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 1.1.1k and 1.1.1l [24 Aug 2021] 11 12 *) Fixed an SM2 Decryption Buffer Overflow. 13 14 In order to decrypt SM2 encrypted data an application is expected to call the 15 API function EVP_PKEY_decrypt(). Typically an application will call this 16 function twice. The first time, on entry, the "out" parameter can be NULL and, 17 on exit, the "outlen" parameter is populated with the buffer size required to 18 hold the decrypted plaintext. The application can then allocate a sufficiently 19 sized buffer and call EVP_PKEY_decrypt() again, but this time passing a non-NULL 20 value for the "out" parameter. 21 22 A bug in the implementation of the SM2 decryption code means that the 23 calculation of the buffer size required to hold the plaintext returned by the 24 first call to EVP_PKEY_decrypt() can be smaller than the actual size required by 25 the second call. This can lead to a buffer overflow when EVP_PKEY_decrypt() is 26 called by the application a second time with a buffer that is too small. 27 28 A malicious attacker who is able present SM2 content for decryption to an 29 application could cause attacker chosen data to overflow the buffer by up to a 30 maximum of 62 bytes altering the contents of other data held after the 31 buffer, possibly changing application behaviour or causing the application to 32 crash. The location of the buffer is application dependent but is typically 33 heap allocated. 34 (CVE-2021-3711) 35 [Matt Caswell] 36 37 *) Fixed various read buffer overruns processing ASN.1 strings 38 39 ASN.1 strings are represented internally within OpenSSL as an ASN1_STRING 40 structure which contains a buffer holding the string data and a field holding 41 the buffer length. This contrasts with normal C strings which are repesented as 42 a buffer for the string data which is terminated with a NUL (0) byte. 43 44 Although not a strict requirement, ASN.1 strings that are parsed using OpenSSL's 45 own "d2i" functions (and other similar parsing functions) as well as any string 46 whose value has been set with the ASN1_STRING_set() function will additionally 47 NUL terminate the byte array in the ASN1_STRING structure. 48 49 However, it is possible for applications to directly construct valid ASN1_STRING 50 structures which do not NUL terminate the byte array by directly setting the 51 "data" and "length" fields in the ASN1_STRING array. This can also happen by 52 using the ASN1_STRING_set0() function. 53 54 Numerous OpenSSL functions that print ASN.1 data have been found to assume that 55 the ASN1_STRING byte array will be NUL terminated, even though this is not 56 guaranteed for strings that have been directly constructed. Where an application 57 requests an ASN.1 structure to be printed, and where that ASN.1 structure 58 contains ASN1_STRINGs that have been directly constructed by the application 59 without NUL terminating the "data" field, then a read buffer overrun can occur. 60 61 The same thing can also occur during name constraints processing of certificates 62 (for example if a certificate has been directly constructed by the application 63 instead of loading it via the OpenSSL parsing functions, and the certificate 64 contains non NUL terminated ASN1_STRING structures). It can also occur in the 65 X509_get1_email(), X509_REQ_get1_email() and X509_get1_ocsp() functions. 66 67 If a malicious actor can cause an application to directly construct an 68 ASN1_STRING and then process it through one of the affected OpenSSL functions 69 then this issue could be hit. This might result in a crash (causing a Denial of 70 Service attack). It could also result in the disclosure of private memory 71 contents (such as private keys, or sensitive plaintext). 72 (CVE-2021-3712) 73 [Matt Caswell] 74 75 Changes between 1.1.1j and 1.1.1k [25 Mar 2021] 76 77 *) Fixed a problem with verifying a certificate chain when using the 78 X509_V_FLAG_X509_STRICT flag. This flag enables additional security checks 79 of the certificates present in a certificate chain. It is not set by 80 default. 81 82 Starting from OpenSSL version 1.1.1h a check to disallow certificates in 83 the chain that have explicitly encoded elliptic curve parameters was added 84 as an additional strict check. 85 86 An error in the implementation of this check meant that the result of a 87 previous check to confirm that certificates in the chain are valid CA 88 certificates was overwritten. This effectively bypasses the check 89 that non-CA certificates must not be able to issue other certificates. 90 91 If a "purpose" has been configured then there is a subsequent opportunity 92 for checks that the certificate is a valid CA. All of the named "purpose" 93 values implemented in libcrypto perform this check. Therefore, where 94 a purpose is set the certificate chain will still be rejected even when the 95 strict flag has been used. A purpose is set by default in libssl client and 96 server certificate verification routines, but it can be overridden or 97 removed by an application. 98 99 In order to be affected, an application must explicitly set the 100 X509_V_FLAG_X509_STRICT verification flag and either not set a purpose 101 for the certificate verification or, in the case of TLS client or server 102 applications, override the default purpose. 103 (CVE-2021-3450) 104 [Tomáš Mráz] 105 106 *) Fixed an issue where an OpenSSL TLS server may crash if sent a maliciously 107 crafted renegotiation ClientHello message from a client. If a TLSv1.2 108 renegotiation ClientHello omits the signature_algorithms extension (where 109 it was present in the initial ClientHello), but includes a 110 signature_algorithms_cert extension then a NULL pointer dereference will 111 result, leading to a crash and a denial of service attack. 112 113 A server is only vulnerable if it has TLSv1.2 and renegotiation enabled 114 (which is the default configuration). OpenSSL TLS clients are not impacted 115 by this issue. 116 (CVE-2021-3449) 117 [Peter Kästle and Samuel Sapalski] 118 119 Changes between 1.1.1i and 1.1.1j [16 Feb 2021] 120 121 *) Fixed the X509_issuer_and_serial_hash() function. It attempts to 122 create a unique hash value based on the issuer and serial number data 123 contained within an X509 certificate. However it was failing to correctly 124 handle any errors that may occur while parsing the issuer field (which might 125 occur if the issuer field is maliciously constructed). This may subsequently 126 result in a NULL pointer deref and a crash leading to a potential denial of 127 service attack. 128 (CVE-2021-23841) 129 [Matt Caswell] 130 131 *) Fixed the RSA_padding_check_SSLv23() function and the RSA_SSLV23_PADDING 132 padding mode to correctly check for rollback attacks. This is considered a 133 bug in OpenSSL 1.1.1 because it does not support SSLv2. In 1.0.2 this is 134 CVE-2021-23839. 135 [Matt Caswell] 136 137 *) Fixed the EVP_CipherUpdate, EVP_EncryptUpdate and EVP_DecryptUpdate 138 functions. Previously they could overflow the output length argument in some 139 cases where the input length is close to the maximum permissable length for 140 an integer on the platform. In such cases the return value from the function 141 call would be 1 (indicating success), but the output length value would be 142 negative. This could cause applications to behave incorrectly or crash. 143 (CVE-2021-23840) 144 [Matt Caswell] 145 146 *) Fixed SRP_Calc_client_key so that it runs in constant time. The previous 147 implementation called BN_mod_exp without setting BN_FLG_CONSTTIME. This 148 could be exploited in a side channel attack to recover the password. Since 149 the attack is local host only this is outside of the current OpenSSL 150 threat model and therefore no CVE is assigned. 151 152 Thanks to Mohammed Sabt and Daniel De Almeida Braga for reporting this 153 issue. 154 [Matt Caswell] 155 156 Changes between 1.1.1h and 1.1.1i [8 Dec 2020] 157 158 *) Fixed NULL pointer deref in the GENERAL_NAME_cmp function 159 This function could crash if both GENERAL_NAMEs contain an EDIPARTYNAME. 160 If an attacker can control both items being compared then this could lead 161 to a possible denial of service attack. OpenSSL itself uses the 162 GENERAL_NAME_cmp function for two purposes: 163 1) Comparing CRL distribution point names between an available CRL and a 164 CRL distribution point embedded in an X509 certificate 165 2) When verifying that a timestamp response token signer matches the 166 timestamp authority name (exposed via the API functions 167 TS_RESP_verify_response and TS_RESP_verify_token) 168 (CVE-2020-1971) 169 [Matt Caswell] 170 171 *) Add support for Apple Silicon M1 Macs with the darwin64-arm64-cc target. 172 [Stuart Carnie] 173 174 *) The security callback, which can be customised by application code, supports 175 the security operation SSL_SECOP_TMP_DH. This is defined to take an EVP_PKEY 176 in the "other" parameter. In most places this is what is passed. All these 177 places occur server side. However there was one client side call of this 178 security operation and it passed a DH object instead. This is incorrect 179 according to the definition of SSL_SECOP_TMP_DH, and is inconsistent with all 180 of the other locations. Therefore this client side call has been changed to 181 pass an EVP_PKEY instead. 182 [Matt Caswell] 183 184 *) In 1.1.1h, an expired trusted (root) certificate was not anymore rejected 185 when validating a certificate path. This check is restored in 1.1.1i. 186 [David von Oheimb] 187 188 Changes between 1.1.1g and 1.1.1h [22 Sep 2020] 189 190 *) Certificates with explicit curve parameters are now disallowed in 191 verification chains if the X509_V_FLAG_X509_STRICT flag is used. 192 [Tomas Mraz] 193 194 *) The 'MinProtocol' and 'MaxProtocol' configuration commands now silently 195 ignore TLS protocol version bounds when configuring DTLS-based contexts, and 196 conversely, silently ignore DTLS protocol version bounds when configuring 197 TLS-based contexts. The commands can be repeated to set bounds of both 198 types. The same applies with the corresponding "min_protocol" and 199 "max_protocol" command-line switches, in case some application uses both TLS 200 and DTLS. 201 202 SSL_CTX instances that are created for a fixed protocol version (e.g. 203 TLSv1_server_method()) also silently ignore version bounds. Previously 204 attempts to apply bounds to these protocol versions would result in an 205 error. Now only the "version-flexible" SSL_CTX instances are subject to 206 limits in configuration files in command-line options. 207 [Viktor Dukhovni] 208 209 *) Handshake now fails if Extended Master Secret extension is dropped 210 on renegotiation. 211 [Tomas Mraz] 212 213 *) Accidentally, an expired trusted (root) certificate is not anymore rejected 214 when validating a certificate path. 215 [David von Oheimb] 216 217 *) The Oracle Developer Studio compiler will start reporting deprecated APIs 218 219 Changes between 1.1.1f and 1.1.1g [21 Apr 2020] 220 221 *) Fixed segmentation fault in SSL_check_chain() 222 Server or client applications that call the SSL_check_chain() function 223 during or after a TLS 1.3 handshake may crash due to a NULL pointer 224 dereference as a result of incorrect handling of the 225 "signature_algorithms_cert" TLS extension. The crash occurs if an invalid 226 or unrecognised signature algorithm is received from the peer. This could 227 be exploited by a malicious peer in a Denial of Service attack. 228 (CVE-2020-1967) 229 [Benjamin Kaduk] 230 231 *) Added AES consttime code for no-asm configurations 232 an optional constant time support for AES was added 233 when building openssl for no-asm. 234 Enable with: ./config no-asm -DOPENSSL_AES_CONST_TIME 235 Disable with: ./config no-asm -DOPENSSL_NO_AES_CONST_TIME 236 At this time this feature is by default disabled. 237 It will be enabled by default in 3.0. 238 [Bernd Edlinger] 239 240 Changes between 1.1.1e and 1.1.1f [31 Mar 2020] 241 242 *) Revert the change of EOF detection while reading in libssl to avoid 243 regressions in applications depending on the current way of reporting 244 the EOF. As the existing method is not fully accurate the change to 245 reporting the EOF via SSL_ERROR_SSL is kept on the current development 246 branch and will be present in the 3.0 release. 247 [Tomas Mraz] 248 249 *) Revised BN_generate_prime_ex to not avoid factors 3..17863 in p-1 250 when primes for RSA keys are computed. 251 Since we previously always generated primes == 2 (mod 3) for RSA keys, 252 the 2-prime and 3-prime RSA modules were easy to distinguish, since 253 N = p*q = 1 (mod 3), but N = p*q*r = 2 (mod 3). Therefore fingerprinting 254 2-prime vs. 3-prime RSA keys was possible by computing N mod 3. 255 This avoids possible fingerprinting of newly generated RSA modules. 256 [Bernd Edlinger] 257 258 Changes between 1.1.1d and 1.1.1e [17 Mar 2020] 259 *) Properly detect EOF while reading in libssl. Previously if we hit an EOF 260 while reading in libssl then we would report an error back to the 261 application (SSL_ERROR_SYSCALL) but errno would be 0. We now add 262 an error to the stack (which means we instead return SSL_ERROR_SSL) and 263 therefore give a hint as to what went wrong. 264 [Matt Caswell] 265 266 *) Check that ed25519 and ed448 are allowed by the security level. Previously 267 signature algorithms not using an MD were not being checked that they were 268 allowed by the security level. 269 [Kurt Roeckx] 270 271 *) Fixed SSL_get_servername() behaviour. The behaviour of SSL_get_servername() 272 was not quite right. The behaviour was not consistent between resumption 273 and normal handshakes, and also not quite consistent with historical 274 behaviour. The behaviour in various scenarios has been clarified and 275 it has been updated to make it match historical behaviour as closely as 276 possible. 277 [Matt Caswell] 278 279 *) [VMS only] The header files that the VMS compilers include automatically, 280 __DECC_INCLUDE_PROLOGUE.H and __DECC_INCLUDE_EPILOGUE.H, use pragmas that 281 the C++ compiler doesn't understand. This is a shortcoming in the 282 compiler, but can be worked around with __cplusplus guards. 283 284 C++ applications that use OpenSSL libraries must be compiled using the 285 qualifier '/NAMES=(AS_IS,SHORTENED)' to be able to use all the OpenSSL 286 functions. Otherwise, only functions with symbols of less than 31 287 characters can be used, as the linker will not be able to successfully 288 resolve symbols with longer names. 289 [Richard Levitte] 290 291 *) Corrected the documentation of the return values from the EVP_DigestSign* 292 set of functions. The documentation mentioned negative values for some 293 errors, but this was never the case, so the mention of negative values 294 was removed. 295 296 Code that followed the documentation and thereby check with something 297 like 'EVP_DigestSignInit(...) <= 0' will continue to work undisturbed. 298 [Richard Levitte] 299 300 *) Fixed an an overflow bug in the x64_64 Montgomery squaring procedure 301 used in exponentiation with 512-bit moduli. No EC algorithms are 302 affected. Analysis suggests that attacks against 2-prime RSA1024, 303 3-prime RSA1536, and DSA1024 as a result of this defect would be very 304 difficult to perform and are not believed likely. Attacks against DH512 305 are considered just feasible. However, for an attack the target would 306 have to re-use the DH512 private key, which is not recommended anyway. 307 Also applications directly using the low level API BN_mod_exp may be 308 affected if they use BN_FLG_CONSTTIME. 309 (CVE-2019-1551) 310 [Andy Polyakov] 311 312 *) Added a new method to gather entropy on VMS, based on SYS$GET_ENTROPY. 313 The presence of this system service is determined at run-time. 314 [Richard Levitte] 315 316 *) Added newline escaping functionality to a filename when using openssl dgst. 317 This output format is to replicate the output format found in the '*sum' 318 checksum programs. This aims to preserve backward compatibility. 319 [Matt Eaton, Richard Levitte, and Paul Dale] 320 321 *) Print all values for a PKCS#12 attribute with 'openssl pkcs12', not just 322 the first value. 323 [Jon Spillett] 324 325 Changes between 1.1.1c and 1.1.1d [10 Sep 2019] 326 327 *) Fixed a fork protection issue. OpenSSL 1.1.1 introduced a rewritten random 328 number generator (RNG). This was intended to include protection in the 329 event of a fork() system call in order to ensure that the parent and child 330 processes did not share the same RNG state. However this protection was not 331 being used in the default case. 332 333 A partial mitigation for this issue is that the output from a high 334 precision timer is mixed into the RNG state so the likelihood of a parent 335 and child process sharing state is significantly reduced. 336 337 If an application already calls OPENSSL_init_crypto() explicitly using 338 OPENSSL_INIT_ATFORK then this problem does not occur at all. 339 (CVE-2019-1549) 340 [Matthias St. Pierre] 341 342 *) For built-in EC curves, ensure an EC_GROUP built from the curve name is 343 used even when parsing explicit parameters, when loading a serialized key 344 or calling `EC_GROUP_new_from_ecpkparameters()`/ 345 `EC_GROUP_new_from_ecparameters()`. 346 This prevents bypass of security hardening and performance gains, 347 especially for curves with specialized EC_METHODs. 348 By default, if a key encoded with explicit parameters is loaded and later 349 serialized, the output is still encoded with explicit parameters, even if 350 internally a "named" EC_GROUP is used for computation. 351 [Nicola Tuveri] 352 353 *) Compute ECC cofactors if not provided during EC_GROUP construction. Before 354 this change, EC_GROUP_set_generator would accept order and/or cofactor as 355 NULL. After this change, only the cofactor parameter can be NULL. It also 356 does some minimal sanity checks on the passed order. 