() is 83 called by the application a second time with a buffer that is too small. 84 85 A malicious attacker who is able present SM2 content for decryption to an 86 application could cause attacker chosen data to overflow the buffer by up to a 87 maximum of 62 bytes altering the contents of other data held after the 88 buffer, possibly changing application behaviour or causing the application to 89 crash. The location of the buffer is application dependent but is typically 90 heap allocated. 91 (CVE-2021-3711) 92 [Matt Caswell] 93 94 *) Fixed various read buffer overruns processing ASN.1 strings 95 96 ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING 97 structure which contains a buffer holding the string data and a field holding 98 the buffer length. This contrasts with normal C strings which are repesented as 99 a buffer for the string data which is terminated with a NUL (0) byte. 100 101 Although not a strict requirement, ASN.1 strings that are parsed using OpenSSL's 102 own "d2i" functions (and other similar parsing functions) as well as any string 103 whose value has been set with the ASN1_STRING_set() function will additionally 104 NUL terminate the byte array in the ASN1_STRING structure. 105 106 However, it is possible for applications to directly construct valid ASN1_STRING 107 structures which do not NUL terminate the byte array by directly setting the 108 "data" and "length" fields in the ASN1_STRING array. This can also happen by 109 using the ASN1_STRING_set0() function. 110 111 Numerous OpenSSL functions that print ASN.1 data have been found to assume that 112 the ASN1_STRING byte array will be NUL terminated, even though this is not 113 guaranteed for strings that have been directly constructed. Where an application 114 requests an ASN.1 structure to be printed, and where that ASN.1 structure 115 contains ASN1_STRINGs that have been directly constructed by the application 116 without NUL terminating the "data" field, then a read buffer overrun can occur. 117 118 The same thing can also occur during name constraints processing of certificates 119 (for example if a certificate has been directly constructed by the application 120 instead of loading it via the OpenSSL parsing functions, and the certificate 121 contains non NUL terminated ASN1_STRING structures). It can also occur in the 122 X509_get1_email(), X509_REQ_get1_email() and X509_get1_ocsp() functions. 123 124 If a malicious actor can cause an application to directly construct an 125 ASN1_STRING and then process it through one of the affected OpenSSL functions 126 then this issue could be hit. This might result in a crash (causing a Denial of 127 Service attack). It could also result in the disclosure of private memory 128 contents (such as private keys, or sensitive plaintext). 129 (CVE-2021-3712) 130 [Matt Caswell] 131 132 Changes between 1.1.1j and 1.1.1k [25 Mar 2021] 133 134 *) Fixed a problem with verifying a certificate chain when using the 135 X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks 136 of the certificates present in a certificate chain. It is not set by 137 default. 138 139 Starting from OpenSSL version 1.1.1h a check to disallow certificates in 140 the chain that have explicitly encoded elliptic curve parameters was added 141 as an additional strict check. 142 143 An error in the implementation of this check meant that the result of a 144 previous check to confirm that certificates in the chain are valid CA 145 certificates was overwritten. This effectively bypasses the check 146 that non-CA certificates must not be able to issue other certificates. 147 148 If a "purpose" has been configured then there is a subsequent opportunity 149 for checks that the certificate is a valid CA. All of the named "purpose" 150 values implemented in libcrypto perform this check. Therefore, where 151 a purpose is set the certificate chain will still be rejected even when the 152 strict flag has been used. A purpose is set by default in libssl client and 153 server certificate verification routines, but it can be overridden or 154 removed by an application. 155 156 In order to be affected, an application must explicitly set the 157 X509_V_FLAG_X509_STRICT verification flag and either not set a purpose 158 for the certificate verification or, in the case of TLS client or server 159 applications, override the default purpose. 160 (CVE-2021-3450) 161 [Tomáš Mráz] 162 163 *) Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously 164 crafted renegotiation ClientHello message from a client. If a TLSv1.2 165 renegotiation ClientHello omits the signature_algorithms extension (where 166 it was present in the initial ClientHello), but includes a 167 signature_algorithms_cert extension then a NULL pointer dereference will 168 result, leading to a crash and a denial of service attack. 169 170 A server is only vulnerable if it has TLSv1.2 and renegotiation enabled 171 (which is the default configuration). OpenSSL TLS clients are not impacted 172 by this issue. 173 (CVE-2021-3449) 174 [Peter Kästle and Samuel Sapalski] 175 176 Changes between 1.1.1i and 1.1.1j [16 Feb 2021] 177 178 *) Fixed the X509_issuer_and_serial_hash() function. It attempts to 179 create a unique hash value based on the issuer and serial number data 180 contained within an X509 certificate. However it was failing to correctly 181 handle any errors that may occur while parsing the issuer field (which might 182 occur if the issuer field is maliciously constructed). This may subsequently 183 result in a NULL pointer deref and a crash leading to a potential denial of 184 service attack. 185 (CVE-2021-23841) 186 [Matt Caswell] 187 188 *) Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING 189 padding mode to correctly check for rollback attacks. This is considered a 190 bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is 191 CVE-2021-23839. 192 [Matt Caswell] 193 194 *) Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate 195 functions. Previously they could overflow the output length argument in some 196 cases where the input length is close to the maximum permissable length for 197 an integer on the platform. In such cases the return value from the function 198 call would be 1 (indicating success), but the output length value would be 199 negative. This could cause applications to behave incorrectly or crash. 200 (CVE-2021-23840) 201 [Matt Caswell] 202 203 *) Fixed SRP_Calc_client_key so that it runs in constant time. The previous 204 implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This 205 could be exploited in a side channel attack to recover the password. Since 206 the attack is local host only this is outside of the current OpenSSL 207 threat model and therefore no CVE is assigned. 208 209 Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this 210 issue. 211 [Matt Caswell] 212 213 Changes between 1.1.1h and 1.1.1i [8 Dec 2020] 214 215 *) Fixed NULL pointer deref in the GENERAL_NAME_cmp function 216 This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME. 217 If an attacker can control both items being compared then this could lead 218 to a possible denial of service attack. OpenSSL itself uses the 219 GENERAL_NAME_cmp function for two purposes: 220 1) Comparing CRL distribution point names between an available CRL and a 221 CRL distribution point embedded in an X509 certificate 222 2) When verifying that a timestamp response token signer matches the 223 timestamp authority name (exposed via the API functions 224 TS_RESP_verify_response and TS_RESP_verify_token) 225 (CVE-2020-1971) 226 [Matt Caswell] 227 228 *) Add support for Apple Silicon M1 Macs with the darwin64-arm64-cc target. 229 [Stuart Carnie] 230 231 *) The security callback, which can be customised by application code, supports 232 the security operation SSL_SECOP_TMP_DH. This is defined to take an EVP_PKEY 233 in the "other" parameter. In most places this is what is passed. All these 234 places occur server side. However there was one client side call of this 235 security operation and it passed a DH object instead. This is incorrect 236 according to the definition of SSL_SECOP_TMP_DH, and is inconsistent with all 237 of the other locations. Therefore this client side call has been changed to 238 pass an EVP_PKEY instead. 239 [Matt Caswell] 240 241 *) In 1.1.1h, an expired trusted (root) certificate was not anymore rejected 242 when validating a certificate path. This check is restored in 1.1.1i. 243 [David von Oheimb] 244 245 Changes between 1.1.1g and 1.1.1h [22 Sep 2020] 246 247 *) Certificates with explicit curve parameters are now disallowed in 248 verification chains if the X509_V_FLAG_X509_STRICT flag is used. 249 [Tomas Mraz] 250 251 *) The 'MinProtocol' and 'MaxProtocol' configuration commands now silently 252 ignore TLS protocol version bounds when configuring DTLS-based contexts, and 253 conversely, silently ignore DTLS protocol version bounds when configuring 254 TLS-based contexts. The commands can be repeated to set bounds of both 255 types. The same applies with the corresponding "min_protocol" and 256 "max_protocol" command-line switches, in case some application uses both TLS 257 and DTLS. 258 259 SSL_CTX instances that are created for a fixed protocol version (e.g. 260 TLSv1_server_method()) also silently ignore version bounds. Previously 261 attempts to apply bounds to these protocol versions would result in an 262 error. Now only the "version-flexible" SSL_CTX instances are subject to 263 limits in configuration files in command-line options. 264 [Viktor Dukhovni] 265 266 *) Handshake now fails if Extended Master Secret extension is dropped 267 on renegotiation. 268 [Tomas Mraz] 269 270 *) Accidentally, an expired trusted (root) certificate is not anymore rejected 271 when validating a certificate path. 272 [David von Oheimb] 273 274 *) The Oracle Developer Studio compiler will start reporting deprecated APIs 275 276 Changes between 1.1.1f and 1.1.1g [21 Apr 2020] 277 278 *) Fixed segmentation fault in SSL_check_chain() 279 Server or client applications that call the SSL_check_chain() function 280 during or after a TLS 1.3 handshake may crash due to a NULL pointer 281 dereference as a result of incorrect handling of the 282 "signature_algorithms_cert" TLS extension. The crash occurs if an invalid 283 or unrecognised signature algorithm is received from the peer. This could 284 be exploited by a malicious peer in a Denial of Service attack. 285 (CVE-2020-1967) 286 [Benjamin Kaduk] 287 288 *) Added AES consttime code for no-asm configurations 289 an optional constant time support for AES was added 290 when building openssl for no-asm. 291 Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME 292 Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME 293 At this time this feature is by default disabled. 294 It will be enabled by default in 3.0. 295 [Bernd Edlinger] 296 297 Changes between 1.1.1e and 1.1.1f [31 Mar 2020] 298 299 *) Revert the change of EOF detection while reading in libssl to avoid 300 regressions in applications depending on the current way of reporting 301 the EOF. As the existing method is not fully accurate the change to 302 reporting the EOF via SSL_ERROR_SSL is kept on the current development 303 branch and will be present in the 3.0 release. 304 [Tomas Mraz] 305 306 *) Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1 307 when primes for RSA keys are computed. 308 Since we previously always generated primes == 2 (mod 3) for RSA keys, 309 the 2-prime and 3-prime RSA modules were easy to distinguish, since 310 N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting 311 2-prime vs. 3-prime RSA keys was possible by computing N mod 3. 312 This avoids possible fingerprinting of newly generated RSA modules. 313 [Bernd Edlinger] 314 315 Changes between 1.1.1d and 1.1.1e [17 Mar 2020] 316 *) Properly detect EOF while reading in libssl. Previously if we hit an EOF 317 while reading in libssl then we would report an error back to the 318 application (SSL_ERROR_SYSCALL) but errno would be 0. We now add 319 an error to the stack (which means we instead return SSL_ERROR_SSL) and 320 therefore give a hint as to what went wrong. 321 [Matt Caswell] 322 323 *) Check that ed25519 and ed448 are allowed by the security level. Previously 324 signature algorithms not using an MD were not being checked that they were 325 allowed by the security level. 326 [Kurt Roeckx] 327 328 *) Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername() 329 was not quite right. The behaviour was not consistent between resumption 330 and normal handshakes, and also not quite consistent with historical 331 behaviour. The behaviour in various scenarios has been clarified and 332 it has been updated to make it match historical behaviour as closely as 333 possible. 334 [Matt Caswell] 335 336 *) [VMS only] The header files that the VMS compilers include automatically, 337 __DECC_INCLUDE_PROLOGUE.H and __DECC_INCLUDE_EPILOGUE.H, use pragmas that 338 the C++ compiler doesn't understand. This is a shortcoming in the 339 compiler, but can be worked around with __cplusplus guards. 340 341 C++ applications that use OpenSSL libraries must be compiled using the 342 qualifier '/NAMES=(AS_IS,SHORTENED)' to be able to use all the OpenSSL 343 functions. Otherwise, only functions with symbols of less than 31 344 characters can be used, as the linker will not be able to successfully 345 resolve symbols with longer names. 346 [Richard Levitte] 347 348 *) Corrected the documentation of the return values from the EVP_DigestSign* 349 set of functions. The documentation mentioned negative values for some 350 errors, but this was never the case, so the mention of negative values 351 was removed. 352 353 Code that followed the documentation and thereby check with something 354 like 'EVP_DigestSignInit(...) <= 0' will continue to work undisturbed. 355 [Richard Levitte] 356 357 *) Fixed an an overflow bug in the x64_64 Montgomery squaring procedure 358 used in exponentiation with 512-bit moduli. No EC algorithms are 359 affected. Analysis suggests that attacks against 2-prime RSA1024, 360 3-prime RSA1536, and DSA1024 as a result of this defect would be very 361 difficult to perform and are not believed likely. Attacks against DH512 362 are considered just feasible. However, for an attack the target would 363 have to re-use the DH512 private key, which is not recommended anyway. 364 Also applications directly using the low level API BN_mod_exp may be 365 affected if they use BN_FLG_CONSTTIME. 366 (CVE-2019-1551) 367 [Andy Polyakov] 368 369 *) Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY. 370 The presence of this system service is determined at run-time. 371 [Richard Levitte] 372 373 *) Added newline escaping functionality to a filename when using openssl dgst. 374 This output format is to replicate the output format found in the '*sum' 375 checksum programs. This aims to preserve backward compatibility. 376 [Matt Eaton, Richard Levitte, and Paul Dale] 377 378 *) Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just 379 the first value. 380 [Jon Spillett] 381 382 Changes between 1.1.1c and 1.1.1d [10 Sep 2019] 383 384 *) Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random 385 number generator (RNG). This was intended to include protection in the 386 event of a fork() system call in order to ensure that the parent and child 387 processes did not share the same RNG state. However this protection was not 388 being used in the default case. 389 390 A partial mitigation for this issue is that the output from a high 391 precision timer is mixed into the RNG state so the likelihood of a parent 392 and child process sharing state is significantly reduced. 393 394 If an application already calls OPENSSL_init_crypto() explicitly using 395 OPENSSL_INIT_ATFORK then this problem does not occur at all. 396 (CVE-2019-1549) 397 [Matthias St. Pierre] 398 399 *) For built-in EC curves, ensure an EC_GROUP built from the curve name is 400 used even when parsing explicit parameters, when loading a serialized key 401 or calling `EC_GROUP_new_from_ecpkparameters()`/ 402 `EC_GROUP_new_from_ecparameters()`. 403 This prevents bypass of security hardening and performance gains, 404 especially for curves with specialized EC_METHODs. 405 By default, if a key encoded with explicit parameters is loaded and later 406 serialized, the output is still encoded with explicit parameters, even if 407 internally a "named" EC_GROUP is used for computation. 408 [Nicola Tuveri] 409 410 *) Compute ECC cofactors if not provided during EC_GROUP construction. Before 411 this change, EC_GROUP_set_generator would accept order and/or cofactor as 412 NULL. After this change, only the cofactor parameter can be NULL. It also 413 does some minimal sanity checks on the passed order. 414 (CVE-2019-1547) 415 [Billy Bob Brumley] 416 417 *) Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey. 418 An attack is simple, if the first CMS_recipientInfo is valid but the 419 second CMS_recipientInfo is chosen ciphertext. If the second 420 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct 421 encryption key will be replaced by garbage, and the message cannot be 422 decoded, but if the RSA decryption fails, the correct encryption key is 423 used and the recipient will not notice the attack. 424 As a work around for this potential attack the length of the decrypted 425 key must be equal to the cipher default key length, in case the 426 certifiate is not given and all recipientInfo are tried out. 427 The old behaviour can be re-enabled in the CMS code by setting the 428 CMS_DEBUG_DECRYPT flag. 429 (CVE-2019-1563) 430 [Bernd Edlinger] 431 432 *) Early start up entropy quality from the DEVRANDOM seed source has been 433 improved for older Linux systems. The RAND subsystem will wait for 434 /dev/random to be producing output before seeding from /dev/urandom. 435 The seeded state is stored for future library initialisations using 436 a system global shared memory segment. The shared memory identifier 437 can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to 438 the desired value. The default identifier is 114. 439 [Paul Dale] 440 441 *) Correct the extended master secret constant on EBCDIC systems. Without this 442 fix TLS connections between an EBCDIC system and a non-EBCDIC system that 443 negotiate EMS will fail. Unfortunately this also means that TLS connections 444 between EBCDIC systems with this fix, and EBCDIC systems without this 445 fix will fail if they negotiate EMS. 446 [Matt Caswell] 447 448 *) Use Windows installation paths in the mingw builds 449 450 Mingw isn't a POSIX environment per se, which means that Windows 451 paths should be used for installation. 452 (CVE-2019-1552) 453 [Richard Levitte] 454 455 *) Changed DH_check to accept parameters with order q and 2q subgroups. 456 With order 2q subgroups the bit 0 of the private key is not secret 457 but DH_generate_key works around that by clearing bit 0 of the 458 private key for those. This avoids leaking bit 0 of the private key. 459 [Bernd Edlinger] 460 461 *) Significantly reduce secure memory usage by the randomness pools. 462 [Paul Dale] 463 464 *) Revert the DEVRANDOM_WAIT feature for Linux systems 465 466 The DEVRANDOM_WAIT feature added a select() call to wait for the 467 /dev/random device to become readable before reading from the 468 /dev/urandom device. 469 470 It turned out that this change had negative side effects on 471 performance which were not acceptable. After some discussion it 472 was decided to revert this feature and leave it up to the OS 473 resp. the platform maintainer to ensure a proper initialization 474 during early boot time. 475 [Matthias St. Pierre] 476 477 Changes between 1.1.1b and 1.1.1c [28 May 2019] 478 479 *) Add build tests for C++. These are generated files that only do one 480 thing, to include one public OpenSSL head file each. This tests that 481 the public header files can be usefully included in a C++ application. 482 483 This test isn't enabled by default. It can be enabled with the option 484 'enable-buildtest-c++'. 485 [Richard Levitte] 486 487 *) Enable SHA3 pre-hashing for ECDSA and DSA. 488 [Patrick Steuer] 489 490 *) Change the default RSA, DSA and DH size to 2048 bit instead of 1024. 491 This changes the size when using the genpkey app when no size is given. It 492 fixes an omission in earlier changes that changed all RSA, DSA and DH 493 generation apps to use 2048 bits by default. 494 [Kurt Roeckx] 495 496 *) Reorganize the manual pages to consistently have RETURN VALUES, 497 EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust 498 util/fix-doc-nits accordingly. 499 [Paul Yang, Joshua Lock] 500 501 *) Add the missing accessor EVP_PKEY_get0_engine() 502 [Matt Caswell] 503 504 *) Have apps like 's_client' and 's_server' output the signature scheme 505 along with other cipher suite parameters when debugging. 506 [Lorinczy Zsigmond] 507 508 *) Make OPENSSL_config() error agnostic again. 509 [Richard Levitte] 510 511 *) Do the error handling in RSA decryption constant time. 512 [Bernd Edlinger] 513 514 *) Prevent over long nonces in ChaCha20-Poly1305. 515 516 ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input 517 for every encryption operation. RFC 7539 specifies that the nonce value 518 (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length 519 and front pads the nonce with 0 bytes if it is less than 12 520 bytes. However it also incorrectly allows a nonce to be set of up to 16 521 bytes. In this case only the last 12 bytes are significant and any 522 additional leading bytes are ignored. 523 524 It is a requirement of using this cipher that nonce values are 525 unique. Messages encrypted using a reused nonce value are susceptible to 526 serious confidentiality and integrity attacks. If an application changes 527 the default nonce length to be longer than 12 bytes and then makes a 528 change to the leading bytes of the nonce expecting the new value to be a 529 new unique nonce then such an application could inadvertently encrypt 530 messages with a reused nonce. 531 532 Additionally the ignored bytes in a long nonce are not covered by the 533 integrity guarantee of this cipher. Any application that relies on the 534 integrity of these ignored leading bytes of a long nonce may be further 535 affected. Any OpenSSL internal use of this cipher, including in SSL/TLS, 536 is safe because no such use sets such a long nonce value. However user 537 applications that use this cipher directly and set a non-default nonce 538 length to be longer than 12 bytes may be vulnerable. 539 540 This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk 541 Greef of Ronomon. 542 (CVE-2019-1543) 543 [Matt Caswell] 544 545 *) Add DEVRANDOM_WAIT feature for Linux systems 546 547 On older Linux systems where the getrandom() system call is not available, 548 OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG. 549 Contrary to getrandom(), the /dev/urandom device will not block during 550 early boot when the kernel CSPRNG has not been seeded yet. 551 552 To mitigate this known weakness, use select() to wait for /dev/random to 553 become readable before reading from /dev/urandom. 554 555 *) Ensure that SM2 only uses SM3 as digest algorithm 556 [Paul Yang] 557 558 Changes between 1.1.1a and 1.1.1b [26 Feb 2019] 559 560 *) Added SCA hardening for modular field inversion in EC_GROUP through 561 a new dedicated field_inv() pointer in EC_METHOD. 562 This also addresses a leakage affecting conversions from projective 563 to affine coordinates. 564 [Billy Bob Brumley, Nicola Tuveri] 565 566 *) Change the info callback signals for the start and end of a post-handshake 567 message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START 568 and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get 569 confused by this and assume that a TLSv1.2 renegotiation has started. This 570 can break KeyUpdate handling. Instead we no longer signal the start and end 571 of a post handshake message exchange (although the messages themselves are 572 still signalled). This could break some applications that were expecting 573 the old signals. However without this KeyUpdate is not usable for many 574 applications. 575 [Matt Caswell] 576 577 *) Fix a bug in the computation of the endpoint-pair shared secret used 578 by DTLS over SCTP. This breaks interoperability with older versions 579 of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime 580 switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling 581 interoperability with such broken implementations. However, enabling 582 this switch breaks interoperability with correct implementations. 583 584 *) Fix a use after free bug in d2i_X509_PUBKEY when overwriting a 585 re-used X509_PUBKEY object if the second PUBKEY is malformed. 586 [Bernd Edlinger] 587 588 *) Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0(). 589 [Richard Levitte] 590 591 *) Remove the 'dist' target and add a tarball building script. The 592 'dist' target has fallen out of use, and it shouldn't be 593 necessary to configure just to create a source distribution. 594 [Richard Levitte] 595 596 *) Added support for Linux Kernel TLS data-path. The Linux Kernel data-path 597 improves application performance by removing data copies and providing 598 applications with zero-copy system calls such as sendfile and splice. 599 [Boris Pismenny] 600 601 Changes between 1.1.1 and 1.1.1a [20 Nov 2018] 602 603 *) Timing vulnerability in DSA signature generation 604 605 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a 606 timing side channel attack. An attacker could use variations in the signing 607 algorithm to recover the private key. 608 609 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser. 610 (CVE-2018-0734) 611 [Paul Dale] 612 613 *) Timing vulnerability in ECDSA signature generation 614 615 The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a 616 timing side channel attack. An attacker could use variations in the signing 617 algorithm to recover the private key. 618 619 This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser. 620 (CVE-2018-0735) 621 [Paul Dale] 622 623 *) Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for 624 the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names 625 are retained for backwards compatibility. 626 [Antoine Salon] 627 628 *) Fixed the issue that RAND_add()/RAND_seed() silently discards random input 629 if its length exceeds 4096 bytes. The limit has been raised to a buffer size 630 of two gigabytes and the error handling improved. 631 632 This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been 633 categorized as a normal bug, not a security issue, because the DRBG reseeds 634 automatically and is fully functional even without additional randomness 635 provided by the application. 636 637 Changes between 1.1.0i and 1.1.1 [11 Sep 2018] 638 639 *) Add a new ClientHello callback. Provides a callback interface that gives 640 the application the ability to adjust the nascent SSL object at the 641 earliest stage of ClientHello processing, immediately after extensions have 642 been collected but before they have been processed. In particular, this 643 callback can adjust the supported TLS versions in response to the contents 644 of the ClientHello 645 [Benjamin Kaduk] 646 647 *) Add SM2 base algorithm support. 648 [Jack Lloyd] 649 650 *) s390x assembly pack: add (improved) hardware-support for the following 651 cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb, 652 aes-cfb/cfb8, aes-ecb. 653 [Patrick Steuer] 654 655 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 656 parameter is no longer accepted, as it leads to a corrupt table. NULL 657 pem_str is reserved for alias entries only. 658 [Richard Levitte] 659 660 *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 661 step for prime curves. The new implementation is based on formulae from 662 differential addition-and-doubling in homogeneous projective coordinates 663 from Izu-Takagi "A fast parallel elliptic curve multiplication resistant 664 against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves 665 and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified 666 to work in projective coordinates. 667 [Billy Bob Brumley, Nicola Tuveri] 668 669 *) Change generating and checking of primes so that the error rate of not 670 being prime depends on the intended use based on the size of the input. 671 For larger primes this will result in more rounds of Miller-Rabin. 672 The maximal error rate for primes with more than 1080 bits is lowered 673 to 2^-128. 674 [Kurt Roeckx, Annie Yousar] 675 676 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 677 [Kurt Roeckx] 678 679 *) The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when 680 moving between systems, and to avoid confusion when a Windows build is 681 done with mingw vs with MSVC. For POSIX installs, there's still a 682 symlink or copy named 'tsget' to avoid that confusion as well. 683 [Richard Levitte] 684 685 *) Revert blinding in ECDSA sign and instead make problematic addition 686 length-invariant. Switch even to fixed-length Montgomery multiplication. 687 [Andy Polyakov] 688 689 *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 690 step for binary curves. The new implementation is based on formulae from 691 differential addition-and-doubling in mixed Lopez-Dahab projective 692 coordinates, modified to independently blind the operands. 693 [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri] 694 695 *) Add a scaffold to optionally enhance the Montgomery ladder implementation 696 for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing 697 EC_METHODs to implement their own specialized "ladder step", to take 698 advantage of more favorable coordinate systems or more efficient 699 differential addition-and-doubling algorithms. 700 [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri] 701 702 *) Modified the random device based seed sources to keep the relevant 703 file descriptors open rather than reopening them on each access. 704 This allows such sources to operate in a chroot() jail without 705 the associated device nodes being available. This behaviour can be 706 controlled using RAND_keep_random_devices_open(). 707 [Paul Dale] 708 709 *) Numerous side-channel attack mitigations have been applied. This may have 710 performance impacts for some algorithms for the benefit of improved 711 security. Specific changes are noted in this change log by their respective 712 authors. 713 [Matt Caswell] 714 715 *) AIX shared library support overhaul. Switch to AIX "natural" way of 716 handling shared libraries, which means collecting shared objects of 717 different versions and bitnesses in one common archive. This allows to 718 mitigate conflict between 1.0 and 1.1 side-by-side installations. It 719 doesn't affect the way 3rd party applications are linked, only how 720 multi-version installation is managed. 721 [Andy Polyakov] 722 723 *) Make ec_group_do_inverse_ord() more robust and available to other 724 EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA 725 mitigations are applied to the fallback BN_mod_inverse(). 726 When using this function rather than BN_mod_inverse() directly, new 727 EC cryptosystem implementations are then safer-by-default. 728 [Billy Bob Brumley] 729 730 *) Add coordinate blinding for EC_POINT and implement projective 731 coordinate blinding for generic prime curves as a countermeasure to 732 chosen point SCA attacks. 733 [Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley] 734 735 *) Add blinding to ECDSA and DSA signatures to protect against side channel 736 attacks discovered by Keegan Ryan (NCC Group). 737 [Matt Caswell] 738 739 *) Enforce checking in the pkeyutl command line app to ensure that the input 740 length does not exceed the maximum supported digest length when performing 741 a sign, verify or verifyrecover operation. 742 [Matt Caswell] 743 744 *) SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking 745 I/O in combination with something like select() or poll() will hang. This 746 can be turned off again using SSL_CTX_clear_mode(). 747 Many applications do not properly handle non-application data records, and 748 TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works 749 around the problems in those applications, but can also break some. 750 It's recommended to read the manpages about SSL_read(), SSL_write(), 751 SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and 752 SSL_CTX_set_read_ahead() again. 753 [Kurt Roeckx] 754 755 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 756 now allow empty (zero character) pass phrases. 757 [Richard Levitte] 758 759 *) Apply blinding to binary field modular inversion and remove patent 760 pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation. 761 [Billy Bob Brumley] 762 763 *) Deprecate ec2_mult.c and unify scalar multiplication code paths for 764 binary and prime elliptic curves. 765 [Billy Bob Brumley] 766 767 *) Remove ECDSA nonce padding: EC_POINT_mul is now responsible for 768 constant time fixed point multiplication. 769 [Billy Bob Brumley] 770 771 *) Revise elliptic curve scalar multiplication with timing attack 772 defenses: ec_wNAF_mul redirects to a constant time implementation 773 when computing fixed point and variable point multiplication (which 774 in OpenSSL are mostly used with secret scalars in keygen, sign, 775 ECDH derive operations). 776 [Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García, 777 Sohaib ul Hassan] 778 779 *) Updated CONTRIBUTING 780 [Rich Salz] 781 782 *) Updated DRBG / RAND to request nonce and additional low entropy 783 randomness from the system. 784 [Matthias St. Pierre] 785 786 *) Updated 'openssl rehash' to use OpenSSL consistent default. 787 [Richard Levitte] 788 789 *) Moved the load of the ssl_conf module to libcrypto, which helps 790 loading engines that libssl uses before libssl is initialised. 791 [Matt Caswell] 792 793 *) Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA 794 [Matt Caswell] 795 796 *) Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases. 797 [Ingo Schwarze, Rich Salz] 798 799 *) Added output of accepting IP address and port for 'openssl s_server' 800 [Richard Levitte] 801 802 *) Added a new API for TLSv1.3 ciphersuites: 803 SSL_CTX_set_ciphersuites() 804 SSL_set_ciphersuites() 805 [Matt Caswell] 806 807 *) Memory allocation failures consistently add an error to the error 808 stack. 809 [Rich Salz] 810 811 *) Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values 812 in libcrypto when run as setuid/setgid. 813 [Bernd Edlinger] 814 815 *) Load any config file by default when libssl is used. 816 [Matt Caswell] 817 818 *) Added new public header file <openssl/rand_drbg.h> and documentation 819 for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview. 820 [Matthias St. Pierre] 821 822 *) QNX support removed (cannot find contributors to get their approval 823 for the license change). 824 [Rich Salz] 825 826 *) TLSv1.3 replay protection for early data has been implemented. See the 827 SSL_read_early_data() man page for further details. 828 [Matt Caswell] 829 830 *) Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite 831 configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and 832 below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3. 833 In order to avoid issues where legacy TLSv1.2 ciphersuite configuration 834 would otherwise inadvertently disable all TLSv1.3 ciphersuites the 835 configuration has been separated out. See the ciphers man page or the 836 SSL_CTX_set_ciphersuites() man page for more information. 837 [Matt Caswell] 838 839 *) On POSIX (BSD, Linux, ...) systems the ocsp(1) command running 840 in responder mode now supports the new "-multi" option, which 841 spawns the specified number of child processes to handle OCSP 842 requests. The "-timeout" option now also limits the OCSP 843 responder's patience to wait to receive the full client request 844 on a newly accepted connection. Child processes are respawned 845 as needed, and the CA index file is automatically reloaded 846 when changed. This makes it possible to run the "ocsp" responder 847 as a long-running service, making the OpenSSL CA somewhat more 848 feature-complete. In this mode, most diagnostic messages logged 849 after entering the event loop are logged via syslog(3) rather than 850 written to stderr. 851 [Viktor Dukhovni] 852 853 *) Added support for X448 and Ed448. Heavily based on original work by 854 Mike Hamburg. 855 [Matt Caswell] 856 857 *) Extend OSSL_STORE with capabilities to search and to narrow the set of 858 objects loaded. This adds the functions OSSL_STORE_expect() and 859 OSSL_STORE_find() as well as needed tools to construct searches and 860 get the search data out of them. 861 [Richard Levitte] 862 863 *) Support for TLSv1.3 added. Note that users upgrading from an earlier 864 version of OpenSSL should review their configuration settings to ensure 865 that they are still appropriate for TLSv1.3. For further information see: 866 https://wiki.openssl.org/index.php/TLS1.3 867 [Matt Caswell] 868 869 *) Grand redesign of the OpenSSL random generator 870 871 The default RAND method now utilizes an AES-CTR DRBG according to 872 NIST standard SP 800-90Ar1. The new random generator is essentially 873 a port of the default random generator from the OpenSSL FIPS 2.0 874 object module. It is a hybrid deterministic random bit generator 875 using an AES-CTR bit stream and which seeds and reseeds itself 876 automatically using trusted system entropy sources. 877 878 Some of its new features are: 879 o Support for multiple DRBG instances with seed chaining. 880 o The default RAND method makes use of a DRBG. 881 o There is a public and private DRBG instance. 882 o The DRBG instances are fork-safe. 883 o Keep all global DRBG instances on the secure heap if it is enabled. 884 o The public and private DRBG instance are per thread for lock free 885 operation 886 [Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre] 887 888 *) Changed Configure so it only says what it does and doesn't dump 889 so much data. Instead, ./configdata.pm should be used as a script 890 to display all sorts of configuration data. 891 [Richard Levitte] 892 893 *) Added processing of "make variables" to Configure. 894 [Richard Levitte] 895 896 *) Added SHA512/224 and SHA512/256 algorithm support. 897 [Paul Dale] 898 899 *) The last traces of Netware support, first removed in 1.1.0, have 900 now been removed. 901 [Rich Salz] 902 903 *) Get rid of Makefile.shared, and in the process, make the processing 904 of certain files (rc.obj, or the .def/.map/.opt files produced from 905 the ordinal files) more visible and hopefully easier to trace and 906 debug (or make silent). 907 [Richard Levitte] 908 909 *) Make it possible to have environment variable assignments as 910 arguments to config / Configure. 911 [Richard Levitte] 912 913 *) Add multi-prime RSA (RFC 8017) support. 914 [Paul Yang] 915 916 *) Add SM3 implemented according to GB/T 32905-2016 917 [ Jack Lloyd <jack.lloyd@ribose.com>, 918 Ronald Tse <ronald.tse@ribose.com>, 919 Erick Borsboom <erick.borsboom@ribose.com> ] 920 921 *) Add 'Maximum Fragment Length' TLS extension negotiation and support 922 as documented in RFC6066. 923 Based on a patch from Tomasz Moń 924 [Filipe Raimundo da Silva] 925 926 *) Add SM4 implemented according to GB/T 32907-2016. 927 [ Jack Lloyd <jack.lloyd@ribose.com>, 928 Ronald Tse <ronald.tse@ribose.com>, 929 Erick Borsboom <erick.borsboom@ribose.com> ] 930 931 *) Reimplement -newreq-nodes and ERR_error_string_n; the 932 original author does not agree with the license change. 933 [Rich Salz] 934 935 *) Add ARIA AEAD TLS support. 936 [Jon Spillett] 937 938 *) Some macro definitions to support VS6 have been removed. Visual 939 Studio 6 has not worked since 1.1.0 940 [Rich Salz] 941 942 *) Add ERR_clear_last_mark(), to allow callers to clear the last mark 943 without clearing the errors. 944 [Richard Levitte] 945 946 *) Add "atfork" functions. If building on a system that without 947 pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application 948 requirements. The RAND facility now uses/requires this. 949 [Rich Salz] 950 951 *) Add SHA3. 952 [Andy Polyakov] 953 954 *) The UI API becomes a permanent and integral part of libcrypto, i.e. 955 not possible to disable entirely. However, it's still possible to 956 disable the console reading UI method, UI_OpenSSL() (use UI_null() 957 as a fallback). 958 959 To disable, configure with 'no-ui-console'. 'no-ui' is still 960 possible to use as an alias. Check at compile time with the 961 macro OPENSSL_NO_UI_CONSOLE. The macro OPENSSL_NO_UI is still 962 possible to check and is an alias for OPENSSL_NO_UI_CONSOLE. 963 [Richard Levitte] 964 965 *) Add a STORE module, which implements a uniform and URI based reader of 966 stores that can contain keys, certificates, CRLs and numerous other 967 objects. The main API is loosely based on a few stdio functions, 968 and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof, 969 OSSL_STORE_error and OSSL_STORE_close. 970 The implementation uses backends called "loaders" to implement arbitrary 971 URI schemes. There is one built in "loader" for the 'file' scheme. 972 [Richard Levitte] 973 974 *) Add devcrypto engine. This has been implemented against cryptodev-linux, 975 then adjusted to work on FreeBSD 8.4 as well. 976 Enable by configuring with 'enable-devcryptoeng'. This is done by default 977 on BSD implementations, as cryptodev.h is assumed to exist on all of them. 978 [Richard Levitte] 979 980 *) Module names can prefixed with OSSL_ or OPENSSL_. This affects 981 util/mkerr.pl, which is adapted to allow those prefixes, leading to 982 error code calls like this: 983 984 OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER); 985 986 With this change, we claim the namespaces OSSL and OPENSSL in a manner 987 that can be encoded in C. For the foreseeable future, this will only 988 affect new modules. 989 [Richard Levitte and Tim Hudson] 990 991 *) Removed BSD cryptodev engine. 992 [Rich Salz] 993 994 *) Add a build target 'build_all_generated', to build all generated files 995 and only that. This can be used to prepare everything that requires 996 things like perl for a system that lacks perl and then move everything 997 to that system and do the rest of the build there. 998 [Richard Levitte] 999 1000 *) In the UI interface, make it possible to duplicate the user data. This 1001 can be used by engines that need to retain the data for a longer time 1002 than just the call where this user data is passed. 1003 [Richard Levitte] 1004 1005 *) Ignore the '-named_curve auto' value for compatibility of applications 1006 with OpenSSL 1.0.2. 1007 [Tomas Mraz <tmraz@fedoraproject.org>] 1008 1009 *) Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2 1010 bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such 1011 alerts across multiple records (some of which could be empty). In practice 1012 it make no sense to send an empty alert record, or to fragment one. TLSv1.3 1013 prohibits this altogether and other libraries (BoringSSL, NSS) do not 1014 support this at all. Supporting it adds significant complexity to the 1015 record layer, and its removal is unlikely to cause interoperability 1016 issues. 1017 [Matt Caswell] 1018 1019 *) Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed 1020 with Z. These are meant to replace LONG and ZLONG and to be size safe. 1021 The use of LONG and ZLONG is discouraged and scheduled for deprecation 1022 in OpenSSL 1.2.0. 1023 [Richard Levitte] 1024 1025 *) Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string, 1026 'z' is to be used for [s]size_t, and 'j' - with [u]int64_t. 1027 [Richard Levitte, Andy Polyakov] 1028 1029 *) Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine() 1030 does for RSA, etc. 1031 [Richard Levitte] 1032 1033 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 1034 platform rather than 'mingw'. 1035 [Richard Levitte] 1036 1037 *) The functions X509_STORE_add_cert and X509_STORE_add_crl return 1038 success if they are asked to add an object which already exists 1039 in the store. This change cascades to other functions which load 1040 certificates and CRLs. 1041 [Paul Dale] 1042 1043 *) x86_64 assembly pack: annotate code with DWARF CFI directives to 1044 facilitate stack unwinding even from assembly subroutines. 1045 [Andy Polyakov] 1046 1047 *) Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN. 1048 Also remove OPENSSL_GLOBAL entirely, as it became a no-op. 1049 [Richard Levitte] 1050 1051 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 1052 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 1053 which is the minimum version we support. 1054 [Richard Levitte] 1055 1056 *) Certificate time validation (X509_cmp_time) enforces stricter 1057 compliance with RFC 5280. Fractional seconds and timezone offsets 1058 are no longer allowed. 1059 [Emilia Käsper] 1060 1061 *) Add support for ARIA 1062 [Paul Dale] 1063 1064 *) s_client will now send the Server Name Indication (SNI) extension by 1065 default unless the new "-noservername" option is used. The server name is 1066 based on the host provided to the "-connect" option unless overridden by 1067 using "-servername". 1068 [Matt Caswell] 1069 1070 *) Add support for SipHash 1071 [Todd Short] 1072 1073 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0 1074 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 1075 prevent issues where no progress is being made and the peer continually 1076 sends unrecognised record types, using up resources processing them. 1077 [Matt Caswell] 1078 1079 *) 'openssl passwd' can now produce SHA256 and SHA512 based output, 1080 using the algorithm defined in 1081 https://www.akkadia.org/drepper/SHA-crypt.txt 1082 [Richard Levitte] 1083 1084 *) Heartbeat support has been removed; the ABI is changed for now. 1085 [Richard Levitte, Rich Salz] 1086 1087 *) Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd. 1088 [Emilia Käsper] 1089 1090 *) The RSA "null" method, which was partially supported to avoid patent 1091 issues, has been replaced to always returns NULL. 1092 [Rich Salz] 1093 1094 1095 Changes between 1.1.0h and 1.1.0i [xx XXX xxxx] 1096 1097 *) Client DoS due to large DH parameter 1098 1099 During key agreement in a TLS handshake using a DH(E) based ciphersuite a 1100 malicious server can send a very large prime value to the client. This will 1101 cause the client to spend an unreasonably long period of time generating a 1102 key for this prime resulting in a hang until the client has finished. This 1103 could be exploited in a Denial Of Service attack. 1104 1105 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken 1106 (CVE-2018-0732) 1107 [Guido Vranken] 1108 1109 *) Cache timing vulnerability in RSA Key Generation 1110 1111 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to 1112 a cache timing side channel attack. An attacker with sufficient access to 1113 mount cache timing attacks during the RSA key generation process could 1114 recover the private key. 1115 1116 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera 1117 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia. 1118 (CVE-2018-0737) 1119 [Billy Brumley] 1120 1121 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 1122 parameter is no longer accepted, as it leads to a corrupt table. NULL 1123 pem_str is reserved for alias entries only. 1124 [Richard Levitte] 1125 1126 *) Revert blinding in ECDSA sign and instead make problematic addition 1127 length-invariant. Switch even to fixed-length Montgomery multiplication. 1128 [Andy Polyakov] 1129 1130 *) Change generating and checking of primes so that the error rate of not 1131 being prime depends on the intended use based on the size of the input. 1132 For larger primes this will result in more rounds of Miller-Rabin. 1133 The maximal error rate for primes with more than 1080 bits is lowered 1134 to 2^-128. 1135 [Kurt Roeckx, Annie Yousar] 1136 1137 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 1138 [Kurt Roeckx] 1139 1140 *) Add blinding to ECDSA and DSA signatures to protect against side channel 1141 attacks discovered by Keegan Ryan (NCC Group). 1142 [Matt Caswell] 1143 1144 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 1145 now allow empty (zero character) pass phrases. 1146 [Richard Levitte] 1147 1148 *) Certificate time validation (X509_cmp_time) enforces stricter 1149 compliance with RFC 5280. Fractional seconds and timezone offsets 1150 are no longer allowed. 1151 [Emilia Käsper] 1152 1153 *) Fixed a text canonicalisation bug in CMS 1154 1155 Where a CMS detached signature is used with text content the text goes 1156 through a canonicalisation process first prior to signing or verifying a 1157 signature. This process strips trailing space at the end of lines, converts 1158 line terminators to CRLF and removes additional trailing line terminators 1159 at the end of a file. A bug in the canonicalisation process meant that 1160 some characters, such as form-feed, were incorrectly treated as whitespace 1161 and removed. This is contrary to the specification (RFC5485). This fix 1162 could mean that detached text data signed with an earlier version of 1163 OpenSSL 1.1.0 may fail to verify using the fixed version, or text data 1164 signed with a fixed OpenSSL may fail to verify with an earlier version of 1165 OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data 1166 and use the "-binary" flag (for the "cms" command line application) or set 1167 the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()). 1168 [Matt Caswell] 1169 1170 Changes between 1.1.0g and 1.1.0h [27 Mar 2018] 1171 1172 *) Constructed ASN.1 types with a recursive definition could exceed the stack 1173 1174 Constructed ASN.1 types with a recursive definition (such as can be found 1175 in PKCS7) could eventually exceed the stack given malicious input with 1176 excessive recursion. This could result in a Denial Of Service attack. There 1177 are no such structures used within SSL/TLS that come from untrusted sources 1178 so this is considered safe. 1179 1180 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz 1181 project. 1182 (CVE-2018-0739) 1183 [Matt Caswell] 1184 1185 *) Incorrect CRYPTO_memcmp on HP-UX PA-RISC 1186 1187 Because of an implementation bug the PA-RISC CRYPTO_memcmp function is 1188 effectively reduced to only comparing the least significant bit of each 1189 byte. This allows an attacker to forge messages that would be considered as 1190 authenticated in an amount of tries lower than that guaranteed by the 1191 security claims of the scheme. The module can only be compiled by the 1192 HP-UX assembler, so that only HP-UX PA-RISC targets are affected. 1193 1194 This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg 1195 (IBM). 1196 (CVE-2018-0733) 1197 [Andy Polyakov] 1198 1199 *) Add a build target 'build_all_generated', to build all generated files 1200 and only that. This can be used to prepare everything that requires 1201 things like perl for a system that lacks perl and then move everything 1202 to that system and do the rest of the build there. 1203 [Richard Levitte] 1204 1205 *) Backport SSL_OP_NO_RENGOTIATION 1206 1207 OpenSSL 1.0.2 and below had the ability to disable renegotiation using the 1208 (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity 1209 changes this is no longer possible in 1.1.0. Therefore the new 1210 SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to 1211 1.1.0 to provide equivalent functionality. 1212 1213 Note that if an application built against 1.1.0h headers (or above) is run 1214 using an older version of 1.1.0 (prior to 1.1.0h) then the option will be 1215 accepted but nothing will happen, i.e. renegotiation will not be prevented. 1216 [Matt Caswell] 1217 1218 *) Removed the OS390-Unix config target. It relied on a script that doesn't 1219 exist. 1220 [Rich Salz] 1221 1222 *) rsaz_1024_mul_avx2 overflow bug on x86_64 1223 1224 There is an overflow bug in the AVX2 Montgomery multiplication procedure 1225 used in exponentiation with 1024-bit moduli. No EC algorithms are affected. 1226 Analysis suggests that attacks against RSA and DSA as a result of this 1227 defect would be very difficult to perform and are not believed likely. 1228 Attacks against DH1024 are considered just feasible, because most of the 1229 work necessary to deduce information about a private key may be performed 1230 offline. The amount of resources required for such an attack would be 1231 significant. However, for an attack on TLS to be meaningful, the server 1232 would have to share the DH1024 private key among multiple clients, which is 1233 no longer an option since CVE-2016-0701. 1234 1235 This only affects processors that support the AVX2 but not ADX extensions 1236 like Intel Haswell (4th generation). 1237 1238 This issue was reported to OpenSSL by David Benjamin (Google). The issue 1239 was originally found via the OSS-Fuzz project. 1240 (CVE-2017-3738) 1241 [Andy Polyakov] 1242 1243 Changes between 1.1.0f and 1.1.0g [2 Nov 2017] 1244 1245 *) bn_sqrx8x_internal carry bug on x86_64 1246 1247 There is a carry propagating bug in the x86_64 Montgomery squaring 1248 procedure. No EC algorithms are affected. Analysis suggests that attacks 1249 against RSA and DSA as a result of this defect would be very difficult to 1250 perform and are not believed likely. Attacks against DH are considered just 1251 feasible (although very difficult) because most of the work necessary to 1252 deduce information about a private key may be performed offline. The amount 1253 of resources required for such an attack would be very significant and 1254 likely only accessible to a limited number of attackers. An attacker would 1255 additionally need online access to an unpatched system using the target 1256 private key in a scenario with persistent DH parameters and a private 1257 key that is shared between multiple clients. 1258 1259 This only affects processors that support the BMI1, BMI2 and ADX extensions 1260 like Intel Broadwell (5th generation) and later or AMD Ryzen. 1261 1262 This issue was reported to OpenSSL by the OSS-Fuzz project. 1263 (CVE-2017-3736) 1264 [Andy Polyakov] 1265 1266 *) Malformed X.509 IPAddressFamily could cause OOB read 1267 1268 If an X.509 certificate has a malformed IPAddressFamily extension, 1269 OpenSSL could do a one-byte buffer overread. The most likely result 1270 would be an erroneous display of the certificate in text format. 1271 1272 This issue was reported to OpenSSL by the OSS-Fuzz project. 1273 (CVE-2017-3735) 1274 [Rich Salz] 1275 1276 Changes between 1.1.0e and 1.1.0f [25 May 2017] 1277 1278 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 1279 platform rather than 'mingw'. 1280 [Richard Levitte] 1281 1282 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 1283 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 1284 which is the minimum version we support. 1285 [Richard Levitte] 1286 1287 Changes between 1.1.0d and 1.1.0e [16 Feb 2017] 1288 1289 *) Encrypt-Then-Mac renegotiation crash 1290 1291 During a renegotiation handshake if the Encrypt-Then-Mac extension is 1292 negotiated where it was not in the original handshake (or vice-versa) then 1293 this can cause OpenSSL to crash (dependant on ciphersuite). Both clients 1294 and servers are affected. 1295 1296 This issue was reported to OpenSSL by Joe Orton (Red Hat). 1297 (CVE-2017-3733) 1298 [Matt Caswell] 1299 1300 Changes between 1.1.0c and 1.1.0d [26 Jan 2017] 1301 1302 *) Truncated packet could crash via OOB read 1303 1304 If one side of an SSL/TLS path is running on a 32-bit host and a specific 1305 cipher is being used, then a truncated packet can cause that host to 1306 perform an out-of-bounds read, usually resulting in a crash. 1307 1308 This issue was reported to OpenSSL by Robert Święcki of Google. 1309 (CVE-2017-3731) 1310 [Andy Polyakov] 1311 1312 *) Bad (EC)DHE parameters cause a client crash 1313 1314 If a malicious server supplies bad parameters for a DHE or ECDHE key 1315 exchange then this can result in the client attempting to dereference a 1316 NULL pointer leading to a client crash. This could be exploited in a Denial 1317 of Service attack. 1318 1319 This issue was reported to OpenSSL by Guido Vranken. 1320 (CVE-2017-3730) 1321 [Matt Caswell] 1322 1323 *) BN_mod_exp may produce incorrect results on x86_64 1324 1325 There is a carry propagating bug in the x86_64 Montgomery squaring 1326 procedure. No EC algorithms are affected. Analysis suggests that attacks 1327 against RSA and DSA as a result of this defect would be very difficult to 1328 perform and are not believed likely. Attacks against DH are considered just 1329 feasible (although very difficult) because most of the work necessary to 1330 deduce information about a private key may be performed offline. The amount 1331 of resources required for such an attack would be very significant and 1332 likely only accessible to a limited number of attackers. An attacker would 1333 additionally need online access to an unpatched system using the target 1334 private key in a scenario with persistent DH parameters and a private 1335 key that is shared between multiple clients. For example this can occur by 1336 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 1337 similar to CVE-2015-3193 but must be treated as a separate problem. 1338 1339 This issue was reported to OpenSSL by the OSS-Fuzz project. 1340 (CVE-2017-3732) 1341 [Andy Polyakov] 1342 1343 Changes between 1.1.0b and 1.1.0c [10 Nov 2016] 1344 1345 *) ChaCha20/Poly1305 heap-buffer-overflow 1346 1347 TLS connections using *-CHACHA20-POLY1305 ciphersuites are susceptible to 1348 a DoS attack by corrupting larger payloads. This can result in an OpenSSL 1349 crash. This issue is not considered to be exploitable beyond a DoS. 1350 1351 This issue was reported to OpenSSL by Robert Święcki (Google Security Team) 1352 (CVE-2016-7054) 1353 [Richard Levitte] 1354 1355 *) CMS Null dereference 1356 1357 Applications parsing invalid CMS structures can crash with a NULL pointer 1358 dereference. This is caused by a bug in the handling of the ASN.1 CHOICE 1359 type in OpenSSL 1.1.0 which can result in a NULL value being passed to the 1360 structure callback if an attempt is made to free certain invalid encodings. 1361 Only CHOICE structures using a callback which do not handle NULL value are 1362 affected. 1363 1364 This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure. 1365 (CVE-2016-7053) 1366 [Stephen Henson] 1367 1368 *) Montgomery multiplication may produce incorrect results 1369 1370 There is a carry propagating bug in the Broadwell-specific Montgomery 1371 multiplication procedure that handles input lengths divisible by, but 1372 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 1373 and DH private keys are impossible. This is because the subroutine in 1374 question is not used in operations with the private key itself and an input 1375 of the attacker's direct choice. Otherwise the bug can manifest itself as 1376 transient authentication and key negotiation failures or reproducible 1377 erroneous outcome of public-key operations with specially crafted input. 1378 Among EC algorithms only Brainpool P-512 curves are affected and one 1379 presumably can attack ECDH key negotiation. Impact was not analyzed in 1380 detail, because pre-requisites for attack are considered unlikely. Namely 1381 multiple clients have to choose the curve in question and the server has to 1382 share the private key among them, neither of which is default behaviour. 1383 Even then only clients that chose the curve will be affected. 1384 1385 This issue was publicly reported as transient failures and was not 1386 initially recognized as a security issue. Thanks to Richard Morgan for 1387 providing reproducible case. 1388 (CVE-2016-7055) 1389 [Andy Polyakov] 1390 1391 *) Removed automatic addition of RPATH in shared libraries and executables, 1392 as this was a remainder from OpenSSL 1.0.x and isn't needed any more. 1393 [Richard Levitte] 1394 1395 Changes between 1.1.0a and 1.1.0b [26 Sep 2016] 1396 1397 *) Fix Use After Free for large message sizes 1398 1399 The patch applied to address CVE-2016-6307 resulted in an issue where if a 1400 message larger than approx 16k is received then the underlying buffer to 1401 store the incoming message is reallocated and moved. Unfortunately a 1402 dangling pointer to the old location is left which results in an attempt to 1403 write to the previously freed location. This is likely to result in a 1404 crash, however it could potentially lead to execution of arbitrary code. 1405 1406 This issue only affects OpenSSL 1.1.0a. 1407 1408 This issue was reported to OpenSSL by Robert Święcki. 1409 (CVE-2016-6309) 1410 [Matt Caswell] 1411 1412 Changes between 1.1.0 and 1.1.0a [22 Sep 2016] 1413 1414 *) OCSP Status Request extension unbounded memory growth 1415 1416 A malicious client can send an excessively large OCSP Status Request 1417 extension. If that client continually requests renegotiation, sending a 1418 large OCSP Status Request extension each time, then there will be unbounded 1419 memory growth on the server. This will eventually lead to a Denial Of 1420 Service attack through memory exhaustion. Servers with a default 1421 configuration are vulnerable even if they do not support OCSP. Builds using 1422 the "no-ocsp" build time option are not affected. 1423 1424 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 1425 (CVE-2016-6304) 1426 [Matt Caswell] 1427 1428 *) SSL_peek() hang on empty record 1429 1430 OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer 1431 sends an empty record. This could be exploited by a malicious peer in a 1432 Denial Of Service attack. 1433 1434 This issue was reported to OpenSSL by Alex Gaynor. 1435 (CVE-2016-6305) 1436 [Matt Caswell] 1437 1438 *) Excessive allocation of memory in tls_get_message_header() and 1439 dtls1_preprocess_fragment() 1440 1441 A (D)TLS message includes 3 bytes for its length in the header for the 1442 message. This would allow for messages up to 16Mb in length. Messages of 1443 this length are excessive and OpenSSL includes a check to ensure that a 1444 peer is sending reasonably sized messages in order to avoid too much memory 1445 being consumed to service a connection. A flaw in the logic of version 1446 1.1.0 means that memory for the message is allocated too early, prior to 1447 the excessive message length check. Due to way memory is allocated in 1448 OpenSSL this could mean an attacker could force up to 21Mb to be allocated 1449 to service a connection. This could lead to a Denial of Service through 1450 memory exhaustion. However, the excessive message length check still takes 1451 place, and this would cause the connection to immediately fail. Assuming 1452 that the application calls SSL_free() on the failed connection in a timely 1453 manner then the 21Mb of allocated memory will then be immediately freed 1454 again. Therefore the excessive memory allocation will be transitory in 1455 nature. This then means that there is only a security impact if: 1456 1457 1) The application does not call SSL_free() in a timely manner in the event 1458 that the connection fails 1459 or 1460 2) The application is working in a constrained environment where there is 1461 very little free memory 1462 or 1463 3) The attacker initiates multiple connection attempts such that there are 1464 multiple connections in a state where memory has been allocated for the 1465 connection; SSL_free() has not yet been called; and there is insufficient 1466 memory to service the multiple requests. 1467 1468 Except in the instance of (1) above any Denial Of Service is likely to be 1469 transitory because as soon as the connection fails the memory is 1470 subsequently freed again in the SSL_free() call. However there is an 1471 increased risk during this period of application crashes due to the lack of 1472 memory - which would then mean a more serious Denial of Service. 1473 1474 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 1475 (CVE-2016-6307 and CVE-2016-6308) 1476 [Matt Caswell] 1477 1478 *) solaris-x86-cc, i.e. 32-bit configuration with vendor compiler, 1479 had to be removed. Primary reason is that vendor assembler can't 1480 assemble our modules with -KPIC flag. As result it, assembly 1481 support, was not even available as option. But its lack means 1482 lack of side-channel resistant code, which is incompatible with 1483 security by todays standards. Fortunately gcc is readily available 1484 prepackaged option, which we firmly point at... 1485 [Andy Polyakov] 1486 1487 Changes between 1.0.2h and 1.1.0 [25 Aug 2016] 1488 1489 *) Windows command-line tool supports UTF-8 opt-in option for arguments 1490 and console input. Setting OPENSSL_WIN32_UTF8 environment variable 1491 (to any value) allows Windows user to access PKCS#12 file generated 1492 with Windows CryptoAPI and protected with non-ASCII password, as well 1493 as files generated under UTF-8 locale on Linux also protected with 1494 non-ASCII password. 1495 [Andy Polyakov] 1496 1497 *) To mitigate the SWEET32 attack (CVE-2016-2183), 3DES cipher suites 1498 have been disabled by default and removed from DEFAULT, just like RC4. 1499 See the RC4 item below to re-enable both. 1500 [Rich Salz] 1501 1502 *) The method for finding the storage location for the Windows RAND seed file 1503 has changed. First we check %RANDFILE%. If that is not set then we check 1504 the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If 1505 all else fails we fall back to C:\. 1506 [Matt Caswell] 1507 1508 *) The EVP_EncryptUpdate() function has had its return type changed from void 1509 to int. A return of 0 indicates and error while a return of 1 indicates 1510 success. 1511 [Matt Caswell] 1512 1513 *) The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and 1514 DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch 1515 off the constant time implementation for RSA, DSA and DH have been made 1516 no-ops and deprecated. 1517 [Matt Caswell] 1518 1519 *) Windows RAND implementation was simplified to only get entropy by 1520 calling CryptGenRandom(). Various other RAND-related tickets 1521 were also closed. 1522 [Joseph Wylie Yandle, Rich Salz] 1523 1524 *) The stack and lhash API's were renamed to start with OPENSSL_SK_ 1525 and OPENSSL_LH_, respectively. The old names are available 1526 with API compatibility. They new names are now completely documented. 1527 [Rich Salz] 1528 1529 *) Unify TYPE_up_ref(obj) methods signature. 1530 SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(), 1531 X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an 1532 int (instead of void) like all others TYPE_up_ref() methods. 1533 So now these methods also check the return value of CRYPTO_atomic_add(), 1534 and the validity of object reference counter. 1535 [fdasilvayy@gmail.com] 1536 1537 *) With Windows Visual Studio builds, the .pdb files are installed 1538 alongside the installed libraries and executables. For a static 1539 library installation, ossl_static.pdb is the associate compiler 1540 generated .pdb file to be used when linking programs. 1541 [Richard Levitte] 1542 1543 *) Remove openssl.spec. Packaging files belong with the packagers. 1544 [Richard Levitte] 1545 1546 *) Automatic Darwin/OSX configuration has had a refresh, it will now 1547 recognise x86_64 architectures automatically. You can still decide 1548 to build for a different bitness with the environment variable 1549 KERNEL_BITS (can be 32 or 64), for example: 1550 1551 KERNEL_BITS=32 ./config 1552 1553 [Richard Levitte] 1554 1555 *) Change default algorithms in pkcs8 utility to use PKCS#5 v2.0, 1556 256 bit AES and HMAC with SHA256. 1557 [Steve Henson] 1558 1559 *) Remove support for MIPS o32 ABI on IRIX (and IRIX only). 1560 [Andy Polyakov] 1561 1562 *) Triple-DES ciphers have been moved from HIGH to MEDIUM. 1563 [Rich Salz] 1564 1565 *) To enable users to have their own config files and build file templates, 1566 Configure looks in the directory indicated by the environment variable 1567 OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/ 1568 directory. On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical 1569 name and is used as is. 1570 [Richard Levitte] 1571 1572 *) The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX, 1573 X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD. The unused type 1574 X509_CERT_FILE_CTX was removed. 1575 [Rich Salz] 1576 1577 *) "shared" builds are now the default. To create only static libraries use 1578 the "no-shared" Configure option. 1579 [Matt Caswell] 1580 1581 *) Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options. 1582 All of these option have not worked for some while and are fundamental 1583 algorithms. 1584 [Matt Caswell] 1585 1586 *) Make various cleanup routines no-ops and mark them as deprecated. Most 1587 global cleanup functions are no longer required because they are handled 1588 via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages). 1589 Explicitly de-initing can cause problems (e.g. where a library that uses 1590 OpenSSL de-inits, but an application is still using it). The affected 1591 functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(), 1592 EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(), 1593 RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and 1594 COMP_zlib_cleanup(). 1595 [Matt Caswell] 1596 1597 *) --strict-warnings no longer enables runtime debugging options 1598 such as REF_DEBUG. Instead, debug options are automatically 1599 enabled with '--debug' builds. 1600 [Andy Polyakov, Emilia Käsper] 1601 1602 *) Made DH and DH_METHOD opaque. The structures for managing DH objects 1603 have been moved out of the public header files. New functions for managing 1604 these have been added. 1605 [Matt Caswell] 1606 1607 *) Made RSA and RSA_METHOD opaque. The structures for managing RSA 1608 objects have been moved out of the public header files. New 1609 functions for managing these have been added. 1610 [Richard Levitte] 1611 1612 *) Made DSA and DSA_METHOD opaque. The structures for managing DSA objects 1613 have been moved out of the public header files. New functions for managing 1614 these have been added. 1615 [Matt Caswell] 1616 1617 *) Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been 1618 moved out of the public header files. New functions for managing these 1619 have been added. 1620 [Matt Caswell] 1621 1622 *) Removed no-rijndael as a config option. Rijndael is an old name for AES. 1623 [Matt Caswell] 1624 1625 *) Removed the mk1mf build scripts. 1626 [Richard Levitte] 1627 1628 *) Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so 1629 it is always safe to #include a header now. 1630 [Rich Salz] 1631 1632 *) Removed the aged BC-32 config and all its supporting scripts 1633 [Richard Levitte] 1634 1635 *) Removed support for Ultrix, Netware, and OS/2. 1636 [Rich Salz] 1637 1638 *) Add support for HKDF. 1639 [Alessandro Ghedini] 1640 1641 *) Add support for blake2b and blake2s 1642 [Bill Cox] 1643 1644 *) Added support for "pipelining". Ciphers that have the 1645 EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple 1646 encryptions/decryptions simultaneously. There are currently no built-in 1647 ciphers with this property but the expectation is that engines will be able 1648 to offer it to significantly improve throughput. Support has been extended 1649 into libssl so that multiple records for a single connection can be 1650 processed in one go (for >=TLS 1.1). 1651 [Matt Caswell] 1652 1653 *) Added the AFALG engine. This is an async capable engine which is able to 1654 offload work to the Linux kernel. In this initial version it only supports 1655 AES128-CBC. The kernel must be version 4.1.0 or greater. 1656 [Catriona Lucey] 1657 1658 *) OpenSSL now uses a new threading API. It is no longer necessary to 1659 set locking callbacks to use OpenSSL in a multi-threaded environment. There 1660 are two supported threading models: pthreads and windows threads. It is 1661 also possible to configure OpenSSL at compile time for "no-threads". The 1662 old threading API should no longer be used. The functions have been 1663 replaced with "no-op" compatibility macros. 1664 [Alessandro Ghedini, Matt Caswell] 1665 1666 *) Modify behavior of ALPN to invoke callback after SNI/servername 1667 callback, such that updates to the SSL_CTX affect ALPN. 1668 [Todd Short] 1669 1670 *) Add SSL_CIPHER queries for authentication and key-exchange. 1671 [Todd Short] 1672 1673 *) Changes to the DEFAULT cipherlist: 1674 - Prefer (EC)DHE handshakes over plain RSA. 1675 - Prefer AEAD ciphers over legacy ciphers. 1676 - Prefer ECDSA over RSA when both certificates are available. 1677 - Prefer TLSv1.2 ciphers/PRF. 1678 - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the 1679 default cipherlist. 1680 [Emilia Käsper] 1681 1682 *) Change the ECC default curve list to be this, in order: x25519, 1683 secp256r1, secp521r1, secp384r1. 1684 [Rich Salz] 1685 1686 *) RC4 based libssl ciphersuites are now classed as "weak" ciphers and are 1687 disabled by default. They can be re-enabled using the 1688 enable-weak-ssl-ciphers option to Configure. 1689 [Matt Caswell] 1690 1691 *) If the server has ALPN configured, but supports no protocols that the 1692 client advertises, send a fatal "no_application_protocol" alert. 1693 This behaviour is SHALL in RFC 7301, though it isn't universally 1694 implemented by other servers. 1695 [Emilia Käsper] 1696 1697 *) Add X25519 support. 1698 Add ASN.1 and EVP_PKEY methods for X25519. This includes support 1699 for public and private key encoding using the format documented in 1700 draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports 1701 key generation and key derivation. 1702 1703 TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses 1704 X25519(29). 1705 [Steve Henson] 1706 1707 *) Deprecate SRP_VBASE_get_by_user. 1708 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 1709 In order to fix an unavoidable memory leak (CVE-2016-0798), 1710 SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP 1711 seed, even if the seed is configured. 