357 (CVE-2019-1547) 358 [Billy Bob Brumley] 359 360 *) Fixed a padding oracle in PKCS7_dataDecode and CMS_decrypt_set1_pkey. 361 An attack is simple, if the first CMS_recipientInfo is valid but the 362 second CMS_recipientInfo is chosen ciphertext. If the second 363 recipientInfo decodes to PKCS #1 v1.5 form plaintext, the correct 364 encryption key will be replaced by garbage, and the message cannot be 365 decoded, but if the RSA decryption fails, the correct encryption key is 366 used and the recipient will not notice the attack. 367 As a work around for this potential attack the length of the decrypted 368 key must be equal to the cipher default key length, in case the 369 certifiate is not given and all recipientInfo are tried out. 370 The old behaviour can be re-enabled in the CMS code by setting the 371 CMS_DEBUG_DECRYPT flag. 372 (CVE-2019-1563) 373 [Bernd Edlinger] 374 375 *) Early start up entropy quality from the DEVRANDOM seed source has been 376 improved for older Linux systems. The RAND subsystem will wait for 377 /dev/random to be producing output before seeding from /dev/urandom. 378 The seeded state is stored for future library initialisations using 379 a system global shared memory segment. The shared memory identifier 380 can be configured by defining OPENSSL_RAND_SEED_DEVRANDOM_SHM_ID to 381 the desired value. The default identifier is 114. 382 [Paul Dale] 383 384 *) Correct the extended master secret constant on EBCDIC systems. Without this 385 fix TLS connections between an EBCDIC system and a non-EBCDIC system that 386 negotiate EMS will fail. Unfortunately this also means that TLS connections 387 between EBCDIC systems with this fix, and EBCDIC systems without this 388 fix will fail if they negotiate EMS. 389 [Matt Caswell] 390 391 *) Use Windows installation paths in the mingw builds 392 393 Mingw isn't a POSIX environment per se, which means that Windows 394 paths should be used for installation. 395 (CVE-2019-1552) 396 [Richard Levitte] 397 398 *) Changed DH_check to accept parameters with order q and 2q subgroups. 399 With order 2q subgroups the bit 0 of the private key is not secret 400 but DH_generate_key works around that by clearing bit 0 of the 401 private key for those. This avoids leaking bit 0 of the private key. 402 [Bernd Edlinger] 403 404 *) Significantly reduce secure memory usage by the randomness pools. 405 [Paul Dale] 406 407 *) Revert the DEVRANDOM_WAIT feature for Linux systems 408 409 The DEVRANDOM_WAIT feature added a select() call to wait for the 410 /dev/random device to become readable before reading from the 411 /dev/urandom device. 412 413 It turned out that this change had negative side effects on 414 performance which were not acceptable. After some discussion it 415 was decided to revert this feature and leave it up to the OS 416 resp. the platform maintainer to ensure a proper initialization 417 during early boot time. 418 [Matthias St. Pierre] 419 420 Changes between 1.1.1b and 1.1.1c [28 May 2019] 421 422 *) Add build tests for C++. These are generated files that only do one 423 thing, to include one public OpenSSL head file each. This tests that 424 the public header files can be usefully included in a C++ application. 425 426 This test isn't enabled by default. It can be enabled with the option 427 'enable-buildtest-c++'. 428 [Richard Levitte] 429 430 *) Enable SHA3 pre-hashing for ECDSA and DSA. 431 [Patrick Steuer] 432 433 *) Change the default RSA, DSA and DH size to 2048 bit instead of 1024. 434 This changes the size when using the genpkey app when no size is given. It 435 fixes an omission in earlier changes that changed all RSA, DSA and DH 436 generation apps to use 2048 bits by default. 437 [Kurt Roeckx] 438 439 *) Reorganize the manual pages to consistently have RETURN VALUES, 440 EXAMPLES, SEE ALSO and HISTORY come in that order, and adjust 441 util/fix-doc-nits accordingly. 442 [Paul Yang, Joshua Lock] 443 444 *) Add the missing accessor EVP_PKEY_get0_engine() 445 [Matt Caswell] 446 447 *) Have apps like 's_client' and 's_server' output the signature scheme 448 along with other cipher suite parameters when debugging. 449 [Lorinczy Zsigmond] 450 451 *) Make OPENSSL_config() error agnostic again. 452 [Richard Levitte] 453 454 *) Do the error handling in RSA decryption constant time. 455 [Bernd Edlinger] 456 457 *) Prevent over long nonces in ChaCha20-Poly1305. 458 459 ChaCha20-Poly1305 is an AEAD cipher, and requires a unique nonce input 460 for every encryption operation. RFC 7539 specifies that the nonce value 461 (IV) should be 96 bits (12 bytes). OpenSSL allows a variable nonce length 462 and front pads the nonce with 0 bytes if it is less than 12 463 bytes. However it also incorrectly allows a nonce to be set of up to 16 464 bytes. In this case only the last 12 bytes are significant and any 465 additional leading bytes are ignored. 466 467 It is a requirement of using this cipher that nonce values are 468 unique. Messages encrypted using a reused nonce value are susceptible to 469 serious confidentiality and integrity attacks. If an application changes 470 the default nonce length to be longer than 12 bytes and then makes a 471 change to the leading bytes of the nonce expecting the new value to be a 472 new unique nonce then such an application could inadvertently encrypt 473 messages with a reused nonce. 474 475 Additionally the ignored bytes in a long nonce are not covered by the 476 integrity guarantee of this cipher. Any application that relies on the 477 integrity of these ignored leading bytes of a long nonce may be further 478 affected. Any OpenSSL internal use of this cipher, including in SSL/TLS, 479 is safe because no such use sets such a long nonce value. However user 480 applications that use this cipher directly and set a non-default nonce 481 length to be longer than 12 bytes may be vulnerable. 482 483 This issue was reported to OpenSSL on 16th of March 2019 by Joran Dirk 484 Greef of Ronomon. 485 (CVE-2019-1543) 486 [Matt Caswell] 487 488 *) Add DEVRANDOM_WAIT feature for Linux systems 489 490 On older Linux systems where the getrandom() system call is not available, 491 OpenSSL normally uses the /dev/urandom device for seeding its CSPRNG. 492 Contrary to getrandom(), the /dev/urandom device will not block during 493 early boot when the kernel CSPRNG has not been seeded yet. 494 495 To mitigate this known weakness, use select() to wait for /dev/random to 496 become readable before reading from /dev/urandom. 497 498 *) Ensure that SM2 only uses SM3 as digest algorithm 499 [Paul Yang] 500 501 Changes between 1.1.1a and 1.1.1b [26 Feb 2019] 502 503 *) Added SCA hardening for modular field inversion in EC_GROUP through 504 a new dedicated field_inv() pointer in EC_METHOD. 505 This also addresses a leakage affecting conversions from projective 506 to affine coordinates. 507 [Billy Bob Brumley, Nicola Tuveri] 508 509 *) Change the info callback signals for the start and end of a post-handshake 510 message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START 511 and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get 512 confused by this and assume that a TLSv1.2 renegotiation has started. This 513 can break KeyUpdate handling. Instead we no longer signal the start and end 514 of a post handshake message exchange (although the messages themselves are 515 still signalled). This could break some applications that were expecting 516 the old signals. However without this KeyUpdate is not usable for many 517 applications. 518 [Matt Caswell] 519 520 *) Fix a bug in the computation of the endpoint-pair shared secret used 521 by DTLS over SCTP. This breaks interoperability with older versions 522 of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime 523 switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling 524 interoperability with such broken implementations. However, enabling 525 this switch breaks interoperability with correct implementations. 526 527 *) Fix a use after free bug in d2i_X509_PUBKEY when overwriting a 528 re-used X509_PUBKEY object if the second PUBKEY is malformed. 529 [Bernd Edlinger] 530 531 *) Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0(). 532 [Richard Levitte] 533 534 *) Remove the 'dist' target and add a tarball building script. The 535 'dist' target has fallen out of use, and it shouldn't be 536 necessary to configure just to create a source distribution. 537 [Richard Levitte] 538 539 Changes between 1.1.1 and 1.1.1a [20 Nov 2018] 540 541 *) Timing vulnerability in DSA signature generation 542 543 The OpenSSL DSA signature algorithm has been shown to be vulnerable to a 544 timing side channel attack. An attacker could use variations in the signing 545 algorithm to recover the private key. 546 547 This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser. 548 (CVE-2018-0734) 549 [Paul Dale] 550 551 *) Timing vulnerability in ECDSA signature generation 552 553 The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a 554 timing side channel attack. An attacker could use variations in the signing 555 algorithm to recover the private key. 556 557 This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser. 558 (CVE-2018-0735) 559 [Paul Dale] 560 561 *) Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for 562 the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names 563 are retained for backwards compatibility. 564 [Antoine Salon] 565 566 *) Fixed the issue that RAND_add()/RAND_seed() silently discards random input 567 if its length exceeds 4096 bytes. The limit has been raised to a buffer size 568 of two gigabytes and the error handling improved. 569 570 This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been 571 categorized as a normal bug, not a security issue, because the DRBG reseeds 572 automatically and is fully functional even without additional randomness 573 provided by the application. 574 575 Changes between 1.1.0i and 1.1.1 [11 Sep 2018] 576 577 *) Add a new ClientHello callback. Provides a callback interface that gives 578 the application the ability to adjust the nascent SSL object at the 579 earliest stage of ClientHello processing, immediately after extensions have 580 been collected but before they have been processed. In particular, this 581 callback can adjust the supported TLS versions in response to the contents 582 of the ClientHello 583 [Benjamin Kaduk] 584 585 *) Add SM2 base algorithm support. 586 [Jack Lloyd] 587 588 *) s390x assembly pack: add (improved) hardware-support for the following 589 cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb, 590 aes-cfb/cfb8, aes-ecb. 591 [Patrick Steuer] 592 593 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 594 parameter is no longer accepted, as it leads to a corrupt table. NULL 595 pem_str is reserved for alias entries only. 596 [Richard Levitte] 597 598 *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 599 step for prime curves. The new implementation is based on formulae from 600 differential addition-and-doubling in homogeneous projective coordinates 601 from Izu-Takagi "A fast parallel elliptic curve multiplication resistant 602 against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves 603 and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified 604 to work in projective coordinates. 605 [Billy Bob Brumley, Nicola Tuveri] 606 607 *) Change generating and checking of primes so that the error rate of not 608 being prime depends on the intended use based on the size of the input. 609 For larger primes this will result in more rounds of Miller-Rabin. 610 The maximal error rate for primes with more than 1080 bits is lowered 611 to 2^-128. 612 [Kurt Roeckx, Annie Yousar] 613 614 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 615 [Kurt Roeckx] 616 617 *) The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when 618 moving between systems, and to avoid confusion when a Windows build is 619 done with mingw vs with MSVC. For POSIX installs, there's still a 620 symlink or copy named 'tsget' to avoid that confusion as well. 621 [Richard Levitte] 622 623 *) Revert blinding in ECDSA sign and instead make problematic addition 624 length-invariant. Switch even to fixed-length Montgomery multiplication. 625 [Andy Polyakov] 626 627 *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder 628 step for binary curves. The new implementation is based on formulae from 629 differential addition-and-doubling in mixed Lopez-Dahab projective 630 coordinates, modified to independently blind the operands. 631 [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri] 632 633 *) Add a scaffold to optionally enhance the Montgomery ladder implementation 634 for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing 635 EC_METHODs to implement their own specialized "ladder step", to take 636 advantage of more favorable coordinate systems or more efficient 637 differential addition-and-doubling algorithms. 638 [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri] 639 640 *) Modified the random device based seed sources to keep the relevant 641 file descriptors open rather than reopening them on each access. 642 This allows such sources to operate in a chroot() jail without 643 the associated device nodes being available. This behaviour can be 644 controlled using RAND_keep_random_devices_open(). 645 [Paul Dale] 646 647 *) Numerous side-channel attack mitigations have been applied. This may have 648 performance impacts for some algorithms for the benefit of improved 649 security. Specific changes are noted in this change log by their respective 650 authors. 651 [Matt Caswell] 652 653 *) AIX shared library support overhaul. Switch to AIX "natural" way of 654 handling shared libraries, which means collecting shared objects of 655 different versions and bitnesses in one common archive. This allows to 656 mitigate conflict between 1.0 and 1.1 side-by-side installations. It 657 doesn't affect the way 3rd party applications are linked, only how 658 multi-version installation is managed. 659 [Andy Polyakov] 660 661 *) Make ec_group_do_inverse_ord() more robust and available to other 662 EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA 663 mitigations are applied to the fallback BN_mod_inverse(). 664 When using this function rather than BN_mod_inverse() directly, new 665 EC cryptosystem implementations are then safer-by-default. 666 [Billy Bob Brumley] 667 668 *) Add coordinate blinding for EC_POINT and implement projective 669 coordinate blinding for generic prime curves as a countermeasure to 670 chosen point SCA attacks. 671 [Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley] 672 673 *) Add blinding to ECDSA and DSA signatures to protect against side channel 674 attacks discovered by Keegan Ryan (NCC Group). 675 [Matt Caswell] 676 677 *) Enforce checking in the pkeyutl command line app to ensure that the input 678 length does not exceed the maximum supported digest length when performing 679 a sign, verify or verifyrecover operation. 680 [Matt Caswell] 681 682 *) SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking 683 I/O in combination with something like select() or poll() will hang. This 684 can be turned off again using SSL_CTX_clear_mode(). 685 Many applications do not properly handle non-application data records, and 686 TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works 687 around the problems in those applications, but can also break some. 688 It's recommended to read the manpages about SSL_read(), SSL_write(), 689 SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and 690 SSL_CTX_set_read_ahead() again. 691 [Kurt Roeckx] 692 693 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 694 now allow empty (zero character) pass phrases. 695 [Richard Levitte] 696 697 *) Apply blinding to binary field modular inversion and remove patent 698 pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation. 699 [Billy Bob Brumley] 700 701 *) Deprecate ec2_mult.c and unify scalar multiplication code paths for 702 binary and prime elliptic curves. 703 [Billy Bob Brumley] 704 705 *) Remove ECDSA nonce padding: EC_POINT_mul is now responsible for 706 constant time fixed point multiplication. 707 [Billy Bob Brumley] 708 709 *) Revise elliptic curve scalar multiplication with timing attack 710 defenses: ec_wNAF_mul redirects to a constant time implementation 711 when computing fixed point and variable point multiplication (which 712 in OpenSSL are mostly used with secret scalars in keygen, sign, 713 ECDH derive operations). 714 [Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García, 715 Sohaib ul Hassan] 716 717 *) Updated CONTRIBUTING 718 [Rich Salz] 719 720 *) Updated DRBG / RAND to request nonce and additional low entropy 721 randomness from the system. 722 [Matthias St. Pierre] 723 724 *) Updated 'openssl rehash' to use OpenSSL consistent default. 725 [Richard Levitte] 726 727 *) Moved the load of the ssl_conf module to libcrypto, which helps 728 loading engines that libssl uses before libssl is initialised. 729 [Matt Caswell] 730 731 *) Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA 732 [Matt Caswell] 733 734 *) Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases. 735 [Ingo Schwarze, Rich Salz] 736 737 *) Added output of accepting IP address and port for 'openssl s_server' 738 [Richard Levitte] 739 740 *) Added a new API for TLSv1.3 ciphersuites: 741 SSL_CTX_set_ciphersuites() 742 SSL_set_ciphersuites() 743 [Matt Caswell] 744 745 *) Memory allocation failures consistently add an error to the error 746 stack. 747 [Rich Salz] 748 749 *) Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values 750 in libcrypto when run as setuid/setgid. 751 [Bernd Edlinger] 752 753 *) Load any config file by default when libssl is used. 754 [Matt Caswell] 755 756 *) Added new public header file <openssl/rand_drbg.h> and documentation 757 for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview. 758 [Matthias St. Pierre] 759 760 *) QNX support removed (cannot find contributors to get their approval 761 for the license change). 762 [Rich Salz] 763 764 *) TLSv1.3 replay protection for early data has been implemented. See the 765 SSL_read_early_data() man page for further details. 766 [Matt Caswell] 767 768 *) Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite 769 configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and 770 below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3. 771 In order to avoid issues where legacy TLSv1.2 ciphersuite configuration 772 would otherwise inadvertently disable all TLSv1.3 ciphersuites the 773 configuration has been separated out. See the ciphers man page or the 774 SSL_CTX_set_ciphersuites() man page for more information. 775 [Matt Caswell] 776 777 *) On POSIX (BSD, Linux, ...) systems the ocsp(1) command running 778 in responder mode now supports the new "-multi" option, which 779 spawns the specified number of child processes to handle OCSP 780 requests. The "-timeout" option now also limits the OCSP 781 responder's patience to wait to receive the full client request 782 on a newly accepted connection. Child processes are respawned 783 as needed, and the CA index file is automatically reloaded 784 when changed. This makes it possible to run the "ocsp" responder 785 as a long-running service, making the OpenSSL CA somewhat more 786 feature-complete. In this mode, most diagnostic messages logged 787 after entering the event loop are logged via syslog(3) rather than 788 written to stderr. 789 [Viktor Dukhovni] 790 791 *) Added support for X448 and Ed448. Heavily based on original work by 792 Mike Hamburg. 793 [Matt Caswell] 794 795 *) Extend OSSL_STORE with capabilities to search and to narrow the set of 796 objects loaded. This adds the functions OSSL_STORE_expect() and 797 OSSL_STORE_find() as well as needed tools to construct searches and 798 get the search data out of them. 799 [Richard Levitte] 800 801 *) Support for TLSv1.3 added. Note that users upgrading from an earlier 802 version of OpenSSL should review their configuration settings to ensure 803 that they are still appropriate for TLSv1.3. For further information see: 804 https://wiki.openssl.org/index.php/TLS1.3 805 [Matt Caswell] 806 807 *) Grand redesign of the OpenSSL random generator 808 809 The default RAND method now utilizes an AES-CTR DRBG according to 810 NIST standard SP 800-90Ar1. The new random generator is essentially 811 a port of the default random generator from the OpenSSL FIPS 2.0 812 object module. It is a hybrid deterministic random bit generator 813 using an AES-CTR bit stream and which seeds and reseeds itself 814 automatically using trusted system entropy sources. 815 816 Some of its new features are: 817 o Support for multiple DRBG instances with seed chaining. 818 o The default RAND method makes use of a DRBG. 819 o There is a public and private DRBG instance. 820 o The DRBG instances are fork-safe. 821 o Keep all global DRBG instances on the secure heap if it is enabled. 822 o The public and private DRBG instance are per thread for lock free 823 operation 824 [Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre] 825 826 *) Changed Configure so it only says what it does and doesn't dump 827 so much data. Instead, ./configdata.pm should be used as a script 828 to display all sorts of configuration data. 829 [Richard Levitte] 830 831 *) Added processing of "make variables" to Configure. 832 [Richard Levitte] 833 834 *) Added SHA512/224 and SHA512/256 algorithm support. 835 [Paul Dale] 836 837 *) The last traces of Netware support, first removed in 1.1.0, have 838 now been removed. 839 [Rich Salz] 840 841 *) Get rid of Makefile.shared, and in the process, make the processing 842 of certain files (rc.obj, or the .def/.map/.opt files produced from 843 the ordinal files) more visible and hopefully easier to trace and 844 debug (or make silent). 845 [Richard Levitte] 846 847 *) Make it possible to have environment variable assignments as 848 arguments to config / Configure. 849 [Richard Levitte] 850 851 *) Add multi-prime RSA (RFC 8017) support. 852 [Paul Yang] 853 854 *) Add SM3 implemented according to GB/T 32905-2016 855 [ Jack Lloyd <jack.lloyd@ribose.com>, 856 Ronald Tse <ronald.tse@ribose.com>, 857 Erick Borsboom <erick.borsboom@ribose.com> ] 858 859 *) Add 'Maximum Fragment Length' TLS extension negotiation and support 860 as documented in RFC6066. 861 Based on a patch from Tomasz Moń 862 [Filipe Raimundo da Silva] 863 864 *) Add SM4 implemented according to GB/T 32907-2016. 865 [ Jack Lloyd <jack.lloyd@ribose.com>, 866 Ronald Tse <ronald.tse@ribose.com>, 867 Erick Borsboom <erick.borsboom@ribose.com> ] 868 869 *) Reimplement -newreq-nodes and ERR_error_string_n; the 870 original author does not agree with the license change. 871 [Rich Salz] 872 873 *) Add ARIA AEAD TLS support. 874 [Jon Spillett] 875 876 *) Some macro definitions to support VS6 have been removed. Visual 877 Studio 6 has not worked since 1.1.0 878 [Rich Salz] 879 880 *) Add ERR_clear_last_mark(), to allow callers to clear the last mark 881 without clearing the errors. 882 [Richard Levitte] 883 884 *) Add "atfork" functions. If building on a system that without 885 pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application 886 requirements. The RAND facility now uses/requires this. 887 [Rich Salz] 888 889 *) Add SHA3. 890 [Andy Polyakov] 891 892 *) The UI API becomes a permanent and integral part of libcrypto, i.e. 893 not possible to disable entirely. However, it's still possible to 894 disable the console reading UI method, UI_OpenSSL() (use UI_null() 895 as a fallback). 896 897 To disable, configure with 'no-ui-console'. 'no-ui' is still 898 possible to use as an alias. Check at compile time with the 899 macro OPENSSL_NO_UI_CONSOLE. The macro OPENSSL_NO_UI is still 900 possible to check and is an alias for OPENSSL_NO_UI_CONSOLE. 901 [Richard Levitte] 902 903 *) Add a STORE module, which implements a uniform and URI based reader of 904 stores that can contain keys, certificates, CRLs and numerous other 905 objects. The main API is loosely based on a few stdio functions, 906 and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof, 907 OSSL_STORE_error and OSSL_STORE_close. 908 The implementation uses backends called "loaders" to implement arbitrary 909 URI schemes. There is one built in "loader" for the 'file' scheme. 910 [Richard Levitte] 911 912 *) Add devcrypto engine. This has been implemented against cryptodev-linux, 913 then adjusted to work on FreeBSD 8.4 as well. 914 Enable by configuring with 'enable-devcryptoeng'. This is done by default 915 on BSD implementations, as cryptodev.h is assumed to exist on all of them. 916 [Richard Levitte] 917 918 *) Module names can prefixed with OSSL_ or OPENSSL_. This affects 919 util/mkerr.pl, which is adapted to allow those prefixes, leading to 920 error code calls like this: 921 922 OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER); 923 924 With this change, we claim the namespaces OSSL and OPENSSL in a manner 925 that can be encoded in C. For the foreseeable future, this will only 926 affect new modules. 927 [Richard Levitte and Tim Hudson] 928 929 *) Removed BSD cryptodev engine. 930 [Rich Salz] 931 932 *) Add a build target 'build_all_generated', to build all generated files 933 and only that. This can be used to prepare everything that requires 934 things like perl for a system that lacks perl and then move everything 935 to that system and do the rest of the build there. 936 [Richard Levitte] 937 938 *) In the UI interface, make it possible to duplicate the user data. This 939 can be used by engines that need to retain the data for a longer time 940 than just the call where this user data is passed. 941 [Richard Levitte] 942 943 *) Ignore the '-named_curve auto' value for compatibility of applications 944 with OpenSSL 1.0.2. 945 [Tomas Mraz <tmraz@fedoraproject.org>] 946 947 *) Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2 948 bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such 949 alerts across multiple records (some of which could be empty). In practice 950 it make no sense to send an empty alert record, or to fragment one. TLSv1.3 951 prohibits this altogether and other libraries (BoringSSL, NSS) do not 952 support this at all. Supporting it adds significant complexity to the 953 record layer, and its removal is unlikely to cause interoperability 954 issues. 955 [Matt Caswell] 956 957 *) Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed 958 with Z. These are meant to replace LONG and ZLONG and to be size safe. 959 The use of LONG and ZLONG is discouraged and scheduled for deprecation 960 in OpenSSL 1.2.0. 961 [Richard Levitte] 962 963 *) Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string, 964 'z' is to be used for [s]size_t, and 'j' - with [u]int64_t. 965 [Richard Levitte, Andy Polyakov] 966 967 *) Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine() 968 does for RSA, etc. 969 [Richard Levitte] 970 971 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 972 platform rather than 'mingw'. 973 [Richard Levitte] 974 975 *) The functions X509_STORE_add_cert and X509_STORE_add_crl return 976 success if they are asked to add an object which already exists 977 in the store. This change cascades to other functions which load 978 certificates and CRLs. 979 [Paul Dale] 980 981 *) x86_64 assembly pack: annotate code with DWARF CFI directives to 982 facilitate stack unwinding even from assembly subroutines. 983 [Andy Polyakov] 984 985 *) Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN. 986 Also remove OPENSSL_GLOBAL entirely, as it became a no-op. 987 [Richard Levitte] 988 989 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 990 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 991 which is the minimum version we support. 992 [Richard Levitte] 993 994 *) Certificate time validation (X509_cmp_time) enforces stricter 995 compliance with RFC 5280. Fractional seconds and timezone offsets 996 are no longer allowed. 997 [Emilia Käsper] 998 999 *) Add support for ARIA 1000 [Paul Dale] 1001 1002 *) s_client will now send the Server Name Indication (SNI) extension by 1003 default unless the new "-noservername" option is used. The server name is 1004 based on the host provided to the "-connect" option unless overridden by 1005 using "-servername". 1006 [Matt Caswell] 1007 1008 *) Add support for SipHash 1009 [Todd Short] 1010 1011 *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0 1012 or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to 1013 prevent issues where no progress is being made and the peer continually 1014 sends unrecognised record types, using up resources processing them. 1015 [Matt Caswell] 1016 1017 *) 'openssl passwd' can now produce SHA256 and SHA512 based output, 1018 using the algorithm defined in 1019 https://www.akkadia.org/drepper/SHA-crypt.txt 1020 [Richard Levitte] 1021 1022 *) Heartbeat support has been removed; the ABI is changed for now. 1023 [Richard Levitte, Rich Salz] 1024 1025 *) Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd. 1026 [Emilia Käsper] 1027 1028 *) The RSA "null" method, which was partially supported to avoid patent 1029 issues, has been replaced to always returns NULL. 1030 [Rich Salz] 1031 1032 1033 Changes between 1.1.0h and 1.1.0i [xx XXX xxxx] 1034 1035 *) Client DoS due to large DH parameter 1036 1037 During key agreement in a TLS handshake using a DH(E) based ciphersuite a 1038 malicious server can send a very large prime value to the client. This will 1039 cause the client to spend an unreasonably long period of time generating a 1040 key for this prime resulting in a hang until the client has finished. This 1041 could be exploited in a Denial Of Service attack. 1042 1043 This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken 1044 (CVE-2018-0732) 1045 [Guido Vranken] 1046 1047 *) Cache timing vulnerability in RSA Key Generation 1048 1049 The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to 1050 a cache timing side channel attack. An attacker with sufficient access to 1051 mount cache timing attacks during the RSA key generation process could 1052 recover the private key. 1053 1054 This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera 1055 Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia. 1056 (CVE-2018-0737) 1057 [Billy Brumley] 1058 1059 *) Make EVP_PKEY_asn1_new() a bit stricter about its input. A NULL pem_str 1060 parameter is no longer accepted, as it leads to a corrupt table. NULL 1061 pem_str is reserved for alias entries only. 1062 [Richard Levitte] 1063 1064 *) Revert blinding in ECDSA sign and instead make problematic addition 1065 length-invariant. Switch even to fixed-length Montgomery multiplication. 1066 [Andy Polyakov] 1067 1068 *) Change generating and checking of primes so that the error rate of not 1069 being prime depends on the intended use based on the size of the input. 1070 For larger primes this will result in more rounds of Miller-Rabin. 1071 The maximal error rate for primes with more than 1080 bits is lowered 1072 to 2^-128. 1073 [Kurt Roeckx, Annie Yousar] 1074 1075 *) Increase the number of Miller-Rabin rounds for DSA key generating to 64. 1076 [Kurt Roeckx] 1077 1078 *) Add blinding to ECDSA and DSA signatures to protect against side channel 1079 attacks discovered by Keegan Ryan (NCC Group). 1080 [Matt Caswell] 1081 1082 *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we 1083 now allow empty (zero character) pass phrases. 1084 [Richard Levitte] 1085 1086 *) Certificate time validation (X509_cmp_time) enforces stricter 1087 compliance with RFC 5280. Fractional seconds and timezone offsets 1088 are no longer allowed. 1089 [Emilia Käsper] 1090 1091 *) Fixed a text canonicalisation bug in CMS 1092 1093 Where a CMS detached signature is used with text content the text goes 1094 through a canonicalisation process first prior to signing or verifying a 1095 signature. This process strips trailing space at the end of lines, converts 1096 line terminators to CRLF and removes additional trailing line terminators 1097 at the end of a file. A bug in the canonicalisation process meant that 1098 some characters, such as form-feed, were incorrectly treated as whitespace 1099 and removed. This is contrary to the specification (RFC5485). This fix 1100 could mean that detached text data signed with an earlier version of 1101 OpenSSL 1.1.0 may fail to verify using the fixed version, or text data 1102 signed with a fixed OpenSSL may fail to verify with an earlier version of 1103 OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data 1104 and use the "-binary" flag (for the "cms" command line application) or set 1105 the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()). 1106 [Matt Caswell] 1107 1108 Changes between 1.1.0g and 1.1.0h [27 Mar 2018] 1109 1110 *) Constructed ASN.1 types with a recursive definition could exceed the stack 1111 1112 Constructed ASN.1 types with a recursive definition (such as can be found 1113 in PKCS7) could eventually exceed the stack given malicious input with 1114 excessive recursion. This could result in a Denial Of Service attack. There 1115 are no such structures used within SSL/TLS that come from untrusted sources 1116 so this is considered safe. 1117 1118 This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz 1119 project. 1120 (CVE-2018-0739) 1121 [Matt Caswell] 1122 1123 *) Incorrect CRYPTO_memcmp on HP-UX PA-RISC 1124 1125 Because of an implementation bug the PA-RISC CRYPTO_memcmp function is 1126 effectively reduced to only comparing the least significant bit of each 1127 byte. This allows an attacker to forge messages that would be considered as 1128 authenticated in an amount of tries lower than that guaranteed by the 1129 security claims of the scheme. The module can only be compiled by the 1130 HP-UX assembler, so that only HP-UX PA-RISC targets are affected. 1131 1132 This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg 1133 (IBM). 1134 (CVE-2018-0733) 1135 [Andy Polyakov] 1136 1137 *) Add a build target 'build_all_generated', to build all generated files 1138 and only that. This can be used to prepare everything that requires 1139 things like perl for a system that lacks perl and then move everything 1140 to that system and do the rest of the build there. 1141 [Richard Levitte] 1142 1143 *) Backport SSL_OP_NO_RENGOTIATION 1144 1145 OpenSSL 1.0.2 and below had the ability to disable renegotiation using the 1146 (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity 1147 changes this is no longer possible in 1.1.0. Therefore the new 1148 SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to 1149 1.1.0 to provide equivalent functionality. 1150 1151 Note that if an application built against 1.1.0h headers (or above) is run 1152 using an older version of 1.1.0 (prior to 1.1.0h) then the option will be 1153 accepted but nothing will happen, i.e. renegotiation will not be prevented. 1154 [Matt Caswell] 1155 1156 *) Removed the OS390-Unix config target. It relied on a script that doesn't 1157 exist. 1158 [Rich Salz] 1159 1160 *) rsaz_1024_mul_avx2 overflow bug on x86_64 1161 1162 There is an overflow bug in the AVX2 Montgomery multiplication procedure 1163 used in exponentiation with 1024-bit moduli. No EC algorithms are affected. 1164 Analysis suggests that attacks against RSA and DSA as a result of this 1165 defect would be very difficult to perform and are not believed likely. 1166 Attacks against DH1024 are considered just feasible, because most of the 1167 work necessary to deduce information about a private key may be performed 1168 offline. The amount of resources required for such an attack would be 1169 significant. However, for an attack on TLS to be meaningful, the server 1170 would have to share the DH1024 private key among multiple clients, which is 1171 no longer an option since CVE-2016-0701. 