1712 1713 Users should use SRP_VBASE_get1_by_user instead. Note that in 1714 SRP_VBASE_get1_by_user, caller must free the returned value. Note 1715 also that even though configuring the SRP seed attempts to hide 1716 invalid usernames by continuing the handshake with fake 1717 credentials, this behaviour is not constant time and no strong 1718 guarantees are made that the handshake is indistinguishable from 1719 that of a valid user. 1720 [Emilia Käsper] 1721 1722 *) Configuration change; it's now possible to build dynamic engines 1723 without having to build shared libraries and vice versa. This 1724 only applies to the engines in engines/, those in crypto/engine/ 1725 will always be built into libcrypto (i.e. "static"). 1726 1727 Building dynamic engines is enabled by default; to disable, use 1728 the configuration option "disable-dynamic-engine". 1729 1730 The only requirements for building dynamic engines are the 1731 presence of the DSO module and building with position independent 1732 code, so they will also automatically be disabled if configuring 1733 with "disable-dso" or "disable-pic". 1734 1735 The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE 1736 are also taken away from openssl/opensslconf.h, as they are 1737 irrelevant. 1738 [Richard Levitte] 1739 1740 *) Configuration change; if there is a known flag to compile 1741 position independent code, it will always be applied on the 1742 libcrypto and libssl object files, and never on the application 1743 object files. This means other libraries that use routines from 1744 libcrypto / libssl can be made into shared libraries regardless 1745 of how OpenSSL was configured. 1746 1747 If this isn't desirable, the configuration options "disable-pic" 1748 or "no-pic" can be used to disable the use of PIC. This will 1749 also disable building shared libraries and dynamic engines. 1750 [Richard Levitte] 1751 1752 *) Removed JPAKE code. It was experimental and has no wide use. 1753 [Rich Salz] 1754 1755 *) The INSTALL_PREFIX Makefile variable has been renamed to 1756 DESTDIR. That makes for less confusion on what this variable 1757 is for. Also, the configuration option --install_prefix is 1758 removed. 1759 [Richard Levitte] 1760 1761 *) Heartbeat for TLS has been removed and is disabled by default 1762 for DTLS; configure with enable-heartbeats. Code that uses the 1763 old #define's might need to be updated. 1764 [Emilia Käsper, Rich Salz] 1765 1766 *) Rename REF_CHECK to REF_DEBUG. 1767 [Rich Salz] 1768 1769 *) New "unified" build system 1770 1771 The "unified" build system is aimed to be a common system for all 1772 platforms we support. With it comes new support for VMS. 1773 1774 This system builds supports building in a different directory tree 1775 than the source tree. It produces one Makefile (for unix family 1776 or lookalikes), or one descrip.mms (for VMS). 1777 1778 The source of information to make the Makefile / descrip.mms is 1779 small files called 'build.info', holding the necessary 1780 information for each directory with source to compile, and a 1781 template in Configurations, like unix-Makefile.tmpl or 1782 descrip.mms.tmpl. 1783 1784 With this change, the library names were also renamed on Windows 1785 and on VMS. They now have names that are closer to the standard 1786 on Unix, and include the major version number, and in certain 1787 cases, the architecture they are built for. See "Notes on shared 1788 libraries" in INSTALL. 1789 1790 We rely heavily on the perl module Text::Template. 1791 [Richard Levitte] 1792 1793 *) Added support for auto-initialisation and de-initialisation of the library. 1794 OpenSSL no longer requires explicit init or deinit routines to be called, 1795 except in certain circumstances. See the OPENSSL_init_crypto() and 1796 OPENSSL_init_ssl() man pages for further information. 1797 [Matt Caswell] 1798 1799 *) The arguments to the DTLSv1_listen function have changed. Specifically the 1800 "peer" argument is now expected to be a BIO_ADDR object. 1801 1802 *) Rewrite of BIO networking library. The BIO library lacked consistent 1803 support of IPv6, and adding it required some more extensive 1804 modifications. This introduces the BIO_ADDR and BIO_ADDRINFO types, 1805 which hold all types of addresses and chains of address information. 1806 It also introduces a new API, with functions like BIO_socket, 1807 BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept. 1808 The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram 1809 have been adapted accordingly. 1810 [Richard Levitte] 1811 1812 *) RSA_padding_check_PKCS1_type_1 now accepts inputs with and without 1813 the leading 0-byte. 1814 [Emilia Käsper] 1815 1816 *) CRIME protection: disable compression by default, even if OpenSSL is 1817 compiled with zlib enabled. Applications can still enable compression 1818 by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by 1819 using the SSL_CONF library to configure compression. 1820 [Emilia Käsper] 1821 1822 *) The signature of the session callback configured with 1823 SSL_CTX_sess_set_get_cb was changed. The read-only input buffer 1824 was explicitly marked as 'const unsigned char*' instead of 1825 'unsigned char*'. 1826 [Emilia Käsper] 1827 1828 *) Always DPURIFY. Remove the use of uninitialized memory in the 1829 RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op. 1830 [Emilia Käsper] 1831 1832 *) Removed many obsolete configuration items, including 1833 DES_PTR, DES_RISC1, DES_RISC2, DES_INT 1834 MD2_CHAR, MD2_INT, MD2_LONG 1835 BF_PTR, BF_PTR2 1836 IDEA_SHORT, IDEA_LONG 1837 RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX 1838 [Rich Salz, with advice from Andy Polyakov] 1839 1840 *) Many BN internals have been moved to an internal header file. 1841 [Rich Salz with help from Andy Polyakov] 1842 1843 *) Configuration and writing out the results from it has changed. 1844 Files such as Makefile include/openssl/opensslconf.h and are now 1845 produced through general templates, such as Makefile.in and 1846 crypto/opensslconf.h.in and some help from the perl module 1847 Text::Template. 1848 1849 Also, the center of configuration information is no longer 1850 Makefile. Instead, Configure produces a perl module in 1851 configdata.pm which holds most of the config data (in the hash 1852 table %config), the target data that comes from the target 1853 configuration in one of the Configurations/*.conf files (in 1854 %target). 1855 [Richard Levitte] 1856 1857 *) To clarify their intended purposes, the Configure options 1858 --prefix and --openssldir change their semantics, and become more 1859 straightforward and less interdependent. 1860 1861 --prefix shall be used exclusively to give the location INSTALLTOP 1862 where programs, scripts, libraries, include files and manuals are 1863 going to be installed. The default is now /usr/local. 1864 1865 --openssldir shall be used exclusively to give the default 1866 location OPENSSLDIR where certificates, private keys, CRLs are 1867 managed. This is also where the default openssl.cnf gets 1868 installed. 1869 If the directory given with this option is a relative path, the 1870 values of both the --prefix value and the --openssldir value will 1871 be combined to become OPENSSLDIR. 1872 The default for --openssldir is INSTALLTOP/ssl. 1873 1874 Anyone who uses --openssldir to specify where OpenSSL is to be 1875 installed MUST change to use --prefix instead. 1876 [Richard Levitte] 1877 1878 *) The GOST engine was out of date and therefore it has been removed. An up 1879 to date GOST engine is now being maintained in an external repository. 1880 See: https://wiki.openssl.org/index.php/Binaries. Libssl still retains 1881 support for GOST ciphersuites (these are only activated if a GOST engine 1882 is present). 1883 [Matt Caswell] 1884 1885 *) EGD is no longer supported by default; use enable-egd when 1886 configuring. 1887 [Ben Kaduk and Rich Salz] 1888 1889 *) The distribution now has Makefile.in files, which are used to 1890 create Makefile's when Configure is run. *Configure must be run 1891 before trying to build now.* 1892 [Rich Salz] 1893 1894 *) The return value for SSL_CIPHER_description() for error conditions 1895 has changed. 1896 [Rich Salz] 1897 1898 *) Support for RFC6698/RFC7671 DANE TLSA peer authentication. 1899 1900 Obtaining and performing DNSSEC validation of TLSA records is 1901 the application's responsibility. The application provides 1902 the TLSA records of its choice to OpenSSL, and these are then 1903 used to authenticate the peer. 1904 1905 The TLSA records need not even come from DNS. They can, for 1906 example, be used to implement local end-entity certificate or 1907 trust-anchor "pinning", where the "pin" data takes the form 1908 of TLSA records, which can augment or replace verification 1909 based on the usual WebPKI public certification authorities. 1910 [Viktor Dukhovni] 1911 1912 *) Revert default OPENSSL_NO_DEPRECATED setting. Instead OpenSSL 1913 continues to support deprecated interfaces in default builds. 1914 However, applications are strongly advised to compile their 1915 source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides 1916 the declarations of all interfaces deprecated in 0.9.8, 1.0.0 1917 or the 1.1.0 releases. 1918 1919 In environments in which all applications have been ported to 1920 not use any deprecated interfaces OpenSSL's Configure script 1921 should be used with the --api=1.1.0 option to entirely remove 1922 support for the deprecated features from the library and 1923 unconditionally disable them in the installed headers. 1924 Essentially the same effect can be achieved with the "no-deprecated" 1925 argument to Configure, except that this will always restrict 1926 the build to just the latest API, rather than a fixed API 1927 version. 1928 1929 As applications are ported to future revisions of the API, 1930 they should update their compile-time OPENSSL_API_COMPAT define 1931 accordingly, but in most cases should be able to continue to 1932 compile with later releases. 1933 1934 The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are 1935 0x10000000L and 0x00908000L, respectively. However those 1936 versions did not support the OPENSSL_API_COMPAT feature, and 1937 so applications are not typically tested for explicit support 1938 of just the undeprecated features of either release. 1939 [Viktor Dukhovni] 1940 1941 *) Add support for setting the minimum and maximum supported protocol. 1942 It can bet set via the SSL_set_min_proto_version() and 1943 SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and 1944 MaxProtocol. It's recommended to use the new APIs to disable 1945 protocols instead of disabling individual protocols using 1946 SSL_set_options() or SSL_CONF's Protocol. This change also 1947 removes support for disabling TLS 1.2 in the OpenSSL TLS 1948 client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT. 1949 [Kurt Roeckx] 1950 1951 *) Support for ChaCha20 and Poly1305 added to libcrypto and libssl. 1952 [Andy Polyakov] 1953 1954 *) New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD 1955 and integrates ECDSA and ECDH functionality into EC. Implementations can 1956 now redirect key generation and no longer need to convert to or from 1957 ECDSA_SIG format. 1958 1959 Note: the ecdsa.h and ecdh.h headers are now no longer needed and just 1960 include the ec.h header file instead. 1961 [Steve Henson] 1962 1963 *) Remove support for all 40 and 56 bit ciphers. This includes all the export 1964 ciphers who are no longer supported and drops support the ephemeral RSA key 1965 exchange. The LOW ciphers currently doesn't have any ciphers in it. 1966 [Kurt Roeckx] 1967 1968 *) Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX 1969 opaque. For HMAC_CTX, the following constructors and destructors 1970 were added: 1971 1972 HMAC_CTX *HMAC_CTX_new(void); 1973 void HMAC_CTX_free(HMAC_CTX *ctx); 1974 1975 For EVP_MD and EVP_CIPHER, complete APIs to create, fill and 1976 destroy such methods has been added. See EVP_MD_meth_new(3) and 1977 EVP_CIPHER_meth_new(3) for documentation. 1978 1979 Additional changes: 1980 1) EVP_MD_CTX_cleanup(), EVP_CIPHER_CTX_cleanup() and 1981 HMAC_CTX_cleanup() were removed. HMAC_CTX_reset() and 1982 EVP_MD_CTX_reset() should be called instead to reinitialise 1983 an already created structure. 1984 2) For consistency with the majority of our object creators and 1985 destructors, EVP_MD_CTX_(create|destroy) were renamed to 1986 EVP_MD_CTX_(new|free). The old names are retained as macros 1987 for deprecated builds. 1988 [Richard Levitte] 1989 1990 *) Added ASYNC support. Libcrypto now includes the async sub-library to enable 1991 cryptographic operations to be performed asynchronously as long as an 1992 asynchronous capable engine is used. See the ASYNC_start_job() man page for 1993 further details. Libssl has also had this capability integrated with the 1994 introduction of the new mode SSL_MODE_ASYNC and associated error 1995 SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man 1996 pages. This work was developed in partnership with Intel Corp. 1997 [Matt Caswell] 1998 1999 *) SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is 2000 always enabled now. If you want to disable the support you should 2001 exclude it using the list of supported ciphers. This also means that the 2002 "-no_ecdhe" option has been removed from s_server. 2003 [Kurt Roeckx] 2004 2005 *) SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls 2006 SSL_{CTX_}set1_curves() which can set a list. 2007 [Kurt Roeckx] 2008 2009 *) Remove support for SSL_{CTX_}set_tmp_ecdh_callback(). You should set the 2010 curve you want to support using SSL_{CTX_}set1_curves(). 2011 [Kurt Roeckx] 2012 2013 *) State machine rewrite. The state machine code has been significantly 2014 refactored in order to remove much duplication of code and solve issues 2015 with the old code (see ssl/statem/README for further details). This change 2016 does have some associated API changes. Notably the SSL_state() function 2017 has been removed and replaced by SSL_get_state which now returns an 2018 "OSSL_HANDSHAKE_STATE" instead of an int. SSL_set_state() has been removed 2019 altogether. The previous handshake states defined in ssl.h and ssl3.h have 2020 also been removed. 2021 [Matt Caswell] 2022 2023 *) All instances of the string "ssleay" in the public API were replaced 2024 with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's) 2025 Some error codes related to internal RSA_eay API's were renamed. 2026 [Rich Salz] 2027 2028 *) The demo files in crypto/threads were moved to demo/threads. 2029 [Rich Salz] 2030 2031 *) Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp, 2032 sureware and ubsec. 2033 [Matt Caswell, Rich Salz] 2034 2035 *) New ASN.1 embed macro. 2036 2037 New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the 2038 structure is not allocated: it is part of the parent. That is instead of 2039 2040 FOO *x; 2041 2042 it must be: 2043 2044 FOO x; 2045 2046 This reduces memory fragmentation and make it impossible to accidentally 2047 set a mandatory field to NULL. 2048 2049 This currently only works for some fields specifically a SEQUENCE, CHOICE, 2050 or ASN1_STRING type which is part of a parent SEQUENCE. Since it is 2051 equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or 2052 SEQUENCE OF. 2053 [Steve Henson] 2054 2055 *) Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled. 2056 [Emilia Käsper] 2057 2058 *) Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although 2059 in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also 2060 an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add 2061 DES and RC4 ciphersuites. 2062 [Matt Caswell] 2063 2064 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 2065 This changes the decoding behaviour for some invalid messages, 2066 though the change is mostly in the more lenient direction, and 2067 legacy behaviour is preserved as much as possible. 2068 [Emilia Käsper] 2069 2070 *) Fix no-stdio build. 2071 [ David Woodhouse <David.Woodhouse@intel.com> and also 2072 Ivan Nestlerode <ivan.nestlerode@sonos.com> ] 2073 2074 *) New testing framework 2075 The testing framework has been largely rewritten and is now using 2076 perl and the perl modules Test::Harness and an extended variant of 2077 Test::More called OpenSSL::Test to do its work. All test scripts in 2078 test/ have been rewritten into test recipes, and all direct calls to 2079 executables in test/Makefile have become individual recipes using the 2080 simplified testing OpenSSL::Test::Simple. 2081 2082 For documentation on our testing modules, do: 2083 2084 perldoc test/testlib/OpenSSL/Test/Simple.pm 2085 perldoc test/testlib/OpenSSL/Test.pm 2086 2087 [Richard Levitte] 2088 2089 *) Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT 2090 are used; the latter aborts on memory leaks (usually checked on exit). 2091 Some undocumented "set malloc, etc., hooks" functions were removed 2092 and others were changed. All are now documented. 2093 [Rich Salz] 2094 2095 *) In DSA_generate_parameters_ex, if the provided seed is too short, 2096 return an error 2097 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 2098 2099 *) Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites 2100 from RFC4279, RFC4785, RFC5487, RFC5489. 2101 2102 Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the 2103 original RSA_PSK patch. 2104 [Steve Henson] 2105 2106 *) Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay 2107 era flag was never set throughout the codebase (only read). Also removed 2108 SSL3_FLAGS_POP_BUFFER which was only used if 2109 SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set. 2110 [Matt Caswell] 2111 2112 *) Changed the default name options in the "ca", "crl", "req" and "x509" 2113 to be "oneline" instead of "compat". 2114 [Richard Levitte] 2115 2116 *) Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're 2117 not aware of clients that still exhibit this bug, and the workaround 2118 hasn't been working properly for a while. 2119 [Emilia Käsper] 2120 2121 *) The return type of BIO_number_read() and BIO_number_written() as well as 2122 the corresponding num_read and num_write members in the BIO structure has 2123 changed from unsigned long to uint64_t. On platforms where an unsigned 2124 long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is 2125 transferred. 2126 [Matt Caswell] 2127 2128 *) Given the pervasive nature of TLS extensions it is inadvisable to run 2129 OpenSSL without support for them. It also means that maintaining 2130 the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably 2131 not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed. 2132 [Matt Caswell] 2133 2134 *) Removed support for the two export grade static DH ciphersuites 2135 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 2136 were newly added (along with a number of other static DH ciphersuites) to 2137 1.0.2. However the two export ones have *never* worked since they were 2138 introduced. It seems strange in any case to be adding new export 2139 ciphersuites, and given "logjam" it also does not seem correct to fix them. 2140 [Matt Caswell] 2141 2142 *) Version negotiation has been rewritten. In particular SSLv23_method(), 2143 SSLv23_client_method() and SSLv23_server_method() have been deprecated, 2144 and turned into macros which simply call the new preferred function names 2145 TLS_method(), TLS_client_method() and TLS_server_method(). All new code 2146 should use the new names instead. Also as part of this change the ssl23.h 2147 header file has been removed. 2148 [Matt Caswell] 2149 2150 *) Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This 2151 code and the associated standard is no longer considered fit-for-purpose. 2152 [Matt Caswell] 2153 2154 *) RT2547 was closed. When generating a private key, try to make the 2155 output file readable only by the owner. This behavior change might 2156 be noticeable when interacting with other software. 2157 2158 *) Documented all exdata functions. Added CRYPTO_free_ex_index. 2159 Added a test. 2160 [Rich Salz] 2161 2162 *) Added HTTP GET support to the ocsp command. 2163 [Rich Salz] 2164 2165 *) Changed default digest for the dgst and enc commands from MD5 to 2166 sha256 2167 [Rich Salz] 2168 2169 *) RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead. 2170 [Matt Caswell] 2171 2172 *) Added support for TLS extended master secret from 2173 draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an 2174 initial patch which was a great help during development. 2175 [Steve Henson] 2176 2177 *) All libssl internal structures have been removed from the public header 2178 files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is 2179 now redundant). Users should not attempt to access internal structures 2180 directly. Instead they should use the provided API functions. 2181 [Matt Caswell] 2182 2183 *) config has been changed so that by default OPENSSL_NO_DEPRECATED is used. 2184 Access to deprecated functions can be re-enabled by running config with 2185 "enable-deprecated". In addition applications wishing to use deprecated 2186 functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour 2187 will, by default, disable some transitive includes that previously existed 2188 in the header files (e.g. ec.h will no longer, by default, include bn.h) 2189 [Matt Caswell] 2190 2191 *) Added support for OCB mode. OpenSSL has been granted a patent license 2192 compatible with the OpenSSL license for use of OCB. Details are available 2193 at https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf. Support 2194 for OCB can be removed by calling config with no-ocb. 2195 [Matt Caswell] 2196 2197 *) SSLv2 support has been removed. It still supports receiving a SSLv2 2198 compatible client hello. 2199 [Kurt Roeckx] 2200 2201 *) Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz], 2202 done while fixing the error code for the key-too-small case. 2203 [Annie Yousar <a.yousar@informatik.hu-berlin.de>] 2204 2205 *) CA.sh has been removed; use CA.pl instead. 2206 [Rich Salz] 2207 2208 *) Removed old DES API. 2209 [Rich Salz] 2210 2211 *) Remove various unsupported platforms: 2212 Sony NEWS4 2213 BEOS and BEOS_R5 2214 NeXT 2215 SUNOS 2216 MPE/iX 2217 Sinix/ReliantUNIX RM400 2218 DGUX 2219 NCR 2220 Tandem 2221 Cray 2222 16-bit platforms such as WIN16 2223 [Rich Salz] 2224 2225 *) Clean up OPENSSL_NO_xxx #define's 2226 Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF 2227 Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx 2228 OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC 2229 OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160 2230 OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO 2231 Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY 2232 OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP 2233 OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK 2234 OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY 2235 Remove MS_STATIC; it's a relic from platforms <32 bits. 2236 [Rich Salz] 2237 2238 *) Cleaned up dead code 2239 Remove all but one '#ifdef undef' which is to be looked at. 2240 [Rich Salz] 2241 2242 *) Clean up calling of xxx_free routines. 2243 Just like free(), fix most of the xxx_free routines to accept 2244 NULL. Remove the non-null checks from callers. Save much code. 2245 [Rich Salz] 2246 2247 *) Add secure heap for storage of private keys (when possible). 2248 Add BIO_s_secmem(), CBIGNUM, etc. 2249 Contributed by Akamai Technologies under our Corporate CLA. 2250 [Rich Salz] 2251 2252 *) Experimental support for a new, fast, unbiased prime candidate generator, 2253 bn_probable_prime_dh_coprime(). Not currently used by any prime generator. 2254 [Felix Laurie von Massenbach <felix@erbridge.co.uk>] 2255 2256 *) New output format NSS in the sess_id command line tool. This allows 2257 exporting the session id and the master key in NSS keylog format. 2258 [Martin Kaiser <martin@kaiser.cx>] 2259 2260 *) Harmonize version and its documentation. -f flag is used to display 2261 compilation flags. 2262 [mancha <mancha1@zoho.com>] 2263 2264 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 2265 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 2266 [mancha <mancha1@zoho.com>] 2267 2268 *) Fix some double frees. These are not thought to be exploitable. 2269 [mancha <mancha1@zoho.com>] 2270 2271 *) A missing bounds check in the handling of the TLS heartbeat extension 2272 can be used to reveal up to 64k of memory to a connected client or 2273 server. 2274 2275 Thanks for Neel Mehta of Google Security for discovering this bug and to 2276 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 2277 preparing the fix (CVE-2014-0160) 2278 [Adam Langley, Bodo Moeller] 2279 2280 *) Fix for the attack described in the paper "Recovering OpenSSL 2281 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 2282 by Yuval Yarom and Naomi Benger. Details can be obtained from: 2283 http://eprint.iacr.org/2014/140 2284 2285 Thanks to Yuval Yarom and Naomi Benger for discovering this 2286 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 2287 [Yuval Yarom and Naomi Benger] 2288 2289 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 2290 this fixes a limitation in previous versions of OpenSSL. 2291 [Steve Henson] 2292 2293 *) Experimental encrypt-then-mac support. 2294 2295 Experimental support for encrypt then mac from 2296 draft-gutmann-tls-encrypt-then-mac-02.txt 2297 2298 To enable it set the appropriate extension number (0x42 for the test 2299 server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42 2300 2301 For non-compliant peers (i.e. just about everything) this should have no 2302 effect. 2303 2304 WARNING: EXPERIMENTAL, SUBJECT TO CHANGE. 2305 2306 [Steve Henson] 2307 2308 *) Add EVP support for key wrapping algorithms, to avoid problems with 2309 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 2310 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 2311 algorithms and include tests cases. 2312 [Steve Henson] 2313 2314 *) Extend CMS code to support RSA-PSS signatures and RSA-OAEP for 2315 enveloped data. 2316 [Steve Henson] 2317 2318 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 2319 MGF1 digest and OAEP label. 2320 [Steve Henson] 2321 2322 *) Make openssl verify return errors. 2323 [Chris Palmer <palmer@google.com> and Ben Laurie] 2324 2325 *) New function ASN1_TIME_diff to calculate the difference between two 2326 ASN1_TIME structures or one structure and the current time. 2327 [Steve Henson] 2328 2329 *) Update fips_test_suite to support multiple command line options. New 2330 test to induce all self test errors in sequence and check expected 2331 failures. 2332 [Steve Henson] 2333 2334 *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and 2335 sign or verify all in one operation. 2336 [Steve Henson] 2337 2338 *) Add fips_algvs: a multicall fips utility incorporating all the algorithm 2339 test programs and fips_test_suite. Includes functionality to parse 2340 the minimal script output of fipsalgest.pl directly. 2341 [Steve Henson] 2342 2343 *) Add authorisation parameter to FIPS_module_mode_set(). 2344 [Steve Henson] 2345 2346 *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves. 2347 [Steve Henson] 2348 2349 *) Use separate DRBG fields for internal and external flags. New function 2350 FIPS_drbg_health_check() to perform on demand health checking. Add 2351 generation tests to fips_test_suite with reduced health check interval to 2352 demonstrate periodic health checking. Add "nodh" option to 2353 fips_test_suite to skip very slow DH test. 2354 [Steve Henson] 2355 2356 *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers 2357 based on NID. 2358 [Steve Henson] 2359 2360 *) More extensive health check for DRBG checking many more failure modes. 2361 New function FIPS_selftest_drbg_all() to handle every possible DRBG 2362 combination: call this in fips_test_suite. 2363 [Steve Henson] 2364 2365 *) Add support for canonical generation of DSA parameter 'g'. See 2366 FIPS 186-3 A.2.3. 2367 2368 *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and 2369 POST to handle HMAC cases. 2370 [Steve Henson] 2371 2372 *) Add functions FIPS_module_version() and FIPS_module_version_text() 2373 to return numerical and string versions of the FIPS module number. 2374 [Steve Henson] 2375 2376 *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and 2377 FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented 2378 outside the validated module in the FIPS capable OpenSSL. 2379 [Steve Henson] 2380 2381 *) Minor change to DRBG entropy callback semantics. In some cases 2382 there is no multiple of the block length between min_len and 2383 max_len. Allow the callback to return more than max_len bytes 2384 of entropy but discard any extra: it is the callback's responsibility 2385 to ensure that the extra data discarded does not impact the 2386 requested amount of entropy. 2387 [Steve Henson] 2388 2389 *) Add PRNG security strength checks to RSA, DSA and ECDSA using 2390 information in FIPS186-3, SP800-57 and SP800-131A. 2391 [Steve Henson] 2392 2393 *) CCM support via EVP. Interface is very similar to GCM case except we 2394 must supply all data in one chunk (i.e. no update, final) and the 2395 message length must be supplied if AAD is used. Add algorithm test 2396 support. 2397 [Steve Henson] 2398 2399 *) Initial version of POST overhaul. Add POST callback to allow the status 2400 of POST to be monitored and/or failures induced. Modify fips_test_suite 2401 to use callback. Always run all selftests even if one fails. 2402 [Steve Henson] 2403 2404 *) XTS support including algorithm test driver in the fips_gcmtest program. 2405 Note: this does increase the maximum key length from 32 to 64 bytes but 2406 there should be no binary compatibility issues as existing applications 2407 will never use XTS mode. 2408 [Steve Henson] 2409 2410 *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies 2411 to OpenSSL RAND code and replace with a tiny FIPS RAND API which also 2412 performs algorithm blocking for unapproved PRNG types. Also do not 2413 set PRNG type in FIPS_mode_set(): leave this to the application. 2414 Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with 2415 the standard OpenSSL PRNG: set additional data to a date time vector. 2416 [Steve Henson] 2417 2418 *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*. 2419 This shouldn't present any incompatibility problems because applications 2420 shouldn't be using these directly and any that are will need to rethink 2421 anyway as the X9.31 PRNG is now deprecated by FIPS 140-2 2422 [Steve Henson] 2423 2424 *) Extensive self tests and health checking required by SP800-90 DRBG. 2425 Remove strength parameter from FIPS_drbg_instantiate and always 2426 instantiate at maximum supported strength. 2427 [Steve Henson] 2428 2429 *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing. 2430 [Steve Henson] 2431 2432 *) New algorithm test program fips_dhvs to handle DH primitives only testing. 2433 [Steve Henson] 2434 2435 *) New function DH_compute_key_padded() to compute a DH key and pad with 2436 leading zeroes if needed: this complies with SP800-56A et al. 2437 [Steve Henson] 2438 2439 *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by 2440 anything, incomplete, subject to change and largely untested at present. 2441 [Steve Henson] 2442 2443 *) Modify fipscanisteronly build option to only build the necessary object 2444 files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile. 2445 [Steve Henson] 2446 2447 *) Add experimental option FIPSSYMS to give all symbols in 2448 fipscanister.o and FIPS or fips prefix. This will avoid 2449 conflicts with future versions of OpenSSL. Add perl script 2450 util/fipsas.pl to preprocess assembly language source files 2451 and rename any affected symbols. 2452 [Steve Henson] 2453 2454 *) Add selftest checks and algorithm block of non-fips algorithms in 2455 FIPS mode. Remove DES2 from selftests. 2456 [Steve Henson] 2457 2458 *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just 2459 return internal method without any ENGINE dependencies. Add new 2460 tiny fips sign and verify functions. 2461 [Steve Henson] 2462 2463 *) New build option no-ec2m to disable characteristic 2 code. 2464 [Steve Henson] 2465 2466 *) New build option "fipscanisteronly". This only builds fipscanister.o 2467 and (currently) associated fips utilities. Uses the file Makefile.fips 2468 instead of Makefile.org as the prototype. 2469 [Steve Henson] 2470 2471 *) Add some FIPS mode restrictions to GCM. Add internal IV generator. 2472 Update fips_gcmtest to use IV generator. 2473 [Steve Henson] 2474 2475 *) Initial, experimental EVP support for AES-GCM. AAD can be input by 2476 setting output buffer to NULL. The *Final function must be 2477 called although it will not retrieve any additional data. The tag 2478 can be set or retrieved with a ctrl. The IV length is by default 12 2479 bytes (96 bits) but can be set to an alternative value. If the IV 2480 length exceeds the maximum IV length (currently 16 bytes) it cannot be 2481 set before the key. 2482 [Steve Henson] 2483 2484 *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the 2485 underlying do_cipher function handles all cipher semantics itself 2486 including padding and finalisation. This is useful if (for example) 2487 an ENGINE cipher handles block padding itself. The behaviour of 2488 do_cipher is subtly changed if this flag is set: the return value 2489 is the number of characters written to the output buffer (zero is 2490 no longer an error code) or a negative error code. Also if the 2491 input buffer is NULL and length 0 finalisation should be performed. 2492 [Steve Henson] 2493 2494 *) If a candidate issuer certificate is already part of the constructed 2495 path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case. 2496 [Steve Henson] 2497 2498 *) Improve forward-security support: add functions 2499 2500 void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure)) 2501 void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure)) 2502 2503 for use by SSL/TLS servers; the callback function will be called whenever a 2504 new session is created, and gets to decide whether the session may be 2505 cached to make it resumable (return 0) or not (return 1). (As by the 2506 SSL/TLS protocol specifications, the session_id sent by the server will be 2507 empty to indicate that the session is not resumable; also, the server will 2508 not generate RFC 4507 (RFC 5077) session tickets.) 2509 2510 A simple reasonable callback implementation is to return is_forward_secure. 2511 This parameter will be set to 1 or 0 depending on the ciphersuite selected 2512 by the SSL/TLS server library, indicating whether it can provide forward 2513 security. 2514 [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)] 2515 2516 *) New -verify_name option in command line utilities to set verification 2517 parameters by name. 2518 [Steve Henson] 2519 2520 *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE. 2521 Add CMAC pkey methods. 2522 [Steve Henson] 2523 2524 *) Experimental renegotiation in s_server -www mode. If the client 2525 browses /reneg connection is renegotiated. If /renegcert it is 2526 renegotiated requesting a certificate. 2527 [Steve Henson] 2528 2529 *) Add an "external" session cache for debugging purposes to s_server. This 2530 should help trace issues which normally are only apparent in deployed 2531 multi-process servers. 2532 [Steve Henson] 2533 2534 *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where 2535 return value is ignored. NB. The functions RAND_add(), RAND_seed(), 2536 BIO_set_cipher() and some obscure PEM functions were changed so they 2537 can now return an error. The RAND changes required a change to the 2538 RAND_METHOD structure. 2539 [Steve Henson] 2540 2541 *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of 2542 a gcc attribute to warn if the result of a function is ignored. This 2543 is enable if DEBUG_UNUSED is set. Add to several functions in evp.h 2544 whose return value is often ignored. 2545 [Steve Henson] 2546 2547 *) New -noct, -requestct, -requirect and -ctlogfile options for s_client. 2548 These allow SCTs (signed certificate timestamps) to be requested and 2549 validated when establishing a connection. 2550 [Rob Percival <robpercival@google.com>] 2551 2552 Changes between 1.0.2g and 1.0.2h [3 May 2016] 2553 2554 *) Prevent padding oracle in AES-NI CBC MAC check 2555 2556 A MITM attacker can use a padding oracle attack to decrypt traffic 2557 when the connection uses an AES CBC cipher and the server support 2558 AES-NI. 2559 2560 This issue was introduced as part of the fix for Lucky 13 padding 2561 attack (CVE-2013-0169). The padding check was rewritten to be in 2562 constant time by making sure that always the same bytes are read and 2563 compared against either the MAC or padding bytes. But it no longer 2564 checked that there was enough data to have both the MAC and padding 2565 bytes. 2566 2567 This issue was reported by Juraj Somorovsky using TLS-Attacker. 2568 (CVE-2016-2107) 2569 [Kurt Roeckx] 2570 2571 *) Fix EVP_EncodeUpdate overflow 2572 2573 An overflow can occur in the EVP_EncodeUpdate() function which is used for 2574 Base64 encoding of binary data. If an attacker is able to supply very large 2575 amounts of input data then a length check can overflow resulting in a heap 2576 corruption. 2577 2578 Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by 2579 the PEM_write_bio* family of functions. These are mainly used within the 2580 OpenSSL command line applications, so any application which processes data 2581 from an untrusted source and outputs it as a PEM file should be considered 2582 vulnerable to this issue. User applications that call these APIs directly 2583 with large amounts of untrusted data may also be vulnerable. 2584 2585 This issue was reported by Guido Vranken. 2586 (CVE-2016-2105) 2587 [Matt Caswell] 2588 2589 *) Fix EVP_EncryptUpdate overflow 2590 2591 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 2592 is able to supply very large amounts of input data after a previous call to 2593 EVP_EncryptUpdate() with a partial block then a length check can overflow 2594 resulting in a heap corruption. Following an analysis of all OpenSSL 2595 internal usage of the EVP_EncryptUpdate() function all usage is one of two 2596 forms. The first form is where the EVP_EncryptUpdate() call is known to be 2597 the first called function after an EVP_EncryptInit(), and therefore that 2598 specific call must be safe. The second form is where the length passed to 2599 EVP_EncryptUpdate() can be seen from the code to be some small value and 2600 therefore there is no possibility of an overflow. Since all instances are 2601 one of these two forms, it is believed that there can be no overflows in 2602 internal code due to this problem. It should be noted that 2603 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 2604 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 2605 of these calls have also been analysed too and it is believed there are no 2606 instances in internal usage where an overflow could occur. 2607 2608 This issue was reported by Guido Vranken. 2609 (CVE-2016-2106) 2610 [Matt Caswell] 2611 2612 *) Prevent ASN.1 BIO excessive memory allocation 2613 2614 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 2615 a short invalid encoding can cause allocation of large amounts of memory 2616 potentially consuming excessive resources or exhausting memory. 2617 2618 Any application parsing untrusted data through d2i BIO functions is 2619 affected. The memory based functions such as d2i_X509() are *not* affected. 2620 Since the memory based functions are used by the TLS library, TLS 2621 applications are not affected. 2622 2623 This issue was reported by Brian Carpenter. 2624 (CVE-2016-2109) 2625 [Stephen Henson] 2626 2627 *) EBCDIC overread 2628 2629 ASN1 Strings that are over 1024 bytes can cause an overread in applications 2630 using the X509_NAME_oneline() function on EBCDIC systems. This could result 2631 in arbitrary stack data being returned in the buffer. 2632 2633 This issue was reported by Guido Vranken. 2634 (CVE-2016-2176) 2635 [Matt Caswell] 2636 2637 *) Modify behavior of ALPN to invoke callback after SNI/servername 2638 callback, such that updates to the SSL_CTX affect ALPN. 2639 [Todd Short] 2640 2641 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the 2642 default. 2643 [Kurt Roeckx] 2644 2645 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the 2646 methods are enabled and ssl2 is disabled the methods return NULL. 2647 [Kurt Roeckx] 2648 2649 Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 2650 2651 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 2652 Builds that are not configured with "enable-weak-ssl-ciphers" will not 2653 provide any "EXPORT" or "LOW" strength ciphers. 2654 [Viktor Dukhovni] 2655 2656 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 2657 is by default disabled at build-time. Builds that are not configured with 2658 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 2659 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 2660 will need to explicitly call either of: 2661 2662 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 2663 or 2664 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 2665 2666 as appropriate. Even if either of those is used, or the application 2667 explicitly uses the version-specific SSLv2_method() or its client and 2668 server variants, SSLv2 ciphers vulnerable to exhaustive search key 2669 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 2670 ciphers, and SSLv2 56-bit DES are no longer available. 2671 (CVE-2016-0800) 2672 [Viktor Dukhovni] 2673 2674 *) Fix a double-free in DSA code 2675 2676 A double free bug was discovered when OpenSSL parses malformed DSA private 2677 keys and could lead to a DoS attack or memory corruption for applications 2678 that receive DSA private keys from untrusted sources. This scenario is 2679 considered rare. 2680 2681 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 2682 libFuzzer. 2683 (CVE-2016-0705) 2684 [Stephen Henson] 2685 2686 *) Disable SRP fake user seed to address a server memory leak. 2687 2688 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 2689 2690 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 2691 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 2692 was changed to ignore the "fake user" SRP seed, even if the seed 2693 is configured. 2694 2695 Users should use SRP_VBASE_get1_by_user instead. Note that in 2696 SRP_VBASE_get1_by_user, caller must free the returned value. Note 2697 also that even though configuring the SRP seed attempts to hide 2698 invalid usernames by continuing the handshake with fake 2699 credentials, this behaviour is not constant time and no strong 2700 guarantees are made that the handshake is indistinguishable from 2701 that of a valid user. 2702 (CVE-2016-0798) 2703 [Emilia Käsper] 2704 2705 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 2706 2707 In the BN_hex2bn function the number of hex digits is calculated using an 2708 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 2709 large values of |i| this can result in |bn_expand| not allocating any 2710 memory because |i * 4| is negative. This can leave the internal BIGNUM data 2711 field as NULL leading to a subsequent NULL ptr deref. For very large values 2712 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 2713 In this case memory is allocated to the internal BIGNUM data field, but it 2714 is insufficiently sized leading to heap corruption. A similar issue exists 2715 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 2716 is ever called by user applications with very large untrusted hex/dec data. 2717 This is anticipated to be a rare occurrence. 2718 2719 All OpenSSL internal usage of these functions use data that is not expected 2720 to be untrusted, e.g. config file data or application command line 2721 arguments. If user developed applications generate config file data based 2722 on untrusted data then it is possible that this could also lead to security 2723 consequences. This is also anticipated to be rare. 2724 2725 This issue was reported to OpenSSL by Guido Vranken. 2726 (CVE-2016-0797) 2727 [Matt Caswell] 2728 2729 *) Fix memory issues in BIO_*printf functions 2730 2731 The internal |fmtstr| function used in processing a "%s" format string in 2732 the BIO_*printf functions could overflow while calculating the length of a 2733 string and cause an OOB read when printing very long strings. 2734 2735 Additionally the internal |doapr_outch| function can attempt to write to an 2736 OOB memory location (at an offset from the NULL pointer) in the event of a 2737 memory allocation failure. In 1.0.2 and below this could be caused where 2738 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 2739 could be in processing a very long "%s" format string. Memory leaks can 2740 also occur. 2741 2742 The first issue may mask the second issue dependent on compiler behaviour. 2743 These problems could enable attacks where large amounts of untrusted data 2744 is passed to the BIO_*printf functions. If applications use these functions 2745 in this way then they could be vulnerable. OpenSSL itself uses these 2746 functions when printing out human-readable dumps of ASN.1 data. Therefore 2747 applications that print this data could be vulnerable if the data is from 2748 untrusted sources. OpenSSL command line applications could also be 2749 vulnerable where they print out ASN.1 data, or if untrusted data is passed 2750 as command line arguments. 2751 2752 Libssl is not considered directly vulnerable. Additionally certificates etc 2753 received via remote connections via libssl are also unlikely to be able to 2754 trigger these issues because of message size limits enforced within libssl. 2755 2756 This issue was reported to OpenSSL Guido Vranken. 2757 (CVE-2016-0799) 2758 [Matt Caswell] 2759 2760 *) Side channel attack on modular exponentiation 2761 2762 A side-channel attack was found which makes use of cache-bank conflicts on 2763 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 2764 of RSA keys. The ability to exploit this issue is limited as it relies on 2765 an attacker who has control of code in a thread running on the same 2766 hyper-threaded core as the victim thread which is performing decryptions. 2767 2768 This issue was reported to OpenSSL by Yuval Yarom, The University of 2769 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 2770 Nadia Heninger, University of Pennsylvania with more information at 2771 http://cachebleed.info. 2772 (CVE-2016-0702) 2773 [Andy Polyakov] 2774 2775 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 2776 if no keysize is specified with default_bits. This fixes an 2777 omission in an earlier change that changed all RSA/DSA key generation 2778 apps to use 2048 bits by default. 2779 [Emilia Käsper] 2780 2781 Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 2782 *) DH small subgroups 2783 2784 Historically OpenSSL only ever generated DH parameters based on "safe" 2785 primes. More recently (in version 1.0.2) support was provided for 2786 generating X9.42 style parameter files such as those required for RFC 5114 2787 support. The primes used in such files may not be "safe". Where an 2788 application is using DH configured with parameters based on primes that are 2789 not "safe" then an attacker could use this fact to find a peer's private 2790 DH exponent. This attack requires that the attacker complete multiple 2791 handshakes in which the peer uses the same private DH exponent. For example 2792 this could be used to discover a TLS server's private DH exponent if it's 2793 reusing the private DH exponent or it's using a static DH ciphersuite. 2794 2795 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 2796 TLS. It is not on by default. If the option is not set then the server 2797 reuses the same private DH exponent for the life of the server process and 2798 would be vulnerable to this attack. It is believed that many popular 2799 applications do set this option and would therefore not be at risk. 2800 2801 The fix for this issue adds an additional check where a "q" parameter is 2802 available (as is the case in X9.42 based parameters). This detects the 2803 only known attack, and is the only possible defense for static DH 2804 ciphersuites. This could have some performance impact. 2805 2806 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 2807 default and cannot be disabled. This could have some performance impact. 2808 2809 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 2810 (CVE-2016-0701) 2811 [Matt Caswell] 2812 2813 *) SSLv2 doesn't block disabled ciphers 2814 2815 A malicious client can negotiate SSLv2 ciphers that have been disabled on 2816 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 2817 been disabled, provided that the SSLv2 protocol was not also disabled via 2818 SSL_OP_NO_SSLv2. 2819 2820 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 2821 and Sebastian Schinzel. 2822 (CVE-2015-3197) 2823 [Viktor Dukhovni] 2824 2825 Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 2826 2827 *) BN_mod_exp may produce incorrect results on x86_64 2828 2829 There is a carry propagating bug in the x86_64 Montgomery squaring 2830 procedure. No EC algorithms are affected. Analysis suggests that attacks 2831 against RSA and DSA as a result of this defect would be very difficult to 2832 perform and are not believed likely. Attacks against DH are considered just 2833 feasible (although very difficult) because most of the work necessary to 2834 deduce information about a private key may be performed offline. The amount 2835 of resources required for such an attack would be very significant and 2836 likely only accessible to a limited number of attackers. An attacker would 2837 additionally need online access to an unpatched system using the target 2838 private key in a scenario with persistent DH parameters and a private 2839 key that is shared between multiple clients. For example this can occur by 2840 default in OpenSSL DHE based SSL/TLS ciphersuites. 2841 2842 This issue was reported to OpenSSL by Hanno Böck. 2843 (CVE-2015-3193) 2844 [Andy Polyakov] 2845 2846 *) Certificate verify crash with missing PSS parameter 2847 2848 The signature verification routines will crash with a NULL pointer 2849 dereference if presented with an ASN.1 signature using the RSA PSS 2850 algorithm and absent mask generation function parameter. Since these 2851 routines are used to verify certificate signature algorithms this can be 2852 used to crash any certificate verification operation and exploited in a 2853 DoS attack. Any application which performs certificate verification is 2854 vulnerable including OpenSSL clients and servers which enable client 2855 authentication. 2856 2857 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG). 2858 (CVE-2015-3194) 2859 [Stephen Henson] 2860 2861 *) X509_ATTRIBUTE memory leak 2862 2863 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 2864 memory. This structure is used by the PKCS#7 and CMS routines so any 2865 application which reads PKCS#7 or CMS data from untrusted sources is 2866 affected. SSL/TLS is not affected. 2867 2868 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 2869 libFuzzer. 2870 (CVE-2015-3195) 2871 [Stephen Henson] 2872 2873 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 2874 This changes the decoding behaviour for some invalid messages, 2875 though the change is mostly in the more lenient direction, and 2876 legacy behaviour is preserved as much as possible. 2877 [Emilia Käsper] 2878 2879 *) In DSA_generate_parameters_ex, if the provided seed is too short, 2880 return an error 2881 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 2882 2883 Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 2884 2885 *) Alternate chains certificate forgery 2886 2887 During certificate verification, OpenSSL will attempt to find an 2888 alternative certificate chain if the first attempt to build such a chain 2889 fails. An error in the implementation of this logic can mean that an 2890 attacker could cause certain checks on untrusted certificates to be 2891 bypassed, such as the CA flag, enabling them to use a valid leaf 2892 certificate to act as a CA and "issue" an invalid certificate. 2893 2894 This issue was reported to OpenSSL by Adam Langley/David Benjamin 2895 (Google/BoringSSL). 2896 [Matt Caswell] 2897 2898 Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 2899 2900 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 2901 incompatibility in the handling of HMAC. The previous ABI has now been 2902 restored. 2903 [Matt Caswell] 2904 2905 Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 2906 2907 *) Malformed ECParameters causes infinite loop 2908 2909 When processing an ECParameters structure OpenSSL enters an infinite loop 2910 if the curve specified is over a specially malformed binary polynomial 2911 field. 2912 2913 This can be used to perform denial of service against any 2914 system which processes public keys, certificate requests or 2915 certificates. This includes TLS clients and TLS servers with 2916 client authentication enabled. 2917 2918 This issue was reported to OpenSSL by Joseph Barr-Pixton. 2919 (CVE-2015-1788) 2920 [Andy Polyakov] 2921 2922 *) Exploitable out-of-bounds read in X509_cmp_time 2923 2924 X509_cmp_time does not properly check the length of the ASN1_TIME 2925 string and can read a few bytes out of bounds. In addition, 2926 X509_cmp_time accepts an arbitrary number of fractional seconds in the 2927 time string. 2928 2929 An attacker can use this to craft malformed certificates and CRLs of 2930 various sizes and potentially cause a segmentation fault, resulting in 2931 a DoS on applications that verify certificates or CRLs. TLS clients 2932 that verify CRLs are affected. TLS clients and servers with client 2933 authentication enabled may be affected if they use custom verification 2934 callbacks. 2935 2936 This issue was reported to OpenSSL by Robert Swiecki (Google), and 2937 independently by Hanno Böck. 2938 (CVE-2015-1789) 2939 [Emilia Käsper] 2940 2941 *) PKCS7 crash with missing EnvelopedContent 2942 2943 The PKCS#7 parsing code does not handle missing inner EncryptedContent 2944 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 2945 with missing content and trigger a NULL pointer dereference on parsing. 2946 2947 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 2948 structures from untrusted sources are affected. OpenSSL clients and 2949 servers are not affected. 2950 2951 This issue was reported to OpenSSL by Michal Zalewski (Google). 2952 (CVE-2015-1790) 2953 [Emilia Käsper] 2954 2955 *) CMS verify infinite loop with unknown hash function 2956 2957 When verifying a signedData message the CMS code can enter an infinite loop 2958 if presented with an unknown hash function OID. This can be used to perform 2959 denial of service against any system which verifies signedData messages using 2960 the CMS code. 2961 This issue was reported to OpenSSL by Johannes Bauer. 2962 (CVE-2015-1792) 2963 [Stephen Henson] 2964 2965 *) Race condition handling NewSessionTicket 2966 2967 If a NewSessionTicket is received by a multi-threaded client when attempting to 2968 reuse a previous ticket then a race condition can occur potentially leading to 2969 a double free of the ticket data. 2970 (CVE-2015-1791) 2971 [Matt Caswell] 2972 2973 *) Only support 256-bit or stronger elliptic curves with the 2974 'ecdh_auto' setting (server) or by default (client). Of supported 2975 curves, prefer P-256 (both). 2976 [Emilia Kasper] 2977 2978 Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 2979 2980 *) ClientHello sigalgs DoS fix 2981 2982 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 2983 invalid signature algorithms extension a NULL pointer dereference will 2984 occur. This can be exploited in a DoS attack against the server. 2985 2986 This issue was was reported to OpenSSL by David Ramos of Stanford 2987 University. 2988 (CVE-2015-0291) 2989 [Stephen Henson and Matt Caswell] 2990 2991 *) Multiblock corrupted pointer fix 2992 2993 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 2994 feature only applies on 64 bit x86 architecture platforms that support AES 2995 NI instructions. A defect in the implementation of "multiblock" can cause 2996 OpenSSL's internal write buffer to become incorrectly set to NULL when 2997 using non-blocking IO. Typically, when the user application is using a 2998 socket BIO for writing, this will only result in a failed connection. 2999 However if some other BIO is used then it is likely that a segmentation 3000 fault will be triggered, thus enabling a potential DoS attack. 3001 3002 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 3003 (CVE-2015-0290) 3004 [Matt Caswell] 3005 3006 *) Segmentation fault in DTLSv1_listen fix 3007 3008 The DTLSv1_listen function is intended to be stateless and processes the 3009 initial ClientHello from many peers. It is common for user code to loop 3010 over the call to DTLSv1_listen until a valid ClientHello is received with 3011 an associated cookie. A defect in the implementation of DTLSv1_listen means 3012 that state is preserved in the SSL object from one invocation to the next 3013 that can lead to a segmentation fault. Errors processing the initial 3014 ClientHello can trigger this scenario. An example of such an error could be 3015 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 3016 server. 3017 3018 This issue was reported to OpenSSL by Per Allansson. 3019 (CVE-2015-0207) 3020 [Matt Caswell] 3021 3022 *) Segmentation fault in ASN1_TYPE_cmp fix 3023 3024 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 3025 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 3026 certificate signature algorithm consistency this can be used to crash any 3027 certificate verification operation and exploited in a DoS attack. Any 3028 application which performs certificate verification is vulnerable including 3029 OpenSSL clients and servers which enable client authentication. 3030 (CVE-2015-0286) 3031 [Stephen Henson] 3032 3033 *) Segmentation fault for invalid PSS parameters fix 3034 3035 The signature verification routines will crash with a NULL pointer 3036 dereference if presented with an ASN.1 signature using the RSA PSS 3037 algorithm and invalid parameters. Since these routines are used to verify 3038 certificate signature algorithms this can be used to crash any 3039 certificate verification operation and exploited in a DoS attack. Any 3040 application which performs certificate verification is vulnerable including 3041 OpenSSL clients and servers which enable client authentication. 3042 3043 This issue was was reported to OpenSSL by Brian Carpenter. 3044 (CVE-2015-0208) 3045 [Stephen Henson] 3046 3047 *) ASN.1 structure reuse memory corruption fix 3048 3049 Reusing a structure in ASN.1 parsing may allow an attacker to cause 3050 memory corruption via an invalid write. Such reuse is and has been 3051 strongly discouraged and is believed to be rare. 3052 3053 Applications that parse structures containing CHOICE or ANY DEFINED BY 3054 components may be affected. Certificate parsing (d2i_X509 and related 3055 functions) are however not affected. OpenSSL clients and servers are 3056 not affected. 3057 (CVE-2015-0287) 3058 [Stephen Henson] 3059 3060 *) PKCS7 NULL pointer dereferences fix 3061 3062 The PKCS#7 parsing code does not handle missing outer ContentInfo 3063 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 3064 missing content and trigger a NULL pointer dereference on parsing. 3065 3066 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 3067 otherwise parse PKCS#7 structures from untrusted sources are 3068 affected. OpenSSL clients and servers are not affected. 3069 3070 This issue was reported to OpenSSL by Michal Zalewski (Google). 3071 (CVE-2015-0289) 3072 [Emilia Käsper] 3073 3074 *) DoS via reachable assert in SSLv2 servers fix 3075 3076 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 3077 servers that both support SSLv2 and enable export cipher suites by sending 3078 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 3079 3080 This issue was discovered by Sean Burford (Google) and Emilia Käsper 3081 (OpenSSL development team). 3082 (CVE-2015-0293) 3083 [Emilia Käsper] 3084 3085 *) Empty CKE with client auth and DHE fix 3086 3087 If client auth is used then a server can seg fault in the event of a DHE 3088 ciphersuite being selected and a zero length ClientKeyExchange message 3089 being sent by the client. This could be exploited in a DoS attack. 3090 (CVE-2015-1787) 3091 [Matt Caswell] 3092 3093 *) Handshake with unseeded PRNG fix 3094 3095 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 3096 with an unseeded PRNG. The conditions are: 3097 - The client is on a platform where the PRNG has not been seeded 3098 automatically, and the user has not seeded manually 3099 - A protocol specific client method version has been used (i.e. not 3100 SSL_client_methodv23) 3101 - A ciphersuite is used that does not require additional random data from 3102 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 3103 3104 If the handshake succeeds then the client random that has been used will 3105 have been generated from a PRNG with insufficient entropy and therefore the 3106 output may be predictable. 3107 3108 For example using the following command with an unseeded openssl will 3109 succeed on an unpatched platform: 3110 3111 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 3112 (CVE-2015-0285) 3113 [Matt Caswell] 3114 3115 *) Use After Free following d2i_ECPrivatekey error fix 3116 3117 A malformed EC private key file consumed via the d2i_ECPrivateKey function 3118 could cause a use after free condition. This, in turn, could cause a double 3119 free in several private key parsing functions (such as d2i_PrivateKey 3120 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 3121 for applications that receive EC private keys from untrusted 3122 sources. This scenario is considered rare. 3123 3124 This issue was discovered by the BoringSSL project and fixed in their 3125 commit 517073cd4b. 3126 (CVE-2015-0209) 3127 [Matt Caswell] 3128 3129 *) X509_to_X509_REQ NULL pointer deref fix 3130 3131 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 3132 the certificate key is invalid. This function is rarely used in practice. 3133 3134 This issue was discovered by Brian Carpenter. 3135 (CVE-2015-0288) 3136 [Stephen Henson] 3137 3138 *) Removed the export ciphers from the DEFAULT ciphers 3139 [Kurt Roeckx] 3140 3141 Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 3142 3143 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 3144 ARMv5 through ARMv8, as opposite to "locking" it to single one. 3145 So far those who have to target multiple platforms would compromise 3146 and argue that binary targeting say ARMv5 would still execute on 3147 ARMv8. "Universal" build resolves this compromise by providing 3148 near-optimal performance even on newer platforms. 3149 [Andy Polyakov] 3150 3151 *) Accelerated NIST P-256 elliptic curve implementation for x86_64 3152 (other platforms pending). 3153 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov] 3154 3155 *) Add support for the SignedCertificateTimestampList certificate and 3156 OCSP response extensions from RFC6962. 3157 [Rob Stradling] 3158 3159 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 3160 for corner cases. (Certain input points at infinity could lead to 3161 bogus results, with non-infinity inputs mapped to infinity too.) 3162 [Bodo Moeller] 3163 3164 *) Initial support for PowerISA 2.0.7, first implemented in POWER8. 3165 This covers AES, SHA256/512 and GHASH. "Initial" means that most 3166 common cases are optimized and there still is room for further 3167 improvements. Vector Permutation AES for Altivec is also added. 3168 [Andy Polyakov] 3169 3170 *) Add support for little-endian ppc64 Linux target. 3171 [Marcelo Cerri (IBM)] 3172 3173 *) Initial support for AMRv8 ISA crypto extensions. This covers AES, 3174 SHA1, SHA256 and GHASH. "Initial" means that most common cases 3175 are optimized and there still is room for further improvements. 3176 Both 32- and 64-bit modes are supported. 3177 [Andy Polyakov, Ard Biesheuvel (Linaro)] 3178 3179 *) Improved ARMv7 NEON support. 3180 [Andy Polyakov] 3181 3182 *) Support for SPARC Architecture 2011 crypto extensions, first 3183 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 3184 SHA256/512, MD5, GHASH and modular exponentiation. 3185 [Andy Polyakov, David Miller] 3186 3187 *) Accelerated modular exponentiation for Intel processors, a.k.a. 3188 RSAZ. 3189 [Shay Gueron & Vlad Krasnov (Intel Corp)] 3190 3191 *) Support for new and upcoming Intel processors, including AVX2, 3192 BMI and SHA ISA extensions. This includes additional "stitched" 3193 implementations, AESNI-SHA256 and GCM, and multi-buffer support 3194 for TLS encrypt. 3195 3196 This work was sponsored by Intel Corp. 3197 [Andy Polyakov] 3198 3199 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 3200 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 3201 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 3202 [Steve Henson] 3203 3204 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 3205 this fixes a limitation in previous versions of OpenSSL. 3206 [Steve Henson] 3207 3208 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 3209 MGF1 digest and OAEP label. 3210 [Steve Henson] 3211 3212 *) Add EVP support for key wrapping algorithms, to avoid problems with 3213 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 3214 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 3215 algorithms and include tests cases. 3216 [Steve Henson] 3217 3218 *) Add functions to allocate and set the fields of an ECDSA_METHOD 3219 structure. 3220 [Douglas E. Engert, Steve Henson] 3221 3222 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 3223 difference in days and seconds between two tm or ASN1_TIME structures. 3224 [Steve Henson] 3225 3226 *) Add -rev test option to s_server to just reverse order of characters 3227 received by client and send back to server. Also prints an abbreviated 3228 summary of the connection parameters. 3229 [Steve Henson] 3230 3231 *) New option -brief for s_client and s_server to print out a brief summary 3232 of connection parameters. 3233 [Steve Henson] 3234 3235 *) Add callbacks for arbitrary TLS extensions. 3236 [Trevor Perrin <trevp@trevp.net> and Ben Laurie] 3237 3238 *) New option -crl_download in several openssl utilities to download CRLs 3239 from CRLDP extension in certificates. 3240 [Steve Henson] 3241 3242 *) New options -CRL and -CRLform for s_client and s_server for CRLs. 3243 [Steve Henson] 3244 3245 *) New function X509_CRL_diff to generate a delta CRL from the difference 3246 of two full CRLs. Add support to "crl" utility. 3247 [Steve Henson] 3248 3249 *) New functions to set lookup_crls function and to retrieve 3250 X509_STORE from X509_STORE_CTX. 3251 [Steve Henson] 3252 3253 *) Print out deprecated issuer and subject unique ID fields in 3254 certificates. 3255 [Steve Henson] 3256 3257 *) Extend OCSP I/O functions so they can be used for simple general purpose 3258 HTTP as well as OCSP. New wrapper function which can be used to download 3259 CRLs using the OCSP API. 3260 [Steve Henson] 3261 3262 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs. 3263 [Steve Henson] 3264 3265 *) SSL_CONF* functions. These provide a common framework for application 3266 configuration using configuration files or command lines. 3267 [Steve Henson] 3268 3269 *) SSL/TLS tracing code. This parses out SSL/TLS records using the 3270 message callback and prints the results. Needs compile time option 3271 "enable-ssl-trace". New options to s_client and s_server to enable 3272 tracing. 3273 [Steve Henson] 3274 3275 *) New ctrl and macro to retrieve supported points extensions. 3276 Print out extension in s_server and s_client. 3277 [Steve Henson] 3278 3279 *) New functions to retrieve certificate signature and signature 3280 OID NID. 3281 [Steve Henson] 3282 3283 *) Add functions to retrieve and manipulate the raw cipherlist sent by a 3284 client to OpenSSL. 3285 [Steve Henson] 3286 3287 *) New Suite B modes for TLS code. These use and enforce the requirements 3288 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 3289 only use Suite B curves. The Suite B modes can be set by using the 3290 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 3291 [Steve Henson] 3292 3293 *) New chain verification flags for Suite B levels of security. Check 3294 algorithms are acceptable when flags are set in X509_verify_cert. 3295 [Steve Henson] 3296 3297 *) Make tls1_check_chain return a set of flags indicating checks passed 3298 by a certificate chain. Add additional tests to handle client 3299 certificates: checks for matching certificate type and issuer name 3300 comparison. 3301 [Steve Henson] 3302 3303 *) If an attempt is made to use a signature algorithm not in the peer 3304 preference list abort the handshake. If client has no suitable 3305 signature algorithms in response to a certificate request do not 3306 use the certificate. 3307 [Steve Henson] 3308 3309 *) If server EC tmp key is not in client preference list abort handshake. 3310 [Steve Henson] 3311 3312 *) Add support for certificate stores in CERT structure. This makes it 3313 possible to have different stores per SSL structure or one store in 3314 the parent SSL_CTX. Include distinct stores for certificate chain 3315 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 3316 to build and store a certificate chain in CERT structure: returning 3317 an error if the chain cannot be built: this will allow applications 3318 to test if a chain is correctly configured. 3319 3320 Note: if the CERT based stores are not set then the parent SSL_CTX 3321 store is used to retain compatibility with existing behaviour. 3322 3323 [Steve Henson] 3324 3325 *) New function ssl_set_client_disabled to set a ciphersuite disabled 3326 mask based on the current session, check mask when sending client 3327 hello and checking the requested ciphersuite. 3328 [Steve Henson] 3329 3330 *) New ctrls to retrieve and set certificate types in a certificate 3331 request message. Print out received values in s_client. If certificate 3332 types is not set with custom values set sensible values based on 3333 supported signature algorithms. 3334 [Steve Henson] 3335 3336 *) Support for distinct client and server supported signature algorithms. 3337 [Steve Henson] 3338 3339 *) Add certificate callback. If set this is called whenever a certificate 3340 is required by client or server. An application can decide which 3341 certificate chain to present based on arbitrary criteria: for example 3342 supported signature algorithms. Add very simple example to s_server. 3343 This fixes many of the problems and restrictions of the existing client 3344 certificate callback: for example you can now clear an existing 3345 certificate and specify the whole chain. 3346 [Steve Henson] 3347 3348 *) Add new "valid_flags" field to CERT_PKEY structure which determines what 3349 the certificate can be used for (if anything). Set valid_flags field 3350 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 3351 to have similar checks in it. 3352 3353 Add new "cert_flags" field to CERT structure and include a "strict mode". 3354 This enforces some TLS certificate requirements (such as only permitting 3355 certificate signature algorithms contained in the supported algorithms 3356 extension) which some implementations ignore: this option should be used 3357 with caution as it could cause interoperability issues. 3358 [Steve Henson] 3359 3360 *) Update and tidy signature algorithm extension processing. Work out 3361 shared signature algorithms based on preferences and peer algorithms 3362 and print them out in s_client and s_server. Abort handshake if no 3363 shared signature algorithms. 3364 [Steve Henson] 3365 3366 *) Add new functions to allow customised supported signature algorithms 3367 for SSL and SSL_CTX structures. Add options to s_client and s_server 3368 to support them. 3369 [Steve Henson] 3370 3371 *) New function SSL_certs_clear() to delete all references to certificates 3372 from an SSL structure. Before this once a certificate had been added 3373 it couldn't be removed. 3374 [Steve Henson] 3375 3376 *) Integrate hostname, email address and IP address checking with certificate 3377 verification. New verify options supporting checking in openssl utility. 3378 [Steve Henson] 3379 3380 *) Fixes and wildcard matching support to hostname and email checking 3381 functions. Add manual page. 3382 [Florian Weimer (Red Hat Product Security Team)] 3383 3384 *) New functions to check a hostname email or IP address against a 3385 certificate. Add options x509 utility to print results of checks against 3386 a certificate. 3387 [Steve Henson] 3388 3389 *) Fix OCSP checking. 3390 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie] 3391 3392 *) Initial experimental support for explicitly trusted non-root CAs. 3393 OpenSSL still tries to build a complete chain to a root but if an 3394 intermediate CA has a trust setting included that is used. The first 3395 setting is used: whether to trust (e.g., -addtrust option to the x509 3396 utility) or reject. 3397 [Steve Henson] 3398 3399 *) Add -trusted_first option which attempts to find certificates in the 3400 trusted store even if an untrusted chain is also supplied. 3401 [Steve Henson] 3402 3403 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 3404 platform support for Linux and Android. 3405 [Andy Polyakov] 3406 3407 *) Support for linux-x32, ILP32 environment in x86_64 framework. 3408 [Andy Polyakov] 3409 3410 *) Experimental multi-implementation support for FIPS capable OpenSSL. 3411 When in FIPS mode the approved implementations are used as normal, 3412 when not in FIPS mode the internal unapproved versions are used instead. 3413 This means that the FIPS capable OpenSSL isn't forced to use the 3414 (often lower performance) FIPS implementations outside FIPS mode. 3415 [Steve Henson] 3416 3417 *) Transparently support X9.42 DH parameters when calling 3418 PEM_read_bio_DHparameters. This means existing applications can handle 3419 the new parameter format automatically. 3420 [Steve Henson] 3421 3422 *) Initial experimental support for X9.42 DH parameter format: mainly 3423 to support use of 'q' parameter for RFC5114 parameters. 3424 [Steve Henson] 3425 3426 *) Add DH parameters from RFC5114 including test data to dhtest. 3427 [Steve Henson] 3428 3429 *) Support for automatic EC temporary key parameter selection. If enabled 3430 the most preferred EC parameters are automatically used instead of 3431 hardcoded fixed parameters. Now a server just has to call: 3432 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 3433 support ECDH and use the most appropriate parameters. 3434 [Steve Henson] 3435 3436 *) Enhance and tidy EC curve and point format TLS extension code. Use 3437 static structures instead of allocation if default values are used. 3438 New ctrls to set curves we wish to support and to retrieve shared curves. 3439 Print out shared curves in s_server. New options to s_server and s_client 3440 to set list of supported curves. 3441 [Steve Henson] 3442 3443 *) New ctrls to retrieve supported signature algorithms and 3444 supported curve values as an array of NIDs. Extend openssl utility 3445 to print out received values. 3446 [Steve Henson] 3447 3448 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 3449 between NIDs and the more common NIST names such as "P-256". Enhance 3450 ecparam utility and ECC method to recognise the NIST names for curves. 3451 [Steve Henson] 3452 3453 *) Enhance SSL/TLS certificate chain handling to support different 3454 chains for each certificate instead of one chain in the parent SSL_CTX. 3455 [Steve Henson] 3456 3457 *) Support for fixed DH ciphersuite client authentication: where both 3458 server and client use DH certificates with common parameters. 3459 [Steve Henson] 3460 3461 *) Support for fixed DH ciphersuites: those requiring DH server 3462 certificates. 3463 [Steve Henson] 3464 3465 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of 3466 the certificate. 3467 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 3468 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 3469 X509_CINF_get_signature were reverted post internal team review. 3470 3471 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 3472 3473 *) Build fixes for the Windows and OpenVMS platforms 3474 [Matt Caswell and Richard Levitte] 3475 3476 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 3477 3478 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 3479 message can cause a segmentation fault in OpenSSL due to a NULL pointer 3480 dereference. This could lead to a Denial Of Service attack. Thanks to 3481 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 3482 (CVE-2014-3571) 3483 [Steve Henson] 3484 3485 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 3486 dtls1_buffer_record function under certain conditions. In particular this 3487 could occur if an attacker sent repeated DTLS records with the same 3488 sequence number but for the next epoch. The memory leak could be exploited 3489 by an attacker in a Denial of Service attack through memory exhaustion. 3490 Thanks to Chris Mueller for reporting this issue. 3491 (CVE-2015-0206) 3492 [Matt Caswell] 3493 3494 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 3495 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 3496 method would be set to NULL which could later result in a NULL pointer 3497 dereference. Thanks to Frank Schmirler for reporting this issue. 3498 (CVE-2014-3569) 3499 [Kurt Roeckx] 3500 3501 *) Abort handshake if server key exchange message is omitted for ephemeral 3502 ECDH ciphersuites. 3503 3504 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 3505 reporting this issue. 3506 (CVE-2014-3572) 3507 [Steve Henson] 3508 3509 *) Remove non-export ephemeral RSA code on client and server. This code 3510 violated the TLS standard by allowing the use of temporary RSA keys in 3511 non-export ciphersuites and could be used by a server to effectively 3512 downgrade the RSA key length used to a value smaller than the server 3513 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 3514 INRIA or reporting this issue. 3515 (CVE-2015-0204) 3516 [Steve Henson] 3517 3518 *) Fixed issue where DH client certificates are accepted without verification. 3519 An OpenSSL server will accept a DH certificate for client authentication 3520 without the certificate verify message. This effectively allows a client to 3521 authenticate without the use of a private key. This only affects servers 3522 which trust a client certificate authority which issues certificates 3523 containing DH keys: these are extremely rare and hardly ever encountered. 3524 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 3525 this issue. 3526 (CVE-2015-0205) 3527 [Steve Henson] 3528 3529 *) Ensure that the session ID context of an SSL is updated when its 3530 SSL_CTX is updated via SSL_set_SSL_CTX. 3531 3532 The session ID context is typically set from the parent SSL_CTX, 3533 and can vary with the CTX. 3534 [Adam Langley] 3535 3536 *) Fix various certificate fingerprint issues. 3537 3538 By using non-DER or invalid encodings outside the signed portion of a 3539 certificate the fingerprint can be changed without breaking the signature. 3540 Although no details of the signed portion of the certificate can be changed 3541 this can cause problems with some applications: e.g. those using the 3542 certificate fingerprint for blacklists. 3543 3544 1. Reject signatures with non zero unused bits. 3545 3546 If the BIT STRING containing the signature has non zero unused bits reject 3547 the signature. All current signature algorithms require zero unused bits. 3548 3549 2. Check certificate algorithm consistency. 3550 3551 Check the AlgorithmIdentifier inside TBS matches the one in the 3552 certificate signature. NB: this will result in signature failure 3553 errors for some broken certificates. 3554 3555 Thanks to Konrad Kraszewski from Google for reporting this issue. 3556 3557 3. Check DSA/ECDSA signatures use DER. 3558 3559 Re-encode DSA/ECDSA signatures and compare with the original received 3560 signature. Return an error if there is a mismatch. 3561 3562 This will reject various cases including garbage after signature 3563 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 3564 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 3565 (negative or with leading zeroes). 3566 3567 Further analysis was conducted and fixes were developed by Stephen Henson 3568 of the OpenSSL core team. 3569 3570 (CVE-2014-8275) 3571 [Steve Henson] 3572 3573 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 3574 results on some platforms, including x86_64. This bug occurs at random 3575 with a very low probability, and is not known to be exploitable in any 3576 way, though its exact impact is difficult to determine. Thanks to Pieter 3577 Wuille (Blockstream) who reported this issue and also suggested an initial 3578 fix. Further analysis was conducted by the OpenSSL development team and 3579 Adam Langley of Google. The final fix was developed by Andy Polyakov of 3580 the OpenSSL core team. 3581 (CVE-2014-3570) 3582 [Andy Polyakov] 3583 3584 *) Do not resume sessions on the server if the negotiated protocol 3585 version does not match the session's version. Resuming with a different 3586 version, while not strictly forbidden by the RFC, is of questionable 3587 sanity and breaks all known clients. 3588 [David Benjamin, Emilia Käsper] 3589 3590 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 3591 early CCS messages during renegotiation. (Note that because 3592 renegotiation is encrypted, this early CCS was not exploitable.) 3593 [Emilia Käsper] 3594 3595 *) Tighten client-side session ticket handling during renegotiation: 3596 ensure that the client only accepts a session ticket if the server sends 3597 the extension anew in the ServerHello. Previously, a TLS client would 3598 reuse the old extension state and thus accept a session ticket if one was 3599 announced in the initial ServerHello. 3600 3601 Similarly, ensure that the client requires a session ticket if one 3602 was advertised in the ServerHello. Previously, a TLS client would 3603 ignore a missing NewSessionTicket message. 3604 [Emilia Käsper] 3605 3606 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 3607 3608 *) SRTP Memory Leak. 3609 3610 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 3611 sends a carefully crafted handshake message, to cause OpenSSL to fail 3612 to free up to 64k of memory causing a memory leak. This could be 3613 exploited in a Denial Of Service attack. This issue affects OpenSSL 3614 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 3615 whether SRTP is used or configured. Implementations of OpenSSL that 3616 have been compiled with OPENSSL_NO_SRTP defined are not affected. 3617 3618 The fix was developed by the OpenSSL team. 3619 (CVE-2014-3513) 3620 [OpenSSL team] 3621 3622 *) Session Ticket Memory Leak. 3623 3624 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 3625 integrity of that ticket is first verified. In the event of a session 3626 ticket integrity check failing, OpenSSL will fail to free memory 3627 causing a memory leak. By sending a large number of invalid session 3628 tickets an attacker could exploit this issue in a Denial Of Service 3629 attack. 3630 (CVE-2014-3567) 3631 [Steve Henson] 3632 3633 *) Build option no-ssl3 is incomplete. 3634 3635 When OpenSSL is configured with "no-ssl3" as a build option, servers 3636 could accept and complete a SSL 3.0 handshake, and clients could be 3637 configured to send them. 3638 (CVE-2014-3568) 3639 [Akamai and the OpenSSL team] 3640 3641 *) Add support for TLS_FALLBACK_SCSV. 3642 Client applications doing fallback retries should call 3643 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 3644 (CVE-2014-3566) 3645 [Adam Langley, Bodo Moeller] 3646 3647 *) Add additional DigestInfo checks. 3648 3649 Re-encode DigestInto in DER and check against the original when 3650 verifying RSA signature: this will reject any improperly encoded 3651 DigestInfo structures. 3652 3653 Note: this is a precautionary measure and no attacks are currently known. 3654 3655 [Steve Henson] 3656 3657 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 3658 3659 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 3660 SRP code can be overrun an internal buffer. Add sanity check that 3661 g, A, B < N to SRP code. 3662 3663 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 3664 Group for discovering this issue. 3665 (CVE-2014-3512) 3666 [Steve Henson] 3667 3668 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 3669 TLS 1.0 instead of higher protocol versions when the ClientHello message 3670 is badly fragmented. This allows a man-in-the-middle attacker to force a 3671 downgrade to TLS 1.0 even if both the server and the client support a 3672 higher protocol version, by modifying the client's TLS records. 3673 3674 Thanks to David Benjamin and Adam Langley (Google) for discovering and 3675 researching this issue. 3676 (CVE-2014-3511) 3677 [David Benjamin] 3678 3679 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 3680 to a denial of service attack. A malicious server can crash the client 3681 with a null pointer dereference (read) by specifying an anonymous (EC)DH 3682 ciphersuite and sending carefully crafted handshake messages. 3683 3684 Thanks to Felix Gröbert (Google) for discovering and researching this 3685 issue. 3686 (CVE-2014-3510) 3687 [Emilia Käsper] 3688 3689 *) By sending carefully crafted DTLS packets an attacker could cause openssl 3690 to leak memory. This can be exploited through a Denial of Service attack. 3691 Thanks to Adam Langley for discovering and researching this issue. 3692 (CVE-2014-3507) 3693 [Adam Langley] 3694 3695 *) An attacker can force openssl to consume large amounts of memory whilst 3696 processing DTLS handshake messages. This can be exploited through a 3697 Denial of Service attack. 3698 Thanks to Adam Langley for discovering and researching this issue. 3699 (CVE-2014-3506) 3700 [Adam Langley] 3701 3702 *) An attacker can force an error condition which causes openssl to crash 3703 whilst processing DTLS packets due to memory being freed twice. This 3704 can be exploited through a Denial of Service attack. 3705 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 3706 this issue. 3707 (CVE-2014-3505) 3708 [Adam Langley] 3709 3710 *) If a multithreaded client connects to a malicious server using a resumed 3711 session and the server sends an ec point format extension it could write 3712 up to 255 bytes to freed memory. 3713 3714 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 3715 issue. 3716 (CVE-2014-3509) 3717 [Gabor Tyukasz] 3718 3719 *) A malicious server can crash an OpenSSL client with a null pointer 3720 dereference (read) by specifying an SRP ciphersuite even though it was not 3721 properly negotiated with the client. This can be exploited through a 3722 Denial of Service attack. 3723 3724 Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for 3725 discovering and researching this issue. 3726 (CVE-2014-5139) 3727 [Steve Henson] 3728 3729 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 3730 X509_name_oneline, X509_name_print_ex et al. to leak some information 3731 from the stack. Applications may be affected if they echo pretty printing 3732 output to the attacker. 3733 3734 Thanks to Ivan Fratric (Google) for discovering this issue. 3735 (CVE-2014-3508) 3736 [Emilia Käsper, and Steve Henson] 3737 3738 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 3739 for corner cases. (Certain input points at infinity could lead to 3740 bogus results, with non-infinity inputs mapped to infinity too.) 3741 [Bodo Moeller] 3742 3743 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 3744 3745 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 3746 handshake can force the use of weak keying material in OpenSSL 3747 SSL/TLS clients and servers. 3748 3749 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 3750 researching this issue. (CVE-2014-0224) 3751 [KIKUCHI Masashi, Steve Henson] 3752 3753 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 3754 OpenSSL DTLS client the code can be made to recurse eventually crashing 3755 in a DoS attack. 3756 3757 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 3758 (CVE-2014-0221) 3759 [Imre Rad, Steve Henson] 3760 3761 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 3762 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 3763 client or server. This is potentially exploitable to run arbitrary 3764 code on a vulnerable client or server. 3765 3766 Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195) 3767 [Jüri Aedla, Steve Henson] 3768 3769 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 3770 are subject to a denial of service attack. 3771 3772 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering 3773 this issue. (CVE-2014-3470) 3774 [Felix Gröbert, Ivan Fratric, Steve Henson] 3775 3776 *) Harmonize version and its documentation. -f flag is used to display 3777 compilation flags. 3778 [mancha <mancha1@zoho.com>] 3779 3780 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 3781 in i2d_ECPrivateKey. 3782 [mancha <mancha1@zoho.com>] 3783 3784 *) Fix some double frees. These are not thought to be exploitable. 3785 [mancha <mancha1@zoho.com>] 3786 3787 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 3788 3789 *) A missing bounds check in the handling of the TLS heartbeat extension 3790 can be used to reveal up to 64k of memory to a connected client or 3791 server. 3792 3793 Thanks for Neel Mehta of Google Security for discovering this bug and to 3794 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 3795 preparing the fix (CVE-2014-0160) 3796 [Adam Langley, Bodo Moeller] 3797 3798 *) Fix for the attack described in the paper "Recovering OpenSSL 3799 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 3800 by Yuval Yarom and Naomi Benger. Details can be obtained from: 3801 http://eprint.iacr.org/2014/140 3802 3803 Thanks to Yuval Yarom and Naomi Benger for discovering this 3804 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 3805 [Yuval Yarom and Naomi Benger] 3806 3807 *) TLS pad extension: draft-agl-tls-padding-03 3808 3809 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 3810 TLS client Hello record length value would otherwise be > 255 and 3811 less that 512 pad with a dummy extension containing zeroes so it 3812 is at least 512 bytes long. 3813 3814 [Adam Langley, Steve Henson] 3815 3816 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 3817 3818 *) Fix for TLS record tampering bug. A carefully crafted invalid 3819 handshake could crash OpenSSL with a NULL pointer exception. 3820 Thanks to Anton Johansson for reporting this issues. 3821 (CVE-2013-4353) 3822 3823 *) Keep original DTLS digest and encryption contexts in retransmission 3824 structures so we can use the previous session parameters if they need 3825 to be resent. (CVE-2013-6450) 3826 [Steve Henson] 3827 3828 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 3829 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 3830 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 3831 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 3832 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 3833 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 3834 [Rob Stradling, Adam Langley] 3835 3836 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 3837 3838 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 3839 supporting platforms or when small records were transferred. 3840 [Andy Polyakov, Steve Henson] 3841 3842 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 3843 3844 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 3845 3846 This addresses the flaw in CBC record processing discovered by 3847 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 3848 at: http://www.isg.rhul.ac.uk/tls/ 3849 3850 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 3851 Security Group at Royal Holloway, University of London 3852 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 3853 Emilia Käsper for the initial patch. 3854 (CVE-2013-0169) 3855 [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 3856 3857 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 3858 ciphersuites which can be exploited in a denial of service attack. 3859 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 3860 and detecting this bug and to Wolfgang Ettlinger 3861 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 3862 (CVE-2012-2686) 3863 [Adam Langley] 3864 3865 *) Return an error when checking OCSP signatures when key is NULL. 3866 This fixes a DoS attack. (CVE-2013-0166) 3867 [Steve Henson] 3868 3869 *) Make openssl verify return errors. 3870 [Chris Palmer <palmer@google.com> and Ben Laurie] 3871 3872 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 3873 the right response is stapled. Also change SSL_get_certificate() 3874 so it returns the certificate actually sent. 3875 See http://rt.openssl.org/Ticket/Display.html?id=2836. 3876 [Rob Stradling <rob.stradling@comodo.com>] 3877 3878 *) Fix possible deadlock when decoding public keys. 3879 [Steve Henson] 3880 3881 *) Don't use TLS 1.0 record version number in initial client hello 3882 if renegotiating. 3883 [Steve Henson] 3884 3885 Changes between 1.0.1b and 1.0.1c [10 May 2012] 3886 3887 *) Sanity check record length before skipping explicit IV in TLS 3888 1.2, 1.1 and DTLS to fix DoS attack. 3889 3890 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 3891 fuzzing as a service testing platform. 3892 (CVE-2012-2333) 3893 [Steve Henson] 3894 3895 *) Initialise tkeylen properly when encrypting CMS messages. 3896 Thanks to Solar Designer of Openwall for reporting this issue. 3897 [Steve Henson] 3898 3899 *) In FIPS mode don't try to use composite ciphers as they are not 3900 approved. 3901 [Steve Henson] 3902 3903 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 3904 3905 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 3906 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 3907 mean any application compiled against OpenSSL 1.0.0 headers setting 3908 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling 3909 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 3910 0x10000000L Any application which was previously compiled against 3911 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 3912 will need to be recompiled as a result. Letting be results in 3913 inability to disable specifically TLS 1.1 and in client context, 3914 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 3915 [Steve Henson] 3916 3917 *) In order to ensure interoperability SSL_OP_NO_protocolX does not 3918 disable just protocol X, but all protocols above X *if* there are 3919 protocols *below* X still enabled. In more practical terms it means 3920 that if application wants to disable TLS1.0 in favor of TLS1.1 and 3921 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 3922 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 3923 client side. 3924 [Andy Polyakov] 3925 3926 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 3927 3928 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 3929 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 3930 in CRYPTO_realloc_clean. 3931 3932 Thanks to Tavis Ormandy, Google Security Team, for discovering this 3933 issue and to Adam Langley <agl@chromium.org> for fixing it. 3934 (CVE-2012-2110) 3935 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 3936 3937 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 3938 [Adam Langley] 3939 3940 *) Workarounds for some broken servers that "hang" if a client hello 3941 record length exceeds 255 bytes. 3942 3943 1. Do not use record version number > TLS 1.0 in initial client 3944 hello: some (but not all) hanging servers will now work. 3945 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 3946 the number of ciphers sent in the client hello. This should be 3947 set to an even number, such as 50, for example by passing: 3948 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 3949 Most broken servers should now work. 3950 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 3951 TLS 1.2 client support entirely. 3952 [Steve Henson] 3953 3954 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 3955 [Andy Polyakov] 3956 3957 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 3958 3959 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 3960 STRING form instead of a DigestInfo. 3961 [Steve Henson] 3962 3963 *) The format used for MDC2 RSA signatures is inconsistent between EVP 3964 and the RSA_sign/RSA_verify functions. This was made more apparent when 3965 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 3966 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 3967 the correct format in RSA_verify so both forms transparently work. 3968 [Steve Henson] 3969 3970 *) Some servers which support TLS 1.0 can choke if we initially indicate 3971 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 3972 encrypted premaster secret. As a workaround use the maximum permitted 3973 client version in client hello, this should keep such servers happy 3974 and still work with previous versions of OpenSSL. 3975 [Steve Henson] 3976 3977 *) Add support for TLS/DTLS heartbeats. 3978 [Robin Seggelmann <seggelmann@fh-muenster.de>] 3979 3980 *) Add support for SCTP. 3981 [Robin Seggelmann <seggelmann@fh-muenster.de>] 3982 3983 *) Improved PRNG seeding for VOS. 3984 [Paul Green <Paul.Green@stratus.com>] 3985 3986 *) Extensive assembler packs updates, most notably: 3987 3988 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 3989 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 3990 - x86_64: bit-sliced AES implementation; 3991 - ARM: NEON support, contemporary platforms optimizations; 3992 - s390x: z196 support; 3993 - *: GHASH and GF(2^m) multiplication implementations; 3994 3995 [Andy Polyakov] 3996 3997 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 3998 (removal of unnecessary code) 3999 [Peter Sylvester <peter.sylvester@edelweb.fr>] 4000 4001 *) Add TLS key material exporter from RFC 5705. 4002 [Eric Rescorla] 4003 4004 *) Add DTLS-SRTP negotiation from RFC 5764. 4005 [Eric Rescorla] 4006 4007 *) Add Next Protocol Negotiation, 4008 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 4009 disabled with a no-npn flag to config or Configure. Code donated 4010 by Google. 4011 [Adam Langley <agl@google.com> and Ben Laurie] 4012 4013 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 4014 NIST-P256, NIST-P521, with constant-time single point multiplication on 4015 typical inputs. Compiler support for the nonstandard type __uint128_t is 4016 required to use this (present in gcc 4.4 and later, for 64-bit builds). 4017 Code made available under Apache License version 2.0. 4018 4019 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 4020 line to include this in your build of OpenSSL, and run "make depend" (or 4021 "make update"). This enables the following EC_METHODs: 4022 4023 EC_GFp_nistp224_method() 4024 EC_GFp_nistp256_method() 4025 EC_GFp_nistp521_method() 4026 4027 EC_GROUP_new_by_curve_name() will automatically use these (while 4028 EC_GROUP_new_curve_GFp() currently prefers the more flexible 4029 implementations). 4030 [Emilia Käsper, Adam Langley, Bodo Moeller (Google)] 4031 4032 *) Use type ossl_ssize_t instead of ssize_t which isn't available on 4033 all platforms. Move ssize_t definition from e_os.h to the public 4034 header file e_os2.h as it now appears in public header file cms.h 4035 [Steve Henson] 4036 4037 *) New -sigopt option to the ca, req and x509 utilities. Additional 4038 signature parameters can be passed using this option and in 4039 particular PSS. 4040 [Steve Henson] 4041 4042 *) Add RSA PSS signing function. This will generate and set the 4043 appropriate AlgorithmIdentifiers for PSS based on those in the 4044 corresponding EVP_MD_CTX structure. No application support yet. 4045 [Steve Henson] 4046 4047 *) Support for companion algorithm specific ASN1 signing routines. 4048 New function ASN1_item_sign_ctx() signs a pre-initialised 4049 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 4050 the appropriate parameters. 4051 [Steve Henson] 4052 4053 *) Add new algorithm specific ASN1 verification initialisation function 4054 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 4055 handling will be the same no matter what EVP_PKEY_METHOD is used. 4056 Add a PSS handler to support verification of PSS signatures: checked 4057 against a number of sample certificates. 4058 [Steve Henson] 4059 4060 *) Add signature printing for PSS. Add PSS OIDs. 4061 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 4062 4063 *) Add algorithm specific signature printing. An individual ASN1 method 4064 can now print out signatures instead of the standard hex dump. 4065 4066 More complex signatures (e.g. PSS) can print out more meaningful 4067 information. Include DSA version that prints out the signature 4068 parameters r, s. 4069 [Steve Henson] 4070 4071 *) Password based recipient info support for CMS library: implementing 4072 RFC3211. 4073 [Steve Henson] 4074 4075 *) Split password based encryption into PBES2 and PBKDF2 functions. This 4076 neatly separates the code into cipher and PBE sections and is required 4077 for some algorithms that split PBES2 into separate pieces (such as 4078 password based CMS). 4079 [Steve Henson] 4080 4081 *) Session-handling fixes: 4082 - Fix handling of connections that are resuming with a session ID, 4083 but also support Session Tickets. 4084 - Fix a bug that suppressed issuing of a new ticket if the client 4085 presented a ticket with an expired session. 4086 - Try to set the ticket lifetime hint to something reasonable. 4087 - Make tickets shorter by excluding irrelevant information. 4088 - On the client side, don't ignore renewed tickets. 4089 [Adam Langley, Bodo Moeller (Google)] 4090 4091 *) Fix PSK session representation. 4092 [Bodo Moeller] 4093 4094 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 4095 4096 This work was sponsored by Intel. 4097 [Andy Polyakov] 4098 4099 *) Add GCM support to TLS library. Some custom code is needed to split 4100 the IV between the fixed (from PRF) and explicit (from TLS record) 4101 portions. This adds all GCM ciphersuites supported by RFC5288 and 4102 RFC5289. Generalise some AES* cipherstrings to include GCM and 4103 add a special AESGCM string for GCM only. 4104 [Steve Henson] 4105 4106 *) Expand range of ctrls for AES GCM. Permit setting invocation 4107 field on decrypt and retrieval of invocation field only on encrypt. 4108 [Steve Henson] 4109 4110 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 4111 As required by RFC5289 these ciphersuites cannot be used if for 4112 versions of TLS earlier than 1.2. 4113 [Steve Henson] 4114 4115 *) For FIPS capable OpenSSL interpret a NULL default public key method 4116 as unset and return the appropriate default but do *not* set the default. 4117 This means we can return the appropriate method in applications that 4118 switch between FIPS and non-FIPS modes. 4119 [Steve Henson] 4120 4121 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 4122 ENGINE is used then we cannot handle that in the FIPS module so we 4123 keep original code iff non-FIPS operations are allowed. 4124 [Steve Henson] 4125 4126 *) Add -attime option to openssl utilities. 4127 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 4128 4129 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 4130 [Steve Henson] 4131 4132 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 4133 FIPS EC methods unconditionally for now. 4134 [Steve Henson] 4135 4136 *) New build option no-ec2m to disable characteristic 2 code. 4137 [Steve Henson] 4138 4139 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 4140 all cases can be covered as some introduce binary incompatibilities. 4141 [Steve Henson] 4142 4143 *) Redirect RSA operations to FIPS module including keygen, 4144 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 4145 [Steve Henson] 4146 4147 *) Add similar low level API blocking to ciphers. 4148 [Steve Henson] 4149 4150 *) Low level digest APIs are not approved in FIPS mode: any attempt 4151 to use these will cause a fatal error. Applications that *really* want 4152 to use them can use the private_* version instead. 4153 [Steve Henson] 4154 4155 *) Redirect cipher operations to FIPS module for FIPS builds. 4156 [Steve Henson] 4157 4158 *) Redirect digest operations to FIPS module for FIPS builds. 4159 [Steve Henson] 4160 4161 *) Update build system to add "fips" flag which will link in fipscanister.o 4162 for static and shared library builds embedding a signature if needed. 4163 [Steve Henson] 4164 4165 *) Output TLS supported curves in preference order instead of numerical 4166 order. This is currently hardcoded for the highest order curves first. 4167 This should be configurable so applications can judge speed vs strength. 4168 [Steve Henson] 4169 4170 *) Add TLS v1.2 server support for client authentication. 4171 [Steve Henson] 4172 4173 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 4174 and enable MD5. 4175 [Steve Henson] 4176 4177 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 4178 FIPS modules versions. 4179 [Steve Henson] 4180 4181 *) Add TLS v1.2 client side support for client authentication. Keep cache 4182 of handshake records longer as we don't know the hash algorithm to use 4183 until after the certificate request message is received. 4184 [Steve Henson] 4185 4186 *) Initial TLS v1.2 client support. Add a default signature algorithms 4187 extension including all the algorithms we support. Parse new signature 4188 format in client key exchange. Relax some ECC signing restrictions for 4189 TLS v1.2 as indicated in RFC5246. 4190 [Steve Henson] 4191 4192 *) Add server support for TLS v1.2 signature algorithms extension. Switch 4193 to new signature format when needed using client digest preference. 4194 All server ciphersuites should now work correctly in TLS v1.2. No client 4195 support yet and no support for client certificates. 4196 [Steve Henson] 4197 4198 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 4199 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 4200 ciphersuites. At present only RSA key exchange ciphersuites work with 4201 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 4202 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 4203 and version checking. 4204 [Steve Henson] 4205 4206 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 4207 with this defined it will not be affected by any changes to ssl internal 4208 structures. Add several utility functions to allow openssl application 4209 to work with OPENSSL_NO_SSL_INTERN defined. 4210 [Steve Henson] 4211 4212 *) A long standing patch to add support for SRP from EdelWeb (Peter 4213 Sylvester and Christophe Renou) was integrated. 4214 [Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester 4215 <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and 4216 Ben Laurie] 4217 4218 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 4219 [Steve Henson] 4220 4221 *) Permit abbreviated handshakes when renegotiating using the function 4222 SSL_renegotiate_abbreviated(). 4223 [Robin Seggelmann <seggelmann@fh-muenster.de>] 4224 4225 *) Add call to ENGINE_register_all_complete() to 4226 ENGINE_load_builtin_engines(), so some implementations get used 4227 automatically instead of needing explicit application support. 4228 [Steve Henson] 4229 4230 *) Add support for TLS key exporter as described in RFC5705. 4231 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 4232 4233 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 4234 a few changes are required: 4235 4236 Add SSL_OP_NO_TLSv1_1 flag. 4237 Add TLSv1_1 methods. 4238 Update version checking logic to handle version 1.1. 4239 Add explicit IV handling (ported from DTLS code). 4240 Add command line options to s_client/s_server. 4241 [Steve Henson] 4242 4243 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 4244 4245 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 4246 in CMS and PKCS7 code. When RSA decryption fails use a random key for 4247 content decryption and always return the same error. Note: this attack 4248 needs on average 2^20 messages so it only affects automated senders. The 4249 old behaviour can be re-enabled in the CMS code by setting the 4250 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 4251 an MMA defence is not necessary. 4252 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 4253 this issue. (CVE-2012-0884) 4254 [Steve Henson] 4255 4256 *) Fix CVE-2011-4619: make sure we really are receiving a 4257 client hello before rejecting multiple SGC restarts. Thanks to 4258 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 4259 [Steve Henson] 4260 4261 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 4262 4263 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 4264 Thanks to Antonio Martin, Enterprise Secure Access Research and 4265 Development, Cisco Systems, Inc. for discovering this bug and 4266 preparing a fix. (CVE-2012-0050) 4267 [Antonio Martin] 4268 4269 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 4270 4271 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 4272 of the Vaudenay padding oracle attack on CBC mode encryption 4273 which enables an efficient plaintext recovery attack against 4274 the OpenSSL implementation of DTLS. Their attack exploits timing 4275 differences arising during decryption processing. A research 4276 paper describing this attack can be found at: 4277 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 4278 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 4279 Security Group at Royal Holloway, University of London 4280 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 4281 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 4282 for preparing the fix. (CVE-2011-4108) 4283 [Robin Seggelmann, Michael Tuexen] 4284 4285 *) Clear bytes used for block padding of SSL 3.0 records. 4286 (CVE-2011-4576) 4287 [Adam Langley (Google)] 4288 4289 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 4290 Kadianakis <desnacked@gmail.com> for discovering this issue and 4291 Adam Langley for preparing the fix. (CVE-2011-4619) 4292 [Adam Langley (Google)] 4293 4294 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 4295 [Andrey Kulikov <amdeich@gmail.com>] 4296 4297 *) Prevent malformed RFC3779 data triggering an assertion failure. 4298 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 4299 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 4300 [Rob Austein <sra@hactrn.net>] 4301 4302 *) Improved PRNG seeding for VOS. 4303 [Paul Green <Paul.Green@stratus.com>] 4304 4305 *) Fix ssl_ciph.c set-up race. 4306 [Adam Langley (Google)] 4307 4308 *) Fix spurious failures in ecdsatest.c. 4309 [Emilia Käsper (Google)] 4310 4311 *) Fix the BIO_f_buffer() implementation (which was mixing different 4312 interpretations of the '..._len' fields). 4313 [Adam Langley (Google)] 4314 4315 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 4316 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 4317 threads won't reuse the same blinding coefficients. 4318 4319 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 4320 lock to call BN_BLINDING_invert_ex, and avoids one use of 4321 BN_BLINDING_update for each BN_BLINDING structure (previously, 4322 the last update always remained unused). 4323 [Emilia Käsper (Google)] 4324 4325 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 4326 [Bob Buckholz (Google)] 4327 4328 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 4329 4330 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 4331 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 4332 [Kaspar Brand <ossl@velox.ch>] 4333 4334 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 4335 for multi-threaded use of ECDH. (CVE-2011-3210) 4336 [Adam Langley (Google)] 4337 4338 *) Fix x509_name_ex_d2i memory leak on bad inputs. 4339 [Bodo Moeller] 4340 4341 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 4342 signature public key algorithm by using OID xref utilities instead. 4343 Before this you could only use some ECC ciphersuites with SHA1 only. 4344 [Steve Henson] 4345 4346 *) Add protection against ECDSA timing attacks as mentioned in the paper 4347 by Billy Bob Brumley and Nicola Tuveri, see: 4348 4349 http://eprint.iacr.org/2011/232.pdf 4350 4351 [Billy Bob Brumley and Nicola Tuveri] 4352 4353 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 4354 4355 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 4356 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 4357 4358 *) Fix bug in string printing code: if *any* escaping is enabled we must 4359 escape the escape character (backslash) or the resulting string is 4360 ambiguous. 4361 [Steve Henson] 4362 4363 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 4364 4365 *) Disable code workaround for ancient and obsolete Netscape browsers 4366 and servers: an attacker can use it in a ciphersuite downgrade attack. 4367 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 4368 [Steve Henson] 4369 4370 *) Fixed J-PAKE implementation error, originally discovered by 4371 Sebastien Martini, further info and confirmation from Stefan 4372 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 4373 [Ben Laurie] 4374 4375 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 4376 4377 *) Fix extension code to avoid race conditions which can result in a buffer 4378 overrun vulnerability: resumed sessions must not be modified as they can 4379 be shared by multiple threads. CVE-2010-3864 4380 [Steve Henson] 4381 4382 *) Fix WIN32 build system to correctly link an ENGINE directory into 4383 a DLL. 4384 [Steve Henson] 4385 4386 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 4387 4388 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 4389 (CVE-2010-1633) 4390 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 4391 4392 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 4393 4394 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 4395 context. The operation can be customised via the ctrl mechanism in 4396 case ENGINEs want to include additional functionality. 4397 [Steve Henson] 4398 4399 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 4400 [Steve Henson] 4401 4402 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 4403 output hashes compatible with older versions of OpenSSL. 4404 [Willy Weisz <weisz@vcpc.univie.ac.at>] 4405 4406 *) Fix compression algorithm handling: if resuming a session use the 4407 compression algorithm of the resumed session instead of determining 4408 it from client hello again. Don't allow server to change algorithm. 4409 [Steve Henson] 4410 4411 *) Add load_crls() function to apps tidying load_certs() too. Add option 4412 to verify utility to allow additional CRLs to be included. 4413 [Steve Henson] 4414 4415 *) Update OCSP request code to permit adding custom headers to the request: 4416 some responders need this. 4417 [Steve Henson] 4418 4419 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 4420 correctly. 4421 [Julia Lawall <julia@diku.dk>] 4422 4423 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 4424 needlessly dereferenced structures, used obsolete functions and 4425 didn't handle all updated verify codes correctly. 4426 [Steve Henson] 4427 4428 *) Disable MD2 in the default configuration. 4429 [Steve Henson] 4430 4431 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 4432 indicate the initial BIO being pushed or popped. This makes it possible 4433 to determine whether the BIO is the one explicitly called or as a result 4434 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 4435 it handles reference counts correctly and doesn't zero out the I/O bio 4436 when it is not being explicitly popped. WARNING: applications which 4437 included workarounds for the old buggy behaviour will need to be modified 4438 or they could free up already freed BIOs. 4439 [Steve Henson] 4440 4441 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 4442 renaming to all platforms (within the 0.9.8 branch, this was 4443 done conditionally on Netware platforms to avoid a name clash). 4444 [Guenter <lists@gknw.net>] 4445 4446 *) Add ECDHE and PSK support to DTLS. 4447 [Michael Tuexen <tuexen@fh-muenster.de>] 4448 4449 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 4450 be used on C++. 4451 [Steve Henson] 4452 4453 *) Add "missing" function EVP_MD_flags() (without this the only way to 4454 retrieve a digest flags is by accessing the structure directly. Update 4455 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 4456 or cipher is registered as in the "from" argument. Print out all 4457 registered digests in the dgst usage message instead of manually 4458 attempting to work them out. 4459 [Steve Henson] 4460 4461 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 4462 this allows the use of compression and extensions. Change default cipher 4463 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 4464 by default unless an application cipher string requests it. 4465 [Steve Henson] 4466 4467 *) Alter match criteria in PKCS12_parse(). It used to try to use local 4468 key ids to find matching certificates and keys but some PKCS#12 files 4469 don't follow the (somewhat unwritten) rules and this strategy fails. 4470 Now just gather all certificates together and the first private key 4471 then look for the first certificate that matches the key. 4472 [Steve Henson] 4473 4474 *) Support use of registered digest and cipher names for dgst and cipher 4475 commands instead of having to add each one as a special case. So now 4476 you can do: 4477 4478 openssl sha256 foo 4479 4480 as well as: 4481 4482 openssl dgst -sha256 foo 4483 4484 and this works for ENGINE based algorithms too. 4485 4486 [Steve Henson] 4487 4488 *) Update Gost ENGINE to support parameter files. 4489 [Victor B. Wagner <vitus@cryptocom.ru>] 4490 4491 *) Support GeneralizedTime in ca utility. 4492 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 4493 4494 *) Enhance the hash format used for certificate directory links. The new 4495 form uses the canonical encoding (meaning equivalent names will work 4496 even if they aren't identical) and uses SHA1 instead of MD5. This form 4497 is incompatible with the older format and as a result c_rehash should 4498 be used to rebuild symbolic links. 4499 [Steve Henson] 4500 4501 *) Make PKCS#8 the default write format for private keys, replacing the 4502 traditional format. This form is standardised, more secure and doesn't 4503 include an implicit MD5 dependency. 4504 [Steve Henson] 4505 4506 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 4507 committed to OpenSSL should pass this lot as a minimum. 4508 [Steve Henson] 4509 4510 *) Add session ticket override functionality for use by EAP-FAST. 4511 [Jouni Malinen <j@w1.fi>] 4512 4513 *) Modify HMAC functions to return a value. Since these can be implemented 4514 in an ENGINE errors can occur. 4515 [Steve Henson] 4516 4517 *) Type-checked OBJ_bsearch_ex. 4518 [Ben Laurie] 4519 4520 *) Type-checked OBJ_bsearch. Also some constification necessitated 4521 by type-checking. Still to come: TXT_DB, bsearch(?), 4522 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 4523 CONF_VALUE. 4524 [Ben Laurie] 4525 4526 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 4527 seconds to a tm structure directly, instead of going through OS 4528 specific date routines. This avoids any issues with OS routines such 4529 as the year 2038 bug. New *_adj() functions for ASN1 time structures 4530 and X509_time_adj_ex() to cover the extended range. The existing 4531 X509_time_adj() is still usable and will no longer have any date issues. 4532 [Steve Henson] 4533 4534 *) Delta CRL support. New use deltas option which will attempt to locate 4535 and search any appropriate delta CRLs available. 4536 4537 This work was sponsored by Google. 4538 [Steve Henson] 4539 4540 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 4541 code and add additional score elements. Validate alternate CRL paths 4542 as part of the CRL checking and indicate a new error "CRL path validation 4543 error" in this case. Applications wanting additional details can use 4544 the verify callback and check the new "parent" field. If this is not 4545 NULL CRL path validation is taking place. Existing applications won't 4546 see this because it requires extended CRL support which is off by 4547 default. 4548 4549 This work was sponsored by Google. 4550 [Steve Henson] 4551 4552 *) Support for freshest CRL extension. 4553 4554 This work was sponsored by Google. 4555 [Steve Henson] 4556 4557 *) Initial indirect CRL support. Currently only supported in the CRLs 4558 passed directly and not via lookup. Process certificate issuer 4559 CRL entry extension and lookup CRL entries by bother issuer name 4560 and serial number. Check and process CRL issuer entry in IDP extension. 4561 4562 This work was sponsored by Google. 4563 [Steve Henson] 4564 4565 *) Add support for distinct certificate and CRL paths. The CRL issuer 4566 certificate is validated separately in this case. Only enabled if 4567 an extended CRL support flag is set: this flag will enable additional 4568 CRL functionality in future. 4569 4570 This work was sponsored by Google. 4571 [Steve Henson] 4572 4573 *) Add support for policy mappings extension. 4574 4575 This work was sponsored by Google. 4576 [Steve Henson] 4577 4578 *) Fixes to pathlength constraint, self issued certificate handling, 4579 policy processing to align with RFC3280 and PKITS tests. 4580 4581 This work was sponsored by Google. 4582 [Steve Henson] 4583 4584 *) Support for name constraints certificate extension. DN, email, DNS 4585 and URI types are currently supported. 4586 4587 This work was sponsored by Google. 4588 [Steve Henson] 4589 4590 *) To cater for systems that provide a pointer-based thread ID rather 4591 than numeric, deprecate the current numeric thread ID mechanism and 4592 replace it with a structure and associated callback type. This 4593 mechanism allows a numeric "hash" to be extracted from a thread ID in 4594 either case, and on platforms where pointers are larger than 'long', 4595 mixing is done to help ensure the numeric 'hash' is usable even if it 4596 can't be guaranteed unique. The default mechanism is to use "&errno" 4597 as a pointer-based thread ID to distinguish between threads. 4598 4599 Applications that want to provide their own thread IDs should now use 4600 CRYPTO_THREADID_set_callback() to register a callback that will call 4601 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 4602 4603 Note that ERR_remove_state() is now deprecated, because it is tied 4604 to the assumption that thread IDs are numeric. ERR_remove_state(0) 4605 to free the current thread's error state should be replaced by 4606 ERR_remove_thread_state(NULL). 4607 4608 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 4609 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 4610 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 4611 application was previously providing a numeric thread callback that 4612 was inappropriate for distinguishing threads, then uniqueness might 4613 have been obtained with &errno that happened immediately in the 4614 intermediate development versions of OpenSSL; this is no longer the 4615 case, the numeric thread callback will now override the automatic use 4616 of &errno.) 4617 [Geoff Thorpe, with help from Bodo Moeller] 4618 4619 *) Initial support for different CRL issuing certificates. This covers a 4620 simple case where the self issued certificates in the chain exist and 4621 the real CRL issuer is higher in the existing chain. 4622 4623 This work was sponsored by Google. 4624 [Steve Henson] 4625 4626 *) Removed effectively defunct crypto/store from the build. 4627 [Ben Laurie] 4628 4629 *) Revamp of STACK to provide stronger type-checking. Still to come: 4630 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 4631 ASN1_STRING, CONF_VALUE. 4632 [Ben Laurie] 4633 4634 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 4635 RAM on SSL connections. This option can save about 34k per idle SSL. 4636 [Nick Mathewson] 4637 4638 *) Revamp of LHASH to provide stronger type-checking. Still to come: 4639 STACK, TXT_DB, bsearch, qsort. 4640 [Ben Laurie] 4641 4642 *) Initial support for Cryptographic Message Syntax (aka CMS) based 4643 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 4644 support for data, signedData, compressedData, digestedData and 4645 encryptedData, envelopedData types included. Scripts to check against 4646 RFC4134 examples draft and interop and consistency checks of many 4647 content types and variants. 4648 [Steve Henson] 4649 4650 *) Add options to enc utility to support use of zlib compression BIO. 4651 [Steve Henson] 4652 4653 *) Extend mk1mf to support importing of options and assembly language 4654 files from Configure script, currently only included in VC-WIN32. 4655 The assembly language rules can now optionally generate the source 4656 files from the associated perl scripts. 4657 [Steve Henson] 4658 4659 *) Implement remaining functionality needed to support GOST ciphersuites. 4660 Interop testing has been performed using CryptoPro implementations. 4661 [Victor B. Wagner <vitus@cryptocom.ru>] 4662 4663 *) s390x assembler pack. 4664 [Andy Polyakov] 4665 4666 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 4667 "family." 4668 [Andy Polyakov] 4669 4670 *) Implement Opaque PRF Input TLS extension as specified in 4671 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 4672 official specification yet and no extension type assignment by 4673 IANA exists, this extension (for now) will have to be explicitly 4674 enabled when building OpenSSL by providing the extension number 4675 to use. For example, specify an option 4676 4677 -DTLSEXT_TYPE_opaque_prf_input=0x9527 4678 4679 to the "config" or "Configure" script to enable the extension, 4680 assuming extension number 0x9527 (which is a completely arbitrary 4681 and unofficial assignment based on the MD5 hash of the Internet 4682 Draft). Note that by doing so, you potentially lose 4683 interoperability with other TLS implementations since these might 4684 be using the same extension number for other purposes. 4685 4686 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 4687 opaque PRF input value to use in the handshake. This will create 4688 an internal copy of the length-'len' string at 'src', and will 4689 return non-zero for success. 4690 4691 To get more control and flexibility, provide a callback function 4692 by using 4693 4694 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 4695 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 4696 4697 where 4698 4699 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 4700 void *arg; 4701 4702 Callback function 'cb' will be called in handshakes, and is 4703 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 4704 Argument 'arg' is for application purposes (the value as given to 4705 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 4706 be provided to the callback function). The callback function 4707 has to return non-zero to report success: usually 1 to use opaque 4708 PRF input just if possible, or 2 to enforce use of the opaque PRF 4709 input. In the latter case, the library will abort the handshake 4710 if opaque PRF input is not successfully negotiated. 4711 4712 Arguments 'peerinput' and 'len' given to the callback function 4713 will always be NULL and 0 in the case of a client. A server will 4714 see the client's opaque PRF input through these variables if 4715 available (NULL and 0 otherwise). Note that if the server 4716 provides an opaque PRF input, the length must be the same as the 4717 length of the client's opaque PRF input. 4718 4719 Note that the callback function will only be called when creating 4720 a new session (session resumption can resume whatever was 4721 previously negotiated), and will not be called in SSL 2.0 4722 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 4723 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 4724 for applications that need to enforce opaque PRF input. 4725 4726 [Bodo Moeller] 4727 4728 *) Update ssl code to support digests other than SHA1+MD5 for handshake 4729 MAC. 4730 4731 [Victor B. Wagner <vitus@cryptocom.ru>] 4732 4733 *) Add RFC4507 support to OpenSSL. This includes the corrections in 4734 RFC4507bis. The encrypted ticket format is an encrypted encoded 4735 SSL_SESSION structure, that way new session features are automatically 4736 supported. 4737 4738 If a client application caches session in an SSL_SESSION structure 4739 support is transparent because tickets are now stored in the encoded 4740 SSL_SESSION. 4741 4742 The SSL_CTX structure automatically generates keys for ticket 4743 protection in servers so again support should be possible 4744 with no application modification. 4745 4746 If a client or server wishes to disable RFC4507 support then the option 4747 SSL_OP_NO_TICKET can be set. 4748 4749 Add a TLS extension debugging callback to allow the contents of any client 4750 or server extensions to be examined. 4751 4752 This work was sponsored by Google. 4753 [Steve Henson] 4754 4755 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 4756 OpenSSL should now compile cleanly on gcc 4.2 4757 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 4758 4759 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 4760 support including streaming MAC support: this is required for GOST 4761 ciphersuite support. 4762 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 4763 4764 *) Add option -stream to use PKCS#7 streaming in smime utility. New 4765 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 4766 to output in BER and PEM format. 4767 [Steve Henson] 4768 4769 *) Experimental support for use of HMAC via EVP_PKEY interface. This 4770 allows HMAC to be handled via the EVP_DigestSign*() interface. The 4771 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 4772 ENGINE support for HMAC keys which are unextractable. New -mac and 4773 -macopt options to dgst utility. 4774 [Steve Henson] 4775 4776 *) New option -sigopt to dgst utility. Update dgst to use 4777 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 4778 alternative signing parameters such as X9.31 or PSS in the dgst 4779 utility. 4780 [Steve Henson] 4781 4782 *) Change ssl_cipher_apply_rule(), the internal function that does 4783 the work each time a ciphersuite string requests enabling 4784 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 4785 removing ("!foo+bar") a class of ciphersuites: Now it maintains 4786 the order of disabled ciphersuites such that those ciphersuites 4787 that most recently went from enabled to disabled not only stay 4788 in order with respect to each other, but also have higher priority 4789 than other disabled ciphersuites the next time ciphersuites are 4790 enabled again. 4791 4792 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 4793 the same ciphersuites as with "HIGH" alone, but in a specific 4794 order where the PSK ciphersuites come first (since they are the 4795 most recently disabled ciphersuites when "HIGH" is parsed). 4796 4797 Also, change ssl_create_cipher_list() (using this new 4798 functionality) such that between otherwise identical 4799 ciphersuites, ephemeral ECDH is preferred over ephemeral DH in 4800 the default order. 4801 [Bodo Moeller] 4802 4803 *) Change ssl_create_cipher_list() so that it automatically 4804 arranges the ciphersuites in reasonable order before starting 4805 to process the rule string. Thus, the definition for "DEFAULT" 4806 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 4807 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 4808 This makes it much easier to arrive at a reasonable default order 4809 in applications for which anonymous ciphers are OK (meaning 4810 that you can't actually use DEFAULT). 4811 [Bodo Moeller; suggested by Victor Duchovni] 4812 4813 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 4814 processing) into multiple integers instead of setting 4815 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 4816 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 4817 (These masks as well as the individual bit definitions are hidden 4818 away into the non-exported interface ssl/ssl_locl.h, so this 4819 change to the definition of the SSL_CIPHER structure shouldn't 4820 affect applications.) This give us more bits for each of these 4821 categories, so there is no longer a need to coagulate AES128 and 4822 AES256 into a single algorithm bit, and to coagulate Camellia128 4823 and Camellia256 into a single algorithm bit, which has led to all 4824 kinds of kludges. 4825 4826 Thus, among other things, the kludge introduced in 0.9.7m and 4827 0.9.8e for masking out AES256 independently of AES128 or masking 4828 out Camellia256 independently of AES256 is not needed here in 0.9.9. 4829 4830 With the change, we also introduce new ciphersuite aliases that 4831 so far were missing: "AES128", "AES256", "CAMELLIA128", and 4832 "CAMELLIA256". 4833 [Bodo Moeller] 4834 4835 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 4836 Use the leftmost N bytes of the signature input if the input is 4837 larger than the prime q (with N being the size in bytes of q). 4838 [Nils Larsch] 4839 4840 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 4841 it yet and it is largely untested. 4842 [Steve Henson] 4843 4844 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 4845 [Nils Larsch] 4846 4847 *) Initial incomplete changes to avoid need for function casts in OpenSSL 4848 some compilers (gcc 4.2 and later) reject their use. Safestack is 4849 reimplemented. Update ASN1 to avoid use of legacy functions. 4850 [Steve Henson] 4851 4852 *) Win32/64 targets are linked with Winsock2. 4853 [Andy Polyakov] 4854 4855 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 4856 to external functions. This can be used to increase CRL handling 4857 efficiency especially when CRLs are very large by (for example) storing 4858 the CRL revoked certificates in a database. 4859 [Steve Henson] 4860 4861 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 4862 new CRLs added to a directory can be used. New command line option 4863 -verify_return_error to s_client and s_server. This causes real errors 4864 to be returned by the verify callback instead of carrying on no matter 4865 what. This reflects the way a "real world" verify callback would behave. 4866 [Steve Henson] 4867 4868 *) GOST engine, supporting several GOST algorithms and public key formats. 4869 Kindly donated by Cryptocom. 4870 [Cryptocom] 4871 4872 *) Partial support for Issuing Distribution Point CRL extension. CRLs 4873 partitioned by DP are handled but no indirect CRL or reason partitioning 4874 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 4875 selected via a scoring technique which handles IDP and AKID in CRLs. 4876 [Steve Henson] 4877 4878 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 4879 will ultimately be used for all verify operations: this will remove the 4880 X509_STORE dependency on certificate verification and allow alternative 4881 lookup methods. X509_STORE based implementations of these two callbacks. 4882 [Steve Henson] 4883 4884 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 4885 Modify get_crl() to find a valid (unexpired) CRL if possible. 4886 [Steve Henson] 4887 4888 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 4889 this would be called X509_CRL_cmp() but that name is already used by 4890 a function that just compares CRL issuer names. Cache several CRL 4891 extensions in X509_CRL structure and cache CRLDP in X509. 4892 [Steve Henson] 4893 4894 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 4895 this maps equivalent X509_NAME structures into a consistent structure. 4896 Name comparison can then be performed rapidly using memcmp(). 4897 [Steve Henson] 4898 4899 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 4900 utility. 4901 [Steve Henson] 4902 4903 *) Allow digests to supply their own micalg string for S/MIME type using 4904 the ctrl EVP_MD_CTRL_MICALG. 4905 [Steve Henson] 4906 4907 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 4908 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 4909 ctrl. It can then customise the structure before and/or after signing 4910 if necessary. 4911 [Steve Henson] 4912 4913 *) New function OBJ_add_sigid() to allow application defined signature OIDs 4914 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 4915 to free up any added signature OIDs. 4916 [Steve Henson] 4917 4918 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 4919 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 4920 digest and cipher tables. New options added to openssl utility: 4921 list-message-digest-algorithms and list-cipher-algorithms. 4922 [Steve Henson] 4923 4924 *) Change the array representation of binary polynomials: the list 4925 of degrees of non-zero coefficients is now terminated with -1. 4926 Previously it was terminated with 0, which was also part of the 4927 value; thus, the array representation was not applicable to 4928 polynomials where t^0 has coefficient zero. This change makes 4929 the array representation useful in a more general context. 4930 [Douglas Stebila] 4931 4932 *) Various modifications and fixes to SSL/TLS cipher string 4933 handling. For ECC, the code now distinguishes between fixed ECDH 4934 with RSA certificates on the one hand and with ECDSA certificates 4935 on the other hand, since these are separate ciphersuites. The 4936 unused code for Fortezza ciphersuites has been removed. 4937 4938 For consistency with EDH, ephemeral ECDH is now called "EECDH" 4939 (not "ECDHE"). For consistency with the code for DH 4940 certificates, use of ECDH certificates is now considered ECDH 4941 authentication, not RSA or ECDSA authentication (the latter is 4942 merely the CA's signing algorithm and not actively used in the 4943 protocol). 4944 4945 The temporary ciphersuite alias "ECCdraft" is no longer 4946 available, and ECC ciphersuites are no longer excluded from "ALL" 4947 and "DEFAULT". The following aliases now exist for RFC 4492 4948 ciphersuites, most of these by analogy with the DH case: 4949 4950 kECDHr - ECDH cert, signed with RSA 4951 kECDHe - ECDH cert, signed with ECDSA 4952 kECDH - ECDH cert (signed with either RSA or ECDSA) 4953 kEECDH - ephemeral ECDH 4954 ECDH - ECDH cert or ephemeral ECDH 4955 4956 aECDH - ECDH cert 4957 aECDSA - ECDSA cert 4958 ECDSA - ECDSA cert 4959 4960 AECDH - anonymous ECDH 4961 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 4962 4963 [Bodo Moeller] 4964 4965 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 4966 Use correct micalg parameters depending on digest(s) in signed message. 4967 [Steve Henson] 4968 4969 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 4970 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 4971 [Steve Henson] 4972 4973 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 4974 an engine to register a method. Add ENGINE lookups for methods and 4975 functional reference processing. 4976 [Steve Henson] 4977 4978 *) New functions EVP_Digest{Sign,Verify)*. These are enhanced versions of 4979 EVP_{Sign,Verify}* which allow an application to customise the signature 4980 process. 4981 [Steve Henson] 4982 4983 *) New -resign option to smime utility. This adds one or more signers 4984 to an existing PKCS#7 signedData structure. Also -md option to use an 4985 alternative message digest algorithm for signing. 4986 [Steve Henson] 4987 4988 *) Tidy up PKCS#7 routines and add new functions to make it easier to 4989 create PKCS7 structures containing multiple signers. Update smime 4990 application to support multiple signers. 4991 [Steve Henson] 4992 4993 *) New -macalg option to pkcs12 utility to allow setting of an alternative 4994 digest MAC. 4995 [Steve Henson] 4996 4997 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 4998 Reorganize PBE internals to lookup from a static table using NIDs, 4999 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 5000 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 5001 PRF which will be automatically used with PBES2. 5002 [Steve Henson] 5003 5004 *) Replace the algorithm specific calls to generate keys in "req" with the 5005 new API. 5006 [Steve Henson] 5007 5008 *) Update PKCS#7 enveloped data routines to use new API. This is now 5009 supported by any public key method supporting the encrypt operation. A 5010 ctrl is added to allow the public key algorithm to examine or modify 5011 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 5012 a no op. 5013 [Steve Henson] 5014 5015 *) Add a ctrl to asn1 method to allow a public key algorithm to express 5016 a default digest type to use. In most cases this will be SHA1 but some 5017 algorithms (such as GOST) need to specify an alternative digest. The 5018 return value indicates how strong the preference is 1 means optional and 5019 2 is mandatory (that is it is the only supported type). Modify 5020 ASN1_item_sign() to accept a NULL digest argument to indicate it should 5021 use the default md. Update openssl utilities to use the default digest 5022 type for signing if it is not explicitly indicated. 5023 [Steve Henson] 5024 5025 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 5026 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 5027 signing method from the key type. This effectively removes the link 5028 between digests and public key types. 5029 [Steve Henson] 5030 5031 *) Add an OID cross reference table and utility functions. Its purpose is to 5032 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 5033 rsaEncryption. This will allow some of the algorithm specific hackery 5034 needed to use the correct OID to be removed. 5035 [Steve Henson] 5036 5037 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 5038 structures for PKCS7_sign(). They are now set up by the relevant public 5039 key ASN1 method. 5040 [Steve Henson] 5041 5042 *) Add provisional EC pkey method with support for ECDSA and ECDH. 5043 [Steve Henson] 5044 5045 *) Add support for key derivation (agreement) in the API, DH method and 5046 pkeyutl. 5047 [Steve Henson] 5048 5049 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 5050 public and private key formats. As a side effect these add additional 5051 command line functionality not previously available: DSA signatures can be 5052 generated and verified using pkeyutl and DH key support and generation in 5053 pkey, genpkey. 5054 [Steve Henson] 5055 5056 *) BeOS support. 5057 [Oliver Tappe <zooey@hirschkaefer.de>] 5058 5059 *) New make target "install_html_docs" installs HTML renditions of the 5060 manual pages. 5061 [Oliver Tappe <zooey@hirschkaefer.de>] 5062 5063 *) New utility "genpkey" this is analogous to "genrsa" etc except it can 5064 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 5065 support key and parameter generation and add initial key generation 5066 functionality for RSA. 5067 [Steve Henson] 5068 5069 *) Add functions for main EVP_PKEY_method operations. The undocumented 5070 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 5071 EVP_PKEY_{encrypt,decrypt}_old. 5072 [Steve Henson] 5073 5074 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 5075 key API, doesn't do much yet. 5076 [Steve Henson] 5077 5078 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 5079 public key algorithms. New option to openssl utility: 5080 "list-public-key-algorithms" to print out info. 5081 [Steve Henson] 5082 5083 *) Implement the Supported Elliptic Curves Extension for 5084 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 5085 [Douglas Stebila] 5086 5087 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 5088 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 5089 [Steve Henson] 5090 5091 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 5092 utilities such as rsa, dsa, dsaparam etc except they process any key 5093 type. 5094 [Steve Henson] 5095 5096 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 5097 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 5098 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 5099 structure. 5100 [Steve Henson] 5101 5102 *) Initial support for pluggable public key ASN1. 5103 De-spaghettify the public key ASN1 handling. Move public and private 5104 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 5105 algorithm specific handling to a single module within the relevant 5106 algorithm directory. Add functions to allow (near) opaque processing 5107 of public and private key structures. 5108 [Steve Henson] 5109 5110 *) Implement the Supported Point Formats Extension for 5111 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 5112 [Douglas Stebila] 5113 5114 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 5115 for the psk identity [hint] and the psk callback functions to the 5116 SSL_SESSION, SSL and SSL_CTX structure. 5117 5118 New ciphersuites: 5119 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 5120 PSK-AES256-CBC-SHA 5121 5122 New functions: 5123 SSL_CTX_use_psk_identity_hint 5124 SSL_get_psk_identity_hint 5125 SSL_get_psk_identity 5126 SSL_use_psk_identity_hint 5127 5128 [Mika Kousa and Pasi Eronen of Nokia Corporation] 5129 5130 *) Add RFC 3161 compliant time stamp request creation, response generation 5131 and response verification functionality. 5132 [Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project] 5133 5134 *) Add initial support for TLS extensions, specifically for the server_name 5135 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 5136 have new members for a host name. The SSL data structure has an 5137 additional member SSL_CTX *initial_ctx so that new sessions can be 5138 stored in that context to allow for session resumption, even after the 5139 SSL has been switched to a new SSL_CTX in reaction to a client's 5140 server_name extension. 5141 5142 New functions (subject to change): 5143 5144 SSL_get_servername() 5145 SSL_get_servername_type() 5146 SSL_set_SSL_CTX() 5147 5148 New CTRL codes and macros (subject to change): 5149 5150 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 5151 - SSL_CTX_set_tlsext_servername_callback() 5152 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 5153 - SSL_CTX_set_tlsext_servername_arg() 5154 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 5155 5156 openssl s_client has a new '-servername ...' option. 5157 5158 openssl s_server has new options '-servername_host ...', '-cert2 ...', 5159 '-key2 ...', '-servername_fatal' (subject to change). This allows 5160 testing the HostName extension for a specific single host name ('-cert' 5161 and '-key' remain fallbacks for handshakes without HostName 5162 negotiation). If the unrecognized_name alert has to be sent, this by 5163 default is a warning; it becomes fatal with the '-servername_fatal' 5164 option. 5165 5166 [Peter Sylvester, Remy Allais, Christophe Renou] 5167 5168 *) Whirlpool hash implementation is added. 5169 [Andy Polyakov] 5170 5171 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 5172 bn(64,32). Because of instruction set limitations it doesn't have 5173 any negative impact on performance. This was done mostly in order 5174 to make it possible to share assembler modules, such as bn_mul_mont 5175 implementations, between 32- and 64-bit builds without hassle. 