1172 1173 This only affects processors that support the AVX2 but not ADX extensions 1174 like Intel Haswell (4th generation). 1175 1176 This issue was reported to OpenSSL by David Benjamin (Google). The issue 1177 was originally found via the OSS-Fuzz project. 1178 (CVE-2017-3738) 1179 [Andy Polyakov] 1180 1181 Changes between 1.1.0f and 1.1.0g [2 Nov 2017] 1182 1183 *) bn_sqrx8x_internal carry bug on x86_64 1184 1185 There is a carry propagating bug in the x86_64 Montgomery squaring 1186 procedure. No EC algorithms are affected. Analysis suggests that attacks 1187 against RSA and DSA as a result of this defect would be very difficult to 1188 perform and are not believed likely. Attacks against DH are considered just 1189 feasible (although very difficult) because most of the work necessary to 1190 deduce information about a private key may be performed offline. The amount 1191 of resources required for such an attack would be very significant and 1192 likely only accessible to a limited number of attackers. An attacker would 1193 additionally need online access to an unpatched system using the target 1194 private key in a scenario with persistent DH parameters and a private 1195 key that is shared between multiple clients. 1196 1197 This only affects processors that support the BMI1, BMI2 and ADX extensions 1198 like Intel Broadwell (5th generation) and later or AMD Ryzen. 1199 1200 This issue was reported to OpenSSL by the OSS-Fuzz project. 1201 (CVE-2017-3736) 1202 [Andy Polyakov] 1203 1204 *) Malformed X.509 IPAddressFamily could cause OOB read 1205 1206 If an X.509 certificate has a malformed IPAddressFamily extension, 1207 OpenSSL could do a one-byte buffer overread. The most likely result 1208 would be an erroneous display of the certificate in text format. 1209 1210 This issue was reported to OpenSSL by the OSS-Fuzz project. 1211 (CVE-2017-3735) 1212 [Rich Salz] 1213 1214 Changes between 1.1.0e and 1.1.0f [25 May 2017] 1215 1216 *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target 1217 platform rather than 'mingw'. 1218 [Richard Levitte] 1219 1220 *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c. 1221 VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1, 1222 which is the minimum version we support. 1223 [Richard Levitte] 1224 1225 Changes between 1.1.0d and 1.1.0e [16 Feb 2017] 1226 1227 *) Encrypt-Then-Mac renegotiation crash 1228 1229 During a renegotiation handshake if the Encrypt-Then-Mac extension is 1230 negotiated where it was not in the original handshake (or vice-versa) then 1231 this can cause OpenSSL to crash (dependant on ciphersuite). Both clients 1232 and servers are affected. 1233 1234 This issue was reported to OpenSSL by Joe Orton (Red Hat). 1235 (CVE-2017-3733) 1236 [Matt Caswell] 1237 1238 Changes between 1.1.0c and 1.1.0d [26 Jan 2017] 1239 1240 *) Truncated packet could crash via OOB read 1241 1242 If one side of an SSL/TLS path is running on a 32-bit host and a specific 1243 cipher is being used, then a truncated packet can cause that host to 1244 perform an out-of-bounds read, usually resulting in a crash. 1245 1246 This issue was reported to OpenSSL by Robert Święcki of Google. 1247 (CVE-2017-3731) 1248 [Andy Polyakov] 1249 1250 *) Bad (EC)DHE parameters cause a client crash 1251 1252 If a malicious server supplies bad parameters for a DHE or ECDHE key 1253 exchange then this can result in the client attempting to dereference a 1254 NULL pointer leading to a client crash. This could be exploited in a Denial 1255 of Service attack. 1256 1257 This issue was reported to OpenSSL by Guido Vranken. 1258 (CVE-2017-3730) 1259 [Matt Caswell] 1260 1261 *) BN_mod_exp may produce incorrect results on x86_64 1262 1263 There is a carry propagating bug in the x86_64 Montgomery squaring 1264 procedure. No EC algorithms are affected. Analysis suggests that attacks 1265 against RSA and DSA as a result of this defect would be very difficult to 1266 perform and are not believed likely. Attacks against DH are considered just 1267 feasible (although very difficult) because most of the work necessary to 1268 deduce information about a private key may be performed offline. The amount 1269 of resources required for such an attack would be very significant and 1270 likely only accessible to a limited number of attackers. An attacker would 1271 additionally need online access to an unpatched system using the target 1272 private key in a scenario with persistent DH parameters and a private 1273 key that is shared between multiple clients. For example this can occur by 1274 default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very 1275 similar to CVE-2015-3193 but must be treated as a separate problem. 1276 1277 This issue was reported to OpenSSL by the OSS-Fuzz project. 1278 (CVE-2017-3732) 1279 [Andy Polyakov] 1280 1281 Changes between 1.1.0b and 1.1.0c [10 Nov 2016] 1282 1283 *) ChaCha20/Poly1305 heap-buffer-overflow 1284 1285 TLS connections using *-CHACHA20-POLY1305 ciphersuites are susceptible to 1286 a DoS attack by corrupting larger payloads. This can result in an OpenSSL 1287 crash. This issue is not considered to be exploitable beyond a DoS. 1288 1289 This issue was reported to OpenSSL by Robert Święcki (Google Security Team) 1290 (CVE-2016-7054) 1291 [Richard Levitte] 1292 1293 *) CMS Null dereference 1294 1295 Applications parsing invalid CMS structures can crash with a NULL pointer 1296 dereference. This is caused by a bug in the handling of the ASN.1 CHOICE 1297 type in OpenSSL 1.1.0 which can result in a NULL value being passed to the 1298 structure callback if an attempt is made to free certain invalid encodings. 1299 Only CHOICE structures using a callback which do not handle NULL value are 1300 affected. 1301 1302 This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure. 1303 (CVE-2016-7053) 1304 [Stephen Henson] 1305 1306 *) Montgomery multiplication may produce incorrect results 1307 1308 There is a carry propagating bug in the Broadwell-specific Montgomery 1309 multiplication procedure that handles input lengths divisible by, but 1310 longer than 256 bits. Analysis suggests that attacks against RSA, DSA 1311 and DH private keys are impossible. This is because the subroutine in 1312 question is not used in operations with the private key itself and an input 1313 of the attacker's direct choice. Otherwise the bug can manifest itself as 1314 transient authentication and key negotiation failures or reproducible 1315 erroneous outcome of public-key operations with specially crafted input. 1316 Among EC algorithms only Brainpool P-512 curves are affected and one 1317 presumably can attack ECDH key negotiation. Impact was not analyzed in 1318 detail, because pre-requisites for attack are considered unlikely. Namely 1319 multiple clients have to choose the curve in question and the server has to 1320 share the private key among them, neither of which is default behaviour. 1321 Even then only clients that chose the curve will be affected. 1322 1323 This issue was publicly reported as transient failures and was not 1324 initially recognized as a security issue. Thanks to Richard Morgan for 1325 providing reproducible case. 1326 (CVE-2016-7055) 1327 [Andy Polyakov] 1328 1329 *) Removed automatic addition of RPATH in shared libraries and executables, 1330 as this was a remainder from OpenSSL 1.0.x and isn't needed any more. 1331 [Richard Levitte] 1332 1333 Changes between 1.1.0a and 1.1.0b [26 Sep 2016] 1334 1335 *) Fix Use After Free for large message sizes 1336 1337 The patch applied to address CVE-2016-6307 resulted in an issue where if a 1338 message larger than approx 16k is received then the underlying buffer to 1339 store the incoming message is reallocated and moved. Unfortunately a 1340 dangling pointer to the old location is left which results in an attempt to 1341 write to the previously freed location. This is likely to result in a 1342 crash, however it could potentially lead to execution of arbitrary code. 1343 1344 This issue only affects OpenSSL 1.1.0a. 1345 1346 This issue was reported to OpenSSL by Robert Święcki. 1347 (CVE-2016-6309) 1348 [Matt Caswell] 1349 1350 Changes between 1.1.0 and 1.1.0a [22 Sep 2016] 1351 1352 *) OCSP Status Request extension unbounded memory growth 1353 1354 A malicious client can send an excessively large OCSP Status Request 1355 extension. If that client continually requests renegotiation, sending a 1356 large OCSP Status Request extension each time, then there will be unbounded 1357 memory growth on the server. This will eventually lead to a Denial Of 1358 Service attack through memory exhaustion. Servers with a default 1359 configuration are vulnerable even if they do not support OCSP. Builds using 1360 the "no-ocsp" build time option are not affected. 1361 1362 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 1363 (CVE-2016-6304) 1364 [Matt Caswell] 1365 1366 *) SSL_peek() hang on empty record 1367 1368 OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer 1369 sends an empty record. This could be exploited by a malicious peer in a 1370 Denial Of Service attack. 1371 1372 This issue was reported to OpenSSL by Alex Gaynor. 1373 (CVE-2016-6305) 1374 [Matt Caswell] 1375 1376 *) Excessive allocation of memory in tls_get_message_header() and 1377 dtls1_preprocess_fragment() 1378 1379 A (D)TLS message includes 3 bytes for its length in the header for the 1380 message. This would allow for messages up to 16Mb in length. Messages of 1381 this length are excessive and OpenSSL includes a check to ensure that a 1382 peer is sending reasonably sized messages in order to avoid too much memory 1383 being consumed to service a connection. A flaw in the logic of version 1384 1.1.0 means that memory for the message is allocated too early, prior to 1385 the excessive message length check. Due to way memory is allocated in 1386 OpenSSL this could mean an attacker could force up to 21Mb to be allocated 1387 to service a connection. This could lead to a Denial of Service through 1388 memory exhaustion. However, the excessive message length check still takes 1389 place, and this would cause the connection to immediately fail. Assuming 1390 that the application calls SSL_free() on the failed connection in a timely 1391 manner then the 21Mb of allocated memory will then be immediately freed 1392 again. Therefore the excessive memory allocation will be transitory in 1393 nature. This then means that there is only a security impact if: 1394 1395 1) The application does not call SSL_free() in a timely manner in the event 1396 that the connection fails 1397 or 1398 2) The application is working in a constrained environment where there is 1399 very little free memory 1400 or 1401 3) The attacker initiates multiple connection attempts such that there are 1402 multiple connections in a state where memory has been allocated for the 1403 connection; SSL_free() has not yet been called; and there is insufficient 1404 memory to service the multiple requests. 1405 1406 Except in the instance of (1) above any Denial Of Service is likely to be 1407 transitory because as soon as the connection fails the memory is 1408 subsequently freed again in the SSL_free() call. However there is an 1409 increased risk during this period of application crashes due to the lack of 1410 memory - which would then mean a more serious Denial of Service. 1411 1412 This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.) 1413 (CVE-2016-6307 and CVE-2016-6308) 1414 [Matt Caswell] 1415 1416 *) solaris-x86-cc, i.e. 32-bit configuration with vendor compiler, 1417 had to be removed. Primary reason is that vendor assembler can't 1418 assemble our modules with -KPIC flag. As result it, assembly 1419 support, was not even available as option. But its lack means 1420 lack of side-channel resistant code, which is incompatible with 1421 security by todays standards. Fortunately gcc is readily available 1422 prepackaged option, which we firmly point at... 1423 [Andy Polyakov] 1424 1425 Changes between 1.0.2h and 1.1.0 [25 Aug 2016] 1426 1427 *) Windows command-line tool supports UTF-8 opt-in option for arguments 1428 and console input. Setting OPENSSL_WIN32_UTF8 environment variable 1429 (to any value) allows Windows user to access PKCS#12 file generated 1430 with Windows CryptoAPI and protected with non-ASCII password, as well 1431 as files generated under UTF-8 locale on Linux also protected with 1432 non-ASCII password. 1433 [Andy Polyakov] 1434 1435 *) To mitigate the SWEET32 attack (CVE-2016-2183), 3DES cipher suites 1436 have been disabled by default and removed from DEFAULT, just like RC4. 1437 See the RC4 item below to re-enable both. 1438 [Rich Salz] 1439 1440 *) The method for finding the storage location for the Windows RAND seed file 1441 has changed. First we check %RANDFILE%. If that is not set then we check 1442 the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If 1443 all else fails we fall back to C:\. 1444 [Matt Caswell] 1445 1446 *) The EVP_EncryptUpdate() function has had its return type changed from void 1447 to int. A return of 0 indicates and error while a return of 1 indicates 1448 success. 1449 [Matt Caswell] 1450 1451 *) The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and 1452 DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch 1453 off the constant time implementation for RSA, DSA and DH have been made 1454 no-ops and deprecated. 1455 [Matt Caswell] 1456 1457 *) Windows RAND implementation was simplified to only get entropy by 1458 calling CryptGenRandom(). Various other RAND-related tickets 1459 were also closed. 1460 [Joseph Wylie Yandle, Rich Salz] 1461 1462 *) The stack and lhash API's were renamed to start with OPENSSL_SK_ 1463 and OPENSSL_LH_, respectively. The old names are available 1464 with API compatibility. They new names are now completely documented. 1465 [Rich Salz] 1466 1467 *) Unify TYPE_up_ref(obj) methods signature. 1468 SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(), 1469 X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an 1470 int (instead of void) like all others TYPE_up_ref() methods. 1471 So now these methods also check the return value of CRYPTO_atomic_add(), 1472 and the validity of object reference counter. 1473 [fdasilvayy@gmail.com] 1474 1475 *) With Windows Visual Studio builds, the .pdb files are installed 1476 alongside the installed libraries and executables. For a static 1477 library installation, ossl_static.pdb is the associate compiler 1478 generated .pdb file to be used when linking programs. 1479 [Richard Levitte] 1480 1481 *) Remove openssl.spec. Packaging files belong with the packagers. 1482 [Richard Levitte] 1483 1484 *) Automatic Darwin/OSX configuration has had a refresh, it will now 1485 recognise x86_64 architectures automatically. You can still decide 1486 to build for a different bitness with the environment variable 1487 KERNEL_BITS (can be 32 or 64), for example: 1488 1489 KERNEL_BITS=32 ./config 1490 1491 [Richard Levitte] 1492 1493 *) Change default algorithms in pkcs8 utility to use PKCS#5 v2.0, 1494 256 bit AES and HMAC with SHA256. 1495 [Steve Henson] 1496 1497 *) Remove support for MIPS o32 ABI on IRIX (and IRIX only). 1498 [Andy Polyakov] 1499 1500 *) Triple-DES ciphers have been moved from HIGH to MEDIUM. 1501 [Rich Salz] 1502 1503 *) To enable users to have their own config files and build file templates, 1504 Configure looks in the directory indicated by the environment variable 1505 OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/ 1506 directory. On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical 1507 name and is used as is. 1508 [Richard Levitte] 1509 1510 *) The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX, 1511 X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD. The unused type 1512 X509_CERT_FILE_CTX was removed. 1513 [Rich Salz] 1514 1515 *) "shared" builds are now the default. To create only static libraries use 1516 the "no-shared" Configure option. 1517 [Matt Caswell] 1518 1519 *) Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options. 1520 All of these option have not worked for some while and are fundamental 1521 algorithms. 1522 [Matt Caswell] 1523 1524 *) Make various cleanup routines no-ops and mark them as deprecated. Most 1525 global cleanup functions are no longer required because they are handled 1526 via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages). 1527 Explicitly de-initing can cause problems (e.g. where a library that uses 1528 OpenSSL de-inits, but an application is still using it). The affected 1529 functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(), 1530 EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(), 1531 RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and 1532 COMP_zlib_cleanup(). 1533 [Matt Caswell] 1534 1535 *) --strict-warnings no longer enables runtime debugging options 1536 such as REF_DEBUG. Instead, debug options are automatically 1537 enabled with '--debug' builds. 1538 [Andy Polyakov, Emilia Käsper] 1539 1540 *) Made DH and DH_METHOD opaque. The structures for managing DH objects 1541 have been moved out of the public header files. New functions for managing 1542 these have been added. 1543 [Matt Caswell] 1544 1545 *) Made RSA and RSA_METHOD opaque. The structures for managing RSA 1546 objects have been moved out of the public header files. New 1547 functions for managing these have been added. 1548 [Richard Levitte] 1549 1550 *) Made DSA and DSA_METHOD opaque. The structures for managing DSA objects 1551 have been moved out of the public header files. New functions for managing 1552 these have been added. 1553 [Matt Caswell] 1554 1555 *) Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been 1556 moved out of the public header files. New functions for managing these 1557 have been added. 1558 [Matt Caswell] 1559 1560 *) Removed no-rijndael as a config option. Rijndael is an old name for AES. 1561 [Matt Caswell] 1562 1563 *) Removed the mk1mf build scripts. 