5176 [Andy Polyakov] 5177 5178 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 5179 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 5180 macro. 5181 [Bodo Moeller] 5182 5183 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 5184 dedicated Montgomery multiplication procedure, is introduced. 5185 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 5186 "64-bit" performance on certain 32-bit targets. 5187 [Andy Polyakov] 5188 5189 *) New option SSL_OP_NO_COMP to disable use of compression selectively 5190 in SSL structures. New SSL ctrl to set maximum send fragment size. 5191 Save memory by setting the I/O buffer sizes dynamically instead of 5192 using the maximum available value. 5193 [Steve Henson] 5194 5195 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 5196 in addition to the text details. 5197 [Bodo Moeller] 5198 5199 *) Very, very preliminary EXPERIMENTAL support for printing of general 5200 ASN1 structures. This currently produces rather ugly output and doesn't 5201 handle several customised structures at all. 5202 [Steve Henson] 5203 5204 *) Integrated support for PVK file format and some related formats such 5205 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 5206 these in the 'rsa' and 'dsa' utilities. 5207 [Steve Henson] 5208 5209 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 5210 [Steve Henson] 5211 5212 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 5213 place for the (very old) "NETSCAPE" format certificates which are now 5214 handled using new ASN1 code equivalents. 5215 [Steve Henson] 5216 5217 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 5218 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 5219 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 5220 [Nils Larsch] 5221 5222 *) Modify CRL distribution points extension code to print out previously 5223 unsupported fields. Enhance extension setting code to allow setting of 5224 all fields. 5225 [Steve Henson] 5226 5227 *) Add print and set support for Issuing Distribution Point CRL extension. 5228 [Steve Henson] 5229 5230 *) Change 'Configure' script to enable Camellia by default. 5231 [NTT] 5232 5233 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 5234 5235 *) When rejecting SSL/TLS records due to an incorrect version number, never 5236 update s->server with a new major version number. As of 5237 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 5238 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 5239 the previous behavior could result in a read attempt at NULL when 5240 receiving specific incorrect SSL/TLS records once record payload 5241 protection is active. (CVE-2010-0740) 5242 [Bodo Moeller, Adam Langley <agl@chromium.org>] 5243 5244 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 5245 could be crashed if the relevant tables were not present (e.g. chrooted). 5246 [Tomas Hoger <thoger@redhat.com>] 5247 5248 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 5249 5250 *) Always check bn_wexpand() return values for failure. (CVE-2009-3245) 5251 [Martin Olsson, Neel Mehta] 5252 5253 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 5254 accommodate for stack sorting, always a write lock!). 5255 [Bodo Moeller] 5256 5257 *) On some versions of WIN32 Heap32Next is very slow. This can cause 5258 excessive delays in the RAND_poll(): over a minute. As a workaround 5259 include a time check in the inner Heap32Next loop too. 5260 [Steve Henson] 5261 5262 *) The code that handled flushing of data in SSL/TLS originally used the 5263 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 5264 the problem outlined in PR#1949. The fix suggested there however can 5265 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 5266 of Apache). So instead simplify the code to flush unconditionally. 5267 This should be fine since flushing with no data to flush is a no op. 5268 [Steve Henson] 5269 5270 *) Handle TLS versions 2.0 and later properly and correctly use the 5271 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 5272 off ancient servers have a habit of sticking around for a while... 5273 [Steve Henson] 5274 5275 *) Modify compression code so it frees up structures without using the 5276 ex_data callbacks. This works around a problem where some applications 5277 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 5278 restarting) then use compression (e.g. SSL with compression) later. 5279 This results in significant per-connection memory leaks and 5280 has caused some security issues including CVE-2008-1678 and 5281 CVE-2009-4355. 5282 [Steve Henson] 5283 5284 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 5285 change when encrypting or decrypting. 5286 [Bodo Moeller] 5287 5288 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 5289 connect and renegotiate with servers which do not support RI. 5290 Until RI is more widely deployed this option is enabled by default. 5291 [Steve Henson] 5292 5293 *) Add "missing" ssl ctrls to clear options and mode. 5294 [Steve Henson] 5295 5296 *) If client attempts to renegotiate and doesn't support RI respond with 5297 a no_renegotiation alert as required by RFC5746. Some renegotiating 5298 TLS clients will continue a connection gracefully when they receive 5299 the alert. Unfortunately OpenSSL mishandled this alert and would hang 5300 waiting for a server hello which it will never receive. Now we treat a 5301 received no_renegotiation alert as a fatal error. This is because 5302 applications requesting a renegotiation might well expect it to succeed 5303 and would have no code in place to handle the server denying it so the 5304 only safe thing to do is to terminate the connection. 5305 [Steve Henson] 5306 5307 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 5308 peer supports secure renegotiation and 0 otherwise. Print out peer 5309 renegotiation support in s_client/s_server. 5310 [Steve Henson] 5311 5312 *) Replace the highly broken and deprecated SPKAC certification method with 5313 the updated NID creation version. This should correctly handle UTF8. 5314 [Steve Henson] 5315 5316 *) Implement RFC5746. Re-enable renegotiation but require the extension 5317 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 5318 turns out to be a bad idea. It has been replaced by 5319 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 5320 SSL_CTX_set_options(). This is really not recommended unless you 5321 know what you are doing. 5322 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 5323 5324 *) Fixes to stateless session resumption handling. Use initial_ctx when 5325 issuing and attempting to decrypt tickets in case it has changed during 5326 servername handling. Use a non-zero length session ID when attempting 5327 stateless session resumption: this makes it possible to determine if 5328 a resumption has occurred immediately after receiving server hello 5329 (several places in OpenSSL subtly assume this) instead of later in 5330 the handshake. 5331 [Steve Henson] 5332 5333 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 5334 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 5335 fixes for a few places where the return code is not checked 5336 correctly. 5337 [Julia Lawall <julia@diku.dk>] 5338 5339 *) Add --strict-warnings option to Configure script to include devteam 5340 warnings in other configurations. 5341 [Steve Henson] 5342 5343 *) Add support for --libdir option and LIBDIR variable in makefiles. This 5344 makes it possible to install openssl libraries in locations which 5345 have names other than "lib", for example "/usr/lib64" which some 5346 systems need. 5347 [Steve Henson, based on patch from Jeremy Utley] 5348 5349 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 5350 X690 8.9.12 and can produce some misleading textual output of OIDs. 5351 [Steve Henson, reported by Dan Kaminsky] 5352 5353 *) Delete MD2 from algorithm tables. This follows the recommendation in 5354 several standards that it is not used in new applications due to 5355 several cryptographic weaknesses. For binary compatibility reasons 5356 the MD2 API is still compiled in by default. 5357 [Steve Henson] 5358 5359 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 5360 and restored. 5361 [Steve Henson] 5362 5363 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 5364 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 5365 clash. 5366 [Guenter <lists@gknw.net>] 5367 5368 *) Fix the server certificate chain building code to use X509_verify_cert(), 5369 it used to have an ad-hoc builder which was unable to cope with anything 5370 other than a simple chain. 5371 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 5372 5373 *) Don't check self signed certificate signatures in X509_verify_cert() 5374 by default (a flag can override this): it just wastes time without 5375 adding any security. As a useful side effect self signed root CAs 5376 with non-FIPS digests are now usable in FIPS mode. 5377 [Steve Henson] 5378 5379 *) In dtls1_process_out_of_seq_message() the check if the current message 5380 is already buffered was missing. For every new message was memory 5381 allocated, allowing an attacker to perform an denial of service attack 5382 with sending out of seq handshake messages until there is no memory 5383 left. Additionally every future message was buffered, even if the 5384 sequence number made no sense and would be part of another handshake. 5385 So only messages with sequence numbers less than 10 in advance will be 5386 buffered. (CVE-2009-1378) 5387 [Robin Seggelmann, discovered by Daniel Mentz] 5388 5389 *) Records are buffered if they arrive with a future epoch to be 5390 processed after finishing the corresponding handshake. There is 5391 currently no limitation to this buffer allowing an attacker to perform 5392 a DOS attack with sending records with future epochs until there is no 5393 memory left. This patch adds the pqueue_size() function to determine 5394 the size of a buffer and limits the record buffer to 100 entries. 5395 (CVE-2009-1377) 5396 [Robin Seggelmann, discovered by Daniel Mentz] 5397 5398 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 5399 parent structure is freed. (CVE-2009-1379) 5400 [Daniel Mentz] 5401 5402 *) Handle non-blocking I/O properly in SSL_shutdown() call. 5403 [Darryl Miles <darryl-mailinglists@netbauds.net>] 5404 5405 *) Add 2.5.4.* OIDs 5406 [Ilya O. <vrghost@gmail.com>] 5407 5408 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 5409 5410 *) Disable renegotiation completely - this fixes a severe security 5411 problem (CVE-2009-3555) at the cost of breaking all 5412 renegotiation. Renegotiation can be re-enabled by setting 5413 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 5414 run-time. This is really not recommended unless you know what 5415 you're doing. 5416 [Ben Laurie] 5417 5418 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 5419 5420 *) Don't set val to NULL when freeing up structures, it is freed up by 5421 underlying code. If sizeof(void *) > sizeof(long) this can result in 5422 zeroing past the valid field. (CVE-2009-0789) 5423 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 5424 5425 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 5426 checked correctly. This would allow some invalid signed attributes to 5427 appear to verify correctly. (CVE-2009-0591) 5428 [Ivan Nestlerode <inestlerode@us.ibm.com>] 5429 5430 *) Reject UniversalString and BMPString types with invalid lengths. This 5431 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 5432 a legal length. (CVE-2009-0590) 5433 [Steve Henson] 5434 5435 *) Set S/MIME signing as the default purpose rather than setting it 5436 unconditionally. This allows applications to override it at the store 5437 level. 5438 [Steve Henson] 5439 5440 *) Permit restricted recursion of ASN1 strings. This is needed in practice 5441 to handle some structures. 5442 [Steve Henson] 5443 5444 *) Improve efficiency of mem_gets: don't search whole buffer each time 5445 for a '\n' 5446 [Jeremy Shapiro <jnshapir@us.ibm.com>] 5447 5448 *) New -hex option for openssl rand. 5449 [Matthieu Herrb] 5450 5451 *) Print out UTF8String and NumericString when parsing ASN1. 5452 [Steve Henson] 5453 5454 *) Support NumericString type for name components. 5455 [Steve Henson] 5456 5457 *) Allow CC in the environment to override the automatically chosen 5458 compiler. Note that nothing is done to ensure flags work with the 5459 chosen compiler. 5460 [Ben Laurie] 5461 5462 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 5463 5464 *) Properly check EVP_VerifyFinal() and similar return values 5465 (CVE-2008-5077). 5466 [Ben Laurie, Bodo Moeller, Google Security Team] 5467 5468 *) Enable TLS extensions by default. 5469 [Ben Laurie] 5470 5471 *) Allow the CHIL engine to be loaded, whether the application is 5472 multithreaded or not. (This does not release the developer from the 5473 obligation to set up the dynamic locking callbacks.) 5474 [Sander Temme <sander@temme.net>] 5475 5476 *) Use correct exit code if there is an error in dgst command. 5477 [Steve Henson; problem pointed out by Roland Dirlewanger] 5478 5479 *) Tweak Configure so that you need to say "experimental-jpake" to enable 5480 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 5481 [Bodo Moeller] 5482 5483 *) Add experimental JPAKE support, including demo authentication in 5484 s_client and s_server. 5485 [Ben Laurie] 5486 5487 *) Set the comparison function in v3_addr_canonize(). 5488 [Rob Austein <sra@hactrn.net>] 5489 5490 *) Add support for XMPP STARTTLS in s_client. 5491 [Philip Paeps <philip@freebsd.org>] 5492 5493 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 5494 to ensure that even with this option, only ciphersuites in the 5495 server's preference list will be accepted. (Note that the option 5496 applies only when resuming a session, so the earlier behavior was 5497 just about the algorithm choice for symmetric cryptography.) 5498 [Bodo Moeller] 5499 5500 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 5501 5502 *) Fix NULL pointer dereference if a DTLS server received 5503 ChangeCipherSpec as first record (CVE-2009-1386). 5504 [PR #1679] 5505 5506 *) Fix a state transition in s3_srvr.c and d1_srvr.c 5507 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 5508 [Nagendra Modadugu] 5509 5510 *) The fix in 0.9.8c that supposedly got rid of unsafe 5511 double-checked locking was incomplete for RSA blinding, 5512 addressing just one layer of what turns out to have been 5513 doubly unsafe triple-checked locking. 5514 5515 So now fix this for real by retiring the MONT_HELPER macro 5516 in crypto/rsa/rsa_eay.c. 5517 5518 [Bodo Moeller; problem pointed out by Marius Schilder] 5519 5520 *) Various precautionary measures: 5521 5522 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 5523 5524 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 5525 (NB: This would require knowledge of the secret session ticket key 5526 to exploit, in which case you'd be SOL either way.) 5527 5528 - Change bn_nist.c so that it will properly handle input BIGNUMs 5529 outside the expected range. 5530 5531 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 5532 builds. 5533 5534 [Neel Mehta, Bodo Moeller] 5535 5536 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 5537 the load fails. Useful for distros. 5538 [Ben Laurie and the FreeBSD team] 5539 5540 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 5541 [Steve Henson] 5542 5543 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 5544 [Huang Ying] 5545 5546 *) Expand ENGINE to support engine supplied SSL client certificate functions. 5547 5548 This work was sponsored by Logica. 5549 [Steve Henson] 5550 5551 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 5552 keystores. Support for SSL/TLS client authentication too. 5553 Not compiled unless enable-capieng specified to Configure. 5554 5555 This work was sponsored by Logica. 5556 [Steve Henson] 5557 5558 *) Fix bug in X509_ATTRIBUTE creation: don't set attribute using 5559 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 5560 attribute creation routines such as certificate requests and PKCS#12 5561 files. 5562 [Steve Henson] 5563 5564 Changes between 0.9.8g and 0.9.8h [28 May 2008] 5565 5566 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 5567 handshake which could lead to a client crash as found using the 5568 Codenomicon TLS test suite (CVE-2008-1672) 5569 [Steve Henson, Mark Cox] 5570 5571 *) Fix double free in TLS server name extensions which could lead to 5572 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 5573 [Joe Orton] 5574 5575 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 5576 5577 Clear the error queue to ensure that error entries left from 5578 older function calls do not interfere with the correct operation. 5579 [Lutz Jaenicke, Erik de Castro Lopo] 5580 5581 *) Remove root CA certificates of commercial CAs: 5582 5583 The OpenSSL project does not recommend any specific CA and does not 5584 have any policy with respect to including or excluding any CA. 5585 Therefore it does not make any sense to ship an arbitrary selection 5586 of root CA certificates with the OpenSSL software. 5587 [Lutz Jaenicke] 5588 5589 *) RSA OAEP patches to fix two separate invalid memory reads. 5590 The first one involves inputs when 'lzero' is greater than 5591 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 5592 before the beginning of from). The second one involves inputs where 5593 the 'db' section contains nothing but zeroes (there is a one-byte 5594 invalid read after the end of 'db'). 5595 [Ivan Nestlerode <inestlerode@us.ibm.com>] 5596 5597 *) Partial backport from 0.9.9-dev: 5598 5599 Introduce bn_mul_mont (dedicated Montgomery multiplication 5600 procedure) as a candidate for BIGNUM assembler implementation. 5601 While 0.9.9-dev uses assembler for various architectures, only 5602 x86_64 is available by default here in the 0.9.8 branch, and 5603 32-bit x86 is available through a compile-time setting. 5604 5605 To try the 32-bit x86 assembler implementation, use Configure 5606 option "enable-montasm" (which exists only for this backport). 5607 5608 As "enable-montasm" for 32-bit x86 disclaims code stability 5609 anyway, in this constellation we activate additional code 5610 backported from 0.9.9-dev for further performance improvements, 5611 namely BN_from_montgomery_word. (To enable this otherwise, 5612 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 5613 5614 [Andy Polyakov (backport partially by Bodo Moeller)] 5615 5616 *) Add TLS session ticket callback. This allows an application to set 5617 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 5618 values. This is useful for key rollover for example where several key 5619 sets may exist with different names. 5620 [Steve Henson] 5621 5622 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 5623 This was broken until now in 0.9.8 releases, such that the only way 5624 a registered ENGINE could be used (assuming it initialises 5625 successfully on the host) was to explicitly set it as the default 5626 for the relevant algorithms. This is in contradiction with 0.9.7 5627 behaviour and the documentation. With this fix, when an ENGINE is 5628 registered into a given algorithm's table of implementations, the 5629 'uptodate' flag is reset so that auto-discovery will be used next 5630 time a new context for that algorithm attempts to select an 5631 implementation. 5632 [Ian Lister (tweaked by Geoff Thorpe)] 5633 5634 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 5635 implementation in the following ways: 5636 5637 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 5638 hard coded. 5639 5640 Lack of BER streaming support means one pass streaming processing is 5641 only supported if data is detached: setting the streaming flag is 5642 ignored for embedded content. 5643 5644 CMS support is disabled by default and must be explicitly enabled 5645 with the enable-cms configuration option. 5646 [Steve Henson] 5647 5648 *) Update the GMP engine glue to do direct copies between BIGNUM and 5649 mpz_t when openssl and GMP use the same limb size. Otherwise the 5650 existing "conversion via a text string export" trick is still used. 5651 [Paul Sheer <paulsheer@gmail.com>] 5652 5653 *) Zlib compression BIO. This is a filter BIO which compressed and 5654 uncompresses any data passed through it. 5655 [Steve Henson] 5656 5657 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 5658 RFC3394 compatible AES key wrapping. 5659 [Steve Henson] 5660 5661 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 5662 sets string data without copying. X509_ALGOR_set0() and 5663 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 5664 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 5665 from an X509_ATTRIBUTE structure optionally checking it occurs only 5666 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 5667 data. 5668 [Steve Henson] 5669 5670 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 5671 to get the expected BN_FLG_CONSTTIME behavior. 5672 [Bodo Moeller (Google)] 5673 5674 *) Netware support: 5675 5676 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 5677 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 5678 - added some more tests to do_tests.pl 5679 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 5680 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 5681 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 5682 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 5683 - various changes to netware.pl to enable gcc-cross builds on Win32 5684 platform 5685 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 5686 - various changes to fix missing prototype warnings 5687 - fixed x86nasm.pl to create correct asm files for NASM COFF output 5688 - added AES, WHIRLPOOL and CPUID assembler code to build files 5689 - added missing AES assembler make rules to mk1mf.pl 5690 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 5691 [Guenter Knauf <eflash@gmx.net>] 5692 5693 *) Implement certificate status request TLS extension defined in RFC3546. 5694 A client can set the appropriate parameters and receive the encoded 5695 OCSP response via a callback. A server can query the supplied parameters 5696 and set the encoded OCSP response in the callback. Add simplified examples 5697 to s_client and s_server. 5698 [Steve Henson] 5699 5700 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 5701 5702 *) Fix various bugs: 5703 + Binary incompatibility of ssl_ctx_st structure 5704 + DTLS interoperation with non-compliant servers 5705 + Don't call get_session_cb() without proposed session 5706 + Fix ia64 assembler code 5707 [Andy Polyakov, Steve Henson] 5708 5709 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 5710 5711 *) DTLS Handshake overhaul. There were longstanding issues with 5712 OpenSSL DTLS implementation, which were making it impossible for 5713 RFC 4347 compliant client to communicate with OpenSSL server. 5714 Unfortunately just fixing these incompatibilities would "cut off" 5715 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 5716 server keeps tolerating non RFC compliant syntax. The opposite is 5717 not true, 0.9.8f client can not communicate with earlier server. 5718 This update even addresses CVE-2007-4995. 5719 [Andy Polyakov] 5720 5721 *) Changes to avoid need for function casts in OpenSSL: some compilers 5722 (gcc 4.2 and later) reject their use. 5723 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 5724 Steve Henson] 5725 5726 *) Add RFC4507 support to OpenSSL. This includes the corrections in 5727 RFC4507bis. The encrypted ticket format is an encrypted encoded 5728 SSL_SESSION structure, that way new session features are automatically 5729 supported. 5730 5731 If a client application caches session in an SSL_SESSION structure 5732 support is transparent because tickets are now stored in the encoded 5733 SSL_SESSION. 5734 5735 The SSL_CTX structure automatically generates keys for ticket 5736 protection in servers so again support should be possible 5737 with no application modification. 5738 5739 If a client or server wishes to disable RFC4507 support then the option 5740 SSL_OP_NO_TICKET can be set. 5741 5742 Add a TLS extension debugging callback to allow the contents of any client 5743 or server extensions to be examined. 5744 5745 This work was sponsored by Google. 5746 [Steve Henson] 5747 5748 *) Add initial support for TLS extensions, specifically for the server_name 5749 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 5750 have new members for a host name. The SSL data structure has an 5751 additional member SSL_CTX *initial_ctx so that new sessions can be 5752 stored in that context to allow for session resumption, even after the 5753 SSL has been switched to a new SSL_CTX in reaction to a client's 5754 server_name extension. 5755 5756 New functions (subject to change): 5757 5758 SSL_get_servername() 5759 SSL_get_servername_type() 5760 SSL_set_SSL_CTX() 5761 5762 New CTRL codes and macros (subject to change): 5763 5764 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 5765 - SSL_CTX_set_tlsext_servername_callback() 5766 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 5767 - SSL_CTX_set_tlsext_servername_arg() 5768 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 5769 5770 openssl s_client has a new '-servername ...' option. 5771 5772 openssl s_server has new options '-servername_host ...', '-cert2 ...', 5773 '-key2 ...', '-servername_fatal' (subject to change). This allows 5774 testing the HostName extension for a specific single host name ('-cert' 5775 and '-key' remain fallbacks for handshakes without HostName 5776 negotiation). If the unrecognized_name alert has to be sent, this by 5777 default is a warning; it becomes fatal with the '-servername_fatal' 5778 option. 5779 5780 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 5781 5782 *) Add AES and SSE2 assembly language support to VC++ build. 5783 [Steve Henson] 5784 5785 *) Mitigate attack on final subtraction in Montgomery reduction. 5786 [Andy Polyakov] 5787 5788 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 5789 (which previously caused an internal error). 5790 [Bodo Moeller] 5791 5792 *) Squeeze another 10% out of IGE mode when in != out. 5793 [Ben Laurie] 5794 5795 *) AES IGE mode speedup. 5796 [Dean Gaudet (Google)] 5797 5798 *) Add the Korean symmetric 128-bit cipher SEED (see 5799 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 5800 add SEED ciphersuites from RFC 4162: 5801 5802 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 5803 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 5804 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 5805 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 5806 5807 To minimize changes between patchlevels in the OpenSSL 0.9.8 5808 series, SEED remains excluded from compilation unless OpenSSL 5809 is configured with 'enable-seed'. 5810 [KISA, Bodo Moeller] 5811 5812 *) Mitigate branch prediction attacks, which can be practical if a 5813 single processor is shared, allowing a spy process to extract 5814 information. For detailed background information, see 5815 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 5816 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 5817 and Necessary Software Countermeasures"). The core of the change 5818 are new versions BN_div_no_branch() and 5819 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 5820 respectively, which are slower, but avoid the security-relevant 5821 conditional branches. These are automatically called by BN_div() 5822 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 5823 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 5824 remove a conditional branch. 5825 5826 BN_FLG_CONSTTIME is the new name for the previous 5827 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 5828 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 5829 in the exponent causes BN_mod_exp_mont() to use the alternative 5830 implementation in BN_mod_exp_mont_consttime().) The old name 5831 remains as a deprecated alias. 5832 5833 Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 5834 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 5835 constant-time implementations for more than just exponentiation. 5836 Here too the old name is kept as a deprecated alias. 5837 5838 BN_BLINDING_new() will now use BN_dup() for the modulus so that 5839 the BN_BLINDING structure gets an independent copy of the 5840 modulus. This means that the previous "BIGNUM *m" argument to 5841 BN_BLINDING_new() and to BN_BLINDING_create_param() now 5842 essentially becomes "const BIGNUM *m", although we can't actually 5843 change this in the header file before 0.9.9. It allows 5844 RSA_setup_blinding() to use BN_with_flags() on the modulus to 5845 enable BN_FLG_CONSTTIME. 5846 5847 [Matthew D Wood (Intel Corp)] 5848 5849 *) In the SSL/TLS server implementation, be strict about session ID 5850 context matching (which matters if an application uses a single 5851 external cache for different purposes). Previously, 5852 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 5853 set. This did ensure strict client verification, but meant that, 5854 with applications using a single external cache for quite 5855 different requirements, clients could circumvent ciphersuite 5856 restrictions for a given session ID context by starting a session 5857 in a different context. 5858 [Bodo Moeller] 5859 5860 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 5861 a ciphersuite string such as "DEFAULT:RSA" cannot enable 5862 authentication-only ciphersuites. 5863 [Bodo Moeller] 5864 5865 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 5866 not complete and could lead to a possible single byte overflow 5867 (CVE-2007-5135) [Ben Laurie] 5868 5869 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 5870 5871 *) Since AES128 and AES256 (and similarly Camellia128 and 5872 Camellia256) share a single mask bit in the logic of 5873 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 5874 kludge to work properly if AES128 is available and AES256 isn't 5875 (or if Camellia128 is available and Camellia256 isn't). 5876 [Victor Duchovni] 5877 5878 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 5879 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 5880 When a point or a seed is encoded in a BIT STRING, we need to 5881 prevent the removal of trailing zero bits to get the proper DER 5882 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 5883 of a NamedBitList, for which trailing 0 bits need to be removed.) 5884 [Bodo Moeller] 5885 5886 *) Have SSL/TLS server implementation tolerate "mismatched" record 5887 protocol version while receiving ClientHello even if the 5888 ClientHello is fragmented. (The server can't insist on the 5889 particular protocol version it has chosen before the ServerHello 5890 message has informed the client about his choice.) 5891 [Bodo Moeller] 5892 5893 *) Add RFC 3779 support. 5894 [Rob Austein for ARIN, Ben Laurie] 5895 5896 *) Load error codes if they are not already present instead of using a 5897 static variable. This allows them to be cleanly unloaded and reloaded. 5898 Improve header file function name parsing. 5899 [Steve Henson] 5900 5901 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 5902 or CAPABILITY handshake as required by RFCs. 5903 [Goetz Babin-Ebell] 5904 5905 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 5906 5907 *) Introduce limits to prevent malicious keys being able to 5908 cause a denial of service. (CVE-2006-2940) 5909 [Steve Henson, Bodo Moeller] 5910 5911 *) Fix ASN.1 parsing of certain invalid structures that can result 5912 in a denial of service. (CVE-2006-2937) [Steve Henson] 5913 5914 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 5915 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 5916 5917 *) Fix SSL client code which could crash if connecting to a 5918 malicious SSLv2 server. (CVE-2006-4343) 5919 [Tavis Ormandy and Will Drewry, Google Security Team] 5920 5921 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 5922 match only those. Before that, "AES256-SHA" would be interpreted 5923 as a pattern and match "AES128-SHA" too (since AES128-SHA got 5924 the same strength classification in 0.9.7h) as we currently only 5925 have a single AES bit in the ciphersuite description bitmap. 5926 That change, however, also applied to ciphersuite strings such as 5927 "RC4-MD5" that intentionally matched multiple ciphersuites -- 5928 namely, SSL 2.0 ciphersuites in addition to the more common ones 5929 from SSL 3.0/TLS 1.0. 5930 5931 So we change the selection algorithm again: Naming an explicit 5932 ciphersuite selects this one ciphersuite, and any other similar 5933 ciphersuite (same bitmap) from *other* protocol versions. 5934 Thus, "RC4-MD5" again will properly select both the SSL 2.0 5935 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 5936 5937 Since SSL 2.0 does not have any ciphersuites for which the 5938 128/256 bit distinction would be relevant, this works for now. 5939 The proper fix will be to use different bits for AES128 and 5940 AES256, which would have avoided the problems from the beginning; 5941 however, bits are scarce, so we can only do this in a new release 5942 (not just a patchlevel) when we can change the SSL_CIPHER 5943 definition to split the single 'unsigned long mask' bitmap into 5944 multiple values to extend the available space. 5945 5946 [Bodo Moeller] 5947 5948 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 5949 5950 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 5951 (CVE-2006-4339) [Ben Laurie and Google Security Team] 5952 5953 *) Add AES IGE and biIGE modes. 5954 [Ben Laurie] 5955 5956 *) Change the Unix randomness entropy gathering to use poll() when 5957 possible instead of select(), since the latter has some 5958 undesirable limitations. 5959 [Darryl Miles via Richard Levitte and Bodo Moeller] 5960 5961 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 5962 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 5963 cannot be implicitly activated as part of, e.g., the "AES" alias. 5964 However, please upgrade to OpenSSL 0.9.9[-dev] for 5965 non-experimental use of the ECC ciphersuites to get TLS extension 5966 support, which is required for curve and point format negotiation 5967 to avoid potential handshake problems. 5968 [Bodo Moeller] 5969 5970 *) Disable rogue ciphersuites: 5971 5972 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 5973 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 5974 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 5975 5976 The latter two were purportedly from 5977 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 5978 appear there. 5979 5980 Also deactivate the remaining ciphersuites from 5981 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 5982 unofficial, and the ID has long expired. 5983 [Bodo Moeller] 5984 5985 *) Fix RSA blinding Heisenbug (problems sometimes occurred on 5986 dual-core machines) and other potential thread-safety issues. 5987 [Bodo Moeller] 5988 5989 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 5990 versions), which is now available for royalty-free use 5991 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 5992 Also, add Camellia TLS ciphersuites from RFC 4132. 5993 5994 To minimize changes between patchlevels in the OpenSSL 0.9.8 5995 series, Camellia remains excluded from compilation unless OpenSSL 5996 is configured with 'enable-camellia'. 5997 [NTT] 5998 5999 *) Disable the padding bug check when compression is in use. The padding 6000 bug check assumes the first packet is of even length, this is not 6001 necessarily true if compression is enabled and can result in false 6002 positives causing handshake failure. The actual bug test is ancient 6003 code so it is hoped that implementations will either have fixed it by 6004 now or any which still have the bug do not support compression. 6005 [Steve Henson] 6006 6007 Changes between 0.9.8a and 0.9.8b [04 May 2006] 6008 6009 *) When applying a cipher rule check to see if string match is an explicit 6010 cipher suite and only match that one cipher suite if it is. 6011 [Steve Henson] 6012 6013 *) Link in manifests for VC++ if needed. 6014 [Austin Ziegler <halostatue@gmail.com>] 6015 6016 *) Update support for ECC-based TLS ciphersuites according to 6017 draft-ietf-tls-ecc-12.txt with proposed changes (but without 6018 TLS extensions, which are supported starting with the 0.9.9 6019 branch, not in the OpenSSL 0.9.8 branch). 6020 [Douglas Stebila] 6021 6022 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 6023 opaque EVP_CIPHER_CTX handling. 6024 [Steve Henson] 6025 6026 *) Fixes and enhancements to zlib compression code. We now only use 6027 "zlib1.dll" and use the default __cdecl calling convention on Win32 6028 to conform with the standards mentioned here: 6029 http://www.zlib.net/DLL_FAQ.txt 6030 Static zlib linking now works on Windows and the new --with-zlib-include 6031 --with-zlib-lib options to Configure can be used to supply the location 6032 of the headers and library. Gracefully handle case where zlib library 6033 can't be loaded. 6034 [Steve Henson] 6035 6036 *) Several fixes and enhancements to the OID generation code. The old code 6037 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 6038 handle numbers larger than ULONG_MAX, truncated printing and had a 6039 non standard OBJ_obj2txt() behaviour. 6040 [Steve Henson] 6041 6042 *) Add support for building of engines under engine/ as shared libraries 6043 under VC++ build system. 6044 [Steve Henson] 6045 6046 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 6047 Hopefully, we will not see any false combination of paths any more. 6048 [Richard Levitte] 6049 6050 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 6051 6052 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 6053 (part of SSL_OP_ALL). This option used to disable the 6054 countermeasure against man-in-the-middle protocol-version 6055 rollback in the SSL 2.0 server implementation, which is a bad 6056 idea. (CVE-2005-2969) 6057 6058 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 6059 for Information Security, National Institute of Advanced Industrial 6060 Science and Technology [AIST], Japan)] 6061 6062 *) Add two function to clear and return the verify parameter flags. 6063 [Steve Henson] 6064 6065 *) Keep cipherlists sorted in the source instead of sorting them at 6066 runtime, thus removing the need for a lock. 6067 [Nils Larsch] 6068 6069 *) Avoid some small subgroup attacks in Diffie-Hellman. 6070 [Nick Mathewson and Ben Laurie] 6071 6072 *) Add functions for well-known primes. 6073 [Nick Mathewson] 6074 6075 *) Extended Windows CE support. 6076 [Satoshi Nakamura and Andy Polyakov] 6077 6078 *) Initialize SSL_METHOD structures at compile time instead of during 6079 runtime, thus removing the need for a lock. 6080 [Steve Henson] 6081 6082 *) Make PKCS7_decrypt() work even if no certificate is supplied by 6083 attempting to decrypt each encrypted key in turn. Add support to 6084 smime utility. 6085 [Steve Henson] 6086 6087 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 6088 6089 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 6090 OpenSSL 0.9.8.] 6091 6092 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 6093 [Richard Levitte] 6094 6095 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 6096 key into the same file any more. 6097 [Richard Levitte] 6098 6099 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 6100 [Andy Polyakov] 6101 6102 *) Add -utf8 command line and config file option to 'ca'. 6103 [Stefan <stf@udoma.org] 6104 6105 *) Removed the macro des_crypt(), as it seems to conflict with some 6106 libraries. Use DES_crypt(). 6107 [Richard Levitte] 6108 6109 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 6110 involves renaming the source and generated shared-libs for 6111 both. The engines will accept the corrected or legacy ids 6112 ('ncipher' and '4758_cca' respectively) when binding. NB, 6113 this only applies when building 'shared'. 6114 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 6115 6116 *) Add attribute functions to EVP_PKEY structure. Modify 6117 PKCS12_create() to recognize a CSP name attribute and 6118 use it. Make -CSP option work again in pkcs12 utility. 6119 [Steve Henson] 6120 6121 *) Add new functionality to the bn blinding code: 6122 - automatic re-creation of the BN_BLINDING parameters after 6123 a fixed number of uses (currently 32) 6124 - add new function for parameter creation 6125 - introduce flags to control the update behaviour of the 6126 BN_BLINDING parameters 6127 - hide BN_BLINDING structure 6128 Add a second BN_BLINDING slot to the RSA structure to improve 6129 performance when a single RSA object is shared among several 6130 threads. 6131 [Nils Larsch] 6132 6133 *) Add support for DTLS. 6134 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 6135 6136 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 6137 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 6138 [Walter Goulet] 6139 6140 *) Remove buggy and incomplete DH cert support from 6141 ssl/ssl_rsa.c and ssl/s3_both.c 6142 [Nils Larsch] 6143 6144 *) Use SHA-1 instead of MD5 as the default digest algorithm for 6145 the apps/openssl applications. 6146 [Nils Larsch] 6147 6148 *) Compile clean with "-Wall -Wmissing-prototypes 6149 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 6150 DEBUG_SAFESTACK must also be set. 6151 [Ben Laurie] 6152 6153 *) Change ./Configure so that certain algorithms can be disabled by default. 6154 The new counterpiece to "no-xxx" is "enable-xxx". 6155 6156 The patented RC5 and MDC2 algorithms will now be disabled unless 6157 "enable-rc5" and "enable-mdc2", respectively, are specified. 6158 6159 (IDEA remains enabled despite being patented. This is because IDEA 6160 is frequently required for interoperability, and there is no license 6161 fee for non-commercial use. As before, "no-idea" can be used to 6162 avoid this algorithm.) 6163 6164 [Bodo Moeller] 6165 6166 *) Add processing of proxy certificates (see RFC 3820). This work was 6167 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 6168 EGEE (Enabling Grids for E-science in Europe). 6169 [Richard Levitte] 6170 6171 *) RC4 performance overhaul on modern architectures/implementations, such 6172 as Intel P4, IA-64 and AMD64. 6173 [Andy Polyakov] 6174 6175 *) New utility extract-section.pl. This can be used specify an alternative 6176 section number in a pod file instead of having to treat each file as 6177 a separate case in Makefile. This can be done by adding two lines to the 6178 pod file: 6179 6180 =for comment openssl_section:XXX 6181 6182 The blank line is mandatory. 6183 6184 [Steve Henson] 6185 6186 *) New arguments -certform, -keyform and -pass for s_client and s_server 6187 to allow alternative format key and certificate files and passphrase 6188 sources. 6189 [Steve Henson] 6190 6191 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 6192 update associated structures and add various utility functions. 6193 6194 Add new policy related verify parameters, include policy checking in 6195 standard verify code. Enhance 'smime' application with extra parameters 6196 to support policy checking and print out. 6197 [Steve Henson] 6198 6199 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 6200 Nehemiah processors. These extensions support AES encryption in hardware 6201 as well as RNG (though RNG support is currently disabled). 6202 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 6203 6204 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 6205 [Geoff Thorpe] 6206 6207 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 6208 [Andy Polyakov and a number of other people] 6209 6210 *) Improved PowerPC platform support. Most notably BIGNUM assembler 6211 implementation contributed by IBM. 6212 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 6213 6214 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 6215 exponent rather than 'unsigned long'. There is a corresponding change to 6216 the new 'rsa_keygen' element of the RSA_METHOD structure. 6217 [Jelte Jansen, Geoff Thorpe] 6218 6219 *) Functionality for creating the initial serial number file is now 6220 moved from CA.pl to the 'ca' utility with a new option -create_serial. 6221 6222 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 6223 number file to 1, which is bound to cause problems. To avoid 6224 the problems while respecting compatibility between different 0.9.7 6225 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 6226 CA.pl for serial number initialization. With the new release 0.9.8, 6227 we can fix the problem directly in the 'ca' utility.) 6228 [Steve Henson] 6229 6230 *) Reduced header interdependencies by declaring more opaque objects in 6231 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 6232 give fewer recursive includes, which could break lazy source code - so 6233 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 6234 developers should define this symbol when building and using openssl to 6235 ensure they track the recommended behaviour, interfaces, [etc], but 6236 backwards-compatible behaviour prevails when this isn't defined. 6237 [Geoff Thorpe] 6238 6239 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 6240 [Steve Henson] 6241 6242 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 6243 This will generate a random key of the appropriate length based on the 6244 cipher context. The EVP_CIPHER can provide its own random key generation 6245 routine to support keys of a specific form. This is used in the des and 6246 3des routines to generate a key of the correct parity. Update S/MIME 6247 code to use new functions and hence generate correct parity DES keys. 6248 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 6249 valid (weak or incorrect parity). 6250 [Steve Henson] 6251 6252 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 6253 as looking them up. This is useful when the verified structure may contain 6254 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 6255 present unless the new PKCS7_NO_CRL flag is asserted. 6256 [Steve Henson] 6257 6258 *) Extend ASN1 oid configuration module. It now additionally accepts the 6259 syntax: 6260 6261 shortName = some long name, 1.2.3.4 6262 [Steve Henson] 6263 6264 *) Reimplemented the BN_CTX implementation. There is now no more static 6265 limitation on the number of variables it can handle nor the depth of the 6266 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 6267 information can now expand as required, and rather than having a single 6268 static array of bignums, BN_CTX now uses a linked-list of such arrays 6269 allowing it to expand on demand whilst maintaining the usefulness of 6270 BN_CTX's "bundling". 6271 [Geoff Thorpe] 6272 6273 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 6274 to allow all RSA operations to function using a single BN_CTX. 6275 [Geoff Thorpe] 6276 6277 *) Preliminary support for certificate policy evaluation and checking. This 6278 is initially intended to pass the tests outlined in "Conformance Testing 6279 of Relying Party Client Certificate Path Processing Logic" v1.07. 6280 [Steve Henson] 6281 6282 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 6283 remained unused and not that useful. A variety of other little bignum 6284 tweaks and fixes have also been made continuing on from the audit (see 6285 below). 6286 [Geoff Thorpe] 6287 6288 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 6289 associated ASN1, EVP and SSL functions and old ASN1 macros. 6290 [Richard Levitte] 6291 6292 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 6293 and this should never fail. So the return value from the use of 6294 BN_set_word() (which can fail due to needless expansion) is now deprecated; 6295 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 6296 [Geoff Thorpe] 6297 6298 *) BN_CTX_get() should return zero-valued bignums, providing the same 6299 initialised value as BN_new(). 6300 [Geoff Thorpe, suggested by Ulf Möller] 6301 6302 *) Support for inhibitAnyPolicy certificate extension. 6303 [Steve Henson] 6304 6305 *) An audit of the BIGNUM code is underway, for which debugging code is 6306 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 6307 is considered valid when processing BIGNUMs, and causes execution to 6308 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 6309 further steps are taken to deliberately pollute unused data in BIGNUM 6310 structures to try and expose faulty code further on. For now, openssl will 6311 (in its default mode of operation) continue to tolerate the inconsistent 6312 forms that it has tolerated in the past, but authors and packagers should 6313 consider trying openssl and their own applications when compiled with 6314 these debugging symbols defined. It will help highlight potential bugs in 6315 their own code, and will improve the test coverage for OpenSSL itself. At 6316 some point, these tighter rules will become openssl's default to improve 6317 maintainability, though the assert()s and other overheads will remain only 6318 in debugging configurations. See bn.h for more details. 6319 [Geoff Thorpe, Nils Larsch, Ulf Möller] 6320 6321 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 6322 that can only be obtained through BN_CTX_new() (which implicitly 6323 initialises it). The presence of this function only made it possible 6324 to overwrite an existing structure (and cause memory leaks). 6325 [Geoff Thorpe] 6326 6327 *) Because of the callback-based approach for implementing LHASH as a 6328 template type, lh_insert() adds opaque objects to hash-tables and 6329 lh_doall() or lh_doall_arg() are typically used with a destructor callback 6330 to clean up those corresponding objects before destroying the hash table 6331 (and losing the object pointers). So some over-zealous constifications in 6332 LHASH have been relaxed so that lh_insert() does not take (nor store) the 6333 objects as "const" and the lh_doall[_arg] callback wrappers are not 6334 prototyped to have "const" restrictions on the object pointers they are 6335 given (and so aren't required to cast them away any more). 6336 [Geoff Thorpe] 6337 6338 *) The tmdiff.h API was so ugly and minimal that our own timing utility 6339 (speed) prefers to use its own implementation. The two implementations 6340 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 6341 its object type properly exposed (MS_TM) instead of casting to/from "char 6342 *". This may still change yet if someone realises MS_TM and "ms_time_***" 6343 aren't necessarily the greatest nomenclatures - but this is what was used 6344 internally to the implementation so I've used that for now. 6345 [Geoff Thorpe] 6346 6347 *) Ensure that deprecated functions do not get compiled when 6348 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 6349 the self-tests were still using deprecated key-generation functions so 6350 these have been updated also. 6351 [Geoff Thorpe] 6352 6353 *) Reorganise PKCS#7 code to separate the digest location functionality 6354 into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest(). 6355 New function PKCS7_set_digest() to set the digest type for PKCS#7 6356 digestedData type. Add additional code to correctly generate the 6357 digestedData type and add support for this type in PKCS7 initialization 6358 functions. 6359 [Steve Henson] 6360 6361 *) New function PKCS7_set0_type_other() this initializes a PKCS7 6362 structure of type "other". 6363 [Steve Henson] 6364 6365 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 6366 sure the loop does correctly stop and breaking ("division by zero") 6367 modulus operations are not performed. The (pre-generated) prime 6368 table crypto/bn/bn_prime.h was already correct, but it could not be 6369 re-generated on some platforms because of the "division by zero" 6370 situation in the script. 6371 [Ralf S. Engelschall] 6372 6373 *) Update support for ECC-based TLS ciphersuites according to 6374 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 6375 SHA-1 now is only used for "small" curves (where the 6376 representation of a field element takes up to 24 bytes); for 6377 larger curves, the field element resulting from ECDH is directly 6378 used as premaster secret. 6379 [Douglas Stebila (Sun Microsystems Laboratories)] 6380 6381 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 6382 curve secp160r1 to the tests. 6383 [Douglas Stebila (Sun Microsystems Laboratories)] 6384 6385 *) Add the possibility to load symbols globally with DSO. 6386 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 6387 6388 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 6389 control of the error stack. 6390 [Richard Levitte] 6391 6392 *) Add support for STORE in ENGINE. 6393 [Richard Levitte] 6394 6395 *) Add the STORE type. The intention is to provide a common interface 6396 to certificate and key stores, be they simple file-based stores, or 6397 HSM-type store, or LDAP stores, or... 6398 NOTE: The code is currently UNTESTED and isn't really used anywhere. 6399 [Richard Levitte] 6400 6401 *) Add a generic structure called OPENSSL_ITEM. This can be used to 6402 pass a list of arguments to any function as well as provide a way 6403 for a function to pass data back to the caller. 6404 [Richard Levitte] 6405 6406 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 6407 works like BUF_strdup() but can be used to duplicate a portion of 6408 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 6409 a memory area. 6410 [Richard Levitte] 6411 6412 *) Add the function sk_find_ex() which works like sk_find(), but will 6413 return an index to an element even if an exact match couldn't be 6414 found. The index is guaranteed to point at the element where the 6415 searched-for key would be inserted to preserve sorting order. 6416 [Richard Levitte] 6417 6418 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 6419 takes an extra flags argument for optional functionality. Currently, 6420 the following flags are defined: 6421 6422 OBJ_BSEARCH_VALUE_ON_NOMATCH 6423 This one gets OBJ_bsearch_ex() to return a pointer to the first 6424 element where the comparing function returns a negative or zero 6425 number. 6426 6427 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 6428 This one gets OBJ_bsearch_ex() to return a pointer to the first 6429 element where the comparing function returns zero. This is useful 6430 if there are more than one element where the comparing function 6431 returns zero. 6432 [Richard Levitte] 6433 6434 *) Make it possible to create self-signed certificates with 'openssl ca' 6435 in such a way that the self-signed certificate becomes part of the 6436 CA database and uses the same mechanisms for serial number generation 6437 as all other certificate signing. The new flag '-selfsign' enables 6438 this functionality. Adapt CA.sh and CA.pl.in. 6439 [Richard Levitte] 6440 6441 *) Add functionality to check the public key of a certificate request 6442 against a given private. This is useful to check that a certificate 6443 request can be signed by that key (self-signing). 6444 [Richard Levitte] 6445 6446 *) Make it possible to have multiple active certificates with the same 6447 subject in the CA index file. This is done only if the keyword 6448 'unique_subject' is set to 'no' in the main CA section (default 6449 if 'CA_default') of the configuration file. The value is saved 6450 with the database itself in a separate index attribute file, 6451 named like the index file with '.attr' appended to the name. 6452 [Richard Levitte] 6453 6454 *) Generate multi-valued AVAs using '+' notation in config files for 6455 req and dirName. 6456 [Steve Henson] 6457 6458 *) Support for nameConstraints certificate extension. 6459 [Steve Henson] 6460 6461 *) Support for policyConstraints certificate extension. 6462 [Steve Henson] 6463 6464 *) Support for policyMappings certificate extension. 6465 [Steve Henson] 6466 6467 *) Make sure the default DSA_METHOD implementation only uses its 6468 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 6469 and change its own handlers to be NULL so as to remove unnecessary 6470 indirection. This lets alternative implementations fallback to the 6471 default implementation more easily. 6472 [Geoff Thorpe] 6473 6474 *) Support for directoryName in GeneralName related extensions 6475 in config files. 6476 [Steve Henson] 6477 6478 *) Make it possible to link applications using Makefile.shared. 6479 Make that possible even when linking against static libraries! 6480 [Richard Levitte] 6481 6482 *) Support for single pass processing for S/MIME signing. This now 6483 means that S/MIME signing can be done from a pipe, in addition 6484 cleartext signing (multipart/signed type) is effectively streaming 6485 and the signed data does not need to be all held in memory. 6486 6487 This is done with a new flag PKCS7_STREAM. When this flag is set 6488 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 6489 is done after the data is output (and digests calculated) in 6490 SMIME_write_PKCS7(). 6491 [Steve Henson] 6492 6493 *) Add full support for -rpath/-R, both in shared libraries and 6494 applications, at least on the platforms where it's known how 6495 to do it. 6496 [Richard Levitte] 6497 6498 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 6499 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 6500 will now compute a table of multiples of the generator that 6501 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 6502 faster (notably in the case of a single point multiplication, 6503 scalar * generator). 6504 [Nils Larsch, Bodo Moeller] 6505 6506 *) IPv6 support for certificate extensions. The various extensions 6507 which use the IP:a.b.c.d can now take IPv6 addresses using the 6508 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 6509 correctly. 6510 [Steve Henson] 6511 6512 *) Added an ENGINE that implements RSA by performing private key 6513 exponentiations with the GMP library. The conversions to and from 6514 GMP's mpz_t format aren't optimised nor are any montgomery forms 6515 cached, and on x86 it appears OpenSSL's own performance has caught up. 6516 However there are likely to be other architectures where GMP could 6517 provide a boost. This ENGINE is not built in by default, but it can be 6518 specified at Configure time and should be accompanied by the necessary 6519 linker additions, eg; 6520 ./config -DOPENSSL_USE_GMP -lgmp 6521 [Geoff Thorpe] 6522 6523 *) "openssl engine" will not display ENGINE/DSO load failure errors when 6524 testing availability of engines with "-t" - the old behaviour is 6525 produced by increasing the feature's verbosity with "-tt". 6526 [Geoff Thorpe] 6527 6528 *) ECDSA routines: under certain error conditions uninitialized BN objects 6529 could be freed. Solution: make sure initialization is performed early 6530 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 6531 via PR#459) 6532 [Lutz Jaenicke] 6533 6534 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 6535 and DH_METHOD (eg. by ENGINE implementations) to override the normal 6536 software implementations. For DSA and DH, parameter generation can 6537 also be overridden by providing the appropriate method callbacks. 6538 [Geoff Thorpe] 6539 6540 *) Change the "progress" mechanism used in key-generation and 6541 primality testing to functions that take a new BN_GENCB pointer in 6542 place of callback/argument pairs. The new API functions have "_ex" 6543 postfixes and the older functions are reimplemented as wrappers for 6544 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 6545 declarations of the old functions to help (graceful) attempts to 6546 migrate to the new functions. Also, the new key-generation API 6547 functions operate on a caller-supplied key-structure and return 6548 success/failure rather than returning a key or NULL - this is to 6549 help make "keygen" another member function of RSA_METHOD etc. 6550 6551 Example for using the new callback interface: 6552 6553 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 6554 void *my_arg = ...; 6555 BN_GENCB my_cb; 6556 6557 BN_GENCB_set(&my_cb, my_callback, my_arg); 6558 6559 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 6560 /* For the meaning of a, b in calls to my_callback(), see the 6561 * documentation of the function that calls the callback. 6562 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 6563 * my_callback should return 1 if it wants BN_is_prime_ex() 6564 * to continue, or 0 to stop. 6565 */ 6566 6567 [Geoff Thorpe] 6568 6569 *) Change the ZLIB compression method to be stateful, and make it 6570 available to TLS with the number defined in 6571 draft-ietf-tls-compression-04.txt. 6572 [Richard Levitte] 6573 6574 *) Add the ASN.1 structures and functions for CertificatePair, which 6575 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 6576 6577 CertificatePair ::= SEQUENCE { 6578 forward [0] Certificate OPTIONAL, 6579 reverse [1] Certificate OPTIONAL, 6580 -- at least one of the pair shall be present -- } 6581 6582 Also implement the PEM functions to read and write certificate 6583 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 6584 6585 This needed to be defined, mostly for the sake of the LDAP 6586 attribute crossCertificatePair, but may prove useful elsewhere as 6587 well. 6588 [Richard Levitte] 6589 6590 *) Make it possible to inhibit symlinking of shared libraries in 6591 Makefile.shared, for Cygwin's sake. 6592 [Richard Levitte] 6593 6594 *) Extend the BIGNUM API by creating a function 6595 void BN_set_negative(BIGNUM *a, int neg); 6596 and a macro that behave like 6597 int BN_is_negative(const BIGNUM *a); 6598 6599 to avoid the need to access 'a->neg' directly in applications. 6600 [Nils Larsch] 6601 6602 *) Implement fast modular reduction for pseudo-Mersenne primes 6603 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 6604 EC_GROUP_new_curve_GFp() will now automatically use this 6605 if applicable. 6606 [Nils Larsch <nla@trustcenter.de>] 6607 6608 *) Add new lock type (CRYPTO_LOCK_BN). 6609 [Bodo Moeller] 6610 6611 *) Change the ENGINE framework to automatically load engines 6612 dynamically from specific directories unless they could be 6613 found to already be built in or loaded. Move all the 6614 current engines except for the cryptodev one to a new 6615 directory engines/. 6616 The engines in engines/ are built as shared libraries if 6617 the "shared" options was given to ./Configure or ./config. 6618 Otherwise, they are inserted in libcrypto.a. 6619 /usr/local/ssl/engines is the default directory for dynamic 6620 engines, but that can be overridden at configure time through 6621 the usual use of --prefix and/or --openssldir, and at run 6622 time with the environment variable OPENSSL_ENGINES. 6623 [Geoff Thorpe and Richard Levitte] 6624 6625 *) Add Makefile.shared, a helper makefile to build shared 6626 libraries. Adapt Makefile.org. 6627 [Richard Levitte] 6628 6629 *) Add version info to Win32 DLLs. 6630 [Peter 'Luna' Runestig" <peter@runestig.com>] 6631 6632 *) Add new 'medium level' PKCS#12 API. Certificates and keys 6633 can be added using this API to created arbitrary PKCS#12 6634 files while avoiding the low level API. 6635 6636 New options to PKCS12_create(), key or cert can be NULL and 6637 will then be omitted from the output file. The encryption 6638 algorithm NIDs can be set to -1 for no encryption, the mac 6639 iteration count can be set to 0 to omit the mac. 6640 6641 Enhance pkcs12 utility by making the -nokeys and -nocerts 6642 options work when creating a PKCS#12 file. New option -nomac 6643 to omit the mac, NONE can be set for an encryption algorithm. 6644 New code is modified to use the enhanced PKCS12_create() 6645 instead of the low level API. 6646 [Steve Henson] 6647 6648 *) Extend ASN1 encoder to support indefinite length constructed 6649 encoding. This can output sequences tags and octet strings in 6650 this form. Modify pk7_asn1.c to support indefinite length 6651 encoding. This is experimental and needs additional code to 6652 be useful, such as an ASN1 bio and some enhanced streaming 6653 PKCS#7 code. 6654 6655 Extend template encode functionality so that tagging is passed 6656 down to the template encoder. 6657 [Steve Henson] 6658 6659 *) Let 'openssl req' fail if an argument to '-newkey' is not 6660 recognized instead of using RSA as a default. 6661 [Bodo Moeller] 6662 6663 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 6664 As these are not official, they are not included in "ALL"; 6665 the "ECCdraft" ciphersuite group alias can be used to select them. 6666 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 6667 6668 *) Add ECDH engine support. 6669 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 6670 6671 *) Add ECDH in new directory crypto/ecdh/. 6672 [Douglas Stebila (Sun Microsystems Laboratories)] 6673 6674 *) Let BN_rand_range() abort with an error after 100 iterations 6675 without success (which indicates a broken PRNG). 6676 [Bodo Moeller] 6677 6678 *) Change BN_mod_sqrt() so that it verifies that the input value 6679 is really the square of the return value. (Previously, 6680 BN_mod_sqrt would show GIGO behaviour.) 6681 [Bodo Moeller] 6682 6683 *) Add named elliptic curves over binary fields from X9.62, SECG, 6684 and WAP/WTLS; add OIDs that were still missing. 6685 6686 [Sheueling Chang Shantz and Douglas Stebila 6687 (Sun Microsystems Laboratories)] 6688 6689 *) Extend the EC library for elliptic curves over binary fields 6690 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 6691 New EC_METHOD: 6692 6693 EC_GF2m_simple_method 6694 6695 New API functions: 6696 6697 EC_GROUP_new_curve_GF2m 6698 EC_GROUP_set_curve_GF2m 6699 EC_GROUP_get_curve_GF2m 6700 EC_POINT_set_affine_coordinates_GF2m 6701 EC_POINT_get_affine_coordinates_GF2m 6702 EC_POINT_set_compressed_coordinates_GF2m 6703 6704 Point compression for binary fields is disabled by default for 6705 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 6706 enable it). 6707 6708 As binary polynomials are represented as BIGNUMs, various members 6709 of the EC_GROUP and EC_POINT data structures can be shared 6710 between the implementations for prime fields and binary fields; 6711 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 6712 are essentially identical to their ..._GFp counterparts. 6713 (For simplicity, the '..._GFp' prefix has been dropped from 6714 various internal method names.) 6715 6716 An internal 'field_div' method (similar to 'field_mul' and 6717 'field_sqr') has been added; this is used only for binary fields. 6718 6719 [Sheueling Chang Shantz and Douglas Stebila 6720 (Sun Microsystems Laboratories)] 6721 6722 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 6723 through methods ('mul', 'precompute_mult'). 6724 6725 The generic implementations (now internally called 'ec_wNAF_mul' 6726 and 'ec_wNAF_precomputed_mult') remain the default if these 6727 methods are undefined. 6728 6729 [Sheueling Chang Shantz and Douglas Stebila 6730 (Sun Microsystems Laboratories)] 6731 6732 *) New function EC_GROUP_get_degree, which is defined through 6733 EC_METHOD. For curves over prime fields, this returns the bit 6734 length of the modulus. 6735 6736 [Sheueling Chang Shantz and Douglas Stebila 6737 (Sun Microsystems Laboratories)] 6738 6739 *) New functions EC_GROUP_dup, EC_POINT_dup. 6740 (These simply call ..._new and ..._copy). 6741 6742 [Sheueling Chang Shantz and Douglas Stebila 6743 (Sun Microsystems Laboratories)] 6744 6745 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 6746 Polynomials are represented as BIGNUMs (where the sign bit is not 6747 used) in the following functions [macros]: 6748 6749 BN_GF2m_add 6750 BN_GF2m_sub [= BN_GF2m_add] 6751 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 6752 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 6753 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 6754 BN_GF2m_mod_inv 6755 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 6756 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 6757 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 6758 BN_GF2m_cmp [= BN_ucmp] 6759 6760 (Note that only the 'mod' functions are actually for fields GF(2^m). 6761 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 6762 6763 For some functions, an the irreducible polynomial defining a 6764 field can be given as an 'unsigned int[]' with strictly 6765 decreasing elements giving the indices of those bits that are set; 6766 i.e., p[] represents the polynomial 6767 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 6768 where 6769 p[0] > p[1] > ... > p[k] = 0. 6770 This applies to the following functions: 6771 6772 BN_GF2m_mod_arr 6773 BN_GF2m_mod_mul_arr 6774 BN_GF2m_mod_sqr_arr 6775 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 6776 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 6777 BN_GF2m_mod_exp_arr 6778 BN_GF2m_mod_sqrt_arr 6779 BN_GF2m_mod_solve_quad_arr 6780 BN_GF2m_poly2arr 6781 BN_GF2m_arr2poly 6782 6783 Conversion can be performed by the following functions: 6784 6785 BN_GF2m_poly2arr 6786 BN_GF2m_arr2poly 6787 6788 bntest.c has additional tests for binary polynomial arithmetic. 6789 6790 Two implementations for BN_GF2m_mod_div() are available. 6791 The default algorithm simply uses BN_GF2m_mod_inv() and 6792 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 6793 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 6794 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 6795 6796 [Sheueling Chang Shantz and Douglas Stebila 6797 (Sun Microsystems Laboratories)] 6798 6799 *) Add new error code 'ERR_R_DISABLED' that can be used when some 6800 functionality is disabled at compile-time. 6801 [Douglas Stebila <douglas.stebila@sun.com>] 6802 6803 *) Change default behaviour of 'openssl asn1parse' so that more 6804 information is visible when viewing, e.g., a certificate: 6805 6806 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 6807 mode the content of non-printable OCTET STRINGs is output in a 6808 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 6809 avoid the appearance of a printable string. 6810 [Nils Larsch <nla@trustcenter.de>] 6811 6812 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 6813 functions 6814 EC_GROUP_set_asn1_flag() 6815 EC_GROUP_get_asn1_flag() 6816 EC_GROUP_set_point_conversion_form() 6817 EC_GROUP_get_point_conversion_form() 6818 These control ASN1 encoding details: 6819 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 6820 has been set to OPENSSL_EC_NAMED_CURVE. 6821 - Points are encoded in uncompressed form by default; options for 6822 asn1_for are as for point2oct, namely 6823 POINT_CONVERSION_COMPRESSED 6824 POINT_CONVERSION_UNCOMPRESSED 6825 POINT_CONVERSION_HYBRID 6826 6827 Also add 'seed' and 'seed_len' members to EC_GROUP with access 6828 functions 6829 EC_GROUP_set_seed() 6830 EC_GROUP_get0_seed() 6831 EC_GROUP_get_seed_len() 6832 This is used only for ASN1 purposes (so far). 6833 [Nils Larsch <nla@trustcenter.de>] 6834 6835 *) Add 'field_type' member to EC_METHOD, which holds the NID 6836 of the appropriate field type OID. The new function 6837 EC_METHOD_get_field_type() returns this value. 6838 [Nils Larsch <nla@trustcenter.de>] 6839 6840 *) Add functions 6841 EC_POINT_point2bn() 6842 EC_POINT_bn2point() 6843 EC_POINT_point2hex() 6844 EC_POINT_hex2point() 6845 providing useful interfaces to EC_POINT_point2oct() and 6846 EC_POINT_oct2point(). 6847 [Nils Larsch <nla@trustcenter.de>] 6848 6849 *) Change internals of the EC library so that the functions 6850 EC_GROUP_set_generator() 6851 EC_GROUP_get_generator() 6852 EC_GROUP_get_order() 6853 EC_GROUP_get_cofactor() 6854 are implemented directly in crypto/ec/ec_lib.c and not dispatched 6855 to methods, which would lead to unnecessary code duplication when 6856 adding different types of curves. 6857 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 6858 6859 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 6860 arithmetic, and such that modified wNAFs are generated 6861 (which avoid length expansion in many cases). 6862 [Bodo Moeller] 6863 6864 *) Add a function EC_GROUP_check_discriminant() (defined via 6865 EC_METHOD) that verifies that the curve discriminant is non-zero. 6866 6867 Add a function EC_GROUP_check() that makes some sanity tests 6868 on a EC_GROUP, its generator and order. This includes 6869 EC_GROUP_check_discriminant(). 6870 [Nils Larsch <nla@trustcenter.de>] 6871 6872 *) Add ECDSA in new directory crypto/ecdsa/. 6873 6874 Add applications 'openssl ecparam' and 'openssl ecdsa' 6875 (these are based on 'openssl dsaparam' and 'openssl dsa'). 6876 6877 ECDSA support is also included in various other files across the 6878 library. Most notably, 6879 - 'openssl req' now has a '-newkey ecdsa:file' option; 6880 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 6881 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 6882 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 6883 them suitable for ECDSA where domain parameters must be 6884 extracted before the specific public key; 6885 - ECDSA engine support has been added. 6886 [Nils Larsch <nla@trustcenter.de>] 6887 6888 *) Include some named elliptic curves, and add OIDs from X9.62, 6889 SECG, and WAP/WTLS. Each curve can be obtained from the new 6890 function 6891 EC_GROUP_new_by_curve_name(), 6892 and the list of available named curves can be obtained with 6893 EC_get_builtin_curves(). 6894 Also add a 'curve_name' member to EC_GROUP objects, which can be 6895 accessed via 6896 EC_GROUP_set_curve_name() 6897 EC_GROUP_get_curve_name() 6898 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 6899 6900 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 6901 was actually never needed) and in BN_mul(). The removal in BN_mul() 6902 required a small change in bn_mul_part_recursive() and the addition 6903 of the functions bn_cmp_part_words(), bn_sub_part_words() and 6904 bn_add_part_words(), which do the same thing as bn_cmp_words(), 6905 bn_sub_words() and bn_add_words() except they take arrays with 6906 differing sizes. 