1564 [Richard Levitte] 1565 1566 *) Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so 1567 it is always safe to #include a header now. 1568 [Rich Salz] 1569 1570 *) Removed the aged BC-32 config and all its supporting scripts 1571 [Richard Levitte] 1572 1573 *) Removed support for Ultrix, Netware, and OS/2. 1574 [Rich Salz] 1575 1576 *) Add support for HKDF. 1577 [Alessandro Ghedini] 1578 1579 *) Add support for blake2b and blake2s 1580 [Bill Cox] 1581 1582 *) Added support for "pipelining". Ciphers that have the 1583 EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple 1584 encryptions/decryptions simultaneously. There are currently no built-in 1585 ciphers with this property but the expectation is that engines will be able 1586 to offer it to significantly improve throughput. Support has been extended 1587 into libssl so that multiple records for a single connection can be 1588 processed in one go (for >=TLS 1.1). 1589 [Matt Caswell] 1590 1591 *) Added the AFALG engine. This is an async capable engine which is able to 1592 offload work to the Linux kernel. In this initial version it only supports 1593 AES128-CBC. The kernel must be version 4.1.0 or greater. 1594 [Catriona Lucey] 1595 1596 *) OpenSSL now uses a new threading API. It is no longer necessary to 1597 set locking callbacks to use OpenSSL in a multi-threaded environment. There 1598 are two supported threading models: pthreads and windows threads. It is 1599 also possible to configure OpenSSL at compile time for "no-threads". The 1600 old threading API should no longer be used. The functions have been 1601 replaced with "no-op" compatibility macros. 1602 [Alessandro Ghedini, Matt Caswell] 1603 1604 *) Modify behavior of ALPN to invoke callback after SNI/servername 1605 callback, such that updates to the SSL_CTX affect ALPN. 1606 [Todd Short] 1607 1608 *) Add SSL_CIPHER queries for authentication and key-exchange. 1609 [Todd Short] 1610 1611 *) Changes to the DEFAULT cipherlist: 1612 - Prefer (EC)DHE handshakes over plain RSA. 1613 - Prefer AEAD ciphers over legacy ciphers. 1614 - Prefer ECDSA over RSA when both certificates are available. 1615 - Prefer TLSv1.2 ciphers/PRF. 1616 - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the 1617 default cipherlist. 1618 [Emilia Käsper] 1619 1620 *) Change the ECC default curve list to be this, in order: x25519, 1621 secp256r1, secp521r1, secp384r1. 1622 [Rich Salz] 1623 1624 *) RC4 based libssl ciphersuites are now classed as "weak" ciphers and are 1625 disabled by default. They can be re-enabled using the 1626 enable-weak-ssl-ciphers option to Configure. 1627 [Matt Caswell] 1628 1629 *) If the server has ALPN configured, but supports no protocols that the 1630 client advertises, send a fatal "no_application_protocol" alert. 1631 This behaviour is SHALL in RFC 7301, though it isn't universally 1632 implemented by other servers. 1633 [Emilia Käsper] 1634 1635 *) Add X25519 support. 1636 Add ASN.1 and EVP_PKEY methods for X25519. This includes support 1637 for public and private key encoding using the format documented in 1638 draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports 1639 key generation and key derivation. 1640 1641 TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses 1642 X25519(29). 1643 [Steve Henson] 1644 1645 *) Deprecate SRP_VBASE_get_by_user. 1646 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 1647 In order to fix an unavoidable memory leak (CVE-2016-0798), 1648 SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP 1649 seed, even if the seed is configured. 1650 1651 Users should use SRP_VBASE_get1_by_user instead. Note that in 1652 SRP_VBASE_get1_by_user, caller must free the returned value. Note 1653 also that even though configuring the SRP seed attempts to hide 1654 invalid usernames by continuing the handshake with fake 1655 credentials, this behaviour is not constant time and no strong 1656 guarantees are made that the handshake is indistinguishable from 1657 that of a valid user. 1658 [Emilia Käsper] 1659 1660 *) Configuration change; it's now possible to build dynamic engines 1661 without having to build shared libraries and vice versa. This 1662 only applies to the engines in engines/, those in crypto/engine/ 1663 will always be built into libcrypto (i.e. "static"). 1664 1665 Building dynamic engines is enabled by default; to disable, use 1666 the configuration option "disable-dynamic-engine". 1667 1668 The only requirements for building dynamic engines are the 1669 presence of the DSO module and building with position independent 1670 code, so they will also automatically be disabled if configuring 1671 with "disable-dso" or "disable-pic". 1672 1673 The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE 1674 are also taken away from openssl/opensslconf.h, as they are 1675 irrelevant. 1676 [Richard Levitte] 1677 1678 *) Configuration change; if there is a known flag to compile 1679 position independent code, it will always be applied on the 1680 libcrypto and libssl object files, and never on the application 1681 object files. This means other libraries that use routines from 1682 libcrypto / libssl can be made into shared libraries regardless 1683 of how OpenSSL was configured. 1684 1685 If this isn't desirable, the configuration options "disable-pic" 1686 or "no-pic" can be used to disable the use of PIC. This will 1687 also disable building shared libraries and dynamic engines. 1688 [Richard Levitte] 1689 1690 *) Removed JPAKE code. It was experimental and has no wide use. 1691 [Rich Salz] 1692 1693 *) The INSTALL_PREFIX Makefile variable has been renamed to 1694 DESTDIR. That makes for less confusion on what this variable 1695 is for. Also, the configuration option --install_prefix is 1696 removed. 1697 [Richard Levitte] 1698 1699 *) Heartbeat for TLS has been removed and is disabled by default 1700 for DTLS; configure with enable-heartbeats. Code that uses the 1701 old #define's might need to be updated. 1702 [Emilia Käsper, Rich Salz] 1703 1704 *) Rename REF_CHECK to REF_DEBUG. 1705 [Rich Salz] 1706 1707 *) New "unified" build system 1708 1709 The "unified" build system is aimed to be a common system for all 1710 platforms we support. With it comes new support for VMS. 1711 1712 This system builds supports building in a different directory tree 1713 than the source tree. It produces one Makefile (for unix family 1714 or lookalikes), or one descrip.mms (for VMS). 1715 1716 The source of information to make the Makefile / descrip.mms is 1717 small files called 'build.info', holding the necessary 1718 information for each directory with source to compile, and a 1719 template in Configurations, like unix-Makefile.tmpl or 1720 descrip.mms.tmpl. 1721 1722 With this change, the library names were also renamed on Windows 1723 and on VMS. They now have names that are closer to the standard 1724 on Unix, and include the major version number, and in certain 1725 cases, the architecture they are built for. See "Notes on shared 1726 libraries" in INSTALL. 1727 1728 We rely heavily on the perl module Text::Template. 1729 [Richard Levitte] 1730 1731 *) Added support for auto-initialisation and de-initialisation of the library. 1732 OpenSSL no longer requires explicit init or deinit routines to be called, 1733 except in certain circumstances. See the OPENSSL_init_crypto() and 1734 OPENSSL_init_ssl() man pages for further information. 1735 [Matt Caswell] 1736 1737 *) The arguments to the DTLSv1_listen function have changed. Specifically the 1738 "peer" argument is now expected to be a BIO_ADDR object. 1739 1740 *) Rewrite of BIO networking library. The BIO library lacked consistent 1741 support of IPv6, and adding it required some more extensive 1742 modifications. This introduces the BIO_ADDR and BIO_ADDRINFO types, 1743 which hold all types of addresses and chains of address information. 1744 It also introduces a new API, with functions like BIO_socket, 1745 BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept. 1746 The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram 1747 have been adapted accordingly. 1748 [Richard Levitte] 1749 1750 *) RSA_padding_check_PKCS1_type_1 now accepts inputs with and without 1751 the leading 0-byte. 1752 [Emilia Käsper] 1753 1754 *) CRIME protection: disable compression by default, even if OpenSSL is 1755 compiled with zlib enabled. Applications can still enable compression 1756 by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by 1757 using the SSL_CONF library to configure compression. 1758 [Emilia Käsper] 1759 1760 *) The signature of the session callback configured with 1761 SSL_CTX_sess_set_get_cb was changed. The read-only input buffer 1762 was explicitly marked as 'const unsigned char*' instead of 1763 'unsigned char*'. 1764 [Emilia Käsper] 1765 1766 *) Always DPURIFY. Remove the use of uninitialized memory in the 1767 RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op. 1768 [Emilia Käsper] 1769 1770 *) Removed many obsolete configuration items, including 1771 DES_PTR, DES_RISC1, DES_RISC2, DES_INT 1772 MD2_CHAR, MD2_INT, MD2_LONG 1773 BF_PTR, BF_PTR2 1774 IDEA_SHORT, IDEA_LONG 1775 RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX 1776 [Rich Salz, with advice from Andy Polyakov] 1777 1778 *) Many BN internals have been moved to an internal header file. 1779 [Rich Salz with help from Andy Polyakov] 1780 1781 *) Configuration and writing out the results from it has changed. 1782 Files such as Makefile include/openssl/opensslconf.h and are now 1783 produced through general templates, such as Makefile.in and 1784 crypto/opensslconf.h.in and some help from the perl module 1785 Text::Template. 1786 1787 Also, the center of configuration information is no longer 1788 Makefile. Instead, Configure produces a perl module in 1789 configdata.pm which holds most of the config data (in the hash 1790 table %config), the target data that comes from the target 1791 configuration in one of the Configurations/*.conf files (in 1792 %target). 1793 [Richard Levitte] 1794 1795 *) To clarify their intended purposes, the Configure options 1796 --prefix and --openssldir change their semantics, and become more 1797 straightforward and less interdependent. 1798 1799 --prefix shall be used exclusively to give the location INSTALLTOP 1800 where programs, scripts, libraries, include files and manuals are 1801 going to be installed. The default is now /usr/local. 1802 1803 --openssldir shall be used exclusively to give the default 1804 location OPENSSLDIR where certificates, private keys, CRLs are 1805 managed. This is also where the default openssl.cnf gets 1806 installed. 1807 If the directory given with this option is a relative path, the 1808 values of both the --prefix value and the --openssldir value will 1809 be combined to become OPENSSLDIR. 1810 The default for --openssldir is INSTALLTOP/ssl. 1811 1812 Anyone who uses --openssldir to specify where OpenSSL is to be 1813 installed MUST change to use --prefix instead. 1814 [Richard Levitte] 1815 1816 *) The GOST engine was out of date and therefore it has been removed. An up 1817 to date GOST engine is now being maintained in an external repository. 1818 See: https://wiki.openssl.org/index.php/Binaries. Libssl still retains 1819 support for GOST ciphersuites (these are only activated if a GOST engine 1820 is present). 1821 [Matt Caswell] 1822 1823 *) EGD is no longer supported by default; use enable-egd when 1824 configuring. 1825 [Ben Kaduk and Rich Salz] 1826 1827 *) The distribution now has Makefile.in files, which are used to 1828 create Makefile's when Configure is run. *Configure must be run 1829 before trying to build now.* 1830 [Rich Salz] 1831 1832 *) The return value for SSL_CIPHER_description() for error conditions 1833 has changed. 1834 [Rich Salz] 1835 1836 *) Support for RFC6698/RFC7671 DANE TLSA peer authentication. 1837 1838 Obtaining and performing DNSSEC validation of TLSA records is 1839 the application's responsibility. The application provides 1840 the TLSA records of its choice to OpenSSL, and these are then 1841 used to authenticate the peer. 1842 1843 The TLSA records need not even come from DNS. They can, for 1844 example, be used to implement local end-entity certificate or 1845 trust-anchor "pinning", where the "pin" data takes the form 1846 of TLSA records, which can augment or replace verification 1847 based on the usual WebPKI public certification authorities. 1848 [Viktor Dukhovni] 1849 1850 *) Revert default OPENSSL_NO_DEPRECATED setting. Instead OpenSSL 1851 continues to support deprecated interfaces in default builds. 1852 However, applications are strongly advised to compile their 1853 source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides 1854 the declarations of all interfaces deprecated in 0.9.8, 1.0.0 1855 or the 1.1.0 releases. 1856 1857 In environments in which all applications have been ported to 1858 not use any deprecated interfaces OpenSSL's Configure script 1859 should be used with the --api=1.1.0 option to entirely remove 1860 support for the deprecated features from the library and 1861 unconditionally disable them in the installed headers. 1862 Essentially the same effect can be achieved with the "no-deprecated" 1863 argument to Configure, except that this will always restrict 1864 the build to just the latest API, rather than a fixed API 1865 version. 1866 1867 As applications are ported to future revisions of the API, 1868 they should update their compile-time OPENSSL_API_COMPAT define 1869 accordingly, but in most cases should be able to continue to 1870 compile with later releases. 1871 1872 The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are 1873 0x10000000L and 0x00908000L, respectively. However those 1874 versions did not support the OPENSSL_API_COMPAT feature, and 1875 so applications are not typically tested for explicit support 1876 of just the undeprecated features of either release. 1877 [Viktor Dukhovni] 1878 1879 *) Add support for setting the minimum and maximum supported protocol. 1880 It can bet set via the SSL_set_min_proto_version() and 1881 SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and 1882 MaxProtocol. It's recommended to use the new APIs to disable 1883 protocols instead of disabling individual protocols using 1884 SSL_set_options() or SSL_CONF's Protocol. This change also 1885 removes support for disabling TLS 1.2 in the OpenSSL TLS 1886 client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT. 1887 [Kurt Roeckx] 1888 1889 *) Support for ChaCha20 and Poly1305 added to libcrypto and libssl. 1890 [Andy Polyakov] 1891 1892 *) New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD 1893 and integrates ECDSA and ECDH functionality into EC. Implementations can 1894 now redirect key generation and no longer need to convert to or from 1895 ECDSA_SIG format. 1896 1897 Note: the ecdsa.h and ecdh.h headers are now no longer needed and just 1898 include the ec.h header file instead. 1899 [Steve Henson] 1900 1901 *) Remove support for all 40 and 56 bit ciphers. This includes all the export 1902 ciphers who are no longer supported and drops support the ephemeral RSA key 1903 exchange. The LOW ciphers currently doesn't have any ciphers in it. 1904 [Kurt Roeckx] 1905 1906 *) Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX 1907 opaque. For HMAC_CTX, the following constructors and destructors 1908 were added: 1909 1910 HMAC_CTX *HMAC_CTX_new(void); 1911 void HMAC_CTX_free(HMAC_CTX *ctx); 1912 1913 For EVP_MD and EVP_CIPHER, complete APIs to create, fill and 1914 destroy such methods has been added. See EVP_MD_meth_new(3) and 1915 EVP_CIPHER_meth_new(3) for documentation. 1916 1917 Additional changes: 1918 1) EVP_MD_CTX_cleanup(), EVP_CIPHER_CTX_cleanup() and 1919 HMAC_CTX_cleanup() were removed. HMAC_CTX_reset() and 1920 EVP_MD_CTX_reset() should be called instead to reinitialise 1921 an already created structure. 1922 2) For consistency with the majority of our object creators and 1923 destructors, EVP_MD_CTX_(create|destroy) were renamed to 1924 EVP_MD_CTX_(new|free). The old names are retained as macros 1925 for deprecated builds. 1926 [Richard Levitte] 1927 1928 *) Added ASYNC support. Libcrypto now includes the async sub-library to enable 1929 cryptographic operations to be performed asynchronously as long as an 1930 asynchronous capable engine is used. See the ASYNC_start_job() man page for 1931 further details. Libssl has also had this capability integrated with the 1932 introduction of the new mode SSL_MODE_ASYNC and associated error 1933 SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man 1934 pages. This work was developed in partnership with Intel Corp. 1935 [Matt Caswell] 1936 1937 *) SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is 1938 always enabled now. If you want to disable the support you should 1939 exclude it using the list of supported ciphers. This also means that the 1940 "-no_ecdhe" option has been removed from s_server. 1941 [Kurt Roeckx] 1942 1943 *) SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls 1944 SSL_{CTX_}set1_curves() which can set a list. 1945 [Kurt Roeckx] 1946 1947 *) Remove support for SSL_{CTX_}set_tmp_ecdh_callback(). You should set the 1948 curve you want to support using SSL_{CTX_}set1_curves(). 1949 [Kurt Roeckx] 1950 1951 *) State machine rewrite. The state machine code has been significantly 1952 refactored in order to remove much duplication of code and solve issues 1953 with the old code (see ssl/statem/README for further details). This change 1954 does have some associated API changes. Notably the SSL_state() function 1955 has been removed and replaced by SSL_get_state which now returns an 1956 "OSSL_HANDSHAKE_STATE" instead of an int. SSL_set_state() has been removed 1957 altogether. The previous handshake states defined in ssl.h and ssl3.h have 1958 also been removed. 