6907 [Richard Levitte] 6908 6909 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 6910 6911 *) Cleanse PEM buffers before freeing them since they may contain 6912 sensitive data. 6913 [Benjamin Bennett <ben@psc.edu>] 6914 6915 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 6916 a ciphersuite string such as "DEFAULT:RSA" cannot enable 6917 authentication-only ciphersuites. 6918 [Bodo Moeller] 6919 6920 *) Since AES128 and AES256 share a single mask bit in the logic of 6921 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 6922 kludge to work properly if AES128 is available and AES256 isn't. 6923 [Victor Duchovni] 6924 6925 *) Expand security boundary to match 1.1.1 module. 6926 [Steve Henson] 6927 6928 *) Remove redundant features: hash file source, editing of test vectors 6929 modify fipsld to use external fips_premain.c signature. 6930 [Steve Henson] 6931 6932 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 6933 run algorithm test programs. 6934 [Steve Henson] 6935 6936 *) Make algorithm test programs more tolerant of whitespace. 6937 [Steve Henson] 6938 6939 *) Have SSL/TLS server implementation tolerate "mismatched" record 6940 protocol version while receiving ClientHello even if the 6941 ClientHello is fragmented. (The server can't insist on the 6942 particular protocol version it has chosen before the ServerHello 6943 message has informed the client about his choice.) 6944 [Bodo Moeller] 6945 6946 *) Load error codes if they are not already present instead of using a 6947 static variable. This allows them to be cleanly unloaded and reloaded. 6948 [Steve Henson] 6949 6950 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 6951 6952 *) Introduce limits to prevent malicious keys being able to 6953 cause a denial of service. (CVE-2006-2940) 6954 [Steve Henson, Bodo Moeller] 6955 6956 *) Fix ASN.1 parsing of certain invalid structures that can result 6957 in a denial of service. (CVE-2006-2937) [Steve Henson] 6958 6959 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 6960 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 6961 6962 *) Fix SSL client code which could crash if connecting to a 6963 malicious SSLv2 server. (CVE-2006-4343) 6964 [Tavis Ormandy and Will Drewry, Google Security Team] 6965 6966 *) Change ciphersuite string processing so that an explicit 6967 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 6968 will no longer include "AES128-SHA"), and any other similar 6969 ciphersuite (same bitmap) from *other* protocol versions (so that 6970 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 6971 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 6972 changes from 0.9.8b and 0.9.8d. 6973 [Bodo Moeller] 6974 6975 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 6976 6977 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 6978 (CVE-2006-4339) [Ben Laurie and Google Security Team] 6979 6980 *) Change the Unix randomness entropy gathering to use poll() when 6981 possible instead of select(), since the latter has some 6982 undesirable limitations. 6983 [Darryl Miles via Richard Levitte and Bodo Moeller] 6984 6985 *) Disable rogue ciphersuites: 6986 6987 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 6988 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 6989 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 6990 6991 The latter two were purportedly from 6992 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 6993 appear there. 6994 6995 Also deactivate the remaining ciphersuites from 6996 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 6997 unofficial, and the ID has long expired. 6998 [Bodo Moeller] 6999 7000 *) Fix RSA blinding Heisenbug (problems sometimes occurred on 7001 dual-core machines) and other potential thread-safety issues. 7002 [Bodo Moeller] 7003 7004 Changes between 0.9.7i and 0.9.7j [04 May 2006] 7005 7006 *) Adapt fipsld and the build system to link against the validated FIPS 7007 module in FIPS mode. 7008 [Steve Henson] 7009 7010 *) Fixes for VC++ 2005 build under Windows. 7011 [Steve Henson] 7012 7013 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 7014 from a Windows bash shell such as MSYS. It is autodetected from the 7015 "config" script when run from a VC++ environment. Modify standard VC++ 7016 build to use fipscanister.o from the GNU make build. 7017 [Steve Henson] 7018 7019 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 7020 7021 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 7022 The value now differs depending on if you build for FIPS or not. 7023 BEWARE! A program linked with a shared FIPSed libcrypto can't be 7024 safely run with a non-FIPSed libcrypto, as it may crash because of 7025 the difference induced by this change. 7026 [Andy Polyakov] 7027 7028 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 7029 7030 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 7031 (part of SSL_OP_ALL). This option used to disable the 7032 countermeasure against man-in-the-middle protocol-version 7033 rollback in the SSL 2.0 server implementation, which is a bad 7034 idea. (CVE-2005-2969) 7035 7036 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 7037 for Information Security, National Institute of Advanced Industrial 7038 Science and Technology [AIST], Japan)] 7039 7040 *) Minimal support for X9.31 signatures and PSS padding modes. This is 7041 mainly for FIPS compliance and not fully integrated at this stage. 7042 [Steve Henson] 7043 7044 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 7045 the exponentiation using a fixed-length exponent. (Otherwise, 7046 the information leaked through timing could expose the secret key 7047 after many signatures; cf. Bleichenbacher's attack on DSA with 7048 biased k.) 7049 [Bodo Moeller] 7050 7051 *) Make a new fixed-window mod_exp implementation the default for 7052 RSA, DSA, and DH private-key operations so that the sequence of 7053 squares and multiplies and the memory access pattern are 7054 independent of the particular secret key. This will mitigate 7055 cache-timing and potential related attacks. 7056 7057 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 7058 and this is automatically used by BN_mod_exp_mont() if the new flag 7059 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 7060 will use this BN flag for private exponents unless the flag 7061 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 7062 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 7063 7064 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 7065 7066 *) Change the client implementation for SSLv23_method() and 7067 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 7068 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 7069 (Previously, the SSL 2.0 backwards compatible Client Hello 7070 message format would be used even with SSL_OP_NO_SSLv2.) 7071 [Bodo Moeller] 7072 7073 *) Add support for smime-type MIME parameter in S/MIME messages which some 7074 clients need. 7075 [Steve Henson] 7076 7077 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 7078 a threadsafe manner. Modify rsa code to use new function and add calls 7079 to dsa and dh code (which had race conditions before). 7080 [Steve Henson] 7081 7082 *) Include the fixed error library code in the C error file definitions 7083 instead of fixing them up at runtime. This keeps the error code 7084 structures constant. 7085 [Steve Henson] 7086 7087 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 7088 7089 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 7090 OpenSSL 0.9.8.] 7091 7092 *) Fixes for newer kerberos headers. NB: the casts are needed because 7093 the 'length' field is signed on one version and unsigned on another 7094 with no (?) obvious way to tell the difference, without these VC++ 7095 complains. Also the "definition" of FAR (blank) is no longer included 7096 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 7097 some needed definitions. 7098 [Steve Henson] 7099 7100 *) Undo Cygwin change. 7101 [Ulf Möller] 7102 7103 *) Added support for proxy certificates according to RFC 3820. 7104 Because they may be a security thread to unaware applications, 7105 they must be explicitly allowed in run-time. See 7106 docs/HOWTO/proxy_certificates.txt for further information. 7107 [Richard Levitte] 7108 7109 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 7110 7111 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 7112 server and client random values. Previously 7113 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 7114 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 7115 7116 This change has negligible security impact because: 7117 7118 1. Server and client random values still have 24 bytes of pseudo random 7119 data. 7120 7121 2. Server and client random values are sent in the clear in the initial 7122 handshake. 7123 7124 3. The master secret is derived using the premaster secret (48 bytes in 7125 size for static RSA ciphersuites) as well as client server and random 7126 values. 7127 7128 The OpenSSL team would like to thank the UK NISCC for bringing this issue 7129 to our attention. 7130 7131 [Stephen Henson, reported by UK NISCC] 7132 7133 *) Use Windows randomness collection on Cygwin. 7134 [Ulf Möller] 7135 7136 *) Fix hang in EGD/PRNGD query when communication socket is closed 7137 prematurely by EGD/PRNGD. 7138 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014] 7139 7140 *) Prompt for pass phrases when appropriate for PKCS12 input format. 7141 [Steve Henson] 7142 7143 *) Back-port of selected performance improvements from development 7144 branch, as well as improved support for PowerPC platforms. 7145 [Andy Polyakov] 7146 7147 *) Add lots of checks for memory allocation failure, error codes to indicate 7148 failure and freeing up memory if a failure occurs. 7149 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 7150 7151 *) Add new -passin argument to dgst. 7152 [Steve Henson] 7153 7154 *) Perform some character comparisons of different types in X509_NAME_cmp: 7155 this is needed for some certificates that re-encode DNs into UTF8Strings 7156 (in violation of RFC3280) and can't or won't issue name rollover 7157 certificates. 7158 [Steve Henson] 7159 7160 *) Make an explicit check during certificate validation to see that 7161 the CA setting in each certificate on the chain is correct. As a 7162 side effect always do the following basic checks on extensions, 7163 not just when there's an associated purpose to the check: 7164 7165 - if there is an unhandled critical extension (unless the user 7166 has chosen to ignore this fault) 7167 - if the path length has been exceeded (if one is set at all) 7168 - that certain extensions fit the associated purpose (if one has 7169 been given) 7170 [Richard Levitte] 7171 7172 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 7173 7174 *) Avoid a race condition when CRLs are checked in a multi threaded 7175 environment. This would happen due to the reordering of the revoked 7176 entries during signature checking and serial number lookup. Now the 7177 encoding is cached and the serial number sort performed under a lock. 7178 Add new STACK function sk_is_sorted(). 7179 [Steve Henson] 7180 7181 *) Add Delta CRL to the extension code. 7182 [Steve Henson] 7183 7184 *) Various fixes to s3_pkt.c so alerts are sent properly. 7185 [David Holmes <d.holmes@f5.com>] 7186 7187 *) Reduce the chances of duplicate issuer name and serial numbers (in 7188 violation of RFC3280) using the OpenSSL certificate creation utilities. 7189 This is done by creating a random 64 bit value for the initial serial 7190 number when a serial number file is created or when a self signed 7191 certificate is created using 'openssl req -x509'. The initial serial 7192 number file is created using 'openssl x509 -next_serial' in CA.pl 7193 rather than being initialized to 1. 7194 [Steve Henson] 7195 7196 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 7197 7198 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 7199 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 7200 [Joe Orton, Steve Henson] 7201 7202 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 7203 (CVE-2004-0112) 7204 [Joe Orton, Steve Henson] 7205 7206 *) Make it possible to have multiple active certificates with the same 7207 subject in the CA index file. This is done only if the keyword 7208 'unique_subject' is set to 'no' in the main CA section (default 7209 if 'CA_default') of the configuration file. The value is saved 7210 with the database itself in a separate index attribute file, 7211 named like the index file with '.attr' appended to the name. 7212 [Richard Levitte] 7213 7214 *) X509 verify fixes. Disable broken certificate workarounds when 7215 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 7216 keyUsage extension present. Don't accept CRLs with unhandled critical 7217 extensions: since verify currently doesn't process CRL extensions this 7218 rejects a CRL with *any* critical extensions. Add new verify error codes 7219 for these cases. 7220 [Steve Henson] 7221 7222 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 7223 A clarification of RFC2560 will require the use of OCTET STRINGs and 7224 some implementations cannot handle the current raw format. Since OpenSSL 7225 copies and compares OCSP nonces as opaque blobs without any attempt at 7226 parsing them this should not create any compatibility issues. 7227 [Steve Henson] 7228 7229 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 7230 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 7231 this HMAC (and other) operations are several times slower than OpenSSL 7232 < 0.9.7. 7233 [Steve Henson] 7234 7235 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 7236 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 7237 7238 *) Use the correct content when signing type "other". 7239 [Steve Henson] 7240 7241 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 7242 7243 *) Fix various bugs revealed by running the NISCC test suite: 7244 7245 Stop out of bounds reads in the ASN1 code when presented with 7246 invalid tags (CVE-2003-0543 and CVE-2003-0544). 7247 7248 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 7249 7250 If verify callback ignores invalid public key errors don't try to check 7251 certificate signature with the NULL public key. 7252 7253 [Steve Henson] 7254 7255 *) New -ignore_err option in ocsp application to stop the server 7256 exiting on the first error in a request. 7257 [Steve Henson] 7258 7259 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 7260 if the server requested one: as stated in TLS 1.0 and SSL 3.0 7261 specifications. 7262 [Steve Henson] 7263 7264 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 7265 extra data after the compression methods not only for TLS 1.0 7266 but also for SSL 3.0 (as required by the specification). 7267 [Bodo Moeller; problem pointed out by Matthias Loepfe] 7268 7269 *) Change X509_certificate_type() to mark the key as exported/exportable 7270 when it's 512 *bits* long, not 512 bytes. 7271 [Richard Levitte] 7272 7273 *) Change AES_cbc_encrypt() so it outputs exact multiple of 7274 blocks during encryption. 7275 [Richard Levitte] 7276 7277 *) Various fixes to base64 BIO and non blocking I/O. On write 7278 flushes were not handled properly if the BIO retried. On read 7279 data was not being buffered properly and had various logic bugs. 7280 This also affects blocking I/O when the data being decoded is a 7281 certain size. 7282 [Steve Henson] 7283 7284 *) Various S/MIME bugfixes and compatibility changes: 7285 output correct application/pkcs7 MIME type if 7286 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 7287 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 7288 of files as .eml work). Correctly handle very long lines in MIME 7289 parser. 7290 [Steve Henson] 7291 7292 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 7293 7294 *) Countermeasure against the Klima-Pokorny-Rosa extension of 7295 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 7296 a protocol version number mismatch like a decryption error 7297 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 7298 [Bodo Moeller] 7299 7300 *) Turn on RSA blinding by default in the default implementation 7301 to avoid a timing attack. Applications that don't want it can call 7302 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 7303 They would be ill-advised to do so in most cases. 7304 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 7305 7306 *) Change RSA blinding code so that it works when the PRNG is not 7307 seeded (in this case, the secret RSA exponent is abused as 7308 an unpredictable seed -- if it is not unpredictable, there 7309 is no point in blinding anyway). Make RSA blinding thread-safe 7310 by remembering the creator's thread ID in rsa->blinding and 7311 having all other threads use local one-time blinding factors 7312 (this requires more computation than sharing rsa->blinding, but 7313 avoids excessive locking; and if an RSA object is not shared 7314 between threads, blinding will still be very fast). 7315 [Bodo Moeller] 7316 7317 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 7318 ENGINE as defaults for all supported algorithms irrespective of 7319 the 'flags' parameter. 'flags' is now honoured, so applications 7320 should make sure they are passing it correctly. 7321 [Geoff Thorpe] 7322 7323 *) Target "mingw" now allows native Windows code to be generated in 7324 the Cygwin environment as well as with the MinGW compiler. 7325 [Ulf Moeller] 7326 7327 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 7328 7329 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 7330 via timing by performing a MAC computation even if incorrect 7331 block cipher padding has been found. This is a countermeasure 7332 against active attacks where the attacker has to distinguish 7333 between bad padding and a MAC verification error. (CVE-2003-0078) 7334 7335 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 7336 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 7337 Martin Vuagnoux (EPFL, Ilion)] 7338 7339 *) Make the no-err option work as intended. The intention with no-err 7340 is not to have the whole error stack handling routines removed from 7341 libcrypto, it's only intended to remove all the function name and 7342 reason texts, thereby removing some of the footprint that may not 7343 be interesting if those errors aren't displayed anyway. 7344 7345 NOTE: it's still possible for any application or module to have its 7346 own set of error texts inserted. The routines are there, just not 7347 used by default when no-err is given. 7348 [Richard Levitte] 7349 7350 *) Add support for FreeBSD on IA64. 7351 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 7352 7353 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 7354 Kerberos function mit_des_cbc_cksum(). Before this change, 7355 the value returned by DES_cbc_cksum() was like the one from 7356 mit_des_cbc_cksum(), except the bytes were swapped. 7357 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 7358 7359 *) Allow an application to disable the automatic SSL chain building. 7360 Before this a rather primitive chain build was always performed in 7361 ssl3_output_cert_chain(): an application had no way to send the 7362 correct chain if the automatic operation produced an incorrect result. 7363 7364 Now the chain builder is disabled if either: 7365 7366 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 7367 7368 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 7369 7370 The reasoning behind this is that an application would not want the 7371 auto chain building to take place if extra chain certificates are 7372 present and it might also want a means of sending no additional 7373 certificates (for example the chain has two certificates and the 7374 root is omitted). 7375 [Steve Henson] 7376 7377 *) Add the possibility to build without the ENGINE framework. 7378 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 7379 7380 *) Under Win32 gmtime() can return NULL: check return value in 7381 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 7382 [Steve Henson] 7383 7384 *) DSA routines: under certain error conditions uninitialized BN objects 7385 could be freed. Solution: make sure initialization is performed early 7386 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 7387 Nils Larsch <nla@trustcenter.de> via PR#459) 7388 [Lutz Jaenicke] 7389 7390 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 7391 checked on reconnect on the client side, therefore session resumption 7392 could still fail with a "ssl session id is different" error. This 7393 behaviour is masked when SSL_OP_ALL is used due to 7394 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 7395 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 7396 followup to PR #377. 7397 [Lutz Jaenicke] 7398 7399 *) IA-32 assembler support enhancements: unified ELF targets, support 7400 for SCO/Caldera platforms, fix for Cygwin shared build. 7401 [Andy Polyakov] 7402 7403 *) Add support for FreeBSD on sparc64. As a consequence, support for 7404 FreeBSD on non-x86 processors is separate from x86 processors on 7405 the config script, much like the NetBSD support. 7406 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 7407 7408 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 7409 7410 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 7411 OpenSSL 0.9.7.] 7412 7413 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 7414 code (06) was taken as the first octet of the session ID and the last 7415 octet was ignored consequently. As a result SSLv2 client side session 7416 caching could not have worked due to the session ID mismatch between 7417 client and server. 7418 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 7419 PR #377. 7420 [Lutz Jaenicke] 7421 7422 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 7423 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 7424 removed entirely. 7425 [Richard Levitte] 7426 7427 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 7428 seems that in spite of existing for more than a year, many application 7429 author have done nothing to provide the necessary callbacks, which 7430 means that this particular engine will not work properly anywhere. 7431 This is a very unfortunate situation which forces us, in the name 7432 of usability, to give the hw_ncipher.c a static lock, which is part 7433 of libcrypto. 7434 NOTE: This is for the 0.9.7 series ONLY. This hack will never 7435 appear in 0.9.8 or later. We EXPECT application authors to have 7436 dealt properly with this when 0.9.8 is released (unless we actually 7437 make such changes in the libcrypto locking code that changes will 7438 have to be made anyway). 7439 [Richard Levitte] 7440 7441 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 7442 octets have been read, EOF or an error occurs. Without this change 7443 some truncated ASN1 structures will not produce an error. 7444 [Steve Henson] 7445 7446 *) Disable Heimdal support, since it hasn't been fully implemented. 7447 Still give the possibility to force the use of Heimdal, but with 7448 warnings and a request that patches get sent to openssl-dev. 7449 [Richard Levitte] 7450 7451 *) Add the VC-CE target, introduce the WINCE sysname, and add 7452 INSTALL.WCE and appropriate conditionals to make it build. 7453 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 7454 7455 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 7456 cygssl-x.y.z.dll, where x, y and z are the major, minor and 7457 edit numbers of the version. 7458 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 7459 7460 *) Introduce safe string copy and catenation functions 7461 (BUF_strlcpy() and BUF_strlcat()). 7462 [Ben Laurie (CHATS) and Richard Levitte] 7463 7464 *) Avoid using fixed-size buffers for one-line DNs. 7465 [Ben Laurie (CHATS)] 7466 7467 *) Add BUF_MEM_grow_clean() to avoid information leakage when 7468 resizing buffers containing secrets, and use where appropriate. 7469 [Ben Laurie (CHATS)] 7470 7471 *) Avoid using fixed size buffers for configuration file location. 7472 [Ben Laurie (CHATS)] 7473 7474 *) Avoid filename truncation for various CA files. 7475 [Ben Laurie (CHATS)] 7476 7477 *) Use sizeof in preference to magic numbers. 7478 [Ben Laurie (CHATS)] 7479 7480 *) Avoid filename truncation in cert requests. 7481 [Ben Laurie (CHATS)] 7482 7483 *) Add assertions to check for (supposedly impossible) buffer 7484 overflows. 7485 [Ben Laurie (CHATS)] 7486 7487 *) Don't cache truncated DNS entries in the local cache (this could 7488 potentially lead to a spoofing attack). 7489 [Ben Laurie (CHATS)] 7490 7491 *) Fix various buffers to be large enough for hex/decimal 7492 representations in a platform independent manner. 7493 [Ben Laurie (CHATS)] 7494 7495 *) Add CRYPTO_realloc_clean() to avoid information leakage when 7496 resizing buffers containing secrets, and use where appropriate. 7497 [Ben Laurie (CHATS)] 7498 7499 *) Add BIO_indent() to avoid much slightly worrying code to do 7500 indents. 7501 [Ben Laurie (CHATS)] 7502 7503 *) Convert sprintf()/BIO_puts() to BIO_printf(). 7504 [Ben Laurie (CHATS)] 7505 7506 *) buffer_gets() could terminate with the buffer only half 7507 full. Fixed. 7508 [Ben Laurie (CHATS)] 7509 7510 *) Add assertions to prevent user-supplied crypto functions from 7511 overflowing internal buffers by having large block sizes, etc. 7512 [Ben Laurie (CHATS)] 7513 7514 *) New OPENSSL_assert() macro (similar to assert(), but enabled 7515 unconditionally). 7516 [Ben Laurie (CHATS)] 7517 7518 *) Eliminate unused copy of key in RC4. 7519 [Ben Laurie (CHATS)] 7520 7521 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 7522 [Ben Laurie (CHATS)] 7523 7524 *) Fix off-by-one error in EGD path. 7525 [Ben Laurie (CHATS)] 7526 7527 *) If RANDFILE path is too long, ignore instead of truncating. 7528 [Ben Laurie (CHATS)] 7529 7530 *) Eliminate unused and incorrectly sized X.509 structure 7531 CBCParameter. 7532 [Ben Laurie (CHATS)] 7533 7534 *) Eliminate unused and dangerous function knumber(). 7535 [Ben Laurie (CHATS)] 7536 7537 *) Eliminate unused and dangerous structure, KSSL_ERR. 7538 [Ben Laurie (CHATS)] 7539 7540 *) Protect against overlong session ID context length in an encoded 7541 session object. Since these are local, this does not appear to be 7542 exploitable. 7543 [Ben Laurie (CHATS)] 7544 7545 *) Change from security patch (see 0.9.6e below) that did not affect 7546 the 0.9.6 release series: 7547 7548 Remote buffer overflow in SSL3 protocol - an attacker could 7549 supply an oversized master key in Kerberos-enabled versions. 7550 (CVE-2002-0657) 7551 [Ben Laurie (CHATS)] 7552 7553 *) Change the SSL kerb5 codes to match RFC 2712. 7554 [Richard Levitte] 7555 7556 *) Make -nameopt work fully for req and add -reqopt switch. 7557 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 7558 7559 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 7560 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 7561 7562 *) Make sure tests can be performed even if the corresponding algorithms 7563 have been removed entirely. This was also the last step to make 7564 OpenSSL compilable with DJGPP under all reasonable conditions. 7565 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 7566 7567 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 7568 to allow version independent disabling of normally unselected ciphers, 7569 which may be activated as a side-effect of selecting a single cipher. 7570 7571 (E.g., cipher list string "RSA" enables ciphersuites that are left 7572 out of "ALL" because they do not provide symmetric encryption. 7573 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 7574 [Lutz Jaenicke, Bodo Moeller] 7575 7576 *) Add appropriate support for separate platform-dependent build 7577 directories. The recommended way to make a platform-dependent 7578 build directory is the following (tested on Linux), maybe with 7579 some local tweaks: 7580 7581 # Place yourself outside of the OpenSSL source tree. In 7582 # this example, the environment variable OPENSSL_SOURCE 7583 # is assumed to contain the absolute OpenSSL source directory. 7584 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 7585 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 7586 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 7587 mkdir -p `dirname $F` 7588 ln -s $OPENSSL_SOURCE/$F $F 7589 done 7590 7591 To be absolutely sure not to disturb the source tree, a "make clean" 7592 is a good thing. If it isn't successful, don't worry about it, 7593 it probably means the source directory is very clean. 7594 [Richard Levitte] 7595 7596 *) Make sure any ENGINE control commands make local copies of string 7597 pointers passed to them whenever necessary. Otherwise it is possible 7598 the caller may have overwritten (or deallocated) the original string 7599 data when a later ENGINE operation tries to use the stored values. 7600 [Götz Babin-Ebell <babinebell@trustcenter.de>] 7601 7602 *) Improve diagnostics in file reading and command-line digests. 7603 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 7604 7605 *) Add AES modes CFB and OFB to the object database. Correct an 7606 error in AES-CFB decryption. 7607 [Richard Levitte] 7608 7609 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 7610 allows existing EVP_CIPHER_CTX structures to be reused after 7611 calling EVP_*Final(). This behaviour is used by encryption 7612 BIOs and some applications. This has the side effect that 7613 applications must explicitly clean up cipher contexts with 7614 EVP_CIPHER_CTX_cleanup() or they will leak memory. 7615 [Steve Henson] 7616 7617 *) Check the values of dna and dnb in bn_mul_recursive before calling 7618 bn_mul_comba (a non zero value means the a or b arrays do not contain 7619 n2 elements) and fallback to bn_mul_normal if either is not zero. 7620 [Steve Henson] 7621 7622 *) Fix escaping of non-ASCII characters when using the -subj option 7623 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 7624 [Lutz Jaenicke] 7625 7626 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 7627 form for "surname", serialNumber has no short form. 7628 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 7629 therefore remove "mail" short name for "internet 7". 7630 The OID for unique identifiers in X509 certificates is 7631 x500UniqueIdentifier, not uniqueIdentifier. 7632 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 7633 [Lutz Jaenicke] 7634 7635 *) Add an "init" command to the ENGINE config module and auto initialize 7636 ENGINEs. Without any "init" command the ENGINE will be initialized 7637 after all ctrl commands have been executed on it. If init=1 the 7638 ENGINE is initialized at that point (ctrls before that point are run 7639 on the uninitialized ENGINE and after on the initialized one). If 7640 init=0 then the ENGINE will not be initialized at all. 7641 [Steve Henson] 7642 7643 *) Fix the 'app_verify_callback' interface so that the user-defined 7644 argument is actually passed to the callback: In the 7645 SSL_CTX_set_cert_verify_callback() prototype, the callback 7646 declaration has been changed from 7647 int (*cb)() 7648 into 7649 int (*cb)(X509_STORE_CTX *,void *); 7650 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 7651 i=s->ctx->app_verify_callback(&ctx) 7652 has been changed into 7653 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 7654 7655 To update applications using SSL_CTX_set_cert_verify_callback(), 7656 a dummy argument can be added to their callback functions. 7657 [D. K. Smetters <smetters@parc.xerox.com>] 7658 7659 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 7660 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 7661 7662 *) Add and OPENSSL_LOAD_CONF define which will cause 7663 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 7664 This allows older applications to transparently support certain 7665 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 7666 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 7667 load the config file and OPENSSL_add_all_algorithms_conf() which will 7668 always load it have also been added. 7669 [Steve Henson] 7670 7671 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 7672 Adjust NIDs and EVP layer. 7673 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 7674 7675 *) Config modules support in openssl utility. 7676 7677 Most commands now load modules from the config file, 7678 though in a few (such as version) this isn't done 7679 because it couldn't be used for anything. 7680 7681 In the case of ca and req the config file used is 7682 the same as the utility itself: that is the -config 7683 command line option can be used to specify an 7684 alternative file. 7685 [Steve Henson] 7686 7687 *) Move default behaviour from OPENSSL_config(). If appname is NULL 7688 use "openssl_conf" if filename is NULL use default openssl config file. 7689 [Steve Henson] 7690 7691 *) Add an argument to OPENSSL_config() to allow the use of an alternative 7692 config section name. Add a new flag to tolerate a missing config file 7693 and move code to CONF_modules_load_file(). 7694 [Steve Henson] 7695 7696 *) Support for crypto accelerator cards from Accelerated Encryption 7697 Processing, www.aep.ie. (Use engine 'aep') 7698 The support was copied from 0.9.6c [engine] and adapted/corrected 7699 to work with the new engine framework. 7700 [AEP Inc. and Richard Levitte] 7701 7702 *) Support for SureWare crypto accelerator cards from Baltimore 7703 Technologies. (Use engine 'sureware') 7704 The support was copied from 0.9.6c [engine] and adapted 7705 to work with the new engine framework. 7706 [Richard Levitte] 7707 7708 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 7709 make the newer ENGINE framework commands for the CHIL engine work. 7710 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 7711 7712 *) Make it possible to produce shared libraries on ReliantUNIX. 7713 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 7714 7715 *) Add the configuration target debug-linux-ppro. 7716 Make 'openssl rsa' use the general key loading routines 7717 implemented in apps.c, and make those routines able to 7718 handle the key format FORMAT_NETSCAPE and the variant 7719 FORMAT_IISSGC. 7720 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7721 7722 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 7723 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7724 7725 *) Add -keyform to rsautl, and document -engine. 7726 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 7727 7728 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 7729 BIO_R_NO_SUCH_FILE error code rather than the generic 7730 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 7731 [Ben Laurie] 7732 7733 *) Add new functions 7734 ERR_peek_last_error 7735 ERR_peek_last_error_line 7736 ERR_peek_last_error_line_data. 7737 These are similar to 7738 ERR_peek_error 7739 ERR_peek_error_line 7740 ERR_peek_error_line_data, 7741 but report on the latest error recorded rather than the first one 7742 still in the error queue. 7743 [Ben Laurie, Bodo Moeller] 7744 7745 *) default_algorithms option in ENGINE config module. This allows things 7746 like: 7747 default_algorithms = ALL 7748 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 7749 [Steve Henson] 7750 7751 *) Preliminary ENGINE config module. 7752 [Steve Henson] 7753 7754 *) New experimental application configuration code. 7755 [Steve Henson] 7756 7757 *) Change the AES code to follow the same name structure as all other 7758 symmetric ciphers, and behave the same way. Move everything to 7759 the directory crypto/aes, thereby obsoleting crypto/rijndael. 7760 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 7761 7762 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 7763 [Ben Laurie and Theo de Raadt] 7764 7765 *) Add option to output public keys in req command. 7766 [Massimiliano Pala madwolf@openca.org] 7767 7768 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 7769 (up to about 10% better than before for P-192 and P-224). 7770 [Bodo Moeller] 7771 7772 *) New functions/macros 7773 7774 SSL_CTX_set_msg_callback(ctx, cb) 7775 SSL_CTX_set_msg_callback_arg(ctx, arg) 7776 SSL_set_msg_callback(ssl, cb) 7777 SSL_set_msg_callback_arg(ssl, arg) 7778 7779 to request calling a callback function 7780 7781 void cb(int write_p, int version, int content_type, 7782 const void *buf, size_t len, SSL *ssl, void *arg) 7783 7784 whenever a protocol message has been completely received 7785 (write_p == 0) or sent (write_p == 1). Here 'version' is the 7786 protocol version according to which the SSL library interprets 7787 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 7788 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 7789 the content type as defined in the SSL 3.0/TLS 1.0 protocol 7790 specification (change_cipher_spec(20), alert(21), handshake(22)). 7791 'buf' and 'len' point to the actual message, 'ssl' to the 7792 SSL object, and 'arg' is the application-defined value set by 7793 SSL[_CTX]_set_msg_callback_arg(). 7794 7795 'openssl s_client' and 'openssl s_server' have new '-msg' options 7796 to enable a callback that displays all protocol messages. 7797 [Bodo Moeller] 7798 7799 *) Change the shared library support so shared libraries are built as 7800 soon as the corresponding static library is finished, and thereby get 7801 openssl and the test programs linked against the shared library. 7802 This still only happens when the keyword "shard" has been given to 7803 the configuration scripts. 7804 7805 NOTE: shared library support is still an experimental thing, and 7806 backward binary compatibility is still not guaranteed. 7807 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 7808 7809 *) Add support for Subject Information Access extension. 7810 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 7811 7812 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 7813 additional bytes when new memory had to be allocated, not just 7814 when reusing an existing buffer. 7815 [Bodo Moeller] 7816 7817 *) New command line and configuration option 'utf8' for the req command. 7818 This allows field values to be specified as UTF8 strings. 7819 [Steve Henson] 7820 7821 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 7822 runs for the former and machine-readable output for the latter. 7823 [Ben Laurie] 7824 7825 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 7826 of the e-mail address in the DN (i.e., it will go into a certificate 7827 extension only). The new configuration file option 'email_in_dn = no' 7828 has the same effect. 7829 [Massimiliano Pala madwolf@openca.org] 7830 7831 *) Change all functions with names starting with des_ to be starting 7832 with DES_ instead. Add wrappers that are compatible with libdes, 7833 but are named _ossl_old_des_*. Finally, add macros that map the 7834 des_* symbols to the corresponding _ossl_old_des_* if libdes 7835 compatibility is desired. If OpenSSL 0.9.6c compatibility is 7836 desired, the des_* symbols will be mapped to DES_*, with one 7837 exception. 7838 7839 Since we provide two compatibility mappings, the user needs to 7840 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 7841 compatibility is desired. The default (i.e., when that macro 7842 isn't defined) is OpenSSL 0.9.6c compatibility. 7843 7844 There are also macros that enable and disable the support of old 7845 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 7846 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 7847 are defined, the default will apply: to support the old des routines. 7848 7849 In either case, one must include openssl/des.h to get the correct 7850 definitions. Do not try to just include openssl/des_old.h, that 7851 won't work. 7852 7853 NOTE: This is a major break of an old API into a new one. Software 7854 authors are encouraged to switch to the DES_ style functions. Some 7855 time in the future, des_old.h and the libdes compatibility functions 7856 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 7857 default), and then completely removed. 7858 [Richard Levitte] 7859 7860 *) Test for certificates which contain unsupported critical extensions. 7861 If such a certificate is found during a verify operation it is 7862 rejected by default: this behaviour can be overridden by either 7863 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 7864 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 7865 X509_supported_extension() has also been added which returns 1 if a 7866 particular extension is supported. 7867 [Steve Henson] 7868 7869 *) Modify the behaviour of EVP cipher functions in similar way to digests 7870 to retain compatibility with existing code. 7871 [Steve Henson] 7872 7873 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 7874 compatibility with existing code. In particular the 'ctx' parameter does 7875 not have to be to be initialized before the call to EVP_DigestInit() and 7876 it is tidied up after a call to EVP_DigestFinal(). New function 7877 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 7878 EVP_MD_CTX_copy() changed to not require the destination to be 7879 initialized valid and new function EVP_MD_CTX_copy_ex() added which 7880 requires the destination to be valid. 7881 7882 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 7883 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 7884 [Steve Henson] 7885 7886 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 7887 so that complete 'Handshake' protocol structures are kept in memory 7888 instead of overwriting 'msg_type' and 'length' with 'body' data. 7889 [Bodo Moeller] 7890 7891 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 7892 [Massimo Santin via Richard Levitte] 7893 7894 *) Major restructuring to the underlying ENGINE code. This includes 7895 reduction of linker bloat, separation of pure "ENGINE" manipulation 7896 (initialisation, etc) from functionality dealing with implementations 7897 of specific crypto interfaces. This change also introduces integrated 7898 support for symmetric ciphers and digest implementations - so ENGINEs 7899 can now accelerate these by providing EVP_CIPHER and EVP_MD 7900 implementations of their own. This is detailed in crypto/engine/README 7901 as it couldn't be adequately described here. However, there are a few 7902 API changes worth noting - some RSA, DSA, DH, and RAND functions that 7903 were changed in the original introduction of ENGINE code have now 7904 reverted back - the hooking from this code to ENGINE is now a good 7905 deal more passive and at run-time, operations deal directly with 7906 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 7907 dereferencing through an ENGINE pointer any more. Also, the ENGINE 7908 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 7909 they were not being used by the framework as there is no concept of a 7910 BIGNUM_METHOD and they could not be generalised to the new 7911 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 7912 ENGINE_cpy() has been removed as it cannot be consistently defined in 7913 the new code. 7914 [Geoff Thorpe] 7915 7916 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 7917 [Steve Henson] 7918 7919 *) Change mkdef.pl to sort symbols that get the same entry number, 7920 and make sure the automatically generated functions ERR_load_* 7921 become part of libeay.num as well. 7922 [Richard Levitte] 7923 7924 *) New function SSL_renegotiate_pending(). This returns true once 7925 renegotiation has been requested (either SSL_renegotiate() call 7926 or HelloRequest/ClientHello received from the peer) and becomes 7927 false once a handshake has been completed. 7928 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 7929 sends a HelloRequest, but does not ensure that a handshake takes 7930 place. SSL_renegotiate_pending() is useful for checking if the 7931 client has followed the request.) 7932 [Bodo Moeller] 7933 7934 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 7935 By default, clients may request session resumption even during 7936 renegotiation (if session ID contexts permit); with this option, 7937 session resumption is possible only in the first handshake. 7938 7939 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 7940 more bits available for options that should not be part of 7941 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 7942 [Bodo Moeller] 7943 7944 *) Add some demos for certificate and certificate request creation. 7945 [Steve Henson] 7946 7947 *) Make maximum certificate chain size accepted from the peer application 7948 settable (SSL*_get/set_max_cert_list()), as proposed by 7949 "Douglas E. Engert" <deengert@anl.gov>. 7950 [Lutz Jaenicke] 7951 7952 *) Add support for shared libraries for Unixware-7 7953 (Boyd Lynn Gerber <gerberb@zenez.com>). 7954 [Lutz Jaenicke] 7955 7956 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 7957 be done prior to destruction. Use this to unload error strings from 7958 ENGINEs that load their own error strings. NB: This adds two new API 7959 functions to "get" and "set" this destroy handler in an ENGINE. 7960 [Geoff Thorpe] 7961 7962 *) Alter all existing ENGINE implementations (except "openssl" and 7963 "openbsd") to dynamically instantiate their own error strings. This 7964 makes them more flexible to be built both as statically-linked ENGINEs 7965 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 7966 Also, add stub code to each that makes building them as self-contained 7967 shared-libraries easier (see README.ENGINE). 7968 [Geoff Thorpe] 7969 7970 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 7971 implementations into applications that are completely implemented in 7972 self-contained shared-libraries. The "dynamic" ENGINE exposes control 7973 commands that can be used to configure what shared-library to load and 7974 to control aspects of the way it is handled. Also, made an update to 7975 the README.ENGINE file that brings its information up-to-date and 7976 provides some information and instructions on the "dynamic" ENGINE 7977 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 7978 [Geoff Thorpe] 7979 7980 *) Make it possible to unload ranges of ERR strings with a new 7981 "ERR_unload_strings" function. 7982 [Geoff Thorpe] 7983 7984 *) Add a copy() function to EVP_MD. 7985 [Ben Laurie] 7986 7987 *) Make EVP_MD routines take a context pointer instead of just the 7988 md_data void pointer. 7989 [Ben Laurie] 7990 7991 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 7992 that the digest can only process a single chunk of data 7993 (typically because it is provided by a piece of 7994 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 7995 is only going to provide a single chunk of data, and hence the 7996 framework needn't accumulate the data for oneshot drivers. 7997 [Ben Laurie] 7998 7999 *) As with "ERR", make it possible to replace the underlying "ex_data" 8000 functions. This change also alters the storage and management of global 8001 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 8002 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 8003 index counters. The API functions that use this state have been changed 8004 to take a "class_index" rather than pointers to the class's local STACK 8005 and counter, and there is now an API function to dynamically create new 8006 classes. This centralisation allows us to (a) plug a lot of the 8007 thread-safety problems that existed, and (b) makes it possible to clean 8008 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 8009 such data would previously have always leaked in application code and 8010 workarounds were in place to make the memory debugging turn a blind eye 8011 to it. Application code that doesn't use this new function will still 8012 leak as before, but their memory debugging output will announce it now 8013 rather than letting it slide. 8014 8015 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 8016 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 8017 has a return value to indicate success or failure. 8018 [Geoff Thorpe] 8019 8020 *) Make it possible to replace the underlying "ERR" functions such that the 8021 global state (2 LHASH tables and 2 locks) is only used by the "default" 8022 implementation. This change also adds two functions to "get" and "set" 8023 the implementation prior to it being automatically set the first time 8024 any other ERR function takes place. Ie. an application can call "get", 8025 pass the return value to a module it has just loaded, and that module 8026 can call its own "set" function using that value. This means the 8027 module's "ERR" operations will use (and modify) the error state in the 8028 application and not in its own statically linked copy of OpenSSL code. 8029 [Geoff Thorpe] 8030 8031 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 8032 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 8033 the operation, and provides a more encapsulated way for external code 8034 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 8035 to use these functions rather than manually incrementing the counts. 8036 8037 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 8038 [Geoff Thorpe] 8039 8040 *) Add EVP test program. 8041 [Ben Laurie] 8042 8043 *) Add symmetric cipher support to ENGINE. Expect the API to change! 8044 [Ben Laurie] 8045 8046 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 8047 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 8048 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 8049 These allow a CRL to be built without having to access X509_CRL fields 8050 directly. Modify 'ca' application to use new functions. 8051 [Steve Henson] 8052 8053 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 8054 bug workarounds. Rollback attack detection is a security feature. 8055 The problem will only arise on OpenSSL servers when TLSv1 is not 8056 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 8057 Software authors not wanting to support TLSv1 will have special reasons 8058 for their choice and can explicitly enable this option. 8059 [Bodo Moeller, Lutz Jaenicke] 8060 8061 *) Rationalise EVP so it can be extended: don't include a union of 8062 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 8063 (similar to those existing for EVP_CIPHER_CTX). 8064 Usage example: 8065 8066 EVP_MD_CTX md; 8067 8068 EVP_MD_CTX_init(&md); /* new function call */ 8069 EVP_DigestInit(&md, EVP_sha1()); 8070 EVP_DigestUpdate(&md, in, len); 8071 EVP_DigestFinal(&md, out, NULL); 8072 EVP_MD_CTX_cleanup(&md); /* new function call */ 8073 8074 [Ben Laurie] 8075 8076 *) Make DES key schedule conform to the usual scheme, as well as 8077 correcting its structure. This means that calls to DES functions 8078 now have to pass a pointer to a des_key_schedule instead of a 8079 plain des_key_schedule (which was actually always a pointer 8080 anyway): E.g., 8081 8082 des_key_schedule ks; 8083 8084 des_set_key_checked(..., &ks); 8085 des_ncbc_encrypt(..., &ks, ...); 8086 8087 (Note that a later change renames 'des_...' into 'DES_...'.) 8088 [Ben Laurie] 8089 8090 *) Initial reduction of linker bloat: the use of some functions, such as 8091 PEM causes large amounts of unused functions to be linked in due to 8092 poor organisation. For example pem_all.c contains every PEM function 8093 which has a knock on effect of linking in large amounts of (unused) 8094 ASN1 code. Grouping together similar functions and splitting unrelated 8095 functions prevents this. 8096 [Steve Henson] 8097 8098 *) Cleanup of EVP macros. 8099 [Ben Laurie] 8100 8101 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 8102 correct _ecb suffix. 8103 [Ben Laurie] 8104 8105 *) Add initial OCSP responder support to ocsp application. The 8106 revocation information is handled using the text based index 8107 use by the ca application. The responder can either handle 8108 requests generated internally, supplied in files (for example 8109 via a CGI script) or using an internal minimal server. 8110 [Steve Henson] 8111 8112 *) Add configuration choices to get zlib compression for TLS. 8113 [Richard Levitte] 8114 8115 *) Changes to Kerberos SSL for RFC 2712 compliance: 8116 1. Implemented real KerberosWrapper, instead of just using 8117 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 8118 2. Implemented optional authenticator field of KerberosWrapper. 8119 8120 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 8121 and authenticator structs; see crypto/krb5/. 8122 8123 Generalized Kerberos calls to support multiple Kerberos libraries. 8124 [Vern Staats <staatsvr@asc.hpc.mil>, 8125 Jeffrey Altman <jaltman@columbia.edu> 8126 via Richard Levitte] 8127 8128 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 8129 already does with RSA. testdsa.h now has 'priv_key/pub_key' 8130 values for each of the key sizes rather than having just 8131 parameters (and 'speed' generating keys each time). 8132 [Geoff Thorpe] 8133 8134 *) Speed up EVP routines. 8135 Before: 8136encrypt 8137type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 8138des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 8139des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 8140des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 8141decrypt 8142des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 8143des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 8144des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 8145 After: 8146encrypt 8147des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 8148decrypt 8149des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 8150 [Ben Laurie] 8151 8152 *) Added the OS2-EMX target. 8153 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 8154 8155 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 8156 to support NCONF routines in extension code. New function CONF_set_nconf() 8157 to allow functions which take an NCONF to also handle the old LHASH 8158 structure: this means that the old CONF compatible routines can be 8159 retained (in particular wrt extensions) without having to duplicate the 8160 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 8161 [Steve Henson] 8162 8163 *) Enhance the general user interface with mechanisms for inner control 8164 and with possibilities to have yes/no kind of prompts. 8165 [Richard Levitte] 8166 8167 *) Change all calls to low level digest routines in the library and 8168 applications to use EVP. Add missing calls to HMAC_cleanup() and 8169 don't assume HMAC_CTX can be copied using memcpy(). 8170 [Verdon Walker <VWalker@novell.com>, Steve Henson] 8171 8172 *) Add the possibility to control engines through control names but with 8173 arbitrary arguments instead of just a string. 8174 Change the key loaders to take a UI_METHOD instead of a callback 8175 function pointer. NOTE: this breaks binary compatibility with earlier 8176 versions of OpenSSL [engine]. 8177 Adapt the nCipher code for these new conditions and add a card insertion 8178 callback. 8179 [Richard Levitte] 8180 8181 *) Enhance the general user interface with mechanisms to better support 8182 dialog box interfaces, application-defined prompts, the possibility 8183 to use defaults (for example default passwords from somewhere else) 8184 and interrupts/cancellations. 8185 [Richard Levitte] 8186 8187 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 8188 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 8189 [Steve Henson] 8190 8191 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 8192 tidy up some unnecessarily weird code in 'sk_new()'). 8193 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 8194 8195 *) Change the key loading routines for ENGINEs to use the same kind 8196 callback (pem_password_cb) as all other routines that need this 8197 kind of callback. 8198 [Richard Levitte] 8199 8200 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 8201 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 8202 than this minimum value is recommended. 8203 [Lutz Jaenicke] 8204 8205 *) New random seeder for OpenVMS, using the system process statistics 8206 that are easily reachable. 8207 [Richard Levitte] 8208 8209 *) Windows apparently can't transparently handle global 8210 variables defined in DLLs. Initialisations such as: 8211 8212 const ASN1_ITEM *it = &ASN1_INTEGER_it; 8213 8214 won't compile. This is used by the any applications that need to 8215 declare their own ASN1 modules. This was fixed by adding the option 8216 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 8217 needed for static libraries under Win32. 8218 [Steve Henson] 8219 8220 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 8221 setting of purpose and trust fields. New X509_STORE trust and 8222 purpose functions and tidy up setting in other SSL functions. 8223 [Steve Henson] 8224 8225 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 8226 structure. These are inherited by X509_STORE_CTX when it is 8227 initialised. This allows various defaults to be set in the 8228 X509_STORE structure (such as flags for CRL checking and custom 8229 purpose or trust settings) for functions which only use X509_STORE_CTX 8230 internally such as S/MIME. 8231 8232 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 8233 trust settings if they are not set in X509_STORE. This allows X509_STORE 8234 purposes and trust (in S/MIME for example) to override any set by default. 8235 8236 Add command line options for CRL checking to smime, s_client and s_server 8237 applications. 8238 [Steve Henson] 8239 8240 *) Initial CRL based revocation checking. If the CRL checking flag(s) 8241 are set then the CRL is looked up in the X509_STORE structure and 8242 its validity and signature checked, then if the certificate is found 8243 in the CRL the verify fails with a revoked error. 8244 8245 Various new CRL related callbacks added to X509_STORE_CTX structure. 8246 8247 Command line options added to 'verify' application to support this. 8248 8249 This needs some additional work, such as being able to handle multiple 8250 CRLs with different times, extension based lookup (rather than just 8251 by subject name) and ultimately more complete V2 CRL extension 8252 handling. 8253 [Steve Henson] 8254 8255 *) Add a general user interface API (crypto/ui/). This is designed 8256 to replace things like des_read_password and friends (backward 8257 compatibility functions using this new API are provided). 8258 The purpose is to remove prompting functions from the DES code 8259 section as well as provide for prompting through dialog boxes in 8260 a window system and the like. 8261 [Richard Levitte] 8262 8263 *) Add "ex_data" support to ENGINE so implementations can add state at a 8264 per-structure level rather than having to store it globally. 8265 [Geoff] 8266 8267 *) Make it possible for ENGINE structures to be copied when retrieved by 8268 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 8269 This causes the "original" ENGINE structure to act like a template, 8270 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 8271 operational state can be localised to each ENGINE structure, despite the 8272 fact they all share the same "methods". New ENGINE structures returned in 8273 this case have no functional references and the return value is the single 8274 structural reference. This matches the single structural reference returned 8275 by ENGINE_by_id() normally, when it is incremented on the pre-existing 8276 ENGINE structure. 8277 [Geoff] 8278 8279 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 8280 needs to match any other type at all we need to manually clear the 8281 tag cache. 8282 [Steve Henson] 8283 8284 *) Changes to the "openssl engine" utility to include; 8285 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 8286 about an ENGINE's available control commands. 8287 - executing control commands from command line arguments using the 8288 '-pre' and '-post' switches. '-post' is only used if '-t' is 8289 specified and the ENGINE is successfully initialised. The syntax for 8290 the individual commands are colon-separated, for example; 8291 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 8292 [Geoff] 8293 8294 *) New dynamic control command support for ENGINEs. ENGINEs can now 8295 declare their own commands (numbers), names (strings), descriptions, 8296 and input types for run-time discovery by calling applications. A 8297 subset of these commands are implicitly classed as "executable" 8298 depending on their input type, and only these can be invoked through 8299 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 8300 can be based on user input, config files, etc). The distinction is 8301 that "executable" commands cannot return anything other than a boolean 8302 result and can only support numeric or string input, whereas some 8303 discoverable commands may only be for direct use through 8304 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 8305 pointers, or other custom uses. The "executable" commands are to 8306 support parameterisations of ENGINE behaviour that can be 8307 unambiguously defined by ENGINEs and used consistently across any 8308 OpenSSL-based application. Commands have been added to all the 8309 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 8310 control over shared-library paths without source code alterations. 8311 [Geoff] 8312 8313 *) Changed all ENGINE implementations to dynamically allocate their 8314 ENGINEs rather than declaring them statically. Apart from this being 8315 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 8316 this also allows the implementations to compile without using the 8317 internal engine_int.h header. 8318 [Geoff] 8319 8320 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 8321 'const' value. Any code that should be able to modify a RAND_METHOD 8322 should already have non-const pointers to it (ie. they should only 8323 modify their own ones). 8324 [Geoff] 8325 8326 *) Made a variety of little tweaks to the ENGINE code. 8327 - "atalla" and "ubsec" string definitions were moved from header files 8328 to C code. "nuron" string definitions were placed in variables 8329 rather than hard-coded - allowing parameterisation of these values 8330 later on via ctrl() commands. 8331 - Removed unused "#if 0"'d code. 8332 - Fixed engine list iteration code so it uses ENGINE_free() to release 8333 structural references. 8334 - Constified the RAND_METHOD element of ENGINE structures. 8335 - Constified various get/set functions as appropriate and added 8336 missing functions (including a catch-all ENGINE_cpy that duplicates 8337 all ENGINE values onto a new ENGINE except reference counts/state). 8338 - Removed NULL parameter checks in get/set functions. Setting a method 8339 or function to NULL is a way of cancelling out a previously set 8340 value. Passing a NULL ENGINE parameter is just plain stupid anyway 8341 and doesn't justify the extra error symbols and code. 8342 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 8343 flags from engine_int.h to engine.h. 8344 - Changed prototypes for ENGINE handler functions (init(), finish(), 8345 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 8346 [Geoff] 8347 8348 *) Implement binary inversion algorithm for BN_mod_inverse in addition 8349 to the algorithm using long division. The binary algorithm can be 8350 used only if the modulus is odd. On 32-bit systems, it is faster 8351 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 8352 roughly 5-15% for 256-bit moduli), so we use it only for moduli 8353 up to 450 bits. In 64-bit environments, the binary algorithm 8354 appears to be advantageous for much longer moduli; here we use it 8355 for moduli up to 2048 bits. 8356 [Bodo Moeller] 8357 8358 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 8359 could not support the combine flag in choice fields. 8360 [Steve Henson] 8361 8362 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 8363 extensions from a certificate request to the certificate. 8364 [Steve Henson] 8365 8366 *) Allow multiple 'certopt' and 'nameopt' options to be separated 8367 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 8368 file: this allows the display of the certificate about to be 8369 signed to be customised, to allow certain fields to be included 8370 or excluded and extension details. The old system didn't display 8371 multicharacter strings properly, omitted fields not in the policy 8372 and couldn't display additional details such as extensions. 8373 [Steve Henson] 8374 8375 *) Function EC_POINTs_mul for multiple scalar multiplication 8376 of an arbitrary number of elliptic curve points 8377 \sum scalars[i]*points[i], 8378 optionally including the generator defined for the EC_GROUP: 8379 scalar*generator + \sum scalars[i]*points[i]. 8380 8381 EC_POINT_mul is a simple wrapper function for the typical case 8382 that the point list has just one item (besides the optional 8383 generator). 8384 [Bodo Moeller] 8385 8386 *) First EC_METHODs for curves over GF(p): 8387 8388 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 8389 operations and provides various method functions that can also 8390 operate with faster implementations of modular arithmetic. 8391 8392 EC_GFp_mont_method() reuses most functions that are part of 8393 EC_GFp_simple_method, but uses Montgomery arithmetic. 8394 8395 [Bodo Moeller; point addition and point doubling 8396 implementation directly derived from source code provided by 8397 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 8398 8399 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 8400 crypto/ec/ec_lib.c): 8401 8402 Curves are EC_GROUP objects (with an optional group generator) 8403 based on EC_METHODs that are built into the library. 8404 8405 Points are EC_POINT objects based on EC_GROUP objects. 8406 8407 Most of the framework would be able to handle curves over arbitrary 8408 finite fields, but as there are no obvious types for fields other 8409 than GF(p), some functions are limited to that for now. 8410 [Bodo Moeller] 8411 8412 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 8413 that the file contains a complete HTTP response. 8414 [Richard Levitte] 8415 8416 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 8417 change the def and num file printf format specifier from "%-40sXXX" 8418 to "%-39s XXX". The latter will always guarantee a space after the 8419 field while the former will cause them to run together if the field 8420 is 40 of more characters long. 8421 [Steve Henson] 8422 8423 *) Constify the cipher and digest 'method' functions and structures 8424 and modify related functions to take constant EVP_MD and EVP_CIPHER 8425 pointers. 8426 [Steve Henson] 8427 8428 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 8429 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 8430 [Bodo Moeller] 8431 8432 *) Modify EVP_Digest*() routines so they now return values. Although the 8433 internal software routines can never fail additional hardware versions 8434 might. 8435 [Steve Henson] 8436 8437 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 8438 8439 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 8440 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 8441 8442 ASN1 error codes 8443 ERR_R_NESTED_ASN1_ERROR 8444 ... 8445 ERR_R_MISSING_ASN1_EOS 8446 were 4 .. 9, conflicting with 8447 ERR_LIB_RSA (= ERR_R_RSA_LIB) 8448 ... 8449 ERR_LIB_PEM (= ERR_R_PEM_LIB). 8450 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 8451 8452 Add new error code 'ERR_R_INTERNAL_ERROR'. 8453 [Bodo Moeller] 8454 8455 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 8456 suffices. 8457 [Bodo Moeller] 8458 8459 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 8460 sets the subject name for a new request or supersedes the 8461 subject name in a given request. Formats that can be parsed are 8462 'CN=Some Name, OU=myOU, C=IT' 8463 and 8464 'CN=Some Name/OU=myOU/C=IT'. 8465 8466 Add options '-batch' and '-verbose' to 'openssl req'. 8467 [Massimiliano Pala <madwolf@hackmasters.net>] 8468 8469 *) Introduce the possibility to access global variables through 8470 functions on platform were that's the best way to handle exporting 8471 global variables in shared libraries. To enable this functionality, 8472 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 8473 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 8474 is normally done by Configure or something similar). 8475 8476 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 8477 in the source file (foo.c) like this: 8478 8479 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 8480 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 8481 8482 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 8483 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 8484 8485 OPENSSL_DECLARE_GLOBAL(int,foo); 8486 #define foo OPENSSL_GLOBAL_REF(foo) 8487 OPENSSL_DECLARE_GLOBAL(double,bar); 8488 #define bar OPENSSL_GLOBAL_REF(bar) 8489 8490 The #defines are very important, and therefore so is including the 8491 header file everywhere where the defined globals are used. 8492 8493 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 8494 of ASN.1 items, but that structure is a bit different. 8495 8496 The largest change is in util/mkdef.pl which has been enhanced with 8497 better and easier to understand logic to choose which symbols should 8498 go into the Windows .def files as well as a number of fixes and code 8499 cleanup (among others, algorithm keywords are now sorted 8500 lexicographically to avoid constant rewrites). 8501 [Richard Levitte] 8502 8503 *) In BN_div() keep a copy of the sign of 'num' before writing the 8504 result to 'rm' because if rm==num the value will be overwritten 8505 and produce the wrong result if 'num' is negative: this caused 8506 problems with BN_mod() and BN_nnmod(). 8507 [Steve Henson] 8508 8509 *) Function OCSP_request_verify(). This checks the signature on an 8510 OCSP request and verifies the signer certificate. The signer 8511 certificate is just checked for a generic purpose and OCSP request 8512 trust settings. 8513 [Steve Henson] 8514 8515 *) Add OCSP_check_validity() function to check the validity of OCSP 8516 responses. OCSP responses are prepared in real time and may only 8517 be a few seconds old. Simply checking that the current time lies 8518 between thisUpdate and nextUpdate max reject otherwise valid responses 8519 caused by either OCSP responder or client clock inaccuracy. Instead 8520 we allow thisUpdate and nextUpdate to fall within a certain period of 8521 the current time. The age of the response can also optionally be 8522 checked. Two new options -validity_period and -status_age added to 8523 ocsp utility. 8524 [Steve Henson] 8525 8526 *) If signature or public key algorithm is unrecognized print out its 8527 OID rather that just UNKNOWN. 8528 [Steve Henson] 8529 8530 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 8531 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 8532 ID to be generated from the issuer certificate alone which can then be 8533 passed to OCSP_id_issuer_cmp(). 8534 [Steve Henson] 8535 8536 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 8537 ASN1 modules to export functions returning ASN1_ITEM pointers 8538 instead of the ASN1_ITEM structures themselves. This adds several 8539 new macros which allow the underlying ASN1 function/structure to 8540 be accessed transparently. As a result code should not use ASN1_ITEM 8541 references directly (such as &X509_it) but instead use the relevant 8542 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 8543 use of the new ASN1 code on platforms where exporting structures 8544 is problematical (for example in shared libraries) but exporting 8545 functions returning pointers to structures is not. 8546 [Steve Henson] 8547 8548 *) Add support for overriding the generation of SSL/TLS session IDs. 8549 These callbacks can be registered either in an SSL_CTX or per SSL. 8550 The purpose of this is to allow applications to control, if they wish, 8551 the arbitrary values chosen for use as session IDs, particularly as it 8552 can be useful for session caching in multiple-server environments. A 8553 command-line switch for testing this (and any client code that wishes 8554 to use such a feature) has been added to "s_server". 8555 [Geoff Thorpe, Lutz Jaenicke] 8556 8557 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 8558 of the form '#if defined(...) || defined(...) || ...' and 8559 '#if !defined(...) && !defined(...) && ...'. This also avoids 8560 the growing number of special cases it was previously handling. 8561 [Richard Levitte] 8562 8563 *) Make all configuration macros available for application by making 8564 sure they are available in opensslconf.h, by giving them names starting 8565 with "OPENSSL_" to avoid conflicts with other packages and by making 8566 sure e_os2.h will cover all platform-specific cases together with 8567 opensslconf.h. 8568 Additionally, it is now possible to define configuration/platform- 8569 specific names (called "system identities"). In the C code, these 8570 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 8571 macro with the name beginning with "OPENSSL_SYS_", which is determined 8572 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 8573 what is available. 8574 [Richard Levitte] 8575 8576 *) New option -set_serial to 'req' and 'x509' this allows the serial 8577 number to use to be specified on the command line. Previously self 8578 signed certificates were hard coded with serial number 0 and the 8579 CA options of 'x509' had to use a serial number in a file which was 8580 auto incremented. 8581 [Steve Henson] 8582 8583 *) New options to 'ca' utility to support V2 CRL entry extensions. 8584 Currently CRL reason, invalidity date and hold instruction are 8585 supported. Add new CRL extensions to V3 code and some new objects. 8586 [Steve Henson] 8587 8588 *) New function EVP_CIPHER_CTX_set_padding() this is used to 8589 disable standard block padding (aka PKCS#5 padding) in the EVP 8590 API, which was previously mandatory. This means that the data is 8591 not padded in any way and so the total length much be a multiple 8592 of the block size, otherwise an error occurs. 8593 [Steve Henson] 8594 8595 *) Initial (incomplete) OCSP SSL support. 