1959 [Matt Caswell] 1960 1961 *) All instances of the string "ssleay" in the public API were replaced 1962 with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's) 1963 Some error codes related to internal RSA_eay API's were renamed. 1964 [Rich Salz] 1965 1966 *) The demo files in crypto/threads were moved to demo/threads. 1967 [Rich Salz] 1968 1969 *) Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp, 1970 sureware and ubsec. 1971 [Matt Caswell, Rich Salz] 1972 1973 *) New ASN.1 embed macro. 1974 1975 New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the 1976 structure is not allocated: it is part of the parent. That is instead of 1977 1978 FOO *x; 1979 1980 it must be: 1981 1982 FOO x; 1983 1984 This reduces memory fragmentation and make it impossible to accidentally 1985 set a mandatory field to NULL. 1986 1987 This currently only works for some fields specifically a SEQUENCE, CHOICE, 1988 or ASN1_STRING type which is part of a parent SEQUENCE. Since it is 1989 equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or 1990 SEQUENCE OF. 1991 [Steve Henson] 1992 1993 *) Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled. 1994 [Emilia Käsper] 1995 1996 *) Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although 1997 in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also 1998 an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add 1999 DES and RC4 ciphersuites. 2000 [Matt Caswell] 2001 2002 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 2003 This changes the decoding behaviour for some invalid messages, 2004 though the change is mostly in the more lenient direction, and 2005 legacy behaviour is preserved as much as possible. 2006 [Emilia Käsper] 2007 2008 *) Fix no-stdio build. 2009 [ David Woodhouse <David.Woodhouse@intel.com> and also 2010 Ivan Nestlerode <ivan.nestlerode@sonos.com> ] 2011 2012 *) New testing framework 2013 The testing framework has been largely rewritten and is now using 2014 perl and the perl modules Test::Harness and an extended variant of 2015 Test::More called OpenSSL::Test to do its work. All test scripts in 2016 test/ have been rewritten into test recipes, and all direct calls to 2017 executables in test/Makefile have become individual recipes using the 2018 simplified testing OpenSSL::Test::Simple. 2019 2020 For documentation on our testing modules, do: 2021 2022 perldoc test/testlib/OpenSSL/Test/Simple.pm 2023 perldoc test/testlib/OpenSSL/Test.pm 2024 2025 [Richard Levitte] 2026 2027 *) Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT 2028 are used; the latter aborts on memory leaks (usually checked on exit). 2029 Some undocumented "set malloc, etc., hooks" functions were removed 2030 and others were changed. All are now documented. 2031 [Rich Salz] 2032 2033 *) In DSA_generate_parameters_ex, if the provided seed is too short, 2034 return an error 2035 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 2036 2037 *) Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites 2038 from RFC4279, RFC4785, RFC5487, RFC5489. 2039 2040 Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the 2041 original RSA_PSK patch. 2042 [Steve Henson] 2043 2044 *) Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay 2045 era flag was never set throughout the codebase (only read). Also removed 2046 SSL3_FLAGS_POP_BUFFER which was only used if 2047 SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set. 2048 [Matt Caswell] 2049 2050 *) Changed the default name options in the "ca", "crl", "req" and "x509" 2051 to be "oneline" instead of "compat". 2052 [Richard Levitte] 2053 2054 *) Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're 2055 not aware of clients that still exhibit this bug, and the workaround 2056 hasn't been working properly for a while. 2057 [Emilia Käsper] 2058 2059 *) The return type of BIO_number_read() and BIO_number_written() as well as 2060 the corresponding num_read and num_write members in the BIO structure has 2061 changed from unsigned long to uint64_t. On platforms where an unsigned 2062 long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is 2063 transferred. 2064 [Matt Caswell] 2065 2066 *) Given the pervasive nature of TLS extensions it is inadvisable to run 2067 OpenSSL without support for them. It also means that maintaining 2068 the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably 2069 not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed. 2070 [Matt Caswell] 2071 2072 *) Removed support for the two export grade static DH ciphersuites 2073 EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites 2074 were newly added (along with a number of other static DH ciphersuites) to 2075 1.0.2. However the two export ones have *never* worked since they were 2076 introduced. It seems strange in any case to be adding new export 2077 ciphersuites, and given "logjam" it also does not seem correct to fix them. 2078 [Matt Caswell] 2079 2080 *) Version negotiation has been rewritten. In particular SSLv23_method(), 2081 SSLv23_client_method() and SSLv23_server_method() have been deprecated, 2082 and turned into macros which simply call the new preferred function names 2083 TLS_method(), TLS_client_method() and TLS_server_method(). All new code 2084 should use the new names instead. Also as part of this change the ssl23.h 2085 header file has been removed. 2086 [Matt Caswell] 2087 2088 *) Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This 2089 code and the associated standard is no longer considered fit-for-purpose. 2090 [Matt Caswell] 2091 2092 *) RT2547 was closed. When generating a private key, try to make the 2093 output file readable only by the owner. This behavior change might 2094 be noticeable when interacting with other software. 2095 2096 *) Documented all exdata functions. Added CRYPTO_free_ex_index. 2097 Added a test. 2098 [Rich Salz] 2099 2100 *) Added HTTP GET support to the ocsp command. 2101 [Rich Salz] 2102 2103 *) Changed default digest for the dgst and enc commands from MD5 to 2104 sha256 2105 [Rich Salz] 2106 2107 *) RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead. 2108 [Matt Caswell] 2109 2110 *) Added support for TLS extended master secret from 2111 draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an 2112 initial patch which was a great help during development. 2113 [Steve Henson] 2114 2115 *) All libssl internal structures have been removed from the public header 2116 files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is 2117 now redundant). Users should not attempt to access internal structures 2118 directly. Instead they should use the provided API functions. 2119 [Matt Caswell] 2120 2121 *) config has been changed so that by default OPENSSL_NO_DEPRECATED is used. 2122 Access to deprecated functions can be re-enabled by running config with 2123 "enable-deprecated". In addition applications wishing to use deprecated 2124 functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour 2125 will, by default, disable some transitive includes that previously existed 2126 in the header files (e.g. ec.h will no longer, by default, include bn.h) 2127 [Matt Caswell] 2128 2129 *) Added support for OCB mode. OpenSSL has been granted a patent license 2130 compatible with the OpenSSL license for use of OCB. Details are available 2131 at https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf. Support 2132 for OCB can be removed by calling config with no-ocb. 2133 [Matt Caswell] 2134 2135 *) SSLv2 support has been removed. It still supports receiving a SSLv2 2136 compatible client hello. 2137 [Kurt Roeckx] 2138 2139 *) Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz], 2140 done while fixing the error code for the key-too-small case. 2141 [Annie Yousar <a.yousar@informatik.hu-berlin.de>] 2142 2143 *) CA.sh has been removed; use CA.pl instead. 2144 [Rich Salz] 2145 2146 *) Removed old DES API. 2147 [Rich Salz] 2148 2149 *) Remove various unsupported platforms: 2150 Sony NEWS4 2151 BEOS and BEOS_R5 2152 NeXT 2153 SUNOS 2154 MPE/iX 2155 Sinix/ReliantUNIX RM400 2156 DGUX 2157 NCR 2158 Tandem 2159 Cray 2160 16-bit platforms such as WIN16 2161 [Rich Salz] 2162 2163 *) Clean up OPENSSL_NO_xxx #define's 2164 Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF 2165 Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx 2166 OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC 2167 OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160 2168 OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO 2169 Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY 2170 OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP 2171 OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK 2172 OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY 2173 Remove MS_STATIC; it's a relic from platforms <32 bits. 2174 [Rich Salz] 2175 2176 *) Cleaned up dead code 2177 Remove all but one '#ifdef undef' which is to be looked at. 2178 [Rich Salz] 2179 2180 *) Clean up calling of xxx_free routines. 2181 Just like free(), fix most of the xxx_free routines to accept 2182 NULL. Remove the non-null checks from callers. Save much code. 2183 [Rich Salz] 2184 2185 *) Add secure heap for storage of private keys (when possible). 2186 Add BIO_s_secmem(), CBIGNUM, etc. 2187 Contributed by Akamai Technologies under our Corporate CLA. 2188 [Rich Salz] 2189 2190 *) Experimental support for a new, fast, unbiased prime candidate generator, 2191 bn_probable_prime_dh_coprime(). Not currently used by any prime generator. 2192 [Felix Laurie von Massenbach <felix@erbridge.co.uk>] 2193 2194 *) New output format NSS in the sess_id command line tool. This allows 2195 exporting the session id and the master key in NSS keylog format. 2196 [Martin Kaiser <martin@kaiser.cx>] 2197 2198 *) Harmonize version and its documentation. -f flag is used to display 2199 compilation flags. 2200 [mancha <mancha1@zoho.com>] 2201 2202 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 2203 in i2d_ECPrivateKey. Thanks to Ted Unangst for feedback on this issue. 2204 [mancha <mancha1@zoho.com>] 2205 2206 *) Fix some double frees. These are not thought to be exploitable. 2207 [mancha <mancha1@zoho.com>] 2208 2209 *) A missing bounds check in the handling of the TLS heartbeat extension 2210 can be used to reveal up to 64k of memory to a connected client or 2211 server. 2212 2213 Thanks for Neel Mehta of Google Security for discovering this bug and to 2214 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 2215 preparing the fix (CVE-2014-0160) 2216 [Adam Langley, Bodo Moeller] 2217 2218 *) Fix for the attack described in the paper "Recovering OpenSSL 2219 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 2220 by Yuval Yarom and Naomi Benger. Details can be obtained from: 2221 http://eprint.iacr.org/2014/140 2222 2223 Thanks to Yuval Yarom and Naomi Benger for discovering this 2224 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 2225 [Yuval Yarom and Naomi Benger] 2226 2227 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 2228 this fixes a limitation in previous versions of OpenSSL. 2229 [Steve Henson] 2230 2231 *) Experimental encrypt-then-mac support. 2232 2233 Experimental support for encrypt then mac from 2234 draft-gutmann-tls-encrypt-then-mac-02.txt 2235 2236 To enable it set the appropriate extension number (0x42 for the test 2237 server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42 2238 2239 For non-compliant peers (i.e. just about everything) this should have no 2240 effect. 2241 2242 WARNING: EXPERIMENTAL, SUBJECT TO CHANGE. 2243 2244 [Steve Henson] 2245 2246 *) Add EVP support for key wrapping algorithms, to avoid problems with 2247 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 2248 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 2249 algorithms and include tests cases. 2250 [Steve Henson] 2251 2252 *) Extend CMS code to support RSA-PSS signatures and RSA-OAEP for 2253 enveloped data. 2254 [Steve Henson] 2255 2256 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 2257 MGF1 digest and OAEP label. 2258 [Steve Henson] 2259 2260 *) Make openssl verify return errors. 2261 [Chris Palmer <palmer@google.com> and Ben Laurie] 2262 2263 *) New function ASN1_TIME_diff to calculate the difference between two 2264 ASN1_TIME structures or one structure and the current time. 2265 [Steve Henson] 2266 2267 *) Update fips_test_suite to support multiple command line options. New 2268 test to induce all self test errors in sequence and check expected 2269 failures. 2270 [Steve Henson] 2271 2272 *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and 2273 sign or verify all in one operation. 2274 [Steve Henson] 2275 2276 *) Add fips_algvs: a multicall fips utility incorporating all the algorithm 2277 test programs and fips_test_suite. Includes functionality to parse 2278 the minimal script output of fipsalgest.pl directly. 2279 [Steve Henson] 2280 2281 *) Add authorisation parameter to FIPS_module_mode_set(). 2282 [Steve Henson] 2283 2284 *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves. 2285 [Steve Henson] 2286 2287 *) Use separate DRBG fields for internal and external flags. New function 2288 FIPS_drbg_health_check() to perform on demand health checking. Add 2289 generation tests to fips_test_suite with reduced health check interval to 2290 demonstrate periodic health checking. Add "nodh" option to 2291 fips_test_suite to skip very slow DH test. 2292 [Steve Henson] 2293 2294 *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers 2295 based on NID. 2296 [Steve Henson] 2297 2298 *) More extensive health check for DRBG checking many more failure modes. 2299 New function FIPS_selftest_drbg_all() to handle every possible DRBG 2300 combination: call this in fips_test_suite. 2301 [Steve Henson] 2302 2303 *) Add support for canonical generation of DSA parameter 'g'. See 2304 FIPS 186-3 A.2.3. 2305 2306 *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and 2307 POST to handle HMAC cases. 2308 [Steve Henson] 2309 2310 *) Add functions FIPS_module_version() and FIPS_module_version_text() 2311 to return numerical and string versions of the FIPS module number. 2312 [Steve Henson] 2313 2314 *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and 2315 FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented 2316 outside the validated module in the FIPS capable OpenSSL. 2317 [Steve Henson] 2318 2319 *) Minor change to DRBG entropy callback semantics. In some cases 2320 there is no multiple of the block length between min_len and 2321 max_len. Allow the callback to return more than max_len bytes 2322 of entropy but discard any extra: it is the callback's responsibility 2323 to ensure that the extra data discarded does not impact the 2324 requested amount of entropy. 2325 [Steve Henson] 2326 2327 *) Add PRNG security strength checks to RSA, DSA and ECDSA using 2328 information in FIPS186-3, SP800-57 and SP800-131A. 2329 [Steve Henson] 2330 2331 *) CCM support via EVP. Interface is very similar to GCM case except we 2332 must supply all data in one chunk (i.e. no update, final) and the 2333 message length must be supplied if AAD is used. Add algorithm test 2334 support. 2335 [Steve Henson] 2336 2337 *) Initial version of POST overhaul. Add POST callback to allow the status 2338 of POST to be monitored and/or failures induced. Modify fips_test_suite 2339 to use callback. Always run all selftests even if one fails. 2340 [Steve Henson] 2341 2342 *) XTS support including algorithm test driver in the fips_gcmtest program. 2343 Note: this does increase the maximum key length from 32 to 64 bytes but 2344 there should be no binary compatibility issues as existing applications 2345 will never use XTS mode. 2346 [Steve Henson] 2347 2348 *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies 2349 to OpenSSL RAND code and replace with a tiny FIPS RAND API which also 2350 performs algorithm blocking for unapproved PRNG types. Also do not 2351 set PRNG type in FIPS_mode_set(): leave this to the application. 2352 Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with 2353 the standard OpenSSL PRNG: set additional data to a date time vector. 2354 [Steve Henson] 2355 2356 *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*. 2357 This shouldn't present any incompatibility problems because applications 2358 shouldn't be using these directly and any that are will need to rethink 2359 anyway as the X9.31 PRNG is now deprecated by FIPS 140-2 2360 [Steve Henson] 2361 2362 *) Extensive self tests and health checking required by SP800-90 DRBG. 2363 Remove strength parameter from FIPS_drbg_instantiate and always 2364 instantiate at maximum supported strength. 2365 [Steve Henson] 2366 2367 *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing. 2368 [Steve Henson] 2369 2370 *) New algorithm test program fips_dhvs to handle DH primitives only testing. 2371 [Steve Henson] 2372 2373 *) New function DH_compute_key_padded() to compute a DH key and pad with 2374 leading zeroes if needed: this complies with SP800-56A et al. 2375 [Steve Henson] 2376 2377 *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by 2378 anything, incomplete, subject to change and largely untested at present. 2379 [Steve Henson] 2380 2381 *) Modify fipscanisteronly build option to only build the necessary object 2382 files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile. 2383 [Steve Henson] 2384 2385 *) Add experimental option FIPSSYMS to give all symbols in 2386 fipscanister.o and FIPS or fips prefix. This will avoid 2387 conflicts with future versions of OpenSSL. Add perl script 2388 util/fipsas.pl to preprocess assembly language source files 2389 and rename any affected symbols. 2390 [Steve Henson] 2391 2392 *) Add selftest checks and algorithm block of non-fips algorithms in 2393 FIPS mode. Remove DES2 from selftests. 2394 [Steve Henson] 2395 2396 *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just 2397 return internal method without any ENGINE dependencies. Add new 2398 tiny fips sign and verify functions. 