8596 [Steve Henson] 8597 8598 *) New function OCSP_parse_url(). This splits up a URL into its host, 8599 port and path components: primarily to parse OCSP URLs. New -url 8600 option to ocsp utility. 8601 [Steve Henson] 8602 8603 *) New nonce behavior. The return value of OCSP_check_nonce() now 8604 reflects the various checks performed. Applications can decide 8605 whether to tolerate certain situations such as an absent nonce 8606 in a response when one was present in a request: the ocsp application 8607 just prints out a warning. New function OCSP_add1_basic_nonce() 8608 this is to allow responders to include a nonce in a response even if 8609 the request is nonce-less. 8610 [Steve Henson] 8611 8612 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 8613 skipped when using openssl x509 multiple times on a single input file, 8614 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 8615 [Bodo Moeller] 8616 8617 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 8618 set string type: to handle setting ASN1_TIME structures. Fix ca 8619 utility to correctly initialize revocation date of CRLs. 8620 [Steve Henson] 8621 8622 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 8623 the clients preferred ciphersuites and rather use its own preferences. 8624 Should help to work around M$ SGC (Server Gated Cryptography) bug in 8625 Internet Explorer by ensuring unchanged hash method during stepup. 8626 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 8627 [Lutz Jaenicke] 8628 8629 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 8630 to aes and add a new 'exist' option to print out symbols that don't 8631 appear to exist. 8632 [Steve Henson] 8633 8634 *) Additional options to ocsp utility to allow flags to be set and 8635 additional certificates supplied. 8636 [Steve Henson] 8637 8638 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 8639 OCSP client a number of certificate to only verify the response 8640 signature against. 8641 [Richard Levitte] 8642 8643 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 8644 handle the new API. Currently only ECB, CBC modes supported. Add new 8645 AES OIDs. 8646 8647 Add TLS AES ciphersuites as described in RFC3268, "Advanced 8648 Encryption Standard (AES) Ciphersuites for Transport Layer 8649 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 8650 not enabled by default and were not part of the "ALL" ciphersuite 8651 alias because they were not yet official; they could be 8652 explicitly requested by specifying the "AESdraft" ciphersuite 8653 group alias. In the final release of OpenSSL 0.9.7, the group 8654 alias is called "AES" and is part of "ALL".) 8655 [Ben Laurie, Steve Henson, Bodo Moeller] 8656 8657 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 8658 request to response. 8659 [Steve Henson] 8660 8661 *) Functions for OCSP responders. OCSP_request_onereq_count(), 8662 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 8663 extract information from a certificate request. OCSP_response_create() 8664 creates a response and optionally adds a basic response structure. 8665 OCSP_basic_add1_status() adds a complete single response to a basic 8666 response and returns the OCSP_SINGLERESP structure just added (to allow 8667 extensions to be included for example). OCSP_basic_add1_cert() adds a 8668 certificate to a basic response and OCSP_basic_sign() signs a basic 8669 response with various flags. New helper functions ASN1_TIME_check() 8670 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 8671 (converts ASN1_TIME to GeneralizedTime). 8672 [Steve Henson] 8673 8674 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 8675 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 8676 structure from a certificate. X509_pubkey_digest() digests the public_key 8677 contents: this is used in various key identifiers. 8678 [Steve Henson] 8679 8680 *) Make sk_sort() tolerate a NULL argument. 8681 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 8682 8683 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 8684 passed by the function are trusted implicitly. If any of them signed the 8685 response then it is assumed to be valid and is not verified. 8686 [Steve Henson] 8687 8688 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 8689 to data. This was previously part of the PKCS7 ASN1 code. This 8690 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 8691 [Steve Henson, reported by Kenneth R. Robinette 8692 <support@securenetterm.com>] 8693 8694 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 8695 routines: without these tracing memory leaks is very painful. 8696 Fix leaks in PKCS12 and PKCS7 routines. 8697 [Steve Henson] 8698 8699 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 8700 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 8701 effectively meant GeneralizedTime would never be used. Now it 8702 is initialised to -1 but X509_time_adj() now has to check the value 8703 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 8704 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 8705 [Steve Henson, reported by Kenneth R. Robinette 8706 <support@securenetterm.com>] 8707 8708 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 8709 result in a zero length in the ASN1_INTEGER structure which was 8710 not consistent with the structure when d2i_ASN1_INTEGER() was used 8711 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 8712 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 8713 where it did not print out a minus for negative ASN1_INTEGER. 8714 [Steve Henson] 8715 8716 *) Add summary printout to ocsp utility. The various functions which 8717 convert status values to strings have been renamed to: 8718 OCSP_response_status_str(), OCSP_cert_status_str() and 8719 OCSP_crl_reason_str() and are no longer static. New options 8720 to verify nonce values and to disable verification. OCSP response 8721 printout format cleaned up. 8722 [Steve Henson] 8723 8724 *) Add additional OCSP certificate checks. These are those specified 8725 in RFC2560. This consists of two separate checks: the CA of the 8726 certificate being checked must either be the OCSP signer certificate 8727 or the issuer of the OCSP signer certificate. In the latter case the 8728 OCSP signer certificate must contain the OCSP signing extended key 8729 usage. This check is performed by attempting to match the OCSP 8730 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 8731 in the OCSP_CERTID structures of the response. 8732 [Steve Henson] 8733 8734 *) Initial OCSP certificate verification added to OCSP_basic_verify() 8735 and related routines. This uses the standard OpenSSL certificate 8736 verify routines to perform initial checks (just CA validity) and 8737 to obtain the certificate chain. Then additional checks will be 8738 performed on the chain. Currently the root CA is checked to see 8739 if it is explicitly trusted for OCSP signing. This is used to set 8740 a root CA as a global signing root: that is any certificate that 8741 chains to that CA is an acceptable OCSP signing certificate. 8742 [Steve Henson] 8743 8744 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 8745 extensions from a separate configuration file. 8746 As when reading extensions from the main configuration file, 8747 the '-extensions ...' option may be used for specifying the 8748 section to use. 8749 [Massimiliano Pala <madwolf@comune.modena.it>] 8750 8751 *) New OCSP utility. Allows OCSP requests to be generated or 8752 read. The request can be sent to a responder and the output 8753 parsed, outputted or printed in text form. Not complete yet: 8754 still needs to check the OCSP response validity. 8755 [Steve Henson] 8756 8757 *) New subcommands for 'openssl ca': 8758 'openssl ca -status <serial>' prints the status of the cert with 8759 the given serial number (according to the index file). 8760 'openssl ca -updatedb' updates the expiry status of certificates 8761 in the index file. 8762 [Massimiliano Pala <madwolf@comune.modena.it>] 8763 8764 *) New '-newreq-nodes' command option to CA.pl. This is like 8765 '-newreq', but calls 'openssl req' with the '-nodes' option 8766 so that the resulting key is not encrypted. 8767 [Damien Miller <djm@mindrot.org>] 8768 8769 *) New configuration for the GNU Hurd. 8770 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 8771 8772 *) Initial code to implement OCSP basic response verify. This 8773 is currently incomplete. Currently just finds the signer's 8774 certificate and verifies the signature on the response. 8775 [Steve Henson] 8776 8777 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 8778 value of OPENSSLDIR. This is available via the new '-d' option 8779 to 'openssl version', and is also included in 'openssl version -a'. 8780 [Bodo Moeller] 8781 8782 *) Allowing defining memory allocation callbacks that will be given 8783 file name and line number information in additional arguments 8784 (a const char* and an int). The basic functionality remains, as 8785 well as the original possibility to just replace malloc(), 8786 realloc() and free() by functions that do not know about these 8787 additional arguments. To register and find out the current 8788 settings for extended allocation functions, the following 8789 functions are provided: 8790 8791 CRYPTO_set_mem_ex_functions 8792 CRYPTO_set_locked_mem_ex_functions 8793 CRYPTO_get_mem_ex_functions 8794 CRYPTO_get_locked_mem_ex_functions 8795 8796 These work the same way as CRYPTO_set_mem_functions and friends. 8797 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 8798 extended allocation function is enabled. 8799 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 8800 a conventional allocation function is enabled. 8801 [Richard Levitte, Bodo Moeller] 8802 8803 *) Finish off removing the remaining LHASH function pointer casts. 8804 There should no longer be any prototype-casting required when using 8805 the LHASH abstraction, and any casts that remain are "bugs". See 8806 the callback types and macros at the head of lhash.h for details 8807 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 8808 [Geoff Thorpe] 8809 8810 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 8811 If /dev/[u]random devices are not available or do not return enough 8812 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 8813 be queried. 8814 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 8815 /etc/entropy will be queried once each in this sequence, querying stops 8816 when enough entropy was collected without querying more sockets. 8817 [Lutz Jaenicke] 8818 8819 *) Change the Unix RAND_poll() variant to be able to poll several 8820 random devices, as specified by DEVRANDOM, until a sufficient amount 8821 of data has been collected. We spend at most 10 ms on each file 8822 (select timeout) and read in non-blocking mode. DEVRANDOM now 8823 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 8824 (previously it was just the string "/dev/urandom"), so on typical 8825 platforms the 10 ms delay will never occur. 8826 Also separate out the Unix variant to its own file, rand_unix.c. 8827 For VMS, there's a currently-empty rand_vms.c. 8828 [Richard Levitte] 8829 8830 *) Move OCSP client related routines to ocsp_cl.c. These 8831 provide utility functions which an application needing 8832 to issue a request to an OCSP responder and analyse the 8833 response will typically need: as opposed to those which an 8834 OCSP responder itself would need which will be added later. 8835 8836 OCSP_request_sign() signs an OCSP request with an API similar 8837 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 8838 response. OCSP_response_get1_basic() extracts basic response 8839 from response. OCSP_resp_find_status(): finds and extracts status 8840 information from an OCSP_CERTID structure (which will be created 8841 when the request structure is built). These are built from lower 8842 level functions which work on OCSP_SINGLERESP structures but 8843 won't normally be used unless the application wishes to examine 8844 extensions in the OCSP response for example. 8845 8846 Replace nonce routines with a pair of functions. 8847 OCSP_request_add1_nonce() adds a nonce value and optionally 8848 generates a random value. OCSP_check_nonce() checks the 8849 validity of the nonce in an OCSP response. 8850 [Steve Henson] 8851 8852 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 8853 This doesn't copy the supplied OCSP_CERTID and avoids the 8854 need to free up the newly created id. Change return type 8855 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 8856 This can then be used to add extensions to the request. 8857 Deleted OCSP_request_new(), since most of its functionality 8858 is now in OCSP_REQUEST_new() (and the case insensitive name 8859 clash) apart from the ability to set the request name which 8860 will be added elsewhere. 8861 [Steve Henson] 8862 8863 *) Update OCSP API. Remove obsolete extensions argument from 8864 various functions. Extensions are now handled using the new 8865 OCSP extension code. New simple OCSP HTTP function which 8866 can be used to send requests and parse the response. 8867 [Steve Henson] 8868 8869 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 8870 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 8871 uses the special reorder version of SET OF to sort the attributes 8872 and reorder them to match the encoded order. This resolves a long 8873 standing problem: a verify on a PKCS7 structure just after signing 8874 it used to fail because the attribute order did not match the 8875 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 8876 it uses the received order. This is necessary to tolerate some broken 8877 software that does not order SET OF. This is handled by encoding 8878 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 8879 to produce the required SET OF. 8880 [Steve Henson] 8881 8882 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 8883 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 8884 files to get correct declarations of the ASN.1 item variables. 8885 [Richard Levitte] 8886 8887 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 8888 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 8889 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 8890 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 8891 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 8892 ASN1_ITEM and no wrapper functions. 8893 [Steve Henson] 8894 8895 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 8896 replace the old function pointer based I/O routines. Change most of 8897 the *_d2i_bio() and *_d2i_fp() functions to use these. 8898 [Steve Henson] 8899 8900 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 8901 lines, recognize more "algorithms" that can be deselected, and make 8902 it complain about algorithm deselection that isn't recognised. 8903 [Richard Levitte] 8904 8905 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 8906 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 8907 to use new functions. Add NO_ASN1_OLD which can be set to remove 8908 some old style ASN1 functions: this can be used to determine if old 8909 code will still work when these eventually go away. 8910 [Steve Henson] 8911 8912 *) New extension functions for OCSP structures, these follow the 8913 same conventions as certificates and CRLs. 8914 [Steve Henson] 8915 8916 *) New function X509V3_add1_i2d(). This automatically encodes and 8917 adds an extension. Its behaviour can be customised with various 8918 flags to append, replace or delete. Various wrappers added for 8919 certificates and CRLs. 8920 [Steve Henson] 8921 8922 *) Fix to avoid calling the underlying ASN1 print routine when 8923 an extension cannot be parsed. Correct a typo in the 8924 OCSP_SERVICELOC extension. Tidy up print OCSP format. 8925 [Steve Henson] 8926 8927 *) Make mkdef.pl parse some of the ASN1 macros and add appropriate 8928 entries for variables. 8929 [Steve Henson] 8930 8931 *) Add functionality to apps/openssl.c for detecting locking 8932 problems: As the program is single-threaded, all we have 8933 to do is register a locking callback using an array for 8934 storing which locks are currently held by the program. 8935 [Bodo Moeller] 8936 8937 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 8938 SSL_get_ex_data_X509_STORE_idx(), which is used in 8939 ssl_verify_cert_chain() and thus can be called at any time 8940 during TLS/SSL handshakes so that thread-safety is essential. 8941 Unfortunately, the ex_data design is not at all suited 8942 for multi-threaded use, so it probably should be abolished. 8943 [Bodo Moeller] 8944 8945 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 8946 [Broadcom, tweaked and integrated by Geoff Thorpe] 8947 8948 *) Move common extension printing code to new function 8949 X509V3_print_extensions(). Reorganise OCSP print routines and 8950 implement some needed OCSP ASN1 functions. Add OCSP extensions. 8951 [Steve Henson] 8952 8953 *) New function X509_signature_print() to remove duplication in some 8954 print routines. 8955 [Steve Henson] 8956 8957 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 8958 set (this was treated exactly the same as SET OF previously). This 8959 is used to reorder the STACK representing the structure to match the 8960 encoding. This will be used to get round a problem where a PKCS7 8961 structure which was signed could not be verified because the STACK 8962 order did not reflect the encoded order. 8963 [Steve Henson] 8964 8965 *) Reimplement the OCSP ASN1 module using the new code. 8966 [Steve Henson] 8967 8968 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 8969 for its ASN1 operations. The old style function pointers still exist 8970 for now but they will eventually go away. 8971 [Steve Henson] 8972 8973 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 8974 completely replaces the old ASN1 functionality with a table driven 8975 encoder and decoder which interprets an ASN1_ITEM structure describing 8976 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 8977 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 8978 has also been converted to the new form. 8979 [Steve Henson] 8980 8981 *) Change BN_mod_exp_recp so that negative moduli are tolerated 8982 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 8983 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 8984 for negative moduli. 8985 [Bodo Moeller] 8986 8987 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 8988 of not touching the result's sign bit. 8989 [Bodo Moeller] 8990 8991 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 8992 set. 8993 [Bodo Moeller] 8994 8995 *) Changed the LHASH code to use prototypes for callbacks, and created 8996 macros to declare and implement thin (optionally static) functions 8997 that provide type-safety and avoid function pointer casting for the 8998 type-specific callbacks. 8999 [Geoff Thorpe] 9000 9001 *) Added Kerberos Cipher Suites to be used with TLS, as written in 9002 RFC 2712. 9003 [Veers Staats <staatsvr@asc.hpc.mil>, 9004 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 9005 9006 *) Reformat the FAQ so the different questions and answers can be divided 9007 in sections depending on the subject. 9008 [Richard Levitte] 9009 9010 *) Have the zlib compression code load ZLIB.DLL dynamically under 9011 Windows. 9012 [Richard Levitte] 9013 9014 *) New function BN_mod_sqrt for computing square roots modulo a prime 9015 (using the probabilistic Tonelli-Shanks algorithm unless 9016 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 9017 be handled deterministically). 9018 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 9019 9020 *) Make BN_mod_inverse faster by explicitly handling small quotients 9021 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 9022 512 bits], about 30% for larger ones [1024 or 2048 bits].) 9023 [Bodo Moeller] 9024 9025 *) New function BN_kronecker. 9026 [Bodo Moeller] 9027 9028 *) Fix BN_gcd so that it works on negative inputs; the result is 9029 positive unless both parameters are zero. 9030 Previously something reasonably close to an infinite loop was 9031 possible because numbers could be growing instead of shrinking 9032 in the implementation of Euclid's algorithm. 9033 [Bodo Moeller] 9034 9035 *) Fix BN_is_word() and BN_is_one() macros to take into account the 9036 sign of the number in question. 9037 9038 Fix BN_is_word(a,w) to work correctly for w == 0. 9039 9040 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 9041 because its test if the absolute value of 'a' equals 'w'. 9042 Note that BN_abs_is_word does *not* handle w == 0 reliably; 9043 it exists mostly for use in the implementations of BN_is_zero(), 9044 BN_is_one(), and BN_is_word(). 9045 [Bodo Moeller] 9046 9047 *) New function BN_swap. 9048 [Bodo Moeller] 9049 9050 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 9051 the exponentiation functions are more likely to produce reasonable 9052 results on negative inputs. 9053 [Bodo Moeller] 9054 9055 *) Change BN_mod_mul so that the result is always non-negative. 9056 Previously, it could be negative if one of the factors was negative; 9057 I don't think anyone really wanted that behaviour. 9058 [Bodo Moeller] 9059 9060 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 9061 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 9062 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 9063 and add new functions: 9064 9065 BN_nnmod 9066 BN_mod_sqr 9067 BN_mod_add 9068 BN_mod_add_quick 9069 BN_mod_sub 9070 BN_mod_sub_quick 9071 BN_mod_lshift1 9072 BN_mod_lshift1_quick 9073 BN_mod_lshift 9074 BN_mod_lshift_quick 9075 9076 These functions always generate non-negative results. 9077 9078 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 9079 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 9080 9081 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 9082 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 9083 be reduced modulo m. 9084 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 9085 9086#if 0 9087 The following entry accidentally appeared in the CHANGES file 9088 distributed with OpenSSL 0.9.7. The modifications described in 9089 it do *not* apply to OpenSSL 0.9.7. 9090 9091 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 9092 was actually never needed) and in BN_mul(). The removal in BN_mul() 9093 required a small change in bn_mul_part_recursive() and the addition 9094 of the functions bn_cmp_part_words(), bn_sub_part_words() and 9095 bn_add_part_words(), which do the same thing as bn_cmp_words(), 9096 bn_sub_words() and bn_add_words() except they take arrays with 9097 differing sizes. 9098 [Richard Levitte] 9099#endif 9100 9101 *) In 'openssl passwd', verify passwords read from the terminal 9102 unless the '-salt' option is used (which usually means that 9103 verification would just waste user's time since the resulting 9104 hash is going to be compared with some given password hash) 9105 or the new '-noverify' option is used. 9106 9107 This is an incompatible change, but it does not affect 9108 non-interactive use of 'openssl passwd' (passwords on the command 9109 line, '-stdin' option, '-in ...' option) and thus should not 9110 cause any problems. 9111 [Bodo Moeller] 9112 9113 *) Remove all references to RSAref, since there's no more need for it. 9114 [Richard Levitte] 9115 9116 *) Make DSO load along a path given through an environment variable 9117 (SHLIB_PATH) with shl_load(). 9118 [Richard Levitte] 9119 9120 *) Constify the ENGINE code as a result of BIGNUM constification. 9121 Also constify the RSA code and most things related to it. In a 9122 few places, most notable in the depth of the ASN.1 code, ugly 9123 casts back to non-const were required (to be solved at a later 9124 time) 9125 [Richard Levitte] 9126 9127 *) Make it so the openssl application has all engines loaded by default. 9128 [Richard Levitte] 9129 9130 *) Constify the BIGNUM routines a little more. 9131 [Richard Levitte] 9132 9133 *) Add the following functions: 9134 9135 ENGINE_load_cswift() 9136 ENGINE_load_chil() 9137 ENGINE_load_atalla() 9138 ENGINE_load_nuron() 9139 ENGINE_load_builtin_engines() 9140 9141 That way, an application can itself choose if external engines that 9142 are built-in in OpenSSL shall ever be used or not. The benefit is 9143 that applications won't have to be linked with libdl or other dso 9144 libraries unless it's really needed. 9145 9146 Changed 'openssl engine' to load all engines on demand. 9147 Changed the engine header files to avoid the duplication of some 9148 declarations (they differed!). 9149 [Richard Levitte] 9150 9151 *) 'openssl engine' can now list capabilities. 9152 [Richard Levitte] 9153 9154 *) Better error reporting in 'openssl engine'. 9155 [Richard Levitte] 9156 9157 *) Never call load_dh_param(NULL) in s_server. 9158 [Bodo Moeller] 9159 9160 *) Add engine application. It can currently list engines by name and 9161 identity, and test if they are actually available. 9162 [Richard Levitte] 9163 9164 *) Improve RPM specification file by forcing symbolic linking and making 9165 sure the installed documentation is also owned by root.root. 9166 [Damien Miller <djm@mindrot.org>] 9167 9168 *) Give the OpenSSL applications more possibilities to make use of 9169 keys (public as well as private) handled by engines. 9170 [Richard Levitte] 9171 9172 *) Add OCSP code that comes from CertCo. 9173 [Richard Levitte] 9174 9175 *) Add VMS support for the Rijndael code. 9176 [Richard Levitte] 9177 9178 *) Added untested support for Nuron crypto accelerator. 9179 [Ben Laurie] 9180 9181 *) Add support for external cryptographic devices. This code was 9182 previously distributed separately as the "engine" branch. 9183 [Geoff Thorpe, Richard Levitte] 9184 9185 *) Rework the filename-translation in the DSO code. It is now possible to 9186 have far greater control over how a "name" is turned into a filename 9187 depending on the operating environment and any oddities about the 9188 different shared library filenames on each system. 9189 [Geoff Thorpe] 9190 9191 *) Support threads on FreeBSD-elf in Configure. 9192 [Richard Levitte] 9193 9194 *) Fix for SHA1 assembly problem with MASM: it produces 9195 warnings about corrupt line number information when assembling 9196 with debugging information. This is caused by the overlapping 9197 of two sections. 9198 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 9199 9200 *) NCONF changes. 9201 NCONF_get_number() has no error checking at all. As a replacement, 9202 NCONF_get_number_e() is defined (_e for "error checking") and is 9203 promoted strongly. The old NCONF_get_number is kept around for 9204 binary backward compatibility. 9205 Make it possible for methods to load from something other than a BIO, 9206 by providing a function pointer that is given a name instead of a BIO. 9207 For example, this could be used to load configuration data from an 9208 LDAP server. 9209 [Richard Levitte] 9210 9211 *) Fix for non blocking accept BIOs. Added new I/O special reason 9212 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 9213 with non blocking I/O was not possible because no retry code was 9214 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 9215 this case. 9216 [Steve Henson] 9217 9218 *) Added the beginnings of Rijndael support. 9219 [Ben Laurie] 9220 9221 *) Fix for bug in DirectoryString mask setting. Add support for 9222 X509_NAME_print_ex() in 'req' and X509_print_ex() function 9223 to allow certificate printing to more controllable, additional 9224 'certopt' option to 'x509' to allow new printing options to be 9225 set. 9226 [Steve Henson] 9227 9228 *) Clean old EAY MD5 hack from e_os.h. 9229 [Richard Levitte] 9230 9231 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 9232 9233 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 9234 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 9235 [Joe Orton, Steve Henson] 9236 9237 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 9238 9239 *) Fix additional bug revealed by the NISCC test suite: 9240 9241 Stop bug triggering large recursion when presented with 9242 certain ASN.1 tags (CVE-2003-0851) 9243 [Steve Henson] 9244 9245 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 9246 9247 *) Fix various bugs revealed by running the NISCC test suite: 9248 9249 Stop out of bounds reads in the ASN1 code when presented with 9250 invalid tags (CVE-2003-0543 and CVE-2003-0544). 9251 9252 If verify callback ignores invalid public key errors don't try to check 9253 certificate signature with the NULL public key. 9254 9255 [Steve Henson] 9256 9257 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 9258 if the server requested one: as stated in TLS 1.0 and SSL 3.0 9259 specifications. 9260 [Steve Henson] 9261 9262 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 9263 extra data after the compression methods not only for TLS 1.0 9264 but also for SSL 3.0 (as required by the specification). 9265 [Bodo Moeller; problem pointed out by Matthias Loepfe] 9266 9267 *) Change X509_certificate_type() to mark the key as exported/exportable 9268 when it's 512 *bits* long, not 512 bytes. 9269 [Richard Levitte] 9270 9271 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 9272 9273 *) Countermeasure against the Klima-Pokorny-Rosa extension of 9274 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 9275 a protocol version number mismatch like a decryption error 9276 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 9277 [Bodo Moeller] 9278 9279 *) Turn on RSA blinding by default in the default implementation 9280 to avoid a timing attack. Applications that don't want it can call 9281 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 9282 They would be ill-advised to do so in most cases. 9283 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 9284 9285 *) Change RSA blinding code so that it works when the PRNG is not 9286 seeded (in this case, the secret RSA exponent is abused as 9287 an unpredictable seed -- if it is not unpredictable, there 9288 is no point in blinding anyway). Make RSA blinding thread-safe 9289 by remembering the creator's thread ID in rsa->blinding and 9290 having all other threads use local one-time blinding factors 9291 (this requires more computation than sharing rsa->blinding, but 9292 avoids excessive locking; and if an RSA object is not shared 9293 between threads, blinding will still be very fast). 9294 [Bodo Moeller] 9295 9296 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 9297 9298 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 9299 via timing by performing a MAC computation even if incorrect 9300 block cipher padding has been found. This is a countermeasure 9301 against active attacks where the attacker has to distinguish 9302 between bad padding and a MAC verification error. (CVE-2003-0078) 9303 9304 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 9305 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 9306 Martin Vuagnoux (EPFL, Ilion)] 9307 9308 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 9309 9310 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 9311 memory from its contents. This is done with a counter that will 9312 place alternating values in each byte. This can be used to solve 9313 two issues: 1) the removal of calls to memset() by highly optimizing 9314 compilers, and 2) cleansing with other values than 0, since those can 9315 be read through on certain media, for example a swap space on disk. 9316 [Geoff Thorpe] 9317 9318 *) Bugfix: client side session caching did not work with external caching, 9319 because the session->cipher setting was not restored when reloading 9320 from the external cache. This problem was masked, when 9321 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 9322 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 9323 [Lutz Jaenicke] 9324 9325 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 9326 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 9327 [Zeev Lieber <zeev-l@yahoo.com>] 9328 9329 *) Undo an undocumented change introduced in 0.9.6e which caused 9330 repeated calls to OpenSSL_add_all_ciphers() and 9331 OpenSSL_add_all_digests() to be ignored, even after calling 9332 EVP_cleanup(). 9333 [Richard Levitte] 9334 9335 *) Change the default configuration reader to deal with last line not 9336 being properly terminated. 9337 [Richard Levitte] 9338 9339 *) Change X509_NAME_cmp() so it applies the special rules on handling 9340 DN values that are of type PrintableString, as well as RDNs of type 9341 emailAddress where the value has the type ia5String. 9342 [stefank@valicert.com via Richard Levitte] 9343 9344 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 9345 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 9346 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 9347 the bitwise-OR of the two for use by the majority of applications 9348 wanting this behaviour, and update the docs. The documented 9349 behaviour and actual behaviour were inconsistent and had been 9350 changing anyway, so this is more a bug-fix than a behavioural 9351 change. 9352 [Geoff Thorpe, diagnosed by Nadav Har'El] 9353 9354 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 9355 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 9356 [Bodo Moeller] 9357 9358 *) Fix initialization code race conditions in 9359 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 9360 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 9361 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 9362 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 9363 ssl2_get_cipher_by_char(), 9364 ssl3_get_cipher_by_char(). 9365 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 9366 9367 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 9368 the cached sessions are flushed, as the remove_cb() might use ex_data 9369 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 9370 (see [openssl.org #212]). 9371 [Geoff Thorpe, Lutz Jaenicke] 9372 9373 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 9374 length, instead of the encoding length to d2i_ASN1_OBJECT. 9375 [Steve Henson] 9376 9377 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 9378 9379 *) [In 0.9.6g-engine release:] 9380 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 9381 [Lynn Gazis <lgazis@rainbow.com>] 9382 9383 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 9384 9385 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 9386 and get fix the header length calculation. 9387 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 9388 Alon Kantor <alonk@checkpoint.com> (and others), 9389 Steve Henson] 9390 9391 *) Use proper error handling instead of 'assertions' in buffer 9392 overflow checks added in 0.9.6e. This prevents DoS (the 9393 assertions could call abort()). 9394 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 9395 9396 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 9397 9398 *) Add various sanity checks to asn1_get_length() to reject 9399 the ASN1 length bytes if they exceed sizeof(long), will appear 9400 negative or the content length exceeds the length of the 9401 supplied buffer. 9402 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 9403 9404 *) Fix cipher selection routines: ciphers without encryption had no flags 9405 for the cipher strength set and where therefore not handled correctly 9406 by the selection routines (PR #130). 9407 [Lutz Jaenicke] 9408 9409 *) Fix EVP_dsa_sha macro. 9410 [Nils Larsch] 9411 9412 *) New option 9413 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 9414 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 9415 that was added in OpenSSL 0.9.6d. 9416 9417 As the countermeasure turned out to be incompatible with some 9418 broken SSL implementations, the new option is part of SSL_OP_ALL. 9419 SSL_OP_ALL is usually employed when compatibility with weird SSL 9420 implementations is desired (e.g. '-bugs' option to 's_client' and 9421 's_server'), so the new option is automatically set in many 9422 applications. 9423 [Bodo Moeller] 9424 9425 *) Changes in security patch: 9426 9427 Changes marked "(CHATS)" were sponsored by the Defense Advanced 9428 Research Projects Agency (DARPA) and Air Force Research Laboratory, 9429 Air Force Materiel Command, USAF, under agreement number 9430 F30602-01-2-0537. 9431 9432 *) Add various sanity checks to asn1_get_length() to reject 9433 the ASN1 length bytes if they exceed sizeof(long), will appear 9434 negative or the content length exceeds the length of the 9435 supplied buffer. (CVE-2002-0659) 9436 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 9437 9438 *) Assertions for various potential buffer overflows, not known to 9439 happen in practice. 9440 [Ben Laurie (CHATS)] 9441 9442 *) Various temporary buffers to hold ASCII versions of integers were 9443 too small for 64 bit platforms. (CVE-2002-0655) 9444 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 9445 9446 *) Remote buffer overflow in SSL3 protocol - an attacker could 9447 supply an oversized session ID to a client. (CVE-2002-0656) 9448 [Ben Laurie (CHATS)] 9449 9450 *) Remote buffer overflow in SSL2 protocol - an attacker could 9451 supply an oversized client master key. (CVE-2002-0656) 9452 [Ben Laurie (CHATS)] 9453 9454 Changes between 0.9.6c and 0.9.6d [9 May 2002] 9455 9456 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 9457 encoded as NULL) with id-dsa-with-sha1. 9458 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 9459 9460 *) Check various X509_...() return values in apps/req.c. 9461 [Nils Larsch <nla@trustcenter.de>] 9462 9463 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 9464 an end-of-file condition would erroneously be flagged, when the CRLF 9465 was just at the end of a processed block. The bug was discovered when 9466 processing data through a buffering memory BIO handing the data to a 9467 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 9468 <ptsekov@syntrex.com> and Nedelcho Stanev. 9469 [Lutz Jaenicke] 9470 9471 *) Implement a countermeasure against a vulnerability recently found 9472 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 9473 before application data chunks to avoid the use of known IVs 9474 with data potentially chosen by the attacker. 9475 [Bodo Moeller] 9476 9477 *) Fix length checks in ssl3_get_client_hello(). 9478 [Bodo Moeller] 9479 9480 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 9481 to prevent ssl3_read_internal() from incorrectly assuming that 9482 ssl3_read_bytes() found application data while handshake 9483 processing was enabled when in fact s->s3->in_read_app_data was 9484 merely automatically cleared during the initial handshake. 9485 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 9486 9487 *) Fix object definitions for Private and Enterprise: they were not 9488 recognized in their shortname (=lowercase) representation. Extend 9489 obj_dat.pl to issue an error when using undefined keywords instead 9490 of silently ignoring the problem (Svenning Sorensen 9491 <sss@sss.dnsalias.net>). 9492 [Lutz Jaenicke] 9493 9494 *) Fix DH_generate_parameters() so that it works for 'non-standard' 9495 generators, i.e. generators other than 2 and 5. (Previously, the 9496 code did not properly initialise the 'add' and 'rem' values to 9497 BN_generate_prime().) 9498 9499 In the new general case, we do not insist that 'generator' is 9500 actually a primitive root: This requirement is rather pointless; 9501 a generator of the order-q subgroup is just as good, if not 9502 better. 9503 [Bodo Moeller] 9504 9505 *) Map new X509 verification errors to alerts. Discovered and submitted by 9506 Tom Wu <tom@arcot.com>. 9507 [Lutz Jaenicke] 9508 9509 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 9510 returning non-zero before the data has been completely received 9511 when using non-blocking I/O. 9512 [Bodo Moeller; problem pointed out by John Hughes] 9513 9514 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 9515 [Ben Laurie, Lutz Jaenicke] 9516 9517 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 9518 Yoram Zahavi <YoramZ@gilian.com>). 9519 [Lutz Jaenicke] 9520 9521 *) Add information about CygWin 1.3 and on, and preserve proper 9522 configuration for the versions before that. 9523 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 9524 9525 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 9526 check whether we deal with a copy of a session and do not delete from 9527 the cache in this case. Problem reported by "Izhar Shoshani Levi" 9528 <izhar@checkpoint.com>. 9529 [Lutz Jaenicke] 9530 9531 *) Do not store session data into the internal session cache, if it 9532 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 9533 flag is set). Proposed by Aslam <aslam@funk.com>. 9534 [Lutz Jaenicke] 9535 9536 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 9537 value is 0. 9538 [Richard Levitte] 9539 9540 *) [In 0.9.6d-engine release:] 9541 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 9542 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 9543 9544 *) Add the configuration target linux-s390x. 9545 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 9546 9547 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 9548 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 9549 variable as an indication that a ClientHello message has been 9550 received. As the flag value will be lost between multiple 9551 invocations of ssl3_accept when using non-blocking I/O, the 9552 function may not be aware that a handshake has actually taken 9553 place, thus preventing a new session from being added to the 9554 session cache. 9555 9556 To avoid this problem, we now set s->new_session to 2 instead of 9557 using a local variable. 9558 [Lutz Jaenicke, Bodo Moeller] 9559 9560 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 9561 if the SSL_R_LENGTH_MISMATCH error is detected. 9562 [Geoff Thorpe, Bodo Moeller] 9563 9564 *) New 'shared_ldflag' column in Configure platform table. 9565 [Richard Levitte] 9566 9567 *) Fix EVP_CIPHER_mode macro. 9568 ["Dan S. Camper" <dan@bti.net>] 9569 9570 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 9571 type, we must throw them away by setting rr->length to 0. 9572 [D P Chang <dpc@qualys.com>] 9573 9574 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 9575 9576 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 9577 <Dominikus.Scherkl@biodata.com>. (The previous implementation 9578 worked incorrectly for those cases where range = 10..._2 and 9579 3*range is two bits longer than range.) 9580 [Bodo Moeller] 9581 9582 *) Only add signing time to PKCS7 structures if it is not already 9583 present. 9584 [Steve Henson] 9585 9586 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 9587 OBJ_ld_ce should be OBJ_id_ce. 9588 Also some ip-pda OIDs in crypto/objects/objects.txt were 9589 incorrect (cf. RFC 3039). 9590 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 9591 9592 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 9593 returns early because it has nothing to do. 9594 [Andy Schneider <andy.schneider@bjss.co.uk>] 9595 9596 *) [In 0.9.6c-engine release:] 9597 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 9598 [Andy Schneider <andy.schneider@bjss.co.uk>] 9599 9600 *) [In 0.9.6c-engine release:] 9601 Add support for Cryptographic Appliance's keyserver technology. 9602 (Use engine 'keyclient') 9603 [Cryptographic Appliances and Geoff Thorpe] 9604 9605 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 9606 is called via tools/c89.sh because arguments have to be 9607 rearranged (all '-L' options must appear before the first object 9608 modules). 9609 [Richard Shapiro <rshapiro@abinitio.com>] 9610 9611 *) [In 0.9.6c-engine release:] 9612 Add support for Broadcom crypto accelerator cards, backported 9613 from 0.9.7. 9614 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 9615 9616 *) [In 0.9.6c-engine release:] 9617 Add support for SureWare crypto accelerator cards from 9618 Baltimore Technologies. (Use engine 'sureware') 9619 [Baltimore Technologies and Mark Cox] 9620 9621 *) [In 0.9.6c-engine release:] 9622 Add support for crypto accelerator cards from Accelerated 9623 Encryption Processing, www.aep.ie. (Use engine 'aep') 9624 [AEP Inc. and Mark Cox] 9625 9626 *) Add a configuration entry for gcc on UnixWare. 9627 [Gary Benson <gbenson@redhat.com>] 9628 9629 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 9630 messages are stored in a single piece (fixed-length part and 9631 variable-length part combined) and fix various bugs found on the way. 9632 [Bodo Moeller] 9633 9634 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 9635 instead. BIO_gethostbyname() does not know what timeouts are 9636 appropriate, so entries would stay in cache even when they have 9637 become invalid. 9638 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 9639 9640 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 9641 faced with a pathologically small ClientHello fragment that does 9642 not contain client_version: Instead of aborting with an error, 9643 simply choose the highest available protocol version (i.e., 9644 TLS 1.0 unless it is disabled). In practice, ClientHello 9645 messages are never sent like this, but this change gives us 9646 strictly correct behaviour at least for TLS. 9647 [Bodo Moeller] 9648 9649 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 9650 never resets s->method to s->ctx->method when called from within 9651 one of the SSL handshake functions. 9652 [Bodo Moeller; problem pointed out by Niko Baric] 9653 9654 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 9655 (sent using the client's version number) if client_version is 9656 smaller than the protocol version in use. Also change 9657 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 9658 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 9659 the client will at least see that alert. 9660 [Bodo Moeller] 9661 9662 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 9663 correctly. 9664 [Bodo Moeller] 9665 9666 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 9667 client receives HelloRequest while in a handshake. 9668 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 9669 9670 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 9671 should end in 'break', not 'goto end' which circumvents various 9672 cleanups done in state SSL_ST_OK. But session related stuff 9673 must be disabled for SSL_ST_OK in the case that we just sent a 9674 HelloRequest. 9675 9676 Also avoid some overhead by not calling ssl_init_wbio_buffer() 9677 before just sending a HelloRequest. 9678 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 9679 9680 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 9681 reveal whether illegal block cipher padding was found or a MAC 9682 verification error occurred. (Neither SSLerr() codes nor alerts 9683 are directly visible to potential attackers, but the information 9684 may leak via logfiles.) 9685 9686 Similar changes are not required for the SSL 2.0 implementation 9687 because the number of padding bytes is sent in clear for SSL 2.0, 9688 and the extra bytes are just ignored. However ssl/s2_pkt.c 9689 failed to verify that the purported number of padding bytes is in 9690 the legal range. 9691 [Bodo Moeller] 9692 9693 *) Add OpenUNIX-8 support including shared libraries 9694 (Boyd Lynn Gerber <gerberb@zenez.com>). 9695 [Lutz Jaenicke] 9696 9697 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 9698 'wristwatch attack' using huge encoding parameters (cf. 9699 James H. Manger's CRYPTO 2001 paper). Note that the 9700 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 9701 encoding parameters and hence was not vulnerable. 9702 [Bodo Moeller] 9703 9704 *) BN_sqr() bug fix. 9705 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>] 9706 9707 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 9708 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 9709 followed by modular reduction. 9710 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 9711 9712 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 9713 equivalent based on BN_pseudo_rand() instead of BN_rand(). 9714 [Bodo Moeller] 9715 9716 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 9717 This function was broken, as the check for a new client hello message 9718 to handle SGC did not allow these large messages. 9719 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 9720 [Lutz Jaenicke] 9721 9722 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 9723 [Lutz Jaenicke] 9724 9725 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 9726 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 9727 [Lutz Jaenicke] 9728 9729 *) Rework the configuration and shared library support for Tru64 Unix. 9730 The configuration part makes use of modern compiler features and 9731 still retains old compiler behavior for those that run older versions 9732 of the OS. The shared library support part includes a variant that 9733 uses the RPATH feature, and is available through the special 9734 configuration target "alpha-cc-rpath", which will never be selected 9735 automatically. 9736 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 9737 9738 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 9739 with the same message size as in ssl3_get_certificate_request(). 9740 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 9741 messages might inadvertently be reject as too long. 9742 [Petr Lampa <lampa@fee.vutbr.cz>] 9743 9744 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 9745 [Andy Polyakov] 9746 9747 *) Modified SSL library such that the verify_callback that has been set 9748 specifically for an SSL object with SSL_set_verify() is actually being 9749 used. Before the change, a verify_callback set with this function was 9750 ignored and the verify_callback() set in the SSL_CTX at the time of 9751 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 9752 to allow the necessary settings. 9753 [Lutz Jaenicke] 9754 9755 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 9756 explicitly to NULL, as at least on Solaris 8 this seems not always to be 9757 done automatically (in contradiction to the requirements of the C 9758 standard). This made problems when used from OpenSSH. 9759 [Lutz Jaenicke] 9760 9761 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 9762 dh->length and always used 9763 9764 BN_rand_range(priv_key, dh->p). 9765 9766 BN_rand_range() is not necessary for Diffie-Hellman, and this 9767 specific range makes Diffie-Hellman unnecessarily inefficient if 9768 dh->length (recommended exponent length) is much smaller than the 9769 length of dh->p. We could use BN_rand_range() if the order of 9770 the subgroup was stored in the DH structure, but we only have 9771 dh->length. 9772 9773 So switch back to 9774 9775 BN_rand(priv_key, l, ...) 9776 9777 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 9778 otherwise. 9779 [Bodo Moeller] 9780 9781 *) In 9782 9783 RSA_eay_public_encrypt 9784 RSA_eay_private_decrypt 9785 RSA_eay_private_encrypt (signing) 9786 RSA_eay_public_decrypt (signature verification) 9787 9788 (default implementations for RSA_public_encrypt, 9789 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 9790 always reject numbers >= n. 9791 [Bodo Moeller] 9792 9793 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 9794 to synchronize access to 'locking_thread'. This is necessary on 9795 systems where access to 'locking_thread' (an 'unsigned long' 9796 variable) is not atomic. 9797 [Bodo Moeller] 9798 9799 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 9800 *before* setting the 'crypto_lock_rand' flag. The previous code had 9801 a race condition if 0 is a valid thread ID. 9802 [Travis Vitek <vitek@roguewave.com>] 9803 9804 *) Add support for shared libraries under Irix. 9805 [Albert Chin-A-Young <china@thewrittenword.com>] 9806 9807 *) Add configuration option to build on Linux on both big-endian and 9808 little-endian MIPS. 9809 [Ralf Baechle <ralf@uni-koblenz.de>] 9810 9811 *) Add the possibility to create shared libraries on HP-UX. 9812 [Richard Levitte] 9813 9814 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 9815 9816 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 9817 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 9818 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 9819 PRNG state recovery was possible based on the output of 9820 one PRNG request appropriately sized to gain knowledge on 9821 'md' followed by enough consecutive 1-byte PRNG requests 9822 to traverse all of 'state'. 9823 9824 1. When updating 'md_local' (the current thread's copy of 'md') 9825 during PRNG output generation, hash all of the previous 9826 'md_local' value, not just the half used for PRNG output. 9827 9828 2. Make the number of bytes from 'state' included into the hash 9829 independent from the number of PRNG bytes requested. 9830 9831 The first measure alone would be sufficient to avoid 9832 Markku-Juhani's attack. (Actually it had never occurred 9833 to me that the half of 'md_local' used for chaining was the 9834 half from which PRNG output bytes were taken -- I had always 9835 assumed that the secret half would be used.) The second 9836 measure makes sure that additional data from 'state' is never 9837 mixed into 'md_local' in small portions; this heuristically 9838 further strengthens the PRNG. 9839 [Bodo Moeller] 9840 9841 *) Fix crypto/bn/asm/mips3.s. 9842 [Andy Polyakov] 9843 9844 *) When only the key is given to "enc", the IV is undefined. Print out 9845 an error message in this case. 9846 [Lutz Jaenicke] 9847 9848 *) Handle special case when X509_NAME is empty in X509 printing routines. 9849 [Steve Henson] 9850 9851 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 9852 positive and less than q. 9853 [Bodo Moeller] 9854 9855 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 9856 used: it isn't thread safe and the add_lock_callback should handle 9857 that itself. 9858 [Paul Rose <Paul.Rose@bridge.com>] 9859 9860 *) Verify that incoming data obeys the block size in 9861 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 9862 [Bodo Moeller] 9863 9864 *) Fix OAEP check. 9865 [Ulf Möller, Bodo Möller] 9866 9867 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 9868 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 9869 when fixing the server behaviour for backwards-compatible 'client 9870 hello' messages. (Note that the attack is impractical against 9871 SSL 3.0 and TLS 1.0 anyway because length and version checking 9872 means that the probability of guessing a valid ciphertext is 9873 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 9874 paper.) 9875 9876 Before 0.9.5, the countermeasure (hide the error by generating a 9877 random 'decryption result') did not work properly because 9878 ERR_clear_error() was missing, meaning that SSL_get_error() would 9879 detect the supposedly ignored error. 9880 9881 Both problems are now fixed. 9882 [Bodo Moeller] 9883 9884 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 9885 (previously it was 1024). 9886 [Bodo Moeller] 9887 9888 *) Fix for compatibility mode trust settings: ignore trust settings 9889 unless some valid trust or reject settings are present. 9890 [Steve Henson] 9891 9892 *) Fix for blowfish EVP: its a variable length cipher. 9893 [Steve Henson] 9894 9895 *) Fix various bugs related to DSA S/MIME verification. Handle missing 9896 parameters in DSA public key structures and return an error in the 9897 DSA routines if parameters are absent. 9898 [Steve Henson] 9899 9900 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 9901 in the current directory if neither $RANDFILE nor $HOME was set. 9902 RAND_file_name() in 0.9.6a returned NULL in this case. This has 9903 caused some confusion to Windows users who haven't defined $HOME. 9904 Thus RAND_file_name() is changed again: e_os.h can define a 9905 DEFAULT_HOME, which will be used if $HOME is not set. 9906 For Windows, we use "C:"; on other platforms, we still require 9907 environment variables. 9908 9909 *) Move 'if (!initialized) RAND_poll()' into regions protected by 9910 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 9911 having multiple threads call RAND_poll() concurrently. 9912 [Bodo Moeller] 9913 9914 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 9915 combination of a flag and a thread ID variable. 9916 Otherwise while one thread is in ssleay_rand_bytes (which sets the 9917 flag), *other* threads can enter ssleay_add_bytes without obeying 9918 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 9919 that they do not hold after the first thread unsets add_do_not_lock). 9920 [Bodo Moeller] 9921 9922 *) Change bctest again: '-x' expressions are not available in all 9923 versions of 'test'. 9924 [Bodo Moeller] 9925 9926 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 9927 9928 *) Fix a couple of memory leaks in PKCS7_dataDecode() 9929 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 9930 9931 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 9932 the default extension for executables, if any. Also, make the perl 9933 scripts that use symlink() to test if it really exists and use "cp" 9934 if it doesn't. All this made OpenSSL compilable and installable in 9935 CygWin. 9936 [Richard Levitte] 9937 9938 *) Fix for asn1_GetSequence() for indefinite length constructed data. 9939 If SEQUENCE is length is indefinite just set c->slen to the total 9940 amount of data available. 9941 [Steve Henson, reported by shige@FreeBSD.org] 9942 [This change does not apply to 0.9.7.] 9943 9944 *) Change bctest to avoid here-documents inside command substitution 9945 (workaround for FreeBSD /bin/sh bug). 9946 For compatibility with Ultrix, avoid shell functions (introduced 9947 in the bctest version that searches along $PATH). 9948 [Bodo Moeller] 9949 9950 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 9951 with des_encrypt() defined on some operating systems, like Solaris 9952 and UnixWare. 9953 [Richard Levitte] 9954 9955 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 9956 On the Importance of Eliminating Errors in Cryptographic 9957 Computations, J. Cryptology 14 (2001) 2, 101-119, 9958 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 9959 [Ulf Moeller] 9960 9961 *) MIPS assembler BIGNUM division bug fix. 9962 [Andy Polyakov] 9963 9964 *) Disabled incorrect Alpha assembler code. 9965 [Richard Levitte] 9966 9967 *) Fix PKCS#7 decode routines so they correctly update the length 9968 after reading an EOC for the EXPLICIT tag. 9969 [Steve Henson] 9970 [This change does not apply to 0.9.7.] 9971 9972 *) Fix bug in PKCS#12 key generation routines. This was triggered 9973 if a 3DES key was generated with a 0 initial byte. Include 9974 PKCS12_BROKEN_KEYGEN compilation option to retain the old 9975 (but broken) behaviour. 9976 [Steve Henson] 9977 9978 *) Enhance bctest to search for a working bc along $PATH and print 9979 it when found. 9980 [Tim Rice <tim@multitalents.net> via Richard Levitte] 9981 9982 *) Fix memory leaks in err.c: free err_data string if necessary; 9983 don't write to the wrong index in ERR_set_error_data. 9984 [Bodo Moeller] 9985 9986 *) Implement ssl23_peek (analogous to ssl23_read), which previously 9987 did not exist. 9988 [Bodo Moeller] 9989 9990 *) Replace rdtsc with _emit statements for VC++ version 5. 9991 [Jeremy Cooper <jeremy@baymoo.org>] 9992 9993 *) Make it possible to reuse SSLv2 sessions. 9994 [Richard Levitte] 9995 9996 *) In copy_email() check for >= 0 as a return value for 9997 X509_NAME_get_index_by_NID() since 0 is a valid index. 9998 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 9999 10000 *) Avoid coredump with unsupported or invalid public keys by checking if 10001 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 10002 PKCS7_verify() fails with non detached data. 10003 [Steve Henson] 10004 10005 *) Don't use getenv in library functions when run as setuid/setgid. 10006 New function OPENSSL_issetugid(). 10007 [Ulf Moeller] 10008 10009 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 10010 due to incorrect handling of multi-threading: 10011 10012 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 10013 10014 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 10015 10016 3. Count how many times MemCheck_off() has been called so that 10017 nested use can be treated correctly. This also avoids 10018 inband-signalling in the previous code (which relied on the 10019 assumption that thread ID 0 is impossible). 10020 [Bodo Moeller] 10021 10022 *) Add "-rand" option also to s_client and s_server. 10023 [Lutz Jaenicke] 10024 10025 *) Fix CPU detection on Irix 6.x. 10026 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 10027 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 10028 10029 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 10030 was empty. 10031 [Steve Henson] 10032 [This change does not apply to 0.9.7.] 10033 10034 *) Use the cached encoding of an X509_NAME structure rather than 10035 copying it. This is apparently the reason for the libsafe "errors" 10036 but the code is actually correct. 10037 [Steve Henson] 10038 10039 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 10040 Bleichenbacher's DSA attack. 10041 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 10042 to be set and top=0 forces the highest bit to be set; top=-1 is new 10043 and leaves the highest bit random. 10044 [Ulf Moeller, Bodo Moeller] 10045 10046 *) In the NCONF_...-based implementations for CONF_... queries 10047 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 10048 a temporary CONF structure with the data component set to NULL 10049 (which gives segmentation faults in lh_retrieve). 10050 Instead, use NULL for the CONF pointer in CONF_get_string and 10051 CONF_get_number (which may use environment variables) and directly 10052 return NULL from CONF_get_section. 10053 [Bodo Moeller] 10054 10055 *) Fix potential buffer overrun for EBCDIC. 10056 [Ulf Moeller] 10057 10058 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 10059 keyUsage if basicConstraints absent for a CA. 10060 [Steve Henson] 10061 10062 *) Make SMIME_write_PKCS7() write mail header values with a format that 10063 is more generally accepted (no spaces before the semicolon), since 10064 some programs can't parse those values properly otherwise. Also make 10065 sure BIO's that break lines after each write do not create invalid 10066 headers. 10067 [Richard Levitte] 10068 10069 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 10070 macros previously used would not encode an empty SEQUENCE OF 10071 and break the signature. 10072 [Steve Henson] 10073 [This change does not apply to 0.9.7.] 10074 10075 *) Zero the premaster secret after deriving the master secret in 10076 DH ciphersuites. 10077 [Steve Henson] 10078 10079 *) Add some EVP_add_digest_alias registrations (as found in 10080 OpenSSL_add_all_digests()) to SSL_library_init() 10081 aka OpenSSL_add_ssl_algorithms(). This provides improved 10082 compatibility with peers using X.509 certificates 10083 with unconventional AlgorithmIdentifier OIDs. 10084 [Bodo Moeller] 10085 10086 *) Fix for Irix with NO_ASM. 10087 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 10088 10089 *) ./config script fixes. 10090 [Ulf Moeller, Richard Levitte] 10091 10092 *) Fix 'openssl passwd -1'. 10093 [Bodo Moeller] 10094 10095 *) Change PKCS12_key_gen_asc() so it can cope with non null 10096 terminated strings whose length is passed in the passlen 10097 parameter, for example from PEM callbacks. This was done 10098 by adding an extra length parameter to asc2uni(). 10099 [Steve Henson, reported by <oddissey@samsung.co.kr>] 10100 10101 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 10102 call failed, free the DSA structure. 10103 [Bodo Moeller] 10104 10105 *) Fix to uni2asc() to cope with zero length Unicode strings. 10106 These are present in some PKCS#12 files. 10107 [Steve Henson] 10108 10109 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 10110 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 10111 when writing a 32767 byte record. 10112 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 10113 10114 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 10115 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 10116 10117 (RSA objects have a reference count access to which is protected 10118 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 10119 so they are meant to be shared between threads.) 10120 [Bodo Moeller, Geoff Thorpe; original patch submitted by 10121 "Reddie, Steven" <Steven.Reddie@ca.com>] 10122 10123 *) Fix a deadlock in CRYPTO_mem_leaks(). 10124 [Bodo Moeller] 10125 10126 *) Use better test patterns in bntest. 10127 [Ulf Möller] 10128 10129 *) rand_win.c fix for Borland C. 10130 [Ulf Möller] 10131 10132 *) BN_rshift bugfix for n == 0. 10133 [Bodo Moeller] 10134 10135 *) Add a 'bctest' script that checks for some known 'bc' bugs 10136 so that 'make test' does not abort just because 'bc' is broken. 10137 [Bodo Moeller] 10138 10139 *) Store verify_result within SSL_SESSION also for client side to 10140 avoid potential security hole. (Re-used sessions on the client side 10141 always resulted in verify_result==X509_V_OK, not using the original 10142 result of the server certificate verification.) 10143 [Lutz Jaenicke] 10144 10145 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 10146 SSL3_RT_APPLICATION_DATA, return 0. 10147 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 10148 [Bodo Moeller] 10149 10150 *) Fix SSL_peek: 10151 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 10152 releases, have been re-implemented by renaming the previous 10153 implementations of ssl2_read and ssl3_read to ssl2_read_internal 10154 and ssl3_read_internal, respectively, and adding 'peek' parameters 10155 to them. The new ssl[23]_{read,peek} functions are calls to 10156 ssl[23]_read_internal with the 'peek' flag set appropriately. 10157 A 'peek' parameter has also been added to ssl3_read_bytes, which 10158 does the actual work for ssl3_read_internal. 10159 [Bodo Moeller] 10160 10161 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 10162 the method-specific "init()" handler. Also clean up ex_data after 10163 calling the method-specific "finish()" handler. Previously, this was 10164 happening the other way round. 10165 [Geoff Thorpe] 10166 10167 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 10168 The previous value, 12, was not always sufficient for BN_mod_exp(). 10169 [Bodo Moeller] 10170 10171 *) Make sure that shared libraries get the internal name engine with 10172 the full version number and not just 0. This should mark the 10173 shared libraries as not backward compatible. Of course, this should 10174 be changed again when we can guarantee backward binary compatibility. 10175 [Richard Levitte] 10176 10177 *) Fix typo in get_cert_by_subject() in by_dir.c 10178 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 10179 10180 *) Rework the system to generate shared libraries: 10181 10182 - Make note of the expected extension for the shared libraries and 10183 if there is a need for symbolic links from for example libcrypto.so.0 10184 to libcrypto.so.0.9.7. There is extended info in Configure for 10185 that. 10186 10187 - Make as few rebuilds of the shared libraries as possible. 10188 10189 - Still avoid linking the OpenSSL programs with the shared libraries. 10190 10191 - When installing, install the shared libraries separately from the 10192 static ones. 10193 [Richard Levitte] 10194 10195 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 10196 10197 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 10198 and not in SSL_clear because the latter is also used by the 10199 accept/connect functions; previously, the settings made by 10200 SSL_set_read_ahead would be lost during the handshake. 10201 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 10202 10203 *) Correct util/mkdef.pl to be selective about disabled algorithms. 10204 Previously, it would create entries for disabled algorithms no 10205 matter what. 10206 [Richard Levitte] 10207 10208 *) Added several new manual pages for SSL_* function. 10209 [Lutz Jaenicke] 10210 10211 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 10212 10213 *) In ssl23_get_client_hello, generate an error message when faced 10214 with an initial SSL 3.0/TLS record that is too small to contain the 10215 first two bytes of the ClientHello message, i.e. client_version. 10216 (Note that this is a pathologic case that probably has never happened 10217 in real life.) The previous approach was to use the version number 10218 from the record header as a substitute; but our protocol choice 10219 should not depend on that one because it is not authenticated 10220 by the Finished messages. 10221 [Bodo Moeller] 10222 10223 *) More robust randomness gathering functions for Windows. 10224 [Jeffrey Altman <jaltman@columbia.edu>] 10225 10226 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 10227 not set then we don't setup the error code for issuer check errors 10228 to avoid possibly overwriting other errors which the callback does 10229 handle. If an application does set the flag then we assume it knows 10230 what it is doing and can handle the new informational codes 10231 appropriately. 10232 [Steve Henson] 10233 10234 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 10235 a general "ANY" type, as such it should be able to decode anything 10236 including tagged types. However it didn't check the class so it would 10237 wrongly interpret tagged types in the same way as their universal 10238 counterpart and unknown types were just rejected. Changed so that the 10239 tagged and unknown types are handled in the same way as a SEQUENCE: 10240 that is the encoding is stored intact. There is also a new type 10241 "V_ASN1_OTHER" which is used when the class is not universal, in this 10242 case we have no idea what the actual type is so we just lump them all 10243 together. 10244 [Steve Henson] 10245 10246 *) On VMS, stdout may very well lead to a file that is written to 10247 in a record-oriented fashion. That means that every write() will 10248 write a separate record, which will be read separately by the 10249 programs trying to read from it. This can be very confusing. 10250 10251 The solution is to put a BIO filter in the way that will buffer 10252 text until a linefeed is reached, and then write everything a 10253 line at a time, so every record written will be an actual line, 10254 not chunks of lines and not (usually doesn't happen, but I've 10255 seen it once) several lines in one record. BIO_f_linebuffer() is 10256 the answer. 10257 10258 Currently, it's a VMS-only method, because that's where it has 10259 been tested well enough. 10260 [Richard Levitte] 10261 10262 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 10263 it can return incorrect results. 10264 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 10265 but it was in 0.9.6-beta[12].) 10266 [Bodo Moeller] 10267 10268 *) Disable the check for content being present when verifying detached 10269 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 10270 include zero length content when signing messages. 10271 [Steve Henson] 10272 10273 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 10274 BIO_ctrl (for BIO pairs). 10275 [Bodo Möller] 10276 10277 *) Add DSO method for VMS. 10278 [Richard Levitte] 10279 10280 *) Bug fix: Montgomery multiplication could produce results with the 10281 wrong sign. 10282 [Ulf Möller] 10283 10284 *) Add RPM specification openssl.spec and modify it to build three 10285 packages. The default package contains applications, application 10286 documentation and run-time libraries. The devel package contains 10287 include files, static libraries and function documentation. The 10288 doc package contains the contents of the doc directory. The original 10289 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 10290 [Richard Levitte] 10291 10292 *) Add a large number of documentation files for many SSL routines. 10293 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 10294 10295 *) Add a configuration entry for Sony News 4. 10296 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 10297 10298 *) Don't set the two most significant bits to one when generating a 10299 random number < q in the DSA library. 10300 [Ulf Möller] 10301 10302 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 10303 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 10304 the underlying transport is blocking) if a handshake took place. 10305 (The default behaviour is needed by applications such as s_client 10306 and s_server that use select() to determine when to use SSL_read; 10307 but for applications that know in advance when to expect data, it 10308 just makes things more complicated.) 10309 [Bodo Moeller] 10310 10311 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 10312 from EGD. 10313 [Ben Laurie] 10314 10315 *) Add a few more EBCDIC conditionals that make `req' and `x509' 10316 work better on such systems. 10317 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10318 10319 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 10320 Update PKCS12_parse() so it copies the friendlyName and the 10321 keyid to the certificates aux info. 10322 [Steve Henson] 10323 10324 *) Fix bug in PKCS7_verify() which caused an infinite loop 10325 if there was more than one signature. 10326 [Sven Uszpelkat <su@celocom.de>] 10327 10328 *) Major change in util/mkdef.pl to include extra information 10329 about each symbol, as well as presenting variables as well 10330 as functions. This change means that there's n more need 10331 to rebuild the .num files when some algorithms are excluded. 10332 [Richard Levitte] 10333 10334 *) Allow the verify time to be set by an application, 10335 rather than always using the current time. 10336 [Steve Henson] 10337 10338 *) Phase 2 verify code reorganisation. The certificate 10339 verify code now looks up an issuer certificate by a 10340 number of criteria: subject name, authority key id 10341 and key usage. It also verifies self signed certificates 10342 by the same criteria. The main comparison function is 10343 X509_check_issued() which performs these checks. 