2399 [Steve Henson] 2400 2401 *) New build option no-ec2m to disable characteristic 2 code. 2402 [Steve Henson] 2403 2404 *) New build option "fipscanisteronly". This only builds fipscanister.o 2405 and (currently) associated fips utilities. Uses the file Makefile.fips 2406 instead of Makefile.org as the prototype. 2407 [Steve Henson] 2408 2409 *) Add some FIPS mode restrictions to GCM. Add internal IV generator. 2410 Update fips_gcmtest to use IV generator. 2411 [Steve Henson] 2412 2413 *) Initial, experimental EVP support for AES-GCM. AAD can be input by 2414 setting output buffer to NULL. The *Final function must be 2415 called although it will not retrieve any additional data. The tag 2416 can be set or retrieved with a ctrl. The IV length is by default 12 2417 bytes (96 bits) but can be set to an alternative value. If the IV 2418 length exceeds the maximum IV length (currently 16 bytes) it cannot be 2419 set before the key. 2420 [Steve Henson] 2421 2422 *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the 2423 underlying do_cipher function handles all cipher semantics itself 2424 including padding and finalisation. This is useful if (for example) 2425 an ENGINE cipher handles block padding itself. The behaviour of 2426 do_cipher is subtly changed if this flag is set: the return value 2427 is the number of characters written to the output buffer (zero is 2428 no longer an error code) or a negative error code. Also if the 2429 input buffer is NULL and length 0 finalisation should be performed. 2430 [Steve Henson] 2431 2432 *) If a candidate issuer certificate is already part of the constructed 2433 path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case. 2434 [Steve Henson] 2435 2436 *) Improve forward-security support: add functions 2437 2438 void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure)) 2439 void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure)) 2440 2441 for use by SSL/TLS servers; the callback function will be called whenever a 2442 new session is created, and gets to decide whether the session may be 2443 cached to make it resumable (return 0) or not (return 1). (As by the 2444 SSL/TLS protocol specifications, the session_id sent by the server will be 2445 empty to indicate that the session is not resumable; also, the server will 2446 not generate RFC 4507 (RFC 5077) session tickets.) 2447 2448 A simple reasonable callback implementation is to return is_forward_secure. 2449 This parameter will be set to 1 or 0 depending on the ciphersuite selected 2450 by the SSL/TLS server library, indicating whether it can provide forward 2451 security. 2452 [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)] 2453 2454 *) New -verify_name option in command line utilities to set verification 2455 parameters by name. 2456 [Steve Henson] 2457 2458 *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE. 2459 Add CMAC pkey methods. 2460 [Steve Henson] 2461 2462 *) Experimental renegotiation in s_server -www mode. If the client 2463 browses /reneg connection is renegotiated. If /renegcert it is 2464 renegotiated requesting a certificate. 2465 [Steve Henson] 2466 2467 *) Add an "external" session cache for debugging purposes to s_server. This 2468 should help trace issues which normally are only apparent in deployed 2469 multi-process servers. 2470 [Steve Henson] 2471 2472 *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where 2473 return value is ignored. NB. The functions RAND_add(), RAND_seed(), 2474 BIO_set_cipher() and some obscure PEM functions were changed so they 2475 can now return an error. The RAND changes required a change to the 2476 RAND_METHOD structure. 2477 [Steve Henson] 2478 2479 *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of 2480 a gcc attribute to warn if the result of a function is ignored. This 2481 is enable if DEBUG_UNUSED is set. Add to several functions in evp.h 2482 whose return value is often ignored. 2483 [Steve Henson] 2484 2485 *) New -noct, -requestct, -requirect and -ctlogfile options for s_client. 2486 These allow SCTs (signed certificate timestamps) to be requested and 2487 validated when establishing a connection. 2488 [Rob Percival <robpercival@google.com>] 2489 2490 Changes between 1.0.2g and 1.0.2h [3 May 2016] 2491 2492 *) Prevent padding oracle in AES-NI CBC MAC check 2493 2494 A MITM attacker can use a padding oracle attack to decrypt traffic 2495 when the connection uses an AES CBC cipher and the server support 2496 AES-NI. 2497 2498 This issue was introduced as part of the fix for Lucky 13 padding 2499 attack (CVE-2013-0169). The padding check was rewritten to be in 2500 constant time by making sure that always the same bytes are read and 2501 compared against either the MAC or padding bytes. But it no longer 2502 checked that there was enough data to have both the MAC and padding 2503 bytes. 2504 2505 This issue was reported by Juraj Somorovsky using TLS-Attacker. 2506 (CVE-2016-2107) 2507 [Kurt Roeckx] 2508 2509 *) Fix EVP_EncodeUpdate overflow 2510 2511 An overflow can occur in the EVP_EncodeUpdate() function which is used for 2512 Base64 encoding of binary data. If an attacker is able to supply very large 2513 amounts of input data then a length check can overflow resulting in a heap 2514 corruption. 2515 2516 Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by 2517 the PEM_write_bio* family of functions. These are mainly used within the 2518 OpenSSL command line applications, so any application which processes data 2519 from an untrusted source and outputs it as a PEM file should be considered 2520 vulnerable to this issue. User applications that call these APIs directly 2521 with large amounts of untrusted data may also be vulnerable. 2522 2523 This issue was reported by Guido Vranken. 2524 (CVE-2016-2105) 2525 [Matt Caswell] 2526 2527 *) Fix EVP_EncryptUpdate overflow 2528 2529 An overflow can occur in the EVP_EncryptUpdate() function. If an attacker 2530 is able to supply very large amounts of input data after a previous call to 2531 EVP_EncryptUpdate() with a partial block then a length check can overflow 2532 resulting in a heap corruption. Following an analysis of all OpenSSL 2533 internal usage of the EVP_EncryptUpdate() function all usage is one of two 2534 forms. The first form is where the EVP_EncryptUpdate() call is known to be 2535 the first called function after an EVP_EncryptInit(), and therefore that 2536 specific call must be safe. The second form is where the length passed to 2537 EVP_EncryptUpdate() can be seen from the code to be some small value and 2538 therefore there is no possibility of an overflow. Since all instances are 2539 one of these two forms, it is believed that there can be no overflows in 2540 internal code due to this problem. It should be noted that 2541 EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths. 2542 Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances 2543 of these calls have also been analysed too and it is believed there are no 2544 instances in internal usage where an overflow could occur. 2545 2546 This issue was reported by Guido Vranken. 2547 (CVE-2016-2106) 2548 [Matt Caswell] 2549 2550 *) Prevent ASN.1 BIO excessive memory allocation 2551 2552 When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio() 2553 a short invalid encoding can cause allocation of large amounts of memory 2554 potentially consuming excessive resources or exhausting memory. 2555 2556 Any application parsing untrusted data through d2i BIO functions is 2557 affected. The memory based functions such as d2i_X509() are *not* affected. 2558 Since the memory based functions are used by the TLS library, TLS 2559 applications are not affected. 2560 2561 This issue was reported by Brian Carpenter. 2562 (CVE-2016-2109) 2563 [Stephen Henson] 2564 2565 *) EBCDIC overread 2566 2567 ASN1 Strings that are over 1024 bytes can cause an overread in applications 2568 using the X509_NAME_oneline() function on EBCDIC systems. This could result 2569 in arbitrary stack data being returned in the buffer. 2570 2571 This issue was reported by Guido Vranken. 2572 (CVE-2016-2176) 2573 [Matt Caswell] 2574 2575 *) Modify behavior of ALPN to invoke callback after SNI/servername 2576 callback, such that updates to the SSL_CTX affect ALPN. 2577 [Todd Short] 2578 2579 *) Remove LOW from the DEFAULT cipher list. This removes singles DES from the 2580 default. 2581 [Kurt Roeckx] 2582 2583 *) Only remove the SSLv2 methods with the no-ssl2-method option. When the 2584 methods are enabled and ssl2 is disabled the methods return NULL. 2585 [Kurt Roeckx] 2586 2587 Changes between 1.0.2f and 1.0.2g [1 Mar 2016] 2588 2589 * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL. 2590 Builds that are not configured with "enable-weak-ssl-ciphers" will not 2591 provide any "EXPORT" or "LOW" strength ciphers. 2592 [Viktor Dukhovni] 2593 2594 * Disable SSLv2 default build, default negotiation and weak ciphers. SSLv2 2595 is by default disabled at build-time. Builds that are not configured with 2596 "enable-ssl2" will not support SSLv2. Even if "enable-ssl2" is used, 2597 users who want to negotiate SSLv2 via the version-flexible SSLv23_method() 2598 will need to explicitly call either of: 2599 2600 SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2); 2601 or 2602 SSL_clear_options(ssl, SSL_OP_NO_SSLv2); 2603 2604 as appropriate. Even if either of those is used, or the application 2605 explicitly uses the version-specific SSLv2_method() or its client and 2606 server variants, SSLv2 ciphers vulnerable to exhaustive search key 2607 recovery have been removed. Specifically, the SSLv2 40-bit EXPORT 2608 ciphers, and SSLv2 56-bit DES are no longer available. 2609 (CVE-2016-0800) 2610 [Viktor Dukhovni] 2611 2612 *) Fix a double-free in DSA code 2613 2614 A double free bug was discovered when OpenSSL parses malformed DSA private 2615 keys and could lead to a DoS attack or memory corruption for applications 2616 that receive DSA private keys from untrusted sources. This scenario is 2617 considered rare. 2618 2619 This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using 2620 libFuzzer. 2621 (CVE-2016-0705) 2622 [Stephen Henson] 2623 2624 *) Disable SRP fake user seed to address a server memory leak. 2625 2626 Add a new method SRP_VBASE_get1_by_user that handles the seed properly. 2627 2628 SRP_VBASE_get_by_user had inconsistent memory management behaviour. 2629 In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user 2630 was changed to ignore the "fake user" SRP seed, even if the seed 2631 is configured. 2632 2633 Users should use SRP_VBASE_get1_by_user instead. Note that in 2634 SRP_VBASE_get1_by_user, caller must free the returned value. Note 2635 also that even though configuring the SRP seed attempts to hide 2636 invalid usernames by continuing the handshake with fake 2637 credentials, this behaviour is not constant time and no strong 2638 guarantees are made that the handshake is indistinguishable from 2639 that of a valid user. 2640 (CVE-2016-0798) 2641 [Emilia Käsper] 2642 2643 *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption 2644 2645 In the BN_hex2bn function the number of hex digits is calculated using an 2646 int value |i|. Later |bn_expand| is called with a value of |i * 4|. For 2647 large values of |i| this can result in |bn_expand| not allocating any 2648 memory because |i * 4| is negative. This can leave the internal BIGNUM data 2649 field as NULL leading to a subsequent NULL ptr deref. For very large values 2650 of |i|, the calculation |i * 4| could be a positive value smaller than |i|. 2651 In this case memory is allocated to the internal BIGNUM data field, but it 2652 is insufficiently sized leading to heap corruption. A similar issue exists 2653 in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn 2654 is ever called by user applications with very large untrusted hex/dec data. 2655 This is anticipated to be a rare occurrence. 2656 2657 All OpenSSL internal usage of these functions use data that is not expected 2658 to be untrusted, e.g. config file data or application command line 2659 arguments. If user developed applications generate config file data based 2660 on untrusted data then it is possible that this could also lead to security 2661 consequences. This is also anticipated to be rare. 2662 2663 This issue was reported to OpenSSL by Guido Vranken. 2664 (CVE-2016-0797) 2665 [Matt Caswell] 2666 2667 *) Fix memory issues in BIO_*printf functions 2668 2669 The internal |fmtstr| function used in processing a "%s" format string in 2670 the BIO_*printf functions could overflow while calculating the length of a 2671 string and cause an OOB read when printing very long strings. 2672 2673 Additionally the internal |doapr_outch| function can attempt to write to an 2674 OOB memory location (at an offset from the NULL pointer) in the event of a 2675 memory allocation failure. In 1.0.2 and below this could be caused where 2676 the size of a buffer to be allocated is greater than INT_MAX. E.g. this 2677 could be in processing a very long "%s" format string. Memory leaks can 2678 also occur. 2679 2680 The first issue may mask the second issue dependent on compiler behaviour. 2681 These problems could enable attacks where large amounts of untrusted data 2682 is passed to the BIO_*printf functions. If applications use these functions 2683 in this way then they could be vulnerable. OpenSSL itself uses these 2684 functions when printing out human-readable dumps of ASN.1 data. Therefore 2685 applications that print this data could be vulnerable if the data is from 2686 untrusted sources. OpenSSL command line applications could also be 2687 vulnerable where they print out ASN.1 data, or if untrusted data is passed 2688 as command line arguments. 2689 2690 Libssl is not considered directly vulnerable. Additionally certificates etc 2691 received via remote connections via libssl are also unlikely to be able to 2692 trigger these issues because of message size limits enforced within libssl. 2693 2694 This issue was reported to OpenSSL Guido Vranken. 2695 (CVE-2016-0799) 2696 [Matt Caswell] 2697 2698 *) Side channel attack on modular exponentiation 2699 2700 A side-channel attack was found which makes use of cache-bank conflicts on 2701 the Intel Sandy-Bridge microarchitecture which could lead to the recovery 2702 of RSA keys. The ability to exploit this issue is limited as it relies on 2703 an attacker who has control of code in a thread running on the same 2704 hyper-threaded core as the victim thread which is performing decryptions. 2705 2706 This issue was reported to OpenSSL by Yuval Yarom, The University of 2707 Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and 2708 Nadia Heninger, University of Pennsylvania with more information at 2709 http://cachebleed.info. 2710 (CVE-2016-0702) 2711 [Andy Polyakov] 2712 2713 *) Change the req app to generate a 2048-bit RSA/DSA key by default, 2714 if no keysize is specified with default_bits. This fixes an 2715 omission in an earlier change that changed all RSA/DSA key generation 2716 apps to use 2048 bits by default. 2717 [Emilia Käsper] 2718 2719 Changes between 1.0.2e and 1.0.2f [28 Jan 2016] 2720 *) DH small subgroups 2721 2722 Historically OpenSSL only ever generated DH parameters based on "safe" 2723 primes. More recently (in version 1.0.2) support was provided for 2724 generating X9.42 style parameter files such as those required for RFC 5114 2725 support. The primes used in such files may not be "safe". Where an 2726 application is using DH configured with parameters based on primes that are 2727 not "safe" then an attacker could use this fact to find a peer's private 2728 DH exponent. This attack requires that the attacker complete multiple 2729 handshakes in which the peer uses the same private DH exponent. For example 2730 this could be used to discover a TLS server's private DH exponent if it's 2731 reusing the private DH exponent or it's using a static DH ciphersuite. 2732 2733 OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in 2734 TLS. It is not on by default. If the option is not set then the server 2735 reuses the same private DH exponent for the life of the server process and 2736 would be vulnerable to this attack. It is believed that many popular 2737 applications do set this option and would therefore not be at risk. 2738 2739 The fix for this issue adds an additional check where a "q" parameter is 2740 available (as is the case in X9.42 based parameters). This detects the 2741 only known attack, and is the only possible defense for static DH 2742 ciphersuites. This could have some performance impact. 2743 2744 Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by 2745 default and cannot be disabled. This could have some performance impact. 2746 2747 This issue was reported to OpenSSL by Antonio Sanso (Adobe). 2748 (CVE-2016-0701) 2749 [Matt Caswell] 2750 2751 *) SSLv2 doesn't block disabled ciphers 2752 2753 A malicious client can negotiate SSLv2 ciphers that have been disabled on 2754 the server and complete SSLv2 handshakes even if all SSLv2 ciphers have 2755 been disabled, provided that the SSLv2 protocol was not also disabled via 2756 SSL_OP_NO_SSLv2. 2757 2758 This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram 2759 and Sebastian Schinzel. 2760 (CVE-2015-3197) 2761 [Viktor Dukhovni] 2762 2763 Changes between 1.0.2d and 1.0.2e [3 Dec 2015] 2764 2765 *) BN_mod_exp may produce incorrect results on x86_64 2766 2767 There is a carry propagating bug in the x86_64 Montgomery squaring 2768 procedure. No EC algorithms are affected. Analysis suggests that attacks 2769 against RSA and DSA as a result of this defect would be very difficult to 2770 perform and are not believed likely. Attacks against DH are considered just 2771 feasible (although very difficult) because most of the work necessary to 2772 deduce information about a private key may be performed offline. The amount 2773 of resources required for such an attack would be very significant and 2774 likely only accessible to a limited number of attackers. An attacker would 2775 additionally need online access to an unpatched system using the target 2776 private key in a scenario with persistent DH parameters and a private 2777 key that is shared between multiple clients. For example this can occur by 2778 default in OpenSSL DHE based SSL/TLS ciphersuites. 