10344 10345 Lot of changes were necessary in order to support this 10346 without completely rewriting the lookup code. 10347 10348 Authority and subject key identifier are now cached. 10349 10350 The LHASH 'certs' is X509_STORE has now been replaced 10351 by a STACK_OF(X509_OBJECT). This is mainly because an 10352 LHASH can't store or retrieve multiple objects with 10353 the same hash value. 10354 10355 As a result various functions (which were all internal 10356 use only) have changed to handle the new X509_STORE 10357 structure. This will break anything that messed round 10358 with X509_STORE internally. 10359 10360 The functions X509_STORE_add_cert() now checks for an 10361 exact match, rather than just subject name. 10362 10363 The X509_STORE API doesn't directly support the retrieval 10364 of multiple certificates matching a given criteria, however 10365 this can be worked round by performing a lookup first 10366 (which will fill the cache with candidate certificates) 10367 and then examining the cache for matches. This is probably 10368 the best we can do without throwing out X509_LOOKUP 10369 entirely (maybe later...). 10370 10371 The X509_VERIFY_CTX structure has been enhanced considerably. 10372 10373 All certificate lookup operations now go via a get_issuer() 10374 callback. Although this currently uses an X509_STORE it 10375 can be replaced by custom lookups. This is a simple way 10376 to bypass the X509_STORE hackery necessary to make this 10377 work and makes it possible to use more efficient techniques 10378 in future. A very simple version which uses a simple 10379 STACK for its trusted certificate store is also provided 10380 using X509_STORE_CTX_trusted_stack(). 10381 10382 The verify_cb() and verify() callbacks now have equivalents 10383 in the X509_STORE_CTX structure. 10384 10385 X509_STORE_CTX also has a 'flags' field which can be used 10386 to customise the verify behaviour. 10387 [Steve Henson] 10388 10389 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 10390 excludes S/MIME capabilities. 10391 [Steve Henson] 10392 10393 *) When a certificate request is read in keep a copy of the 10394 original encoding of the signed data and use it when outputting 10395 again. Signatures then use the original encoding rather than 10396 a decoded, encoded version which may cause problems if the 10397 request is improperly encoded. 10398 [Steve Henson] 10399 10400 *) For consistency with other BIO_puts implementations, call 10401 buffer_write(b, ...) directly in buffer_puts instead of calling 10402 BIO_write(b, ...). 10403 10404 In BIO_puts, increment b->num_write as in BIO_write. 10405 [Peter.Sylvester@EdelWeb.fr] 10406 10407 *) Fix BN_mul_word for the case where the word is 0. (We have to use 10408 BN_zero, we may not return a BIGNUM with an array consisting of 10409 words set to zero.) 10410 [Bodo Moeller] 10411 10412 *) Avoid calling abort() from within the library when problems are 10413 detected, except if preprocessor symbols have been defined 10414 (such as REF_CHECK, BN_DEBUG etc.). 10415 [Bodo Moeller] 10416 10417 *) New openssl application 'rsautl'. This utility can be 10418 used for low level RSA operations. DER public key 10419 BIO/fp routines also added. 10420 [Steve Henson] 10421 10422 *) New Configure entry and patches for compiling on QNX 4. 10423 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 10424 10425 *) A demo state-machine implementation was sponsored by 10426 Nuron (http://www.nuron.com/) and is now available in 10427 demos/state_machine. 10428 [Ben Laurie] 10429 10430 *) New options added to the 'dgst' utility for signature 10431 generation and verification. 10432 [Steve Henson] 10433 10434 *) Unrecognized PKCS#7 content types are now handled via a 10435 catch all ASN1_TYPE structure. This allows unsupported 10436 types to be stored as a "blob" and an application can 10437 encode and decode it manually. 10438 [Steve Henson] 10439 10440 *) Fix various signed/unsigned issues to make a_strex.c 10441 compile under VC++. 10442 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 10443 10444 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 10445 length if passed a buffer. ASN1_INTEGER_to_BN failed 10446 if passed a NULL BN and its argument was negative. 10447 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 10448 10449 *) Modification to PKCS#7 encoding routines to output definite 10450 length encoding. Since currently the whole structures are in 10451 memory there's not real point in using indefinite length 10452 constructed encoding. However if OpenSSL is compiled with 10453 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 10454 [Steve Henson] 10455 10456 *) Added BIO_vprintf() and BIO_vsnprintf(). 10457 [Richard Levitte] 10458 10459 *) Added more prefixes to parse for in the strings written 10460 through a logging bio, to cover all the levels that are available 10461 through syslog. The prefixes are now: 10462 10463 PANIC, EMERG, EMR => LOG_EMERG 10464 ALERT, ALR => LOG_ALERT 10465 CRIT, CRI => LOG_CRIT 10466 ERROR, ERR => LOG_ERR 10467 WARNING, WARN, WAR => LOG_WARNING 10468 NOTICE, NOTE, NOT => LOG_NOTICE 10469 INFO, INF => LOG_INFO 10470 DEBUG, DBG => LOG_DEBUG 10471 10472 and as before, if none of those prefixes are present at the 10473 beginning of the string, LOG_ERR is chosen. 10474 10475 On Win32, the LOG_* levels are mapped according to this: 10476 10477 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 10478 LOG_WARNING => EVENTLOG_WARNING_TYPE 10479 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 10480 10481 [Richard Levitte] 10482 10483 *) Made it possible to reconfigure with just the configuration 10484 argument "reconf" or "reconfigure". The command line arguments 10485 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 10486 and are retrieved from there when reconfiguring. 10487 [Richard Levitte] 10488 10489 *) MD4 implemented. 10490 [Assar Westerlund <assar@sics.se>, Richard Levitte] 10491 10492 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 10493 [Richard Levitte] 10494 10495 *) The obj_dat.pl script was messing up the sorting of object 10496 names. The reason was that it compared the quoted version 10497 of strings as a result "OCSP" > "OCSP Signing" because 10498 " > SPACE. Changed script to store unquoted versions of 10499 names and add quotes on output. It was also omitting some 10500 names from the lookup table if they were given a default 10501 value (that is if SN is missing it is given the same 10502 value as LN and vice versa), these are now added on the 10503 grounds that if an object has a name we should be able to 10504 look it up. Finally added warning output when duplicate 10505 short or long names are found. 10506 [Steve Henson] 10507 10508 *) Changes needed for Tandem NSK. 10509 [Scott Uroff <scott@xypro.com>] 10510 10511 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 10512 RSA_padding_check_SSLv23(), special padding was never detected 10513 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 10514 version rollback attacks was not effective. 10515 10516 In s23_clnt.c, don't use special rollback-attack detection padding 10517 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 10518 client; similarly, in s23_srvr.c, don't do the rollback check if 10519 SSL 2.0 is the only protocol enabled in the server. 10520 [Bodo Moeller] 10521 10522 *) Make it possible to get hexdumps of unprintable data with 'openssl 10523 asn1parse'. By implication, the functions ASN1_parse_dump() and 10524 BIO_dump_indent() are added. 10525 [Richard Levitte] 10526 10527 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 10528 these print out strings and name structures based on various 10529 flags including RFC2253 support and proper handling of 10530 multibyte characters. Added options to the 'x509' utility 10531 to allow the various flags to be set. 10532 [Steve Henson] 10533 10534 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 10535 Also change the functions X509_cmp_current_time() and 10536 X509_gmtime_adj() work with an ASN1_TIME structure, 10537 this will enable certificates using GeneralizedTime in validity 10538 dates to be checked. 10539 [Steve Henson] 10540 10541 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 10542 negative public key encodings) on by default, 10543 NO_NEG_PUBKEY_BUG can be set to disable it. 10544 [Steve Henson] 10545 10546 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 10547 content octets. An i2c_ASN1_OBJECT is unnecessary because 10548 the encoding can be trivially obtained from the structure. 10549 [Steve Henson] 10550 10551 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 10552 not read locks (CRYPTO_r_[un]lock). 10553 [Bodo Moeller] 10554 10555 *) A first attempt at creating official support for shared 10556 libraries through configuration. I've kept it so the 10557 default is static libraries only, and the OpenSSL programs 10558 are always statically linked for now, but there are 10559 preparations for dynamic linking in place. 10560 This has been tested on Linux and Tru64. 10561 [Richard Levitte] 10562 10563 *) Randomness polling function for Win9x, as described in: 10564 Peter Gutmann, Software Generation of Practically Strong 10565 Random Numbers. 10566 [Ulf Möller] 10567 10568 *) Fix so PRNG is seeded in req if using an already existing 10569 DSA key. 10570 [Steve Henson] 10571 10572 *) New options to smime application. -inform and -outform 10573 allow alternative formats for the S/MIME message including 10574 PEM and DER. The -content option allows the content to be 10575 specified separately. This should allow things like Netscape 10576 form signing output easier to verify. 10577 [Steve Henson] 10578 10579 *) Fix the ASN1 encoding of tags using the 'long form'. 10580 [Steve Henson] 10581 10582 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 10583 STRING types. These convert content octets to and from the 10584 underlying type. The actual tag and length octets are 10585 already assumed to have been read in and checked. These 10586 are needed because all other string types have virtually 10587 identical handling apart from the tag. By having versions 10588 of the ASN1 functions that just operate on content octets 10589 IMPLICIT tagging can be handled properly. It also allows 10590 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 10591 and ASN1_INTEGER are identical apart from the tag. 10592 [Steve Henson] 10593 10594 *) Change the handling of OID objects as follows: 10595 10596 - New object identifiers are inserted in objects.txt, following 10597 the syntax given in objects.README. 10598 - objects.pl is used to process obj_mac.num and create a new 10599 obj_mac.h. 10600 - obj_dat.pl is used to create a new obj_dat.h, using the data in 10601 obj_mac.h. 10602 10603 This is currently kind of a hack, and the perl code in objects.pl 10604 isn't very elegant, but it works as I intended. The simplest way 10605 to check that it worked correctly is to look in obj_dat.h and 10606 check the array nid_objs and make sure the objects haven't moved 10607 around (this is important!). Additions are OK, as well as 10608 consistent name changes. 10609 [Richard Levitte] 10610 10611 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 10612 [Bodo Moeller] 10613 10614 *) Addition of the command line parameter '-rand file' to 'openssl req'. 10615 The given file adds to whatever has already been seeded into the 10616 random pool through the RANDFILE configuration file option or 10617 environment variable, or the default random state file. 10618 [Richard Levitte] 10619 10620 *) mkstack.pl now sorts each macro group into lexical order. 10621 Previously the output order depended on the order the files 10622 appeared in the directory, resulting in needless rewriting 10623 of safestack.h . 10624 [Steve Henson] 10625 10626 *) Patches to make OpenSSL compile under Win32 again. Mostly 10627 work arounds for the VC++ problem that it treats func() as 10628 func(void). Also stripped out the parts of mkdef.pl that 10629 added extra typesafe functions: these no longer exist. 10630 [Steve Henson] 10631 10632 *) Reorganisation of the stack code. The macros are now all 10633 collected in safestack.h . Each macro is defined in terms of 10634 a "stack macro" of the form SKM_<name>(type, a, b). The 10635 DEBUG_SAFESTACK is now handled in terms of function casts, 10636 this has the advantage of retaining type safety without the 10637 use of additional functions. If DEBUG_SAFESTACK is not defined 10638 then the non typesafe macros are used instead. Also modified the 10639 mkstack.pl script to handle the new form. Needs testing to see 10640 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 10641 the default if no major problems. Similar behaviour for ASN1_SET_OF 10642 and PKCS12_STACK_OF. 10643 [Steve Henson] 10644 10645 *) When some versions of IIS use the 'NET' form of private key the 10646 key derivation algorithm is different. Normally MD5(password) is 10647 used as a 128 bit RC4 key. In the modified case 10648 MD5(MD5(password) + "SGCKEYSALT") is used instead. Added some 10649 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 10650 as the old Netscape_RSA functions except they have an additional 10651 'sgckey' parameter which uses the modified algorithm. Also added 10652 an -sgckey command line option to the rsa utility. Thanks to 10653 Adrian Peck <bertie@ncipher.com> for posting details of the modified 10654 algorithm to openssl-dev. 10655 [Steve Henson] 10656 10657 *) The evp_local.h macros were using 'c.##kname' which resulted in 10658 invalid expansion on some systems (SCO 5.0.5 for example). 10659 Corrected to 'c.kname'. 10660 [Phillip Porch <root@theporch.com>] 10661 10662 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 10663 a STACK of email addresses from a certificate or request, these look 10664 in the subject name and the subject alternative name extensions and 10665 omit any duplicate addresses. 10666 [Steve Henson] 10667 10668 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 10669 This makes DSA verification about 2 % faster. 10670 [Bodo Moeller] 10671 10672 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 10673 (meaning that now 2^5 values will be precomputed, which is only 4 KB 10674 plus overhead for 1024 bit moduli). 10675 This makes exponentiations about 0.5 % faster for 1024 bit 10676 exponents (as measured by "openssl speed rsa2048"). 10677 [Bodo Moeller] 10678 10679 *) Rename memory handling macros to avoid conflicts with other 10680 software: 10681 Malloc => OPENSSL_malloc 10682 Malloc_locked => OPENSSL_malloc_locked 10683 Realloc => OPENSSL_realloc 10684 Free => OPENSSL_free 10685 [Richard Levitte] 10686 10687 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 10688 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 10689 [Bodo Moeller] 10690 10691 *) CygWin32 support. 10692 [John Jarvie <jjarvie@newsguy.com>] 10693 10694 *) The type-safe stack code has been rejigged. It is now only compiled 10695 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 10696 by default all type-specific stack functions are "#define"d back to 10697 standard stack functions. This results in more streamlined output 10698 but retains the type-safety checking possibilities of the original 10699 approach. 10700 [Geoff Thorpe] 10701 10702 *) The STACK code has been cleaned up, and certain type declarations 10703 that didn't make a lot of sense have been brought in line. This has 10704 also involved a cleanup of sorts in safestack.h to more correctly 10705 map type-safe stack functions onto their plain stack counterparts. 10706 This work has also resulted in a variety of "const"ifications of 10707 lots of the code, especially "_cmp" operations which should normally 10708 be prototyped with "const" parameters anyway. 10709 [Geoff Thorpe] 10710 10711 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 10712 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 10713 (The PRNG state consists of two parts, the large pool 'state' and 'md', 10714 where all of 'md' is used each time the PRNG is used, but 'state' 10715 is used only indexed by a cyclic counter. As entropy may not be 10716 well distributed from the beginning, 'md' is important as a 10717 chaining variable. However, the output function chains only half 10718 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 10719 all of 'md', and seeding with STATE_SIZE dummy bytes will result 10720 in all of 'state' being rewritten, with the new values depending 10721 on virtually all of 'md'. This overcomes the 80 bit limitation.) 10722 [Bodo Moeller] 10723 10724 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 10725 the handshake is continued after ssl_verify_cert_chain(); 10726 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 10727 can lead to 'unexplainable' connection aborts later. 10728 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 10729 10730 *) Major EVP API cipher revision. 10731 Add hooks for extra EVP features. This allows various cipher 10732 parameters to be set in the EVP interface. Support added for variable 10733 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 10734 setting of RC2 and RC5 parameters. 10735 10736 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 10737 ciphers. 10738 10739 Remove lots of duplicated code from the EVP library. For example *every* 10740 cipher init() function handles the 'iv' in the same way according to the 10741 cipher mode. They also all do nothing if the 'key' parameter is NULL and 10742 for CFB and OFB modes they zero ctx->num. 10743 10744 New functionality allows removal of S/MIME code RC2 hack. 10745 10746 Most of the routines have the same form and so can be declared in terms 10747 of macros. 10748 10749 By shifting this to the top level EVP_CipherInit() it can be removed from 10750 all individual ciphers. If the cipher wants to handle IVs or keys 10751 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 10752 flags. 10753 10754 Change lots of functions like EVP_EncryptUpdate() to now return a 10755 value: although software versions of the algorithms cannot fail 10756 any installed hardware versions can. 10757 [Steve Henson] 10758 10759 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 10760 this option is set, tolerate broken clients that send the negotiated 10761 protocol version number instead of the requested protocol version 10762 number. 10763 [Bodo Moeller] 10764 10765 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 10766 i.e. non-zero for export ciphersuites, zero otherwise. 10767 Previous versions had this flag inverted, inconsistent with 10768 rsa_tmp_cb (..._TMP_RSA_CB). 10769 [Bodo Moeller; problem reported by Amit Chopra] 10770 10771 *) Add missing DSA library text string. Work around for some IIS 10772 key files with invalid SEQUENCE encoding. 10773 [Steve Henson] 10774 10775 *) Add a document (doc/standards.txt) that list all kinds of standards 10776 and so on that are implemented in OpenSSL. 10777 [Richard Levitte] 10778 10779 *) Enhance c_rehash script. Old version would mishandle certificates 10780 with the same subject name hash and wouldn't handle CRLs at all. 10781 Added -fingerprint option to crl utility, to support new c_rehash 10782 features. 10783 [Steve Henson] 10784 10785 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 10786 [Ulf Möller] 10787 10788 *) Fix for SSL server purpose checking. Server checking was 10789 rejecting certificates which had extended key usage present 10790 but no ssl client purpose. 10791 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 10792 10793 *) Make PKCS#12 code work with no password. The PKCS#12 spec 10794 is a little unclear about how a blank password is handled. 10795 Since the password in encoded as a BMPString with terminating 10796 double NULL a zero length password would end up as just the 10797 double NULL. However no password at all is different and is 10798 handled differently in the PKCS#12 key generation code. NS 10799 treats a blank password as zero length. MSIE treats it as no 10800 password on export: but it will try both on import. We now do 10801 the same: PKCS12_parse() tries zero length and no password if 10802 the password is set to "" or NULL (NULL is now a valid password: 10803 it wasn't before) as does the pkcs12 application. 10804 [Steve Henson] 10805 10806 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 10807 perror when PEM_read_bio_X509_REQ fails, the error message must 10808 be obtained from the error queue. 10809 [Bodo Moeller] 10810 10811 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 10812 it in ERR_remove_state if appropriate, and change ERR_get_state 10813 accordingly to avoid race conditions (this is necessary because 10814 thread_hash is no longer constant once set). 10815 [Bodo Moeller] 10816 10817 *) Bugfix for linux-elf makefile.one. 10818 [Ulf Möller] 10819 10820 *) RSA_get_default_method() will now cause a default 10821 RSA_METHOD to be chosen if one doesn't exist already. 10822 Previously this was only set during a call to RSA_new() 10823 or RSA_new_method(NULL) meaning it was possible for 10824 RSA_get_default_method() to return NULL. 10825 [Geoff Thorpe] 10826 10827 *) Added native name translation to the existing DSO code 10828 that will convert (if the flag to do so is set) filenames 10829 that are sufficiently small and have no path information 10830 into a canonical native form. Eg. "blah" converted to 10831 "libblah.so" or "blah.dll" etc. 10832 [Geoff Thorpe] 10833 10834 *) New function ERR_error_string_n(e, buf, len) which is like 10835 ERR_error_string(e, buf), but writes at most 'len' bytes 10836 including the 0 terminator. For ERR_error_string_n, 'buf' 10837 may not be NULL. 10838 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 10839 10840 *) CONF library reworked to become more general. A new CONF 10841 configuration file reader "class" is implemented as well as a 10842 new functions (NCONF_*, for "New CONF") to handle it. The now 10843 old CONF_* functions are still there, but are reimplemented to 10844 work in terms of the new functions. Also, a set of functions 10845 to handle the internal storage of the configuration data is 10846 provided to make it easier to write new configuration file 10847 reader "classes" (I can definitely see something reading a 10848 configuration file in XML format, for example), called _CONF_*, 10849 or "the configuration storage API"... 10850 10851 The new configuration file reading functions are: 10852 10853 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 10854 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 10855 10856 NCONF_default, NCONF_WIN32 10857 10858 NCONF_dump_fp, NCONF_dump_bio 10859 10860 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 10861 NCONF_new creates a new CONF object. This works in the same way 10862 as other interfaces in OpenSSL, like the BIO interface. 10863 NCONF_dump_* dump the internal storage of the configuration file, 10864 which is useful for debugging. All other functions take the same 10865 arguments as the old CONF_* functions with the exception of the 10866 first that must be a `CONF *' instead of a `LHASH *'. 10867 10868 To make it easier to use the new classes with the old CONF_* functions, 10869 the function CONF_set_default_method is provided. 10870 [Richard Levitte] 10871 10872 *) Add '-tls1' option to 'openssl ciphers', which was already 10873 mentioned in the documentation but had not been implemented. 10874 (This option is not yet really useful because even the additional 10875 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 10876 [Bodo Moeller] 10877 10878 *) Initial DSO code added into libcrypto for letting OpenSSL (and 10879 OpenSSL-based applications) load shared libraries and bind to 10880 them in a portable way. 10881 [Geoff Thorpe, with contributions from Richard Levitte] 10882 10883 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 10884 10885 *) Make sure _lrotl and _lrotr are only used with MSVC. 10886 10887 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 10888 (the default implementation of RAND_status). 10889 10890 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 10891 to '-clrext' (= clear extensions), as intended and documented. 10892 [Bodo Moeller; inconsistency pointed out by Michael Attili 10893 <attili@amaxo.com>] 10894 10895 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 10896 was larger than the MD block size. 10897 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 10898 10899 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 10900 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 10901 using the passed key: if the passed key was a private key the result 10902 of X509_print(), for example, would be to print out all the private key 10903 components. 10904 [Steve Henson] 10905 10906 *) des_quad_cksum() byte order bug fix. 10907 [Ulf Möller, using the problem description in krb4-0.9.7, where 10908 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 10909 10910 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 10911 discouraged. 10912 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 10913 10914 *) For easily testing in shell scripts whether some command 10915 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 10916 returns with exit code 0 iff no command of the given name is available. 10917 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 10918 the output goes to stdout and nothing is printed to stderr. 10919 Additional arguments are always ignored. 10920 10921 Since for each cipher there is a command of the same name, 10922 the 'no-cipher' compilation switches can be tested this way. 10923 10924 ('openssl no-XXX' is not able to detect pseudo-commands such 10925 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 10926 [Bodo Moeller] 10927 10928 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 10929 [Bodo Moeller] 10930 10931 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 10932 is set; it will be thrown away anyway because each handshake creates 10933 its own key. 10934 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 10935 to parameters -- in previous versions (since OpenSSL 0.9.3) the 10936 'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning 10937 you effectively got SSL_OP_SINGLE_DH_USE when using this macro. 10938 [Bodo Moeller] 10939 10940 *) New s_client option -ign_eof: EOF at stdin is ignored, and 10941 'Q' and 'R' lose their special meanings (quit/renegotiate). 10942 This is part of what -quiet does; unlike -quiet, -ign_eof 10943 does not suppress any output. 10944 [Richard Levitte] 10945 10946 *) Add compatibility options to the purpose and trust code. The 10947 purpose X509_PURPOSE_ANY is "any purpose" which automatically 10948 accepts a certificate or CA, this was the previous behaviour, 10949 with all the associated security issues. 10950 10951 X509_TRUST_COMPAT is the old trust behaviour: only and 10952 automatically trust self signed roots in certificate store. A 10953 new trust setting X509_TRUST_DEFAULT is used to specify that 10954 a purpose has no associated trust setting and it should instead 10955 use the value in the default purpose. 10956 [Steve Henson] 10957 10958 *) Fix the PKCS#8 DSA private key code so it decodes keys again 10959 and fix a memory leak. 10960 [Steve Henson] 10961 10962 *) In util/mkerr.pl (which implements 'make errors'), preserve 10963 reason strings from the previous version of the .c file, as 10964 the default to have only downcase letters (and digits) in 10965 automatically generated reasons codes is not always appropriate. 10966 [Bodo Moeller] 10967 10968 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 10969 using strerror. Previously, ERR_reason_error_string() returned 10970 library names as reason strings for SYSerr; but SYSerr is a special 10971 case where small numbers are errno values, not library numbers. 10972 [Bodo Moeller] 10973 10974 *) Add '-dsaparam' option to 'openssl dhparam' application. This 10975 converts DSA parameters into DH parameters. (When creating parameters, 10976 DSA_generate_parameters is used.) 10977 [Bodo Moeller] 10978 10979 *) Include 'length' (recommended exponent length) in C code generated 10980 by 'openssl dhparam -C'. 10981 [Bodo Moeller] 10982 10983 *) The second argument to set_label in perlasm was already being used 10984 so couldn't be used as a "file scope" flag. Moved to third argument 10985 which was free. 10986 [Steve Henson] 10987 10988 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 10989 instead of RAND_bytes for encryption IVs and salts. 10990 [Bodo Moeller] 10991 10992 *) Include RAND_status() into RAND_METHOD instead of implementing 10993 it only for md_rand.c Otherwise replacing the PRNG by calling 10994 RAND_set_rand_method would be impossible. 10995 [Bodo Moeller] 10996 10997 *) Don't let DSA_generate_key() enter an infinite loop if the random 10998 number generation fails. 10999 [Bodo Moeller] 11000 11001 *) New 'rand' application for creating pseudo-random output. 11002 [Bodo Moeller] 11003 11004 *) Added configuration support for Linux/IA64 11005 [Rolf Haberrecker <rolf@suse.de>] 11006 11007 *) Assembler module support for Mingw32. 11008 [Ulf Möller] 11009 11010 *) Shared library support for HPUX (in shlib/). 11011 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 11012 11013 *) Shared library support for Solaris gcc. 11014 [Lutz Behnke <behnke@trustcenter.de>] 11015 11016 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 11017 11018 *) PKCS7_encrypt() was adding text MIME headers twice because they 11019 were added manually and by SMIME_crlf_copy(). 11020 [Steve Henson] 11021 11022 *) In bntest.c don't call BN_rand with zero bits argument. 11023 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 11024 11025 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 11026 case was implemented. This caused BN_div_recp() to fail occasionally. 11027 [Ulf Möller] 11028 11029 *) Add an optional second argument to the set_label() in the perl 11030 assembly language builder. If this argument exists and is set 11031 to 1 it signals that the assembler should use a symbol whose 11032 scope is the entire file, not just the current function. This 11033 is needed with MASM which uses the format label:: for this scope. 11034 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 11035 11036 *) Change the ASN1 types so they are typedefs by default. Before 11037 almost all types were #define'd to ASN1_STRING which was causing 11038 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 11039 for example. 11040 [Steve Henson] 11041 11042 *) Change names of new functions to the new get1/get0 naming 11043 convention: After 'get1', the caller owns a reference count 11044 and has to call ..._free; 'get0' returns a pointer to some 11045 data structure without incrementing reference counters. 11046 (Some of the existing 'get' functions increment a reference 11047 counter, some don't.) 11048 Similarly, 'set1' and 'add1' functions increase reference 11049 counters or duplicate objects. 11050 [Steve Henson] 11051 11052 *) Allow for the possibility of temp RSA key generation failure: 11053 the code used to assume it always worked and crashed on failure. 11054 [Steve Henson] 11055 11056 *) Fix potential buffer overrun problem in BIO_printf(). 11057 [Ulf Möller, using public domain code by Patrick Powell; problem 11058 pointed out by David Sacerdote <das33@cornell.edu>] 11059 11060 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 11061 RAND_egd() and RAND_status(). In the command line application, 11062 the EGD socket can be specified like a seed file using RANDFILE 11063 or -rand. 11064 [Ulf Möller] 11065 11066 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 11067 Some CAs (e.g. Verisign) distribute certificates in this form. 11068 [Steve Henson] 11069 11070 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 11071 list to exclude them. This means that no special compilation option 11072 is needed to use anonymous DH: it just needs to be included in the 11073 cipher list. 11074 [Steve Henson] 11075 11076 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 11077 EVP_MD_type. The old functionality is available in a new macro called 11078 EVP_MD_md(). Change code that uses it and update docs. 11079 [Steve Henson] 11080 11081 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 11082 where the 'void *' argument is replaced by a function pointer argument. 11083 Previously 'void *' was abused to point to functions, which works on 11084 many platforms, but is not correct. As these functions are usually 11085 called by macros defined in OpenSSL header files, most source code 11086 should work without changes. 11087 [Richard Levitte] 11088 11089 *) <openssl/opensslconf.h> (which is created by Configure) now contains 11090 sections with information on -D... compiler switches used for 11091 compiling the library so that applications can see them. To enable 11092 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 11093 must be defined. E.g., 11094 #define OPENSSL_ALGORITHM_DEFINES 11095 #include <openssl/opensslconf.h> 11096 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 11097 [Richard Levitte, Ulf and Bodo Möller] 11098 11099 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 11100 record layer. 11101 [Bodo Moeller] 11102 11103 *) Change the 'other' type in certificate aux info to a STACK_OF 11104 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 11105 the required ASN1 format: arbitrary types determined by an OID. 11106 [Steve Henson] 11107 11108 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 11109 argument to 'req'. This is not because the function is newer or 11110 better than others it just uses the work 'NEW' in the certificate 11111 request header lines. Some software needs this. 11112 [Steve Henson] 11113 11114 *) Reorganise password command line arguments: now passwords can be 11115 obtained from various sources. Delete the PEM_cb function and make 11116 it the default behaviour: i.e. if the callback is NULL and the 11117 usrdata argument is not NULL interpret it as a null terminated pass 11118 phrase. If usrdata and the callback are NULL then the pass phrase 11119 is prompted for as usual. 11120 [Steve Henson] 11121 11122 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 11123 the support is automatically enabled. The resulting binaries will 11124 autodetect the card and use it if present. 11125 [Ben Laurie and Compaq Inc.] 11126 11127 *) Work around for Netscape hang bug. This sends certificate request 11128 and server done in one record. Since this is perfectly legal in the 11129 SSL/TLS protocol it isn't a "bug" option and is on by default. See 11130 the bugs/SSLv3 entry for more info. 11131 [Steve Henson] 11132 11133 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 11134 [Andy Polyakov] 11135 11136 *) Add -rand argument to smime and pkcs12 applications and read/write 11137 of seed file. 11138 [Steve Henson] 11139 11140 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 11141 [Bodo Moeller] 11142 11143 *) Add command line password options to the remaining applications. 11144 [Steve Henson] 11145 11146 *) Bug fix for BN_div_recp() for numerators with an even number of 11147 bits. 11148 [Ulf Möller] 11149 11150 *) More tests in bntest.c, and changed test_bn output. 11151 [Ulf Möller] 11152 11153 *) ./config recognizes MacOS X now. 11154 [Andy Polyakov] 11155 11156 *) Bug fix for BN_div() when the first words of num and divisor are 11157 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 11158 [Ulf Möller] 11159 11160 *) Add support for various broken PKCS#8 formats, and command line 11161 options to produce them. 11162 [Steve Henson] 11163 11164 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 11165 get temporary BIGNUMs from a BN_CTX. 11166 [Ulf Möller] 11167 11168 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 11169 for p == 0. 11170 [Ulf Möller] 11171 11172 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 11173 include a #define from the old name to the new. The original intent 11174 was that statically linked binaries could for example just call 11175 SSLeay_add_all_ciphers() to just add ciphers to the table and not 11176 link with digests. This never worked because SSLeay_add_all_digests() 11177 and SSLeay_add_all_ciphers() were in the same source file so calling 11178 one would link with the other. They are now in separate source files. 11179 [Steve Henson] 11180 11181 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 11182 [Steve Henson] 11183 11184 *) Use a less unusual form of the Miller-Rabin primality test (it used 11185 a binary algorithm for exponentiation integrated into the Miller-Rabin 11186 loop, our standard modexp algorithms are faster). 11187 [Bodo Moeller] 11188 11189 *) Support for the EBCDIC character set completed. 11190 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 11191 11192 *) Source code cleanups: use const where appropriate, eliminate casts, 11193 use void * instead of char * in lhash. 11194 [Ulf Möller] 11195 11196 *) Bugfix: ssl3_send_server_key_exchange was not restartable 11197 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 11198 this the server could overwrite ephemeral keys that the client 11199 has already seen). 11200 [Bodo Moeller] 11201 11202 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 11203 using 50 iterations of the Rabin-Miller test. 11204 11205 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 11206 iterations of the Rabin-Miller test as required by the appendix 11207 to FIPS PUB 186[-1]) instead of DSA_is_prime. 11208 As BN_is_prime_fasttest includes trial division, DSA parameter 11209 generation becomes much faster. 11210 11211 This implies a change for the callback functions in DSA_is_prime 11212 and DSA_generate_parameters: The callback function is called once 11213 for each positive witness in the Rabin-Miller test, not just 11214 occasionally in the inner loop; and the parameters to the 11215 callback function now provide an iteration count for the outer 11216 loop rather than for the current invocation of the inner loop. 11217 DSA_generate_parameters additionally can call the callback 11218 function with an 'iteration count' of -1, meaning that a 11219 candidate has passed the trial division test (when q is generated 11220 from an application-provided seed, trial division is skipped). 11221 [Bodo Moeller] 11222 11223 *) New function BN_is_prime_fasttest that optionally does trial 11224 division before starting the Rabin-Miller test and has 11225 an additional BN_CTX * argument (whereas BN_is_prime always 11226 has to allocate at least one BN_CTX). 11227 'callback(1, -1, cb_arg)' is called when a number has passed the 11228 trial division stage. 11229 [Bodo Moeller] 11230 11231 *) Fix for bug in CRL encoding. The validity dates weren't being handled 11232 as ASN1_TIME. 11233 [Steve Henson] 11234 11235 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 11236 [Steve Henson] 11237 11238 *) New function BN_pseudo_rand(). 11239 [Ulf Möller] 11240 11241 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 11242 bignum version of BN_from_montgomery() with the working code from 11243 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 11244 the comments. 11245 [Ulf Möller] 11246 11247 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 11248 made it impossible to use the same SSL_SESSION data structure in 11249 SSL2 clients in multiple threads. 11250 [Bodo Moeller] 11251 11252 *) The return value of RAND_load_file() no longer counts bytes obtained 11253 by stat(). RAND_load_file(..., -1) is new and uses the complete file 11254 to seed the PRNG (previously an explicit byte count was required). 11255 [Ulf Möller, Bodo Möller] 11256 11257 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 11258 used (char *) instead of (void *) and had casts all over the place. 11259 [Steve Henson] 11260 11261 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 11262 [Ulf Möller] 11263 11264 *) Retain source code compatibility for BN_prime_checks macro: 11265 BN_is_prime(..., BN_prime_checks, ...) now uses 11266 BN_prime_checks_for_size to determine the appropriate number of 11267 Rabin-Miller iterations. 11268 [Ulf Möller] 11269 11270 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 11271 DH_CHECK_P_NOT_SAFE_PRIME. 11272 (Check if this is true? OpenPGP calls them "strong".) 11273 [Ulf Möller] 11274 11275 *) Merge the functionality of "dh" and "gendh" programs into a new program 11276 "dhparam". The old programs are retained for now but will handle DH keys 11277 (instead of parameters) in future. 11278 [Steve Henson] 11279 11280 *) Make the ciphers, s_server and s_client programs check the return values 11281 when a new cipher list is set. 11282 [Steve Henson] 11283 11284 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 11285 ciphers. Before when the 56bit ciphers were enabled the sorting was 11286 wrong. 11287 11288 The syntax for the cipher sorting has been extended to support sorting by 11289 cipher-strength (using the strength_bits hard coded in the tables). 11290 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 11291 11292 Fix a bug in the cipher-command parser: when supplying a cipher command 11293 string with an "undefined" symbol (neither command nor alphanumeric 11294 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 11295 an error is flagged. 11296 11297 Due to the strength-sorting extension, the code of the 11298 ssl_create_cipher_list() function was completely rearranged. I hope that 11299 the readability was also increased :-) 11300 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 11301 11302 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 11303 for the first serial number and places 2 in the serial number file. This 11304 avoids problems when the root CA is created with serial number zero and 11305 the first user certificate has the same issuer name and serial number 11306 as the root CA. 11307 [Steve Henson] 11308 11309 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 11310 the new code. Add documentation for this stuff. 11311 [Steve Henson] 11312 11313 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 11314 X509_*() to X509at_*() on the grounds that they don't handle X509 11315 structures and behave in an analogous way to the X509v3 functions: 11316 they shouldn't be called directly but wrapper functions should be used 11317 instead. 11318 11319 So we also now have some wrapper functions that call the X509at functions 11320 when passed certificate requests. (TO DO: similar things can be done with 11321 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 11322 things. Some of these need some d2i or i2d and print functionality 11323 because they handle more complex structures.) 11324 [Steve Henson] 11325 11326 *) Add missing #ifndefs that caused missing symbols when building libssl 11327 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 11328 NO_RSA in ssl/s2*.c. 11329 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller] 11330 11331 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 11332 has a return value which indicates the quality of the random data 11333 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 11334 error queue. New function RAND_pseudo_bytes() generates output that is 11335 guaranteed to be unique but not unpredictable. RAND_add is like 11336 RAND_seed, but takes an extra argument for an entropy estimate 11337 (RAND_seed always assumes full entropy). 11338 [Ulf Möller] 11339 11340 *) Do more iterations of Rabin-Miller probable prime test (specifically, 11341 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 11342 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 11343 in crypto/bn/bn_prime.c for the complete table). This guarantees a 11344 false-positive rate of at most 2^-80 for random input. 11345 [Bodo Moeller] 11346 11347 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 11348 [Bodo Moeller] 11349 11350 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 11351 in the 0.9.5 release), this returns the chain 11352 from an X509_CTX structure with a dup of the stack and all 11353 the X509 reference counts upped: so the stack will exist 11354 after X509_CTX_cleanup() has been called. Modify pkcs12.c 11355 to use this. 11356 11357 Also make SSL_SESSION_print() print out the verify return 11358 code. 11359 [Steve Henson] 11360 11361 *) Add manpage for the pkcs12 command. Also change the default 11362 behaviour so MAC iteration counts are used unless the new 11363 -nomaciter option is used. This improves file security and 11364 only older versions of MSIE (4.0 for example) need it. 11365 [Steve Henson] 11366 11367 *) Honor the no-xxx Configure options when creating .DEF files. 11368 [Ulf Möller] 11369 11370 *) Add PKCS#10 attributes to field table: challengePassword, 11371 unstructuredName and unstructuredAddress. These are taken from 11372 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 11373 international characters are used. 11374 11375 More changes to X509_ATTRIBUTE code: allow the setting of types 11376 based on strings. Remove the 'loc' parameter when adding 11377 attributes because these will be a SET OF encoding which is sorted 11378 in ASN1 order. 11379 [Steve Henson] 11380 11381 *) Initial changes to the 'req' utility to allow request generation 11382 automation. This will allow an application to just generate a template 11383 file containing all the field values and have req construct the 11384 request. 11385 11386 Initial support for X509_ATTRIBUTE handling. Stacks of these are 11387 used all over the place including certificate requests and PKCS#7 11388 structures. They are currently handled manually where necessary with 11389 some primitive wrappers for PKCS#7. The new functions behave in a 11390 manner analogous to the X509 extension functions: they allow 11391 attributes to be looked up by NID and added. 11392 11393 Later something similar to the X509V3 code would be desirable to 11394 automatically handle the encoding, decoding and printing of the 11395 more complex types. The string types like challengePassword can 11396 be handled by the string table functions. 11397 11398 Also modified the multi byte string table handling. Now there is 11399 a 'global mask' which masks out certain types. The table itself 11400 can use the flag STABLE_NO_MASK to ignore the mask setting: this 11401 is useful when for example there is only one permissible type 11402 (as in countryName) and using the mask might result in no valid 11403 types at all. 11404 [Steve Henson] 11405 11406 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 11407 SSL_get_peer_finished to allow applications to obtain the latest 11408 Finished messages sent to the peer or expected from the peer, 11409 respectively. (SSL_get_peer_finished is usually the Finished message 11410 actually received from the peer, otherwise the protocol will be aborted.) 11411 11412 As the Finished message are message digests of the complete handshake 11413 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 11414 be used for external authentication procedures when the authentication 11415 provided by SSL/TLS is not desired or is not enough. 11416 [Bodo Moeller] 11417 11418 *) Enhanced support for Alpha Linux is added. Now ./config checks if 11419 the host supports BWX extension and if Compaq C is present on the 11420 $PATH. Just exploiting of the BWX extension results in 20-30% 11421 performance kick for some algorithms, e.g. DES and RC4 to mention 11422 a couple. Compaq C in turn generates ~20% faster code for MD5 and 11423 SHA1. 11424 [Andy Polyakov] 11425 11426 *) Add support for MS "fast SGC". This is arguably a violation of the 11427 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 11428 weak crypto and after checking the certificate is SGC a second one 11429 with strong crypto. MS SGC stops the first handshake after receiving 11430 the server certificate message and sends a second client hello. Since 11431 a server will typically do all the time consuming operations before 11432 expecting any further messages from the client (server key exchange 11433 is the most expensive) there is little difference between the two. 11434 11435 To get OpenSSL to support MS SGC we have to permit a second client 11436 hello message after we have sent server done. In addition we have to 11437 reset the MAC if we do get this second client hello. 11438 [Steve Henson] 11439 11440 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 11441 if a DER encoded private key is RSA or DSA traditional format. Changed 11442 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 11443 format DER encoded private key. Newer code should use PKCS#8 format which 11444 has the key type encoded in the ASN1 structure. Added DER private key 11445 support to pkcs8 application. 11446 [Steve Henson] 11447 11448 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 11449 ciphersuites has been selected (as required by the SSL 3/TLS 1 11450 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 11451 is set, we interpret this as a request to violate the specification 11452 (the worst that can happen is a handshake failure, and 'correct' 11453 behaviour would result in a handshake failure anyway). 11454 [Bodo Moeller] 11455 11456 *) In SSL_CTX_add_session, take into account that there might be multiple 11457 SSL_SESSION structures with the same session ID (e.g. when two threads 11458 concurrently obtain them from an external cache). 11459 The internal cache can handle only one SSL_SESSION with a given ID, 11460 so if there's a conflict, we now throw out the old one to achieve 11461 consistency. 11462 [Bodo Moeller] 11463 11464 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 11465 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 11466 some routines that use cipher OIDs: some ciphers do not have OIDs 11467 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 11468 example. 11469 [Steve Henson] 11470 11471 *) Simplify the trust setting structure and code. Now we just have 11472 two sequences of OIDs for trusted and rejected settings. These will 11473 typically have values the same as the extended key usage extension 11474 and any application specific purposes. 11475 11476 The trust checking code now has a default behaviour: it will just 11477 check for an object with the same NID as the passed id. Functions can 11478 be provided to override either the default behaviour or the behaviour 11479 for a given id. SSL client, server and email already have functions 11480 in place for compatibility: they check the NID and also return "trusted" 11481 if the certificate is self signed. 11482 [Steve Henson] 11483 11484 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 11485 traditional format into an EVP_PKEY structure. 11486 [Steve Henson] 11487 11488 *) Add a password callback function PEM_cb() which either prompts for 11489 a password if usr_data is NULL or otherwise assumes it is a null 11490 terminated password. Allow passwords to be passed on command line 11491 environment or config files in a few more utilities. 11492 [Steve Henson] 11493 11494 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 11495 keys. Add some short names for PKCS#8 PBE algorithms and allow them 11496 to be specified on the command line for the pkcs8 and pkcs12 utilities. 11497 Update documentation. 11498 [Steve Henson] 11499 11500 *) Support for ASN1 "NULL" type. This could be handled before by using 11501 ASN1_TYPE but there wasn't any function that would try to read a NULL 11502 and produce an error if it couldn't. For compatibility we also have 11503 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 11504 don't allocate anything because they don't need to. 11505 [Steve Henson] 11506 11507 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 11508 for details. 11509 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 11510 11511 *) Rebuild of the memory allocation routines used by OpenSSL code and 11512 possibly others as well. The purpose is to make an interface that 11513 provide hooks so anyone can build a separate set of allocation and 11514 deallocation routines to be used by OpenSSL, for example memory 11515 pool implementations, or something else, which was previously hard 11516 since Malloc(), Realloc() and Free() were defined as macros having 11517 the values malloc, realloc and free, respectively (except for Win32 11518 compilations). The same is provided for memory debugging code. 11519 OpenSSL already comes with functionality to find memory leaks, but 11520 this gives people a chance to debug other memory problems. 11521 11522 With these changes, a new set of functions and macros have appeared: 11523 11524 CRYPTO_set_mem_debug_functions() [F] 11525 CRYPTO_get_mem_debug_functions() [F] 11526 CRYPTO_dbg_set_options() [F] 11527 CRYPTO_dbg_get_options() [F] 11528 CRYPTO_malloc_debug_init() [M] 11529 11530 The memory debug functions are NULL by default, unless the library 11531 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 11532 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 11533 gives the standard debugging functions that come with OpenSSL) or 11534 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 11535 provided by the library user) must be used. When the standard 11536 debugging functions are used, CRYPTO_dbg_set_options can be used to 11537 request additional information: 11538 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 11539 the CRYPTO_MDEBUG_xxx macro when compiling the library. 11540 11541 Also, things like CRYPTO_set_mem_functions will always give the 11542 expected result (the new set of functions is used for allocation 11543 and deallocation) at all times, regardless of platform and compiler 11544 options. 11545 11546 To finish it up, some functions that were never use in any other 11547 way than through macros have a new API and new semantic: 11548 11549 CRYPTO_dbg_malloc() 11550 CRYPTO_dbg_realloc() 11551 CRYPTO_dbg_free() 11552 11553 All macros of value have retained their old syntax. 11554 [Richard Levitte and Bodo Moeller] 11555 11556 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 11557 ordering of SMIMECapabilities wasn't in "strength order" and there 11558 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 11559 algorithm. 11560 [Steve Henson] 11561 11562 *) Some ASN1 types with illegal zero length encoding (INTEGER, 11563 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 11564 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 11565 11566 *) Merge in my S/MIME library for OpenSSL. This provides a simple 11567 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 11568 functionality to handle multipart/signed properly) and a utility 11569 called 'smime' to call all this stuff. This is based on code I 11570 originally wrote for Celo who have kindly allowed it to be 11571 included in OpenSSL. 11572 [Steve Henson] 11573 11574 *) Add variants des_set_key_checked and des_set_key_unchecked of 11575 des_set_key (aka des_key_sched). Global variable des_check_key 11576 decides which of these is called by des_set_key; this way 11577 des_check_key behaves as it always did, but applications and 11578 the library itself, which was buggy for des_check_key == 1, 11579 have a cleaner way to pick the version they need. 11580 [Bodo Moeller] 11581 11582 *) New function PKCS12_newpass() which changes the password of a 11583 PKCS12 structure. 11584 [Steve Henson] 11585 11586 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 11587 dynamic mix. In both cases the ids can be used as an index into the 11588 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 11589 functions so they accept a list of the field values and the 11590 application doesn't need to directly manipulate the X509_TRUST 11591 structure. 11592 [Steve Henson] 11593 11594 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 11595 need initialising. 11596 [Steve Henson] 11597 11598 *) Modify the way the V3 extension code looks up extensions. This now 11599 works in a similar way to the object code: we have some "standard" 11600 extensions in a static table which is searched with OBJ_bsearch() 11601 and the application can add dynamic ones if needed. The file 11602 crypto/x509v3/ext_dat.h now has the info: this file needs to be 11603 updated whenever a new extension is added to the core code and kept 11604 in ext_nid order. There is a simple program 'tabtest.c' which checks 11605 this. New extensions are not added too often so this file can readily 11606 be maintained manually. 11607 11608 There are two big advantages in doing things this way. The extensions 11609 can be looked up immediately and no longer need to be "added" using 11610 X509V3_add_standard_extensions(): this function now does nothing. 11611 [Side note: I get *lots* of email saying the extension code doesn't 11612 work because people forget to call this function] 11613 Also no dynamic allocation is done unless new extensions are added: 11614 so if we don't add custom extensions there is no need to call 11615 X509V3_EXT_cleanup(). 11616 [Steve Henson] 11617 11618 *) Modify enc utility's salting as follows: make salting the default. Add a 11619 magic header, so unsalted files fail gracefully instead of just decrypting 11620 to garbage. This is because not salting is a big security hole, so people 11621 should be discouraged from doing it. 11622 [Ben Laurie] 11623 11624 *) Fixes and enhancements to the 'x509' utility. It allowed a message 11625 digest to be passed on the command line but it only used this 11626 parameter when signing a certificate. Modified so all relevant 11627 operations are affected by the digest parameter including the 11628 -fingerprint and -x509toreq options. Also -x509toreq choked if a 11629 DSA key was used because it didn't fix the digest. 11630 [Steve Henson] 11631 11632 *) Initial certificate chain verify code. Currently tests the untrusted 11633 certificates for consistency with the verify purpose (which is set 11634 when the X509_STORE_CTX structure is set up) and checks the pathlength. 11635 11636 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 11637 this is because it will reject chains with invalid extensions whereas 11638 every previous version of OpenSSL and SSLeay made no checks at all. 11639 11640 Trust code: checks the root CA for the relevant trust settings. Trust 11641 settings have an initial value consistent with the verify purpose: e.g. 11642 if the verify purpose is for SSL client use it expects the CA to be 11643 trusted for SSL client use. However the default value can be changed to 11644 permit custom trust settings: one example of this would be to only trust 11645 certificates from a specific "secure" set of CAs. 11646 11647 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 11648 which should be used for version portability: especially since the 11649 verify structure is likely to change more often now. 11650 11651 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 11652 to set them. If not set then assume SSL clients will verify SSL servers 11653 and vice versa. 11654 11655 Two new options to the verify program: -untrusted allows a set of 11656 untrusted certificates to be passed in and -purpose which sets the 11657 intended purpose of the certificate. If a purpose is set then the 11658 new chain verify code is used to check extension consistency. 11659 [Steve Henson] 11660 11661 *) Support for the authority information access extension. 11662 [Steve Henson] 11663 11664 *) Modify RSA and DSA PEM read routines to transparently handle 11665 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 11666 public keys in a format compatible with certificate 11667 SubjectPublicKeyInfo structures. Unfortunately there were already 11668 functions called *_PublicKey_* which used various odd formats so 11669 these are retained for compatibility: however the DSA variants were 11670 never in a public release so they have been deleted. Changed dsa/rsa 11671 utilities to handle the new format: note no releases ever handled public 11672 keys so we should be OK. 11673 11674 The primary motivation for this change is to avoid the same fiasco 11675 that dogs private keys: there are several incompatible private key 11676 formats some of which are standard and some OpenSSL specific and 11677 require various evil hacks to allow partial transparent handling and 11678 even then it doesn't work with DER formats. Given the option anything 11679 other than PKCS#8 should be dumped: but the other formats have to 11680 stay in the name of compatibility. 11681 11682 With public keys and the benefit of hindsight one standard format 11683 is used which works with EVP_PKEY, RSA or DSA structures: though 11684 it clearly returns an error if you try to read the wrong kind of key. 11685 11686 Added a -pubkey option to the 'x509' utility to output the public key. 11687 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 11688 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 11689 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 11690 that do the same as the EVP_PKEY_assign_*() except they up the 11691 reference count of the added key (they don't "swallow" the 11692 supplied key). 11693 [Steve Henson] 11694 11695 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 11696 CRLs would fail if the file contained no certificates or no CRLs: 11697 added a new function to read in both types and return the number 11698 read: this means that if none are read it will be an error. The 11699 DER versions of the certificate and CRL reader would always fail 11700 because it isn't possible to mix certificates and CRLs in DER format 11701 without choking one or the other routine. Changed this to just read 11702 a certificate: this is the best we can do. Also modified the code 11703 in apps/verify.c to take notice of return codes: it was previously 11704 attempting to read in certificates from NULL pointers and ignoring 11705 any errors: this is one reason why the cert and CRL reader seemed 11706 to work. It doesn't check return codes from the default certificate 11707 routines: these may well fail if the certificates aren't installed. 11708 [Steve Henson] 11709 11710 *) Code to support otherName option in GeneralName. 11711 [Steve Henson] 11712 11713 *) First update to verify code. Change the verify utility 11714 so it warns if it is passed a self signed certificate: 11715 for consistency with the normal behaviour. X509_verify 11716 has been modified to it will now verify a self signed 11717 certificate if *exactly* the same certificate appears 11718 in the store: it was previously impossible to trust a 11719 single self signed certificate. This means that: 11720 openssl verify ss.pem 11721 now gives a warning about a self signed certificate but 11722 openssl verify -CAfile ss.pem ss.pem 11723 is OK. 11724 [Steve Henson] 11725 11726 *) For servers, store verify_result in SSL_SESSION data structure 11727 (and add it to external session representation). 11728 This is needed when client certificate verifications fails, 11729 but an application-provided verification callback (set by 11730 SSL_CTX_set_cert_verify_callback) allows accepting the session 11731 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 11732 but returns 1): When the session is reused, we have to set 11733 ssl->verify_result to the appropriate error code to avoid 11734 security holes. 11735 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 11736 11737 *) Fix a bug in the new PKCS#7 code: it didn't consider the 11738 case in PKCS7_dataInit() where the signed PKCS7 structure 11739 didn't contain any existing data because it was being created. 11740 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 11741 11742 *) Add a salt to the key derivation routines in enc.c. This 11743 forms the first 8 bytes of the encrypted file. Also add a 11744 -S option to allow a salt to be input on the command line. 11745 [Steve Henson] 11746 11747 *) New function X509_cmp(). Oddly enough there wasn't a function 11748 to compare two certificates. We do this by working out the SHA1 11749 hash and comparing that. X509_cmp() will be needed by the trust 11750 code. 11751 [Steve Henson] 11752 11753 *) SSL_get1_session() is like SSL_get_session(), but increments 11754 the reference count in the SSL_SESSION returned. 11755 [Geoff Thorpe <geoff@eu.c2.net>] 11756 11757 *) Fix for 'req': it was adding a null to request attributes. 11758 Also change the X509_LOOKUP and X509_INFO code to handle 11759 certificate auxiliary information. 11760 [Steve Henson] 11761 11762 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 11763 the 'enc' command. 11764 [Steve Henson] 11765 11766 *) Add the possibility to add extra information to the memory leak 11767 detecting output, to form tracebacks, showing from where each 11768 allocation was originated: CRYPTO_push_info("constant string") adds 11769 the string plus current file name and line number to a per-thread 11770 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 11771 is like calling CYRPTO_pop_info() until the stack is empty. 11772 Also updated memory leak detection code to be multi-thread-safe. 11773 [Richard Levitte] 11774 11775 *) Add options -text and -noout to pkcs7 utility and delete the 11776 encryption options which never did anything. Update docs. 11777 [Steve Henson] 11778 11779 *) Add options to some of the utilities to allow the pass phrase 11780 to be included on either the command line (not recommended on 11781 OSes like Unix) or read from the environment. Update the 11782 manpages and fix a few bugs. 11783 [Steve Henson] 11784 11785 *) Add a few manpages for some of the openssl commands. 11786 [Steve Henson] 11787 11788 *) Fix the -revoke option in ca. It was freeing up memory twice, 11789 leaking and not finding already revoked certificates. 11790 [Steve Henson] 11791 11792 *) Extensive changes to support certificate auxiliary information. 11793 This involves the use of X509_CERT_AUX structure and X509_AUX 11794 functions. An X509_AUX function such as PEM_read_X509_AUX() 11795 can still read in a certificate file in the usual way but it 11796 will also read in any additional "auxiliary information". By 11797 doing things this way a fair degree of compatibility can be 11798 retained: existing certificates can have this information added 11799 using the new 'x509' options. 11800 11801 Current auxiliary information includes an "alias" and some trust 11802 settings. The trust settings will ultimately be used in enhanced 11803 certificate chain verification routines: currently a certificate 11804 can only be trusted if it is self signed and then it is trusted 11805 for all purposes. 11806 [Steve Henson] 11807 11808 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 11809 The problem was that one of the replacement routines had not been working 11810 since SSLeay releases. For now the offending routine has been replaced 11811 with non-optimised assembler. Even so, this now gives around 95% 11812 performance improvement for 1024 bit RSA signs. 11813 [Mark Cox] 11814 11815 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 11816 handling. Most clients have the effective key size in bits equal to 11817 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 11818 A few however don't do this and instead use the size of the decrypted key 11819 to determine the RC2 key length and the AlgorithmIdentifier to determine 11820 the effective key length. In this case the effective key length can still 11821 be 40 bits but the key length can be 168 bits for example. This is fixed 11822 by manually forcing an RC2 key into the EVP_PKEY structure because the 11823 EVP code can't currently handle unusual RC2 key sizes: it always assumes 11824 the key length and effective key length are equal. 11825 [Steve Henson] 11826 11827 *) Add a bunch of functions that should simplify the creation of 11828 X509_NAME structures. Now you should be able to do: 11829 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 11830 and have it automatically work out the correct field type and fill in 11831 the structures. The more adventurous can try: 11832 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 11833 and it will (hopefully) work out the correct multibyte encoding. 11834 [Steve Henson] 11835 11836 *) Change the 'req' utility to use the new field handling and multibyte 11837 copy routines. Before the DN field creation was handled in an ad hoc 11838 way in req, ca, and x509 which was rather broken and didn't support 11839 BMPStrings or UTF8Strings. Since some software doesn't implement 11840 BMPStrings or UTF8Strings yet, they can be enabled using the config file 11841 using the dirstring_type option. See the new comment in the default 11842 openssl.cnf for more info. 11843 [Steve Henson] 11844 11845 *) Make crypto/rand/md_rand.c more robust: 11846 - Assure unique random numbers after fork(). 11847 - Make sure that concurrent threads access the global counter and 11848 md serializably so that we never lose entropy in them 11849 or use exactly the same state in multiple threads. 11850 Access to the large state is not always serializable because 11851 the additional locking could be a performance killer, and 11852 md should be large enough anyway. 11853 [Bodo Moeller] 11854 11855 *) New file apps/app_rand.c with commonly needed functionality 11856 for handling the random seed file. 11857 11858 Use the random seed file in some applications that previously did not: 11859 ca, 11860 dsaparam -genkey (which also ignored its '-rand' option), 11861 s_client, 11862 s_server, 11863 x509 (when signing). 11864 Except on systems with /dev/urandom, it is crucial to have a random 11865 seed file at least for key creation, DSA signing, and for DH exchanges; 11866 for RSA signatures we could do without one. 11867 11868 gendh and gendsa (unlike genrsa) used to read only the first byte 11869 of each file listed in the '-rand' option. The function as previously 11870 found in genrsa is now in app_rand.c and is used by all programs 11871 that support '-rand'. 11872 [Bodo Moeller] 11873 11874 *) In RAND_write_file, use mode 0600 for creating files; 11875 don't just chmod when it may be too late. 11876 [Bodo Moeller] 11877 11878 *) Report an error from X509_STORE_load_locations 11879 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 11880 [Bill Perry] 11881 11882 *) New function ASN1_mbstring_copy() this copies a string in either 11883 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 11884 into an ASN1_STRING type. A mask of permissible types is passed 11885 and it chooses the "minimal" type to use or an error if not type 11886 is suitable. 11887 [Steve Henson] 11888 11889 *) Add function equivalents to the various macros in asn1.h. The old 11890 macros are retained with an M_ prefix. Code inside the library can 11891 use the M_ macros. External code (including the openssl utility) 11892 should *NOT* in order to be "shared library friendly". 11893 [Steve Henson] 11894 11895 *) Add various functions that can check a certificate's extensions 11896 to see if it usable for various purposes such as SSL client, 11897 server or S/MIME and CAs of these types. This is currently 11898 VERY EXPERIMENTAL but will ultimately be used for certificate chain 11899 verification. Also added a -purpose flag to x509 utility to 11900 print out all the purposes. 11901 [Steve Henson] 11902 11903 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 11904 functions. 11905 [Steve Henson] 11906 11907 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 11908 for, obtain and decode and extension and obtain its critical flag. 11909 This allows all the necessary extension code to be handled in a 11910 single function call. 11911 [Steve Henson] 11912 11913 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 11914 platforms. See crypto/rc4/rc4_enc.c for further details. 11915 [Andy Polyakov] 11916 11917 *) New -noout option to asn1parse. This causes no output to be produced 11918 its main use is when combined with -strparse and -out to extract data 11919 from a file (which may not be in ASN.1 format). 11920 [Steve Henson] 11921 11922 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 11923 when producing the local key id. 11924 [Richard Levitte <levitte@stacken.kth.se>] 11925 11926 *) New option -dhparam in s_server. This allows a DH parameter file to be 11927 stated explicitly. If it is not stated then it tries the first server 11928 certificate file. The previous behaviour hard coded the filename 11929 "server.pem". 11930 [Steve Henson] 11931 11932 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 11933 a public key to be input or output. For example: 11934 openssl rsa -in key.pem -pubout -out pubkey.pem 11935 Also added necessary DSA public key functions to handle this. 11936 [Steve Henson] 11937 11938 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 11939 in the message. This was handled by allowing 11940 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 11941 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 11942 11943 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 11944 to the end of the strings whereas this didn't. This would cause problems 11945 if strings read with d2i_ASN1_bytes() were later modified. 11946 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 11947 11948 *) Fix for base64 decode bug. When a base64 bio reads only one line of 11949 data and it contains EOF it will end up returning an error. This is 11950 caused by input 46 bytes long. The cause is due to the way base64 11951 BIOs find the start of base64 encoded data. They do this by trying a 11952 trial decode on each line until they find one that works. When they 11953 do a flag is set and it starts again knowing it can pass all the 11954 data directly through the decoder. Unfortunately it doesn't reset 11955 the context it uses. This means that if EOF is reached an attempt 11956 is made to pass two EOFs through the context and this causes the 11957 resulting error. This can also cause other problems as well. As is 11958 usual with these problems it takes *ages* to find and the fix is 11959 trivial: move one line. 11960 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 11961 11962 *) Ugly workaround to get s_client and s_server working under Windows. The 11963 old code wouldn't work because it needed to select() on sockets and the 11964 tty (for keypresses and to see if data could be written). Win32 only 11965 supports select() on sockets so we select() with a 1s timeout on the 11966 sockets and then see if any characters are waiting to be read, if none 11967 are present then we retry, we also assume we can always write data to 11968 the tty. This isn't nice because the code then blocks until we've 11969 received a complete line of data and it is effectively polling the 11970 keyboard at 1s intervals: however it's quite a bit better than not 11971 working at all :-) A dedicated Windows application might handle this 11972 with an event loop for example. 11973 [Steve Henson] 11974 11975 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 11976 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 11977 will be called when RSA_sign() and RSA_verify() are used. This is useful 11978 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 11979 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 11980 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 11981 This necessitated the support of an extra signature type NID_md5_sha1 11982 for SSL signatures and modifications to the SSL library to use it instead 11983 of calling RSA_public_decrypt() and RSA_private_encrypt(). 11984 [Steve Henson] 11985 11986 *) Add new -verify -CAfile and -CApath options to the crl program, these 11987 will lookup a CRL issuers certificate and verify the signature in a 11988 similar way to the verify program. Tidy up the crl program so it 11989 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 11990 less strict. It will now permit CRL extensions even if it is not 11991 a V2 CRL: this will allow it to tolerate some broken CRLs. 11992 [Steve Henson] 11993 11994 *) Initialize all non-automatic variables each time one of the openssl 11995 sub-programs is started (this is necessary as they may be started 11996 multiple times from the "OpenSSL>" prompt). 11997 [Lennart Bang, Bodo Moeller] 11998 11999 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 12000 removing all other RSA functionality (this is what NO_RSA does). This 12001 is so (for example) those in the US can disable those operations covered 12002 by the RSA patent while allowing storage and parsing of RSA keys and RSA 12003 key generation. 12004 [Steve Henson] 12005 12006 *) Non-copying interface to BIO pairs. 12007 (still largely untested) 12008 [Bodo Moeller] 12009 12010 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 12011 ASCII string. This was handled independently in various places before. 12012 [Steve Henson] 12013 12014 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 12015 UTF8 strings a character at a time. 12016 [Steve Henson] 12017 12018 *) Use client_version from client hello to select the protocol 12019 (s23_srvr.c) and for RSA client key exchange verification 12020 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 12021 [Bodo Moeller] 12022 12023 *) Add various utility functions to handle SPKACs, these were previously 12024 handled by poking round in the structure internals. Added new function 12025 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 12026 print, verify and generate SPKACs. Based on an original idea from 12027 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 12028 [Steve Henson] 12029 12030 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 12031 [Andy Polyakov] 12032 12033 *) Allow the config file extension section to be overwritten on the 12034 command line. Based on an original idea from Massimiliano Pala 12035 <madwolf@comune.modena.it>. The new option is called -extensions 12036 and can be applied to ca, req and x509. Also -reqexts to override 12037 the request extensions in req and -crlexts to override the crl extensions 12038 in ca. 12039 [Steve Henson] 12040 12041 *) Add new feature to the SPKAC handling in ca. Now you can include 12042 the same field multiple times by preceding it by "XXXX." for example: 12043 1.OU="Unit name 1" 12044 2.OU="Unit name 2" 12045 this is the same syntax as used in the req config file. 12046 [Steve Henson] 12047 12048 *) Allow certificate extensions to be added to certificate requests. These 12049 are specified in a 'req_extensions' option of the req section of the 12050 config file. They can be printed out with the -text option to req but 12051 are otherwise ignored at present. 12052 [Steve Henson] 12053 12054 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 12055 data read consists of only the final block it would not decrypted because 12056 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 12057 A misplaced 'break' also meant the decrypted final block might not be 12058 copied until the next read. 12059 [Steve Henson] 12060 12061 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 12062 a few extra parameters to the DH structure: these will be useful if 12063 for example we want the value of 'q' or implement X9.42 DH. 12064 [Steve Henson] 12065 12066 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 12067 provides hooks that allow the default DSA functions or functions on a 12068 "per key" basis to be replaced. This allows hardware acceleration and 12069 hardware key storage to be handled without major modification to the 12070 library. Also added low level modexp hooks and CRYPTO_EX structure and 12071 associated functions. 12072 [Steve Henson] 12073 12074 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 12075 as "read only": it can't be written to and the buffer it points to will 12076 not be freed. Reading from a read only BIO is much more efficient than 12077 a normal memory BIO. This was added because there are several times when 12078 an area of memory needs to be read from a BIO. The previous method was 12079 to create a memory BIO and write the data to it, this results in two 12080 copies of the data and an O(n^2) reading algorithm. There is a new 12081 function BIO_new_mem_buf() which creates a read only memory BIO from 12082 an area of memory. Also modified the PKCS#7 routines to use read only 12083 memory BIOs. 12084 [Steve Henson] 12085 12086 *) Bugfix: ssl23_get_client_hello did not work properly when called in 12087 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 12088 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 12089 but a retry condition occurred while trying to read the rest. 12090 [Bodo Moeller] 12091 12092 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 12093 NID_pkcs7_encrypted by default: this was wrong since this should almost 12094 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 12095 the encrypted data type: this is a more sensible place to put it and it 12096 allows the PKCS#12 code to be tidied up that duplicated this 12097 functionality. 12098 [Steve Henson] 12099 12100 *) Changed obj_dat.pl script so it takes its input and output files on 12101 the command line. This should avoid shell escape redirection problems 12102 under Win32. 12103 [Steve Henson] 12104 12105 *) Initial support for certificate extension requests, these are included 12106 in things like Xenroll certificate requests. Included functions to allow 12107 extensions to be obtained and added. 12108 [Steve Henson] 12109 12110 *) -crlf option to s_client and s_server for sending newlines as 12111 CRLF (as required by many protocols). 12112 [Bodo Moeller] 12113 12114 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 12115 12116 *) Install libRSAglue.a when OpenSSL is built with RSAref. 12117 [Ralf S. Engelschall] 12118 12119 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 12120 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 12121 12122 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 12123 program. 12124 [Steve Henson] 12125 12126 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 12127 DH parameters/keys (q is lost during that conversion, but the resulting 12128 DH parameters contain its length). 12129 12130 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 12131 much faster than DH_generate_parameters (which creates parameters 12132 where p = 2*q + 1), and also the smaller q makes DH computations 12133 much more efficient (160-bit exponentiation instead of 1024-bit 12134 exponentiation); so this provides a convenient way to support DHE 12135 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 12136 utter importance to use 12137 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 12138 or 12139 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 12140 when such DH parameters are used, because otherwise small subgroup 12141 attacks may become possible! 12142 [Bodo Moeller] 12143 12144 *) Avoid memory leak in i2d_DHparams. 12145 [Bodo Moeller] 12146 12147 *) Allow the -k option to be used more than once in the enc program: 12148 this allows the same encrypted message to be read by multiple recipients. 12149 [Steve Henson] 12150 12151 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 12152 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 12153 it will always use the numerical form of the OID, even if it has a short 12154 or long name. 12155 [Steve Henson] 12156 12157 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 12158 method only got called if p,q,dmp1,dmq1,iqmp components were present, 12159 otherwise bn_mod_exp was called. In the case of hardware keys for example 12160 no private key components need be present and it might store extra data 12161 in the RSA structure, which cannot be accessed from bn_mod_exp. 12162 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 12163 private key operations. 12164 [Steve Henson] 12165 12166 *) Added support for SPARC Linux. 12167 [Andy Polyakov] 12168 12169 *) pem_password_cb function type incompatibly changed from 12170 typedef int pem_password_cb(char *buf, int size, int rwflag); 12171 to 12172 ....(char *buf, int size, int rwflag, void *userdata); 12173 so that applications can pass data to their callbacks: 12174 The PEM[_ASN1]_{read,write}... functions and macros now take an 12175 additional void * argument, which is just handed through whenever 12176 the password callback is called. 12177 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 12178 12179 New function SSL_CTX_set_default_passwd_cb_userdata. 12180 12181 Compatibility note: As many C implementations push function arguments 12182 onto the stack in reverse order, the new library version is likely to 12183 interoperate with programs that have been compiled with the old 12184 pem_password_cb definition (PEM_whatever takes some data that 12185 happens to be on the stack as its last argument, and the callback 12186 just ignores this garbage); but there is no guarantee whatsoever that 12187 this will work. 12188 12189 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 12190 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 12191 problems not only on Windows, but also on some Unix platforms. 12192 To avoid problematic command lines, these definitions are now in an 12193 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 12194 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 12195 [Bodo Moeller] 12196 12197 *) MIPS III/IV assembler module is reimplemented. 12198 [Andy Polyakov] 12199 12200 *) More DES library cleanups: remove references to srand/rand and 12201 delete an unused file. 12202 [Ulf Möller] 12203 12204 *) Add support for the free Netwide assembler (NASM) under Win32, 12205 since not many people have MASM (ml) and it can be hard to obtain. 12206 This is currently experimental but it seems to work OK and pass all 12207 the tests. Check out INSTALL.W32 for info. 12208 [Steve Henson] 12209 12210 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 12211 without temporary keys kept an extra copy of the server key, 12212 and connections with temporary keys did not free everything in case 12213 of an error. 12214 [Bodo Moeller] 12215 12216 *) New function RSA_check_key and new openssl rsa option -check 12217 for verifying the consistency of RSA keys. 12218 [Ulf Moeller, Bodo Moeller] 12219 12220 *) Various changes to make Win32 compile work: 12221 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 12222 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 12223 comparison" warnings. 12224 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 12225 [Steve Henson] 12226 12227 *) Add a debugging option to PKCS#5 v2 key generation function: when 12228 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 12229 derived keys are printed to stderr. 12230 [Steve Henson] 12231 12232 *) Copy the flags in ASN1_STRING_dup(). 12233 [Roman E. Pavlov <pre@mo.msk.ru>] 12234 12235 *) The x509 application mishandled signing requests containing DSA 12236 keys when the signing key was also DSA and the parameters didn't match. 12237 12238 It was supposed to omit the parameters when they matched the signing key: 12239 the verifying software was then supposed to automatically use the CA's 12240 parameters if they were absent from the end user certificate. 12241 12242 Omitting parameters is no longer recommended. The test was also 12243 the wrong way round! This was probably due to unusual behaviour in 12244 EVP_cmp_parameters() which returns 1 if the parameters match. 12245 This meant that parameters were omitted when they *didn't* match and 12246 the certificate was useless. Certificates signed with 'ca' didn't have 12247 this bug. 12248 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 12249 12250 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 12251 The interface is as follows: 12252 Applications can use 12253 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 12254 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 12255 "off" is now the default. 12256 The library internally uses 12257 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 12258 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 12259 to disable memory-checking temporarily. 12260 12261 Some inconsistent states that previously were possible (and were 12262 even the default) are now avoided. 12263 12264 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 12265 with each memory chunk allocated; this is occasionally more helpful 12266 than just having a counter. 12267 12268 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 12269 12270 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 12271 extensions. 12272 [Bodo Moeller] 12273 12274 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 12275 which largely parallels "options", but is for changing API behaviour, 12276 whereas "options" are about protocol behaviour. 12277 Initial "mode" flags are: 12278 12279 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 12280 a single record has been written. 12281 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 12282 retries use the same buffer location. 12283 (But all of the contents must be 12284 copied!) 12285 [Bodo Moeller] 12286 12287 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 12288 worked. 12289 12290 *) Fix problems with no-hmac etc. 12291 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>] 12292 12293 *) New functions RSA_get_default_method(), RSA_set_method() and 12294 RSA_get_method(). These allows replacement of RSA_METHODs without having 12295 to mess around with the internals of an RSA structure. 12296 [Steve Henson] 12297 12298 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 12299 Also really enable memory leak checks in openssl.c and in some 12300 test programs. 12301 [Chad C. Mulligan, Bodo Moeller] 12302 12303 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 12304 up the length of negative integers. This has now been simplified to just 12305 store the length when it is first determined and use it later, rather 12306 than trying to keep track of where data is copied and updating it to 12307 point to the end. 12308 [Steve Henson, reported by Brien Wheeler 12309 <bwheeler@authentica-security.com>] 12310 12311 *) Add a new function PKCS7_signatureVerify. This allows the verification 12312 of a PKCS#7 signature but with the signing certificate passed to the 12313 function itself. This contrasts with PKCS7_dataVerify which assumes the 12314 certificate is present in the PKCS#7 structure. This isn't always the 12315 case: certificates can be omitted from a PKCS#7 structure and be 12316 distributed by "out of band" means (such as a certificate database). 12317 [Steve Henson] 12318 12319 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 12320 function prototypes in pem.h, also change util/mkdef.pl to add the 12321 necessary function names. 12322 [Steve Henson] 12323 12324 *) mk1mf.pl (used by Windows builds) did not properly read the 12325 options set by Configure in the top level Makefile, and Configure 12326 was not even able to write more than one option correctly. 12327 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 12328 [Bodo Moeller] 12329 12330 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 12331 file to be loaded from a BIO or FILE pointer. The BIO version will 12332 for example allow memory BIOs to contain config info. 12333 [Steve Henson] 12334 12335 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 12336 Whoever hopes to achieve shared-library compatibility across versions 12337 must use this, not the compile-time macro. 12338 (Exercise 0.9.4: Which is the minimum library version required by 12339 such programs?) 12340 Note: All this applies only to multi-threaded programs, others don't 12341 need locks. 12342 [Bodo Moeller] 12343 12344 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 12345 through a BIO pair triggered the default case, i.e. 12346 SSLerr(...,SSL_R_UNKNOWN_STATE). 12347 [Bodo Moeller] 12348 12349 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 12350 can use the SSL library even if none of the specific BIOs is 12351 appropriate. 12352 [Bodo Moeller] 12353 12354 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 12355 for the encoded length. 12356 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 12357 12358 *) Add initial documentation of the X509V3 functions. 12359 [Steve Henson] 12360 12361 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 12362 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 12363 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 12364 secure PKCS#8 private key format with a high iteration count. 12365 [Steve Henson] 12366 12367 *) Fix determination of Perl interpreter: A perl or perl5 12368 _directory_ in $PATH was also accepted as the interpreter. 12369 [Ralf S. Engelschall] 12370 12371 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 12372 wrong with it but it was very old and did things like calling 12373 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 12374 unusual formatting. 12375 [Steve Henson] 12376 12377 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 12378 to use the new extension code. 12379 [Steve Henson] 12380 12381 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 12382 with macros. This should make it easier to change their form, add extra 12383 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 12384 constant. 12385 [Steve Henson] 12386 12387 *) Add to configuration table a new entry that can specify an alternative 12388 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 12389 according to Mark Crispin <MRC@Panda.COM>. 12390 [Bodo Moeller] 12391 12392#if 0 12393 *) DES CBC did not update the IV. Weird. 12394 [Ben Laurie] 12395#else 12396 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 12397 Changing the behaviour of the former might break existing programs -- 12398 where IV updating is needed, des_ncbc_encrypt can be used. 12399#endif 12400 12401 *) When bntest is run from "make test" it drives bc to check its 12402 calculations, as well as internally checking them. If an internal check 12403 fails, it needs to cause bc to give a non-zero result or make test carries 12404 on without noticing the failure. Fixed. 12405 [Ben Laurie] 12406 12407 *) DES library cleanups. 12408 [Ulf Möller] 12409 12410 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 12411 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 12412 ciphers. NOTE: although the key derivation function has been verified 12413 against some published test vectors it has not been extensively tested 12414 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 12415 of v2.0. 12416 [Steve Henson] 12417 12418 *) Instead of "mkdir -p", which is not fully portable, use new 12419 Perl script "util/mkdir-p.pl". 12420 [Bodo Moeller] 12421 12422 *) Rewrite the way password based encryption (PBE) is handled. It used to 12423 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 12424 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 12425 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 12426 the 'parameter' field of the AlgorithmIdentifier is passed to the 12427 underlying key generation function so it must do its own ASN1 parsing. 12428 This has also changed the EVP_PBE_CipherInit() function which now has a 12429 'parameter' argument instead of literal salt and iteration count values 12430 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 12431 [Steve Henson] 12432 12433 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 12434 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 12435 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 12436 KEY" because this clashed with PKCS#8 unencrypted string. Since this 12437 value was just used as a "magic string" and not used directly its 12438 value doesn't matter. 12439 [Steve Henson] 12440 12441 *) Introduce some semblance of const correctness to BN. Shame C doesn't 12442 support mutable. 12443 [Ben Laurie] 12444 12445 *) "linux-sparc64" configuration (ultrapenguin). 12446 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 12447 "linux-sparc" configuration. 12448 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 12449 12450 *) config now generates no-xxx options for missing ciphers. 12451 [Ulf Möller] 12452 12453 *) Support the EBCDIC character set (work in progress). 12454 File ebcdic.c not yet included because it has a different license. 12455 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 12456 12457 *) Support BS2000/OSD-POSIX. 12458 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 12459 12460 *) Make callbacks for key generation use void * instead of char *. 12461 [Ben Laurie] 12462 12463 *) Make S/MIME samples compile (not yet tested). 12464 [Ben Laurie] 12465 12466 *) Additional typesafe stacks. 12467 [Ben Laurie] 12468 12469 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 12470 [Bodo Moeller] 12471 12472 12473 Changes between 0.9.3 and 0.9.3a [29 May 1999] 12474 12475 *) New configuration variant "sco5-gcc". 12476 12477 *) Updated some demos. 12478 [Sean O Riordain, Wade Scholine] 12479 12480 *) Add missing BIO_free at exit of pkcs12 application. 12481 [Wu Zhigang] 12482 12483 *) Fix memory leak in conf.c. 12484 [Steve Henson] 12485 12486 *) Updates for Win32 to assembler version of MD5. 12487 [Steve Henson] 12488 12489 *) Set #! path to perl in apps/der_chop to where we found it 12490 instead of using a fixed path. 12491 [Bodo Moeller] 12492 12493 *) SHA library changes for irix64-mips4-cc. 12494 [Andy Polyakov] 12495 12496 *) Improvements for VMS support. 12497 [Richard Levitte] 12498 12499 12500 Changes between 0.9.2b and 0.9.3 [24 May 1999] 12501 12502 *) Bignum library bug fix. IRIX 6 passes "make test" now! 12503 This also avoids the problems with SC4.2 and unpatched SC5. 12504 [Andy Polyakov <appro@fy.chalmers.se>] 12505 12506 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 12507 These are required because of the typesafe stack would otherwise break 12508 existing code. If old code used a structure member which used to be STACK 12509 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 12510 sk_num or sk_value it would produce an error because the num, data members 12511 are not present in STACK_OF. Now it just produces a warning. sk_set 12512 replaces the old method of assigning a value to sk_value 12513 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 12514 that does this will no longer work (and should use sk_set instead) but 12515 this could be regarded as a "questionable" behaviour anyway. 12516 [Steve Henson] 12517 12518 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 12519 correctly handle encrypted S/MIME data. 12520 [Steve Henson] 12521 12522 *) Change type of various DES function arguments from des_cblock 12523 (which means, in function argument declarations, pointer to char) 12524 to des_cblock * (meaning pointer to array with 8 char elements), 12525 which allows the compiler to do more typechecking; it was like 12526 that back in SSLeay, but with lots of ugly casts. 12527 12528 Introduce new type const_des_cblock. 12529 [Bodo Moeller] 12530 12531 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 12532 problems: find RecipientInfo structure that matches recipient certificate 12533 and initialise the ASN1 structures properly based on passed cipher. 12534 [Steve Henson] 12535 12536 *) Belatedly make the BN tests actually check the results. 12537 [Ben Laurie] 12538 12539 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 12540 to and from BNs: it was completely broken. New compilation option 12541 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 12542 key elements as negative integers. 12543 [Steve Henson] 12544 12545 *) Reorganize and speed up MD5. 12546 [Andy Polyakov <appro@fy.chalmers.se>] 12547 12548 *) VMS support. 12549 [Richard Levitte <richard@levitte.org>] 12550 12551 *) New option -out to asn1parse to allow the parsed structure to be 12552 output to a file. This is most useful when combined with the -strparse 12553 option to examine the output of things like OCTET STRINGS. 12554 [Steve Henson] 12555 12556 *) Make SSL library a little more fool-proof by not requiring any longer 12557 that SSL_set_{accept,connect}_state be called before 12558 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 12559 in many applications because usually everything *appeared* to work as 12560 intended anyway -- now it really works as intended). 12561 [Bodo Moeller] 12562 12563 *) Move openssl.cnf out of lib/. 12564 [Ulf Möller] 12565 12566 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 12567 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 12568 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 12569 [Ralf S. Engelschall] 12570 12571 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 12572 handle PKCS#7 enveloped data properly. 12573 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 12574 12575 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 12576 copying pointers. The cert_st handling is changed by this in 12577 various ways (and thus what used to be known as ctx->default_cert 12578 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 12579 any longer when s->cert does not give us what we need). 12580 ssl_cert_instantiate becomes obsolete by this change. 12581 As soon as we've got the new code right (possibly it already is?), 12582 we have solved a couple of bugs of the earlier code where s->cert 12583 was used as if it could not have been shared with other SSL structures. 12584 12585 Note that using the SSL API in certain dirty ways now will result 12586 in different behaviour than observed with earlier library versions: 12587 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 12588 does not influence s as it used to. 12589 12590 In order to clean up things more thoroughly, inside SSL_SESSION 12591 we don't use CERT any longer, but a new structure SESS_CERT 12592 that holds per-session data (if available); currently, this is 12593 the peer's certificate chain and, for clients, the server's certificate 12594 and temporary key. CERT holds only those values that can have 12595 meaningful defaults in an SSL_CTX. 12596 [Bodo Moeller] 12597 12598 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 12599 from the internal representation. Various PKCS#7 fixes: remove some 12600 evil casts and set the enc_dig_alg field properly based on the signing 12601 key type. 12602 [Steve Henson] 12603 12604 *) Allow PKCS#12 password to be set from the command line or the 12605 environment. Let 'ca' get its config file name from the environment 12606 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 12607 and 'x509'). 12608 [Steve Henson] 12609 12610 *) Allow certificate policies extension to use an IA5STRING for the 12611 organization field. This is contrary to the PKIX definition but 12612 VeriSign uses it and IE5 only recognises this form. Document 'x509' 12613 extension option. 12614 [Steve Henson] 12615 12616 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 12617 without disallowing inline assembler and the like for non-pedantic builds. 12618 [Ben Laurie] 12619 12620 *) Support Borland C++ builder. 12621 [Janez Jere <jj@void.si>, modified by Ulf Möller] 12622 12623 *) Support Mingw32. 12624 [Ulf Möller] 12625 12626 *) SHA-1 cleanups and performance enhancements. 12627 [Andy Polyakov <appro@fy.chalmers.se>] 12628 12629 *) Sparc v8plus assembler for the bignum library. 12630 [Andy Polyakov <appro@fy.chalmers.se>] 12631 12632 *) Accept any -xxx and +xxx compiler options in Configure. 12633 [Ulf Möller] 12634 12635 *) Update HPUX configuration. 12636 [Anonymous] 12637 12638 *) Add missing sk_<type>_unshift() function to safestack.h 12639 [Ralf S. Engelschall] 12640 12641 *) New function SSL_CTX_use_certificate_chain_file that sets the 12642 "extra_cert"s in addition to the certificate. (This makes sense 12643 only for "PEM" format files, as chains as a whole are not 12644 DER-encoded.) 12645 [Bodo Moeller] 12646 12647 *) Support verify_depth from the SSL API. 12648 x509_vfy.c had what can be considered an off-by-one-error: 12649 Its depth (which was not part of the external interface) 12650 was actually counting the number of certificates in a chain; 12651 now it really counts the depth. 12652 [Bodo Moeller] 12653 12654 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 12655 instead of X509err, which often resulted in confusing error 12656 messages since the error codes are not globally unique 12657 (e.g. an alleged error in ssl3_accept when a certificate 12658 didn't match the private key). 12659 12660 *) New function SSL_CTX_set_session_id_context that allows to set a default 12661 value (so that you don't need SSL_set_session_id_context for each 12662 connection using the SSL_CTX). 12663 [Bodo Moeller] 12664 12665 *) OAEP decoding bug fix. 12666 [Ulf Möller] 12667 12668 *) Support INSTALL_PREFIX for package builders, as proposed by 12669 David Harris. 12670 [Bodo Moeller] 12671 12672 *) New Configure options "threads" and "no-threads". For systems 12673 where the proper compiler options are known (currently Solaris 12674 and Linux), "threads" is the default. 12675 [Bodo Moeller] 12676 12677 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 12678 [Bodo Moeller] 12679 12680 *) Install various scripts to $(OPENSSLDIR)/misc, not to 12681 $(INSTALLTOP)/bin -- they shouldn't clutter directories 12682 such as /usr/local/bin. 12683 [Bodo Moeller] 12684 12685 *) "make linux-shared" to build shared libraries. 12686 [Niels Poppe <niels@netbox.org>] 12687 12688 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 12689 [Ulf Möller] 12690 12691 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 12692 extension adding in x509 utility. 12693 [Steve Henson] 12694 12695 *) Remove NOPROTO sections and error code comments. 12696 [Ulf Möller] 12697 12698 *) Partial rewrite of the DEF file generator to now parse the ANSI 12699 prototypes. 12700 [Steve Henson] 12701 12702 *) New Configure options --prefix=DIR and --openssldir=DIR. 12703 [Ulf Möller] 12704 12705 *) Complete rewrite of the error code script(s). It is all now handled 12706 by one script at the top level which handles error code gathering, 12707 header rewriting and C source file generation. It should be much better 12708 than the old method: it now uses a modified version of Ulf's parser to 12709 read the ANSI prototypes in all header files (thus the old K&R definitions 12710 aren't needed for error creation any more) and do a better job of 12711 translating function codes into names. The old 'ASN1 error code embedded 12712 in a comment' is no longer necessary and it doesn't use .err files which 12713 have now been deleted. Also the error code call doesn't have to appear all 12714 on one line (which resulted in some large lines...). 12715 [Steve Henson] 12716 12717 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 12718 [Bodo Moeller] 12719 12720 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 12721 0 (which usually indicates a closed connection), but continue reading. 12722 [Bodo Moeller] 12723 12724 *) Fix some race conditions. 12725 [Bodo Moeller] 12726 12727 *) Add support for CRL distribution points extension. Add Certificate 12728 Policies and CRL distribution points documentation. 12729 [Steve Henson] 12730 12731 *) Move the autogenerated header file parts to crypto/opensslconf.h. 12732 [Ulf Möller] 12733 12734 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 12735 8 of keying material. Merlin has also confirmed interop with this fix 12736 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 12737 [Merlin Hughes <merlin@baltimore.ie>] 12738 12739 *) Fix lots of warnings. 12740 [Richard Levitte <levitte@stacken.kth.se>] 12741 12742 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 12743 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 12744 [Richard Levitte <levitte@stacken.kth.se>] 12745 12746 *) Fix problems with sizeof(long) == 8. 12747 [Andy Polyakov <appro@fy.chalmers.se>] 12748 12749 *) Change functions to ANSI C. 12750 [Ulf Möller] 12751 12752 *) Fix typos in error codes. 12753 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller] 12754 12755 *) Remove defunct assembler files from Configure. 12756 [Ulf Möller] 12757 12758 *) SPARC v8 assembler BIGNUM implementation. 12759 [Andy Polyakov <appro@fy.chalmers.se>] 12760 12761 *) Support for Certificate Policies extension: both print and set. 12762 Various additions to support the r2i method this uses. 12763 [Steve Henson] 12764 12765 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 12766 return a const string when you are expecting an allocated buffer. 12767 [Ben Laurie] 12768 12769 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 12770 types DirectoryString and DisplayText. 12771 [Steve Henson] 12772 12773 *) Add code to allow r2i extensions to access the configuration database, 12774 add an LHASH database driver and add several ctx helper functions. 12775 [Steve Henson] 12776 12777 *) Fix an evil bug in bn_expand2() which caused various BN functions to 12778 fail when they extended the size of a BIGNUM. 12779 [Steve Henson] 12780 12781 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 12782 support typesafe stack. 12783 [Steve Henson] 12784 12785 *) Fix typo in SSL_[gs]et_options(). 12786 [Nils Frostberg <nils@medcom.se>] 12787 12788 *) Delete various functions and files that belonged to the (now obsolete) 12789 old X509V3 handling code. 12790 [Steve Henson] 12791 12792 *) New Configure option "rsaref". 12793 [Ulf Möller] 12794 12795 *) Don't auto-generate pem.h. 12796 [Bodo Moeller] 12797 12798 *) Introduce type-safe ASN.1 SETs. 12799 [Ben Laurie] 12800 12801 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 12802 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 12803 12804 *) Introduce type-safe STACKs. This will almost certainly break lots of code 12805 that links with OpenSSL (well at least cause lots of warnings), but fear 12806 not: the conversion is trivial, and it eliminates loads of evil casts. A 12807 few STACKed things have been converted already. Feel free to convert more. 12808 In the fullness of time, I'll do away with the STACK type altogether. 12809 [Ben Laurie] 12810 12811 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 12812 specified in <certfile> by updating the entry in the index.txt file. 12813 This way one no longer has to edit the index.txt file manually for 12814 revoking a certificate. The -revoke option does the gory details now. 12815 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 12816 12817 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 12818 `-text' option at all and this way the `-noout -text' combination was 12819 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 12820 [Ralf S. Engelschall] 12821 12822 *) Make sure a corresponding plain text error message exists for the 12823 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 12824 verify callback function determined that a certificate was revoked. 12825 [Ralf S. Engelschall] 12826 12827 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 12828 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 12829 all available ciphers including rc5, which was forgotten until now. 12830 In order to let the testing shell script know which algorithms 12831 are available, a new (up to now undocumented) command 12832 "openssl list-cipher-commands" is used. 12833 [Bodo Moeller] 12834 12835 *) Bugfix: s_client occasionally would sleep in select() when 12836 it should have checked SSL_pending() first. 12837 [Bodo Moeller] 12838 12839 *) New functions DSA_do_sign and DSA_do_verify to provide access to 12840 the raw DSA values prior to ASN.1 encoding. 12841 [Ulf Möller] 12842 12843 *) Tweaks to Configure 12844 [Niels Poppe <niels@netbox.org>] 12845 12846 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 12847 yet... 12848 [Steve Henson] 12849 12850 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 12851 [Ulf Möller] 12852 12853 *) New config option to avoid instructions that are illegal on the 80386. 12854 The default code is faster, but requires at least a 486. 12855 [Ulf Möller] 12856 12857 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 12858 SSL2_SERVER_VERSION (not used at all) macros, which are now the 12859 same as SSL2_VERSION anyway. 12860 [Bodo Moeller] 12861 12862 *) New "-showcerts" option for s_client. 12863 [Bodo Moeller] 12864 12865 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 12866 application. Various cleanups and fixes. 12867 [Steve Henson] 12868 12869 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 12870 modify error routines to work internally. Add error codes and PBE init 12871 to library startup routines. 12872 [Steve Henson] 12873 12874 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 12875 packing functions to asn1 and evp. Changed function names and error 12876 codes along the way. 12877 [Steve Henson] 12878 12879 *) PKCS12 integration: and so it begins... First of several patches to 12880 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 12881 objects to objects.h 12882 [Steve Henson] 12883 12884 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 12885 and display support for Thawte strong extranet extension. 12886 [Steve Henson] 12887 12888 *) Add LinuxPPC support. 12889 [Jeff Dubrule <igor@pobox.org>] 12890 12891 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 12892 bn_div_words in alpha.s. 12893 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 12894 12895 *) Make sure the RSA OAEP test is skipped under -DRSAref because 12896 OAEP isn't supported when OpenSSL is built with RSAref. 12897 [Ulf Moeller <ulf@fitug.de>] 12898 12899 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 12900 so they no longer are missing under -DNOPROTO. 12901 [Soren S. Jorvang <soren@t.dk>] 12902 12903 12904 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 12905 12906 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 12907 doesn't work when the session is reused. Coming soon! 12908 [Ben Laurie] 12909 12910 *) Fix a security hole, that allows sessions to be reused in the wrong 12911 context thus bypassing client cert protection! All software that uses 12912 client certs and session caches in multiple contexts NEEDS PATCHING to 12913 allow session reuse! A fuller solution is in the works. 12914 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 12915 12916 *) Some more source tree cleanups (removed obsolete files 12917 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 12918 permission on "config" script to be executable) and a fix for the INSTALL 12919 document. 12920 [Ulf Moeller <ulf@fitug.de>] 12921 12922 *) Remove some legacy and erroneous uses of malloc, free instead of 12923 Malloc, Free. 12924 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 12925 12926 *) Make rsa_oaep_test return non-zero on error. 12927 [Ulf Moeller <ulf@fitug.de>] 12928 12929 *) Add support for native Solaris shared libraries. Configure 12930 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 12931 if someone would make that last step automatic. 12932 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 12933 12934 *) ctx_size was not built with the right compiler during "make links". Fixed. 12935 [Ben Laurie] 12936 12937 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 12938 except NULL ciphers". This means the default cipher list will no longer 12939 enable NULL ciphers. They need to be specifically enabled e.g. with 12940 the string "DEFAULT:eNULL". 12941 [Steve Henson] 12942 12943 *) Fix to RSA private encryption routines: if p < q then it would 12944 occasionally produce an invalid result. This will only happen with 12945 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 12946 [Steve Henson] 12947 12948 *) Be less restrictive and allow also `perl util/perlpath.pl 12949 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 12950 because this way one can also use an interpreter named `perl5' (which is 12951 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 12952 installed as `perl'). 12953 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 12954 12955 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 12956 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 12957 12958 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 12959 advapi32.lib to Win32 build and change the pem test comparison 12960 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 12961 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 12962 and crypto/des/ede_cbcm_enc.c. 12963 [Steve Henson] 12964 12965 *) DES quad checksum was broken on big-endian architectures. Fixed. 12966 [Ben Laurie] 12967 12968 *) Comment out two functions in bio.h that aren't implemented. Fix up the 12969 Win32 test batch file so it (might) work again. The Win32 test batch file 12970 is horrible: I feel ill.... 12971 [Steve Henson] 12972 12973 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 12974 in e_os.h. Audit of header files to check ANSI and non ANSI 12975 sections: 10 functions were absent from non ANSI section and not exported 12976 from Windows DLLs. Fixed up libeay.num for new functions. 12977 [Steve Henson] 12978 12979 *) Make `openssl version' output lines consistent. 12980 [Ralf S. Engelschall] 12981 12982 *) Fix Win32 symbol export lists for BIO functions: Added 12983 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 12984 to ms/libeay{16,32}.def. 12985 [Ralf S. Engelschall] 12986 12987 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 12988 fine under Unix and passes some trivial tests I've now added. But the 12989 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 12990 added to make sure no one expects that this stuff really works in the 12991 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 12992 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 12993 openssl_bio.xs. 12994 [Ralf S. Engelschall] 12995 12996 *) Fix the generation of two part addresses in perl. 12997 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 12998 12999 *) Add config entry for Linux on MIPS. 13000 [John Tobey <jtobey@channel1.com>] 13001 13002 *) Make links whenever Configure is run, unless we are on Windoze. 13003 [Ben Laurie] 13004 13005 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 13006 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 13007 in CRLs. 13008 [Steve Henson] 13009 13010 *) Add a useful kludge to allow package maintainers to specify compiler and 13011 other platforms details on the command line without having to patch the 13012 Configure script every time: One now can use ``perl Configure 13013 <id>:<details>'', i.e. platform ids are allowed to have details appended 13014 to them (separated by colons). This is treated as there would be a static 13015 pre-configured entry in Configure's %table under key <id> with value 13016 <details> and ``perl Configure <id>'' is called. So, when you want to 13017 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 13018 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 13019 now, which overrides the FreeBSD-elf entry on-the-fly. 13020 [Ralf S. Engelschall] 13021 13022 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 13023 [Ben Laurie] 13024 13025 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 13026 on the `perl Configure ...' command line. This way one can compile 13027 OpenSSL libraries with Position Independent Code (PIC) which is needed 13028 for linking it into DSOs. 13029 [Ralf S. Engelschall] 13030 13031 *) Remarkably, export ciphers were totally broken and no-one had noticed! 13032 Fixed. 13033 [Ben Laurie] 13034 13035 *) Cleaned up the LICENSE document: The official contact for any license 13036 questions now is the OpenSSL core team under openssl-core@openssl.org. 13037 And add a paragraph about the dual-license situation to make sure people 13038 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 13039 to the OpenSSL toolkit. 13040 [Ralf S. Engelschall] 13041 13042 *) General source tree makefile cleanups: Made `making xxx in yyy...' 13043 display consistent in the source tree and replaced `/bin/rm' by `rm'. 13044 Additionally cleaned up the `make links' target: Remove unnecessary 13045 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 13046 to speed processing and no longer clutter the display with confusing 13047 stuff. Instead only the actually done links are displayed. 13048 [Ralf S. Engelschall] 13049 13050 *) Permit null encryption ciphersuites, used for authentication only. It used 13051 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 13052 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 13053 encryption. 13054 [Ben Laurie] 13055 13056 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 13057 signed attributes when verifying signatures (this would break them), 13058 the detached data encoding was wrong and public keys obtained using 13059 X509_get_pubkey() weren't freed. 13060 [Steve Henson] 13061 13062 *) Add text documentation for the BUFFER functions. Also added a work around 13063 to a Win95 console bug. This was triggered by the password read stuff: the 13064 last character typed gets carried over to the next fread(). If you were 13065 generating a new cert request using 'req' for example then the last 13066 character of the passphrase would be CR which would then enter the first 13067 field as blank. 13068 [Steve Henson] 13069 13070 *) Added the new `Includes OpenSSL Cryptography Software' button as 13071 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 13072 button and can be used by applications based on OpenSSL to show the 13073 relationship to the OpenSSL project. 13074 [Ralf S. Engelschall] 13075 13076 *) Remove confusing variables in function signatures in files 13077 ssl/ssl_lib.c and ssl/ssl.h. 13078 [Lennart Bong <lob@kulthea.stacken.kth.se>] 13079 13080 *) Don't install bss_file.c under PREFIX/include/ 13081 [Lennart Bong <lob@kulthea.stacken.kth.se>] 13082 13083 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 13084 functions that return function pointers and has support for NT specific 13085 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 13086 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 13087 unsigned to signed types: this was killing the Win32 compile. 13088 [Steve Henson] 13089 13090 *) Add new certificate file to stack functions, 13091 SSL_add_dir_cert_subjects_to_stack() and 13092 SSL_add_file_cert_subjects_to_stack(). These largely supplant 13093 SSL_load_client_CA_file(), and can be used to add multiple certs easily 13094 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 13095 This means that Apache-SSL and similar packages don't have to mess around 13096 to add as many CAs as they want to the preferred list. 13097 [Ben Laurie] 13098 13099 *) Experiment with doxygen documentation. Currently only partially applied to 13100 ssl/ssl_lib.c. 13101 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 13102 openssl.doxy as the configuration file. 13103 [Ben Laurie] 13104 13105 *) Get rid of remaining C++-style comments which strict C compilers hate. 13106 [Ralf S. Engelschall, pointed out by Carlos Amengual] 13107 13108 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 13109 compiled in by default: it has problems with large keys. 13110 [Steve Henson] 13111 13112 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 13113 DH private keys and/or callback functions which directly correspond to 13114 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 13115 is needed for applications which have to configure certificates on a 13116 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 13117 (e.g. s_server). 13118 For the RSA certificate situation is makes no difference, but 13119 for the DSA certificate situation this fixes the "no shared cipher" 13120 problem where the OpenSSL cipher selection procedure failed because the 13121 temporary keys were not overtaken from the context and the API provided 13122 no way to reconfigure them. 13123 The new functions now let applications reconfigure the stuff and they 13124 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 13125 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 13126 non-public-API function ssl_cert_instantiate() is used as a helper 13127 function and also to reduce code redundancy inside ssl_rsa.c. 13128 [Ralf S. Engelschall] 13129 13130 *) Move s_server -dcert and -dkey options out of the undocumented feature 13131 area because they are useful for the DSA situation and should be 13132 recognized by the users. 13133 [Ralf S. Engelschall] 13134 13135 *) Fix the cipher decision scheme for export ciphers: the export bits are 13136 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 13137 SSL_EXP_MASK. So, the original variable has to be used instead of the 13138 already masked variable. 13139 [Richard Levitte <levitte@stacken.kth.se>] 13140 13141 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 13142 [Richard Levitte <levitte@stacken.kth.se>] 13143 13144 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 13145 from `int' to `unsigned int' because it's a length and initialized by 13146 EVP_DigestFinal() which expects an `unsigned int *'. 13147 [Richard Levitte <levitte@stacken.kth.se>] 13148 13149 *) Don't hard-code path to Perl interpreter on shebang line of Configure 13150 script. Instead use the usual Shell->Perl transition trick. 13151 [Ralf S. Engelschall] 13152 13153 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 13154 (in addition to RSA certificates) to match the behaviour of `openssl dsa 13155 -noout -modulus' as it's already the case for `openssl rsa -noout 13156 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 13157 currently the public key is printed (a decision which was already done by 13158 `openssl dsa -modulus' in the past) which serves a similar purpose. 13159 Additionally the NO_RSA no longer completely removes the whole -modulus 13160 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 13161 now, too. 13162 [Ralf S. Engelschall] 13163 13164 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 13165 BIO. See the source (crypto/evp/bio_ok.c) for more info. 13166 [Arne Ansper <arne@ats.cyber.ee>] 13167 13168 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 13169 to be added. Now both 'req' and 'ca' can use new objects defined in the 13170 config file. 13171 [Steve Henson] 13172 13173 *) Add cool BIO that does syslog (or event log on NT). 13174 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 13175 13176 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 13177 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 13178 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 13179 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 13180 [Ben Laurie] 13181 13182 *) Add preliminary config info for new extension code. 13183 [Steve Henson] 13184 13185 *) Make RSA_NO_PADDING really use no padding. 13186 [Ulf Moeller <ulf@fitug.de>] 13187 13188 *) Generate errors when private/public key check is done. 13189 [Ben Laurie] 13190 13191 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 13192 for some CRL extensions and new objects added. 13193 [Steve Henson] 13194 13195 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 13196 key usage extension and fuller support for authority key id. 13197 [Steve Henson] 13198 13199 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 13200 padding method for RSA, which is recommended for new applications in PKCS 13201 #1 v2.0 (RFC 2437, October 1998). 13202 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 13203 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 13204 against Bleichbacher's attack on RSA. 13205 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 13206 Ben Laurie] 13207 13208 *) Updates to the new SSL compression code 13209 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13210 13211 *) Fix so that the version number in the master secret, when passed 13212 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 13213 (because the server will not accept higher), that the version number 13214 is 0x03,0x01, not 0x03,0x00 13215 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13216 13217 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 13218 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 13219 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 13220 [Steve Henson] 13221 13222 *) Support for RAW extensions where an arbitrary extension can be 13223 created by including its DER encoding. See apps/openssl.cnf for 13224 an example. 13225 [Steve Henson] 13226 13227 *) Make sure latest Perl versions don't interpret some generated C array 13228 code as Perl array code in the crypto/err/err_genc.pl script. 13229 [Lars Weber <3weber@informatik.uni-hamburg.de>] 13230 13231 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 13232 not many people have the assembler. Various Win32 compilation fixes and 13233 update to the INSTALL.W32 file with (hopefully) more accurate Win32 13234 build instructions. 13235 [Steve Henson] 13236 13237 *) Modify configure script 'Configure' to automatically create crypto/date.h 13238 file under Win32 and also build pem.h from pem.org. New script 13239 util/mkfiles.pl to create the MINFO file on environments that can't do a 13240 'make files': perl util/mkfiles.pl >MINFO should work. 13241 [Steve Henson] 13242 13243 *) Major rework of DES function declarations, in the pursuit of correctness 13244 and purity. As a result, many evil casts evaporated, and some weirdness, 13245 too. You may find this causes warnings in your code. Zapping your evil 13246 casts will probably fix them. Mostly. 13247 [Ben Laurie] 13248 13249 *) Fix for a typo in asn1.h. Bug fix to object creation script 13250 obj_dat.pl. It considered a zero in an object definition to mean 13251 "end of object": none of the objects in objects.h have any zeros 13252 so it wasn't spotted. 13253 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 13254 13255 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 13256 Masking (CBCM). In the absence of test vectors, the best I have been able 13257 to do is check that the decrypt undoes the encrypt, so far. Send me test 13258 vectors if you have them. 13259 [Ben Laurie] 13260 13261 *) Correct calculation of key length for export ciphers (too much space was 13262 allocated for null ciphers). This has not been tested! 13263 [Ben Laurie] 13264 13265 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 13266 message is now correct (it understands "crypto" and "ssl" on its 13267 command line). There is also now an "update" option. This will update 13268 the util/ssleay.num and util/libeay.num files with any new functions. 13269 If you do a: 13270 perl util/mkdef.pl crypto ssl update 13271 it will update them. 13272 [Steve Henson] 13273 13274 *) Overhauled the Perl interface (perl/*): 13275 - ported BN stuff to OpenSSL's different BN library 13276 - made the perl/ source tree CVS-aware 13277 - renamed the package from SSLeay to OpenSSL (the files still contain 13278 their history because I've copied them in the repository) 13279 - removed obsolete files (the test scripts will be replaced 13280 by better Test::Harness variants in the future) 13281 [Ralf S. Engelschall] 13282 13283 *) First cut for a very conservative source tree cleanup: 13284 1. merge various obsolete readme texts into doc/ssleay.txt 13285 where we collect the old documents and readme texts. 13286 2. remove the first part of files where I'm already sure that we no 13287 longer need them because of three reasons: either they are just temporary 13288 files which were left by Eric or they are preserved original files where 13289 I've verified that the diff is also available in the CVS via "cvs diff 13290 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 13291 the crypto/md/ stuff). 13292 [Ralf S. Engelschall] 13293 13294 *) More extension code. Incomplete support for subject and issuer alt 13295 name, issuer and authority key id. Change the i2v function parameters 13296 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 13297 what that's for :-) Fix to ASN1 macro which messed up 13298 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 13299 [Steve Henson] 13300 13301 *) Preliminary support for ENUMERATED type. This is largely copied from the 13302 INTEGER code. 13303 [Steve Henson] 13304 13305 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 13306 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13307 13308 *) Make sure `make rehash' target really finds the `openssl' program. 13309 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 13310 13311 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 13312 like to hear about it if this slows down other processors. 13313 [Ben Laurie] 13314 13315 *) Add CygWin32 platform information to Configure script. 13316 [Alan Batie <batie@aahz.jf.intel.com>] 13317 13318 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 13319 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 13320 13321 *) New program nseq to manipulate netscape certificate sequences 13322 [Steve Henson] 13323 13324 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 13325 few typos. 13326 [Steve Henson] 13327 13328 *) Fixes to BN code. Previously the default was to define BN_RECURSION 13329 but the BN code had some problems that would cause failures when 13330 doing certificate verification and some other functions. 13331 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13332 13333 *) Add ASN1 and PEM code to support netscape certificate sequences. 13334 [Steve Henson] 13335 13336 *) Add ASN1 and PEM code to support netscape certificate sequences. 13337 [Steve Henson] 13338 13339 *) Add several PKIX and private extended key usage OIDs. 13340 [Steve Henson] 13341 13342 *) Modify the 'ca' program to handle the new extension code. Modify 13343 openssl.cnf for new extension format, add comments. 13344 [Steve Henson] 13345 13346 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 13347 and add a sample to openssl.cnf so req -x509 now adds appropriate 13348 CA extensions. 13349 [Steve Henson] 13350 13351 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 13352 error code, add initial support to X509_print() and x509 application. 13353 [Steve Henson] 13354 13355 *) Takes a deep breath and start adding X509 V3 extension support code. Add 13356 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 13357 stuff is currently isolated and isn't even compiled yet. 13358 [Steve Henson] 13359 13360 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 13361 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 13362 Removed the versions check from X509 routines when loading extensions: 13363 this allows certain broken certificates that don't set the version 13364 properly to be processed. 13365 [Steve Henson] 13366 13367 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 13368 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 13369 can still be regenerated with "make depend". 13370 [Ben Laurie] 13371 13372 *) Spelling mistake in C version of CAST-128. 13373 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 13374 13375 *) Changes to the error generation code. The perl script err-code.pl 13376 now reads in the old error codes and retains the old numbers, only 13377 adding new ones if necessary. It also only changes the .err files if new 13378 codes are added. The makefiles have been modified to only insert errors 13379 when needed (to avoid needlessly modifying header files). This is done 13380 by only inserting errors if the .err file is newer than the auto generated 13381 C file. To rebuild all the error codes from scratch (the old behaviour) 13382 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 13383 or delete all the .err files. 13384 [Steve Henson] 13385 13386 *) CAST-128 was incorrectly implemented for short keys. The C version has 13387 been fixed, but is untested. The assembler versions are also fixed, but 13388 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 13389 to regenerate it if needed. 13390 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 13391 Hagino <itojun@kame.net>] 13392 13393 *) File was opened incorrectly in randfile.c. 13394 [Ulf Möller <ulf@fitug.de>] 13395 13396 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 13397 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 13398 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 13399 al: it's just almost always a UTCTime. Note this patch adds new error 13400 codes so do a "make errors" if there are problems. 13401 [Steve Henson] 13402 13403 *) Correct Linux 1 recognition in config. 13404 [Ulf Möller <ulf@fitug.de>] 13405 13406 *) Remove pointless MD5 hash when using DSA keys in ca. 13407 [Anonymous <nobody@replay.com>] 13408 13409 *) Generate an error if given an empty string as a cert directory. Also 13410 generate an error if handed NULL (previously returned 0 to indicate an 13411 error, but didn't set one). 13412 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 13413 13414 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 13415 [Ben Laurie] 13416 13417 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 13418 parameters. This was causing a warning which killed off the Win32 compile. 13419 [Steve Henson] 13420 13421 *) Remove C++ style comments from crypto/bn/bn_local.h. 13422 [Neil Costigan <neil.costigan@celocom.com>] 13423 13424 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 13425 based on a text string, looking up short and long names and finally 13426 "dot" format. The "dot" format stuff didn't work. Added new function 13427 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 13428 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 13429 OID is not part of the table. 13430 [Steve Henson] 13431 13432 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 13433 X509_LOOKUP_by_alias(). 13434 [Ben Laurie] 13435 13436 *) Sort openssl functions by name. 13437 [Ben Laurie] 13438 13439 *) Get the gendsa program working (hopefully) and add it to app list. Remove 13440 encryption from sample DSA keys (in case anyone is interested the password 13441 was "1234"). 13442 [Steve Henson] 13443 13444 *) Make _all_ *_free functions accept a NULL pointer. 13445 [Frans Heymans <fheymans@isaserver.be>] 13446 13447 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 13448 NULL pointers. 13449 [Anonymous <nobody@replay.com>] 13450 13451 *) s_server should send the CAfile as acceptable CAs, not its own cert. 13452 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 13453 13454 *) Don't blow it for numeric -newkey arguments to apps/req. 13455 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 13456 13457 *) Temp key "for export" tests were wrong in s3_srvr.c. 13458 [Anonymous <nobody@replay.com>] 13459 13460 *) Add prototype for temp key callback functions 13461 SSL_CTX_set_tmp_{rsa,dh}_callback(). 13462 [Ben Laurie] 13463 13464 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 13465 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 13466 [Steve Henson] 13467 13468 *) X509_name_add_entry() freed the wrong thing after an error. 13469 [Arne Ansper <arne@ats.cyber.ee>] 13470 13471 *) rsa_eay.c would attempt to free a NULL context. 13472 [Arne Ansper <arne@ats.cyber.ee>] 13473 13474 *) BIO_s_socket() had a broken should_retry() on Windoze. 13475 [Arne Ansper <arne@ats.cyber.ee>] 13476 13477 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 13478 [Arne Ansper <arne@ats.cyber.ee>] 13479 13480 *) Make sure the already existing X509_STORE->depth variable is initialized 13481 in X509_STORE_new(), but document the fact that this variable is still 13482 unused in the certificate verification process. 13483 [Ralf S. Engelschall] 13484 13485 *) Fix the various library and apps files to free up pkeys obtained from 13486 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 13487 [Steve Henson] 13488 13489 *) Fix reference counting in X509_PUBKEY_get(). This makes 13490 demos/maurice/example2.c work, amongst others, probably. 13491 [Steve Henson and Ben Laurie] 13492 13493 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 13494 `openssl' and second, the shortcut symlinks for the `openssl <command>' 13495 are no longer created. This way we have a single and consistent command 13496 line interface `openssl <command>', similar to `cvs <command>'. 13497 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 13498 13499 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 13500 BIT STRING wrapper always have zero unused bits. 13501 [Steve Henson] 13502 13503 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 13504 [Steve Henson] 13505 13506 *) Make the top-level INSTALL documentation easier to understand. 13507 [Paul Sutton] 13508 13509 *) Makefiles updated to exit if an error occurs in a sub-directory 13510 make (including if user presses ^C) [Paul Sutton] 13511 13512 *) Make Montgomery context stuff explicit in RSA data structure. 13513 [Ben Laurie] 13514 13515 *) Fix build order of pem and err to allow for generated pem.h. 13516 [Ben Laurie] 13517 13518 *) Fix renumbering bug in X509_NAME_delete_entry(). 13519 [Ben Laurie] 13520 13521 *) Enhanced the err-ins.pl script so it makes the error library number 13522 global and can add a library name. This is needed for external ASN1 and 13523 other error libraries. 13524 [Steve Henson] 13525 13526 *) Fixed sk_insert which never worked properly. 13527 [Steve Henson] 13528 13529 *) Fix ASN1 macros so they can handle indefinite length constructed 13530 EXPLICIT tags. Some non standard certificates use these: they can now 13531 be read in. 13532 [Steve Henson] 13533 13534 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 13535 into a single doc/ssleay.txt bundle. This way the information is still 13536 preserved but no longer messes up this directory. Now it's new room for 13537 the new set of documentation files. 13538 [Ralf S. Engelschall] 13539 13540 *) SETs were incorrectly DER encoded. This was a major pain, because they 13541 shared code with SEQUENCEs, which aren't coded the same. This means that 13542 almost everything to do with SETs or SEQUENCEs has either changed name or 13543 number of arguments. 13544 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 13545 13546 *) Fix test data to work with the above. 13547 [Ben Laurie] 13548 13549 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 13550 was already fixed by Eric for 0.9.1 it seems. 13551 [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>] 13552 13553 *) Autodetect FreeBSD3. 13554 [Ben Laurie] 13555 13556 *) Fix various bugs in Configure. This affects the following platforms: 13557 nextstep 13558 ncr-scde 13559 unixware-2.0 13560 unixware-2.0-pentium 13561 sco5-cc. 13562 [Ben Laurie] 13563 13564 *) Eliminate generated files from CVS. Reorder tests to regenerate files 13565 before they are needed. 13566 [Ben Laurie] 13567 13568 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 13569 [Ben Laurie] 13570 13571 13572 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 13573 13574 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 13575 changed SSLeay to OpenSSL in version strings. 13576 [Ralf S. Engelschall] 13577 13578 *) Some fixups to the top-level documents. 13579 [Paul Sutton] 13580 13581 *) Fixed the nasty bug where rsaref.h was not found under compile-time 13582 because the symlink to include/ was missing. 13583 [Ralf S. Engelschall] 13584 13585 *) Incorporated the popular no-RSA/DSA-only patches 13586 which allow to compile a RSA-free SSLeay. 13587 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 13588 13589 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 13590 when "ssleay" is still not found. 13591 [Ralf S. Engelschall] 13592 13593 *) Added more platforms to Configure: Cray T3E, HPUX 11, 13594 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 13595 13596 *) Updated the README file. 13597 [Ralf S. Engelschall] 13598 13599 *) Added various .cvsignore files in the CVS repository subdirs 13600 to make a "cvs update" really silent. 13601 [Ralf S. Engelschall] 13602 13603 *) Recompiled the error-definition header files and added 13604 missing symbols to the Win32 linker tables. 13605 [Ralf S. Engelschall] 13606 13607 *) Cleaned up the top-level documents; 13608 o new files: CHANGES and LICENSE 13609 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 13610 o merged COPYRIGHT into LICENSE 13611 o removed obsolete TODO file 13612 o renamed MICROSOFT to INSTALL.W32 13613 [Ralf S. Engelschall] 13614 13615 *) Removed dummy files from the 0.9.1b source tree: 13616 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 13617 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 13618 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 13619 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 13620 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 13621 [Ralf S. Engelschall] 13622 13623 *) Added various platform portability fixes. 13624 [Mark J. Cox] 13625 13626 *) The Genesis of the OpenSSL rpject: 13627 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 13628 Young and Tim J. Hudson created while they were working for C2Net until 13629 summer 1998. 13630 [The OpenSSL Project] 13631 13632 13633 Changes between 0.9.0b and 0.9.1b [not released] 13634 13635 *) Updated a few CA certificates under certs/ 13636 [Eric A. Young] 13637 13638 *) Changed some BIGNUM api stuff. 13639 [Eric A. Young] 13640 13641 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 13642 DGUX x86, Linux Alpha, etc. 13643 [Eric A. Young] 13644 13645 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 13646 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 13647 available). 13648 [Eric A. Young] 13649 13650 *) Add -strparse option to asn1pars program which parses nested 13651 binary structures 13652 [Dr Stephen Henson <shenson@bigfoot.com>] 13653 13654 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 13655 [Eric A. Young] 13656 13657 *) DSA fix for "ca" program. 13658 [Eric A. Young] 13659 13660 *) Added "-genkey" option to "dsaparam" program. 13661 [Eric A. Young] 13662 13663 *) Added RIPE MD160 (rmd160) message digest. 13664 [Eric A. Young] 13665 13666 *) Added -a (all) option to "ssleay version" command. 13667 [Eric A. Young] 13668 13669 *) Added PLATFORM define which is the id given to Configure. 13670 [Eric A. Young] 13671 13672 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 13673 [Eric A. Young] 13674 13675 *) Extended the ASN.1 parser routines. 13676 [Eric A. Young] 13677 13678 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 13679 [Eric A. Young] 13680 13681 *) Added a BN_CTX to the BN library. 13682 [Eric A. Young] 13683 13684 *) Fixed the weak key values in DES library 13685 [Eric A. Young] 13686 13687 *) Changed API in EVP library for cipher aliases. 13688 [Eric A. Young] 13689 13690 *) Added support for RC2/64bit cipher. 13691 [Eric A. Young] 13692 13693 *) Converted the lhash library to the crypto/mem.c functions. 13694 [Eric A. Young] 13695 13696 *) Added more recognized ASN.1 object ids. 13697 [Eric A. Young] 13698 13699 *) Added more RSA padding checks for SSL/TLS. 13700 [Eric A. Young] 13701 13702 *) Added BIO proxy/filter functionality. 13703 [Eric A. Young] 13704 13705 *) Added extra_certs to SSL_CTX which can be used 13706 send extra CA certificates to the client in the CA cert chain sending 13707 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 13708 [Eric A. Young] 13709 13710 *) Now Fortezza is denied in the authentication phase because 13711 this is key exchange mechanism is not supported by SSLeay at all. 13712 [Eric A. Young] 13713 13714 *) Additional PKCS1 checks. 13715 [Eric A. Young] 13716 13717 *) Support the string "TLSv1" for all TLS v1 ciphers. 13718 [Eric A. Young] 13719 13720 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 13721 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 13722 [Eric A. Young] 13723 13724 *) Fixed a few memory leaks. 13725 [Eric A. Young] 13726 13727 *) Fixed various code and comment typos. 13728 [Eric A. Young] 13729 13730 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 13731 bytes sent in the client random. 13732 [Edward Bishop <ebishop@spyglass.com>] 13733