2779 2780 This issue was reported to OpenSSL by Hanno Böck. 2781 (CVE-2015-3193) 2782 [Andy Polyakov] 2783 2784 *) Certificate verify crash with missing PSS parameter 2785 2786 The signature verification routines will crash with a NULL pointer 2787 dereference if presented with an ASN.1 signature using the RSA PSS 2788 algorithm and absent mask generation function parameter. Since these 2789 routines are used to verify certificate signature algorithms this can be 2790 used to crash any certificate verification operation and exploited in a 2791 DoS attack. Any application which performs certificate verification is 2792 vulnerable including OpenSSL clients and servers which enable client 2793 authentication. 2794 2795 This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG). 2796 (CVE-2015-3194) 2797 [Stephen Henson] 2798 2799 *) X509_ATTRIBUTE memory leak 2800 2801 When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak 2802 memory. This structure is used by the PKCS#7 and CMS routines so any 2803 application which reads PKCS#7 or CMS data from untrusted sources is 2804 affected. SSL/TLS is not affected. 2805 2806 This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using 2807 libFuzzer. 2808 (CVE-2015-3195) 2809 [Stephen Henson] 2810 2811 *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs. 2812 This changes the decoding behaviour for some invalid messages, 2813 though the change is mostly in the more lenient direction, and 2814 legacy behaviour is preserved as much as possible. 2815 [Emilia Käsper] 2816 2817 *) In DSA_generate_parameters_ex, if the provided seed is too short, 2818 return an error 2819 [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>] 2820 2821 Changes between 1.0.2c and 1.0.2d [9 Jul 2015] 2822 2823 *) Alternate chains certificate forgery 2824 2825 During certificate verification, OpenSSL will attempt to find an 2826 alternative certificate chain if the first attempt to build such a chain 2827 fails. An error in the implementation of this logic can mean that an 2828 attacker could cause certain checks on untrusted certificates to be 2829 bypassed, such as the CA flag, enabling them to use a valid leaf 2830 certificate to act as a CA and "issue" an invalid certificate. 2831 2832 This issue was reported to OpenSSL by Adam Langley/David Benjamin 2833 (Google/BoringSSL). 2834 [Matt Caswell] 2835 2836 Changes between 1.0.2b and 1.0.2c [12 Jun 2015] 2837 2838 *) Fix HMAC ABI incompatibility. The previous version introduced an ABI 2839 incompatibility in the handling of HMAC. The previous ABI has now been 2840 restored. 2841 [Matt Caswell] 2842 2843 Changes between 1.0.2a and 1.0.2b [11 Jun 2015] 2844 2845 *) Malformed ECParameters causes infinite loop 2846 2847 When processing an ECParameters structure OpenSSL enters an infinite loop 2848 if the curve specified is over a specially malformed binary polynomial 2849 field. 2850 2851 This can be used to perform denial of service against any 2852 system which processes public keys, certificate requests or 2853 certificates. This includes TLS clients and TLS servers with 2854 client authentication enabled. 2855 2856 This issue was reported to OpenSSL by Joseph Barr-Pixton. 2857 (CVE-2015-1788) 2858 [Andy Polyakov] 2859 2860 *) Exploitable out-of-bounds read in X509_cmp_time 2861 2862 X509_cmp_time does not properly check the length of the ASN1_TIME 2863 string and can read a few bytes out of bounds. In addition, 2864 X509_cmp_time accepts an arbitrary number of fractional seconds in the 2865 time string. 2866 2867 An attacker can use this to craft malformed certificates and CRLs of 2868 various sizes and potentially cause a segmentation fault, resulting in 2869 a DoS on applications that verify certificates or CRLs. TLS clients 2870 that verify CRLs are affected. TLS clients and servers with client 2871 authentication enabled may be affected if they use custom verification 2872 callbacks. 2873 2874 This issue was reported to OpenSSL by Robert Swiecki (Google), and 2875 independently by Hanno Böck. 2876 (CVE-2015-1789) 2877 [Emilia Käsper] 2878 2879 *) PKCS7 crash with missing EnvelopedContent 2880 2881 The PKCS#7 parsing code does not handle missing inner EncryptedContent 2882 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs 2883 with missing content and trigger a NULL pointer dereference on parsing. 2884 2885 Applications that decrypt PKCS#7 data or otherwise parse PKCS#7 2886 structures from untrusted sources are affected. OpenSSL clients and 2887 servers are not affected. 2888 2889 This issue was reported to OpenSSL by Michal Zalewski (Google). 2890 (CVE-2015-1790) 2891 [Emilia Käsper] 2892 2893 *) CMS verify infinite loop with unknown hash function 2894 2895 When verifying a signedData message the CMS code can enter an infinite loop 2896 if presented with an unknown hash function OID. This can be used to perform 2897 denial of service against any system which verifies signedData messages using 2898 the CMS code. 2899 This issue was reported to OpenSSL by Johannes Bauer. 2900 (CVE-2015-1792) 2901 [Stephen Henson] 2902 2903 *) Race condition handling NewSessionTicket 2904 2905 If a NewSessionTicket is received by a multi-threaded client when attempting to 2906 reuse a previous ticket then a race condition can occur potentially leading to 2907 a double free of the ticket data. 2908 (CVE-2015-1791) 2909 [Matt Caswell] 2910 2911 *) Only support 256-bit or stronger elliptic curves with the 2912 'ecdh_auto' setting (server) or by default (client). Of supported 2913 curves, prefer P-256 (both). 2914 [Emilia Kasper] 2915 2916 Changes between 1.0.2 and 1.0.2a [19 Mar 2015] 2917 2918 *) ClientHello sigalgs DoS fix 2919 2920 If a client connects to an OpenSSL 1.0.2 server and renegotiates with an 2921 invalid signature algorithms extension a NULL pointer dereference will 2922 occur. This can be exploited in a DoS attack against the server. 2923 2924 This issue was was reported to OpenSSL by David Ramos of Stanford 2925 University. 2926 (CVE-2015-0291) 2927 [Stephen Henson and Matt Caswell] 2928 2929 *) Multiblock corrupted pointer fix 2930 2931 OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This 2932 feature only applies on 64 bit x86 architecture platforms that support AES 2933 NI instructions. A defect in the implementation of "multiblock" can cause 2934 OpenSSL's internal write buffer to become incorrectly set to NULL when 2935 using non-blocking IO. Typically, when the user application is using a 2936 socket BIO for writing, this will only result in a failed connection. 2937 However if some other BIO is used then it is likely that a segmentation 2938 fault will be triggered, thus enabling a potential DoS attack. 2939 2940 This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller. 2941 (CVE-2015-0290) 2942 [Matt Caswell] 2943 2944 *) Segmentation fault in DTLSv1_listen fix 2945 2946 The DTLSv1_listen function is intended to be stateless and processes the 2947 initial ClientHello from many peers. It is common for user code to loop 2948 over the call to DTLSv1_listen until a valid ClientHello is received with 2949 an associated cookie. A defect in the implementation of DTLSv1_listen means 2950 that state is preserved in the SSL object from one invocation to the next 2951 that can lead to a segmentation fault. Errors processing the initial 2952 ClientHello can trigger this scenario. An example of such an error could be 2953 that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only 2954 server. 2955 2956 This issue was reported to OpenSSL by Per Allansson. 2957 (CVE-2015-0207) 2958 [Matt Caswell] 2959 2960 *) Segmentation fault in ASN1_TYPE_cmp fix 2961 2962 The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is 2963 made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check 2964 certificate signature algorithm consistency this can be used to crash any 2965 certificate verification operation and exploited in a DoS attack. Any 2966 application which performs certificate verification is vulnerable including 2967 OpenSSL clients and servers which enable client authentication. 2968 (CVE-2015-0286) 2969 [Stephen Henson] 2970 2971 *) Segmentation fault for invalid PSS parameters fix 2972 2973 The signature verification routines will crash with a NULL pointer 2974 dereference if presented with an ASN.1 signature using the RSA PSS 2975 algorithm and invalid parameters. Since these routines are used to verify 2976 certificate signature algorithms this can be used to crash any 2977 certificate verification operation and exploited in a DoS attack. Any 2978 application which performs certificate verification is vulnerable including 2979 OpenSSL clients and servers which enable client authentication. 2980 2981 This issue was was reported to OpenSSL by Brian Carpenter. 2982 (CVE-2015-0208) 2983 [Stephen Henson] 2984 2985 *) ASN.1 structure reuse memory corruption fix 2986 2987 Reusing a structure in ASN.1 parsing may allow an attacker to cause 2988 memory corruption via an invalid write. Such reuse is and has been 2989 strongly discouraged and is believed to be rare. 2990 2991 Applications that parse structures containing CHOICE or ANY DEFINED BY 2992 components may be affected. Certificate parsing (d2i_X509 and related 2993 functions) are however not affected. OpenSSL clients and servers are 2994 not affected. 2995 (CVE-2015-0287) 2996 [Stephen Henson] 2997 2998 *) PKCS7 NULL pointer dereferences fix 2999 3000 The PKCS#7 parsing code does not handle missing outer ContentInfo 3001 correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with 3002 missing content and trigger a NULL pointer dereference on parsing. 3003 3004 Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or 3005 otherwise parse PKCS#7 structures from untrusted sources are 3006 affected. OpenSSL clients and servers are not affected. 3007 3008 This issue was reported to OpenSSL by Michal Zalewski (Google). 3009 (CVE-2015-0289) 3010 [Emilia Käsper] 3011 3012 *) DoS via reachable assert in SSLv2 servers fix 3013 3014 A malicious client can trigger an OPENSSL_assert (i.e., an abort) in 3015 servers that both support SSLv2 and enable export cipher suites by sending 3016 a specially crafted SSLv2 CLIENT-MASTER-KEY message. 3017 3018 This issue was discovered by Sean Burford (Google) and Emilia Käsper 3019 (OpenSSL development team). 3020 (CVE-2015-0293) 3021 [Emilia Käsper] 3022 3023 *) Empty CKE with client auth and DHE fix 3024 3025 If client auth is used then a server can seg fault in the event of a DHE 3026 ciphersuite being selected and a zero length ClientKeyExchange message 3027 being sent by the client. This could be exploited in a DoS attack. 3028 (CVE-2015-1787) 3029 [Matt Caswell] 3030 3031 *) Handshake with unseeded PRNG fix 3032 3033 Under certain conditions an OpenSSL 1.0.2 client can complete a handshake 3034 with an unseeded PRNG. The conditions are: 3035 - The client is on a platform where the PRNG has not been seeded 3036 automatically, and the user has not seeded manually 3037 - A protocol specific client method version has been used (i.e. not 3038 SSL_client_methodv23) 3039 - A ciphersuite is used that does not require additional random data from 3040 the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA). 3041 3042 If the handshake succeeds then the client random that has been used will 3043 have been generated from a PRNG with insufficient entropy and therefore the 3044 output may be predictable. 3045 3046 For example using the following command with an unseeded openssl will 3047 succeed on an unpatched platform: 3048 3049 openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA 3050 (CVE-2015-0285) 3051 [Matt Caswell] 3052 3053 *) Use After Free following d2i_ECPrivatekey error fix 3054 3055 A malformed EC private key file consumed via the d2i_ECPrivateKey function 3056 could cause a use after free condition. This, in turn, could cause a double 3057 free in several private key parsing functions (such as d2i_PrivateKey 3058 or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption 3059 for applications that receive EC private keys from untrusted 3060 sources. This scenario is considered rare. 3061 3062 This issue was discovered by the BoringSSL project and fixed in their 3063 commit 517073cd4b. 3064 (CVE-2015-0209) 3065 [Matt Caswell] 3066 3067 *) X509_to_X509_REQ NULL pointer deref fix 3068 3069 The function X509_to_X509_REQ will crash with a NULL pointer dereference if 3070 the certificate key is invalid. This function is rarely used in practice. 3071 3072 This issue was discovered by Brian Carpenter. 3073 (CVE-2015-0288) 3074 [Stephen Henson] 3075 3076 *) Removed the export ciphers from the DEFAULT ciphers 3077 [Kurt Roeckx] 3078 3079 Changes between 1.0.1l and 1.0.2 [22 Jan 2015] 3080 3081 *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g. 3082 ARMv5 through ARMv8, as opposite to "locking" it to single one. 3083 So far those who have to target multiple platforms would compromise 3084 and argue that binary targeting say ARMv5 would still execute on 3085 ARMv8. "Universal" build resolves this compromise by providing 3086 near-optimal performance even on newer platforms. 3087 [Andy Polyakov] 3088 3089 *) Accelerated NIST P-256 elliptic curve implementation for x86_64 3090 (other platforms pending). 3091 [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov] 3092 3093 *) Add support for the SignedCertificateTimestampList certificate and 3094 OCSP response extensions from RFC6962. 3095 [Rob Stradling] 3096 3097 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 3098 for corner cases. (Certain input points at infinity could lead to 3099 bogus results, with non-infinity inputs mapped to infinity too.) 3100 [Bodo Moeller] 3101 3102 *) Initial support for PowerISA 2.0.7, first implemented in POWER8. 3103 This covers AES, SHA256/512 and GHASH. "Initial" means that most 3104 common cases are optimized and there still is room for further 3105 improvements. Vector Permutation AES for Altivec is also added. 3106 [Andy Polyakov] 3107 3108 *) Add support for little-endian ppc64 Linux target. 3109 [Marcelo Cerri (IBM)] 3110 3111 *) Initial support for AMRv8 ISA crypto extensions. This covers AES, 3112 SHA1, SHA256 and GHASH. "Initial" means that most common cases 3113 are optimized and there still is room for further improvements. 3114 Both 32- and 64-bit modes are supported. 3115 [Andy Polyakov, Ard Biesheuvel (Linaro)] 3116 3117 *) Improved ARMv7 NEON support. 3118 [Andy Polyakov] 3119 3120 *) Support for SPARC Architecture 2011 crypto extensions, first 3121 implemented in SPARC T4. This covers AES, DES, Camellia, SHA1, 3122 SHA256/512, MD5, GHASH and modular exponentiation. 3123 [Andy Polyakov, David Miller] 3124 3125 *) Accelerated modular exponentiation for Intel processors, a.k.a. 3126 RSAZ. 3127 [Shay Gueron & Vlad Krasnov (Intel Corp)] 3128 3129 *) Support for new and upcoming Intel processors, including AVX2, 3130 BMI and SHA ISA extensions. This includes additional "stitched" 3131 implementations, AESNI-SHA256 and GCM, and multi-buffer support 3132 for TLS encrypt. 3133 3134 This work was sponsored by Intel Corp. 3135 [Andy Polyakov] 3136 3137 *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method() 3138 supports both DTLS 1.2 and 1.0 and should use whatever version the peer 3139 supports and DTLSv1_2_*_method() which supports DTLS 1.2 only. 3140 [Steve Henson] 3141 3142 *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file(): 3143 this fixes a limitation in previous versions of OpenSSL. 3144 [Steve Henson] 3145 3146 *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest, 3147 MGF1 digest and OAEP label. 3148 [Steve Henson] 3149 3150 *) Add EVP support for key wrapping algorithms, to avoid problems with 3151 existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in 3152 the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap 3153 algorithms and include tests cases. 3154 [Steve Henson] 3155 3156 *) Add functions to allocate and set the fields of an ECDSA_METHOD 3157 structure. 3158 [Douglas E. Engert, Steve Henson] 3159 3160 *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the 3161 difference in days and seconds between two tm or ASN1_TIME structures. 3162 [Steve Henson] 3163 3164 *) Add -rev test option to s_server to just reverse order of characters 3165 received by client and send back to server. Also prints an abbreviated 3166 summary of the connection parameters. 3167 [Steve Henson] 3168 3169 *) New option -brief for s_client and s_server to print out a brief summary 3170 of connection parameters. 3171 [Steve Henson] 3172 3173 *) Add callbacks for arbitrary TLS extensions. 3174 [Trevor Perrin <trevp@trevp.net> and Ben Laurie] 3175 3176 *) New option -crl_download in several openssl utilities to download CRLs 3177 from CRLDP extension in certificates. 3178 [Steve Henson] 3179 3180 *) New options -CRL and -CRLform for s_client and s_server for CRLs. 3181 [Steve Henson] 3182 3183 *) New function X509_CRL_diff to generate a delta CRL from the difference 3184 of two full CRLs. Add support to "crl" utility. 3185 [Steve Henson] 3186 3187 *) New functions to set lookup_crls function and to retrieve 3188 X509_STORE from X509_STORE_CTX. 3189 [Steve Henson] 3190 3191 *) Print out deprecated issuer and subject unique ID fields in 3192 certificates. 3193 [Steve Henson] 3194 3195 *) Extend OCSP I/O functions so they can be used for simple general purpose 3196 HTTP as well as OCSP. New wrapper function which can be used to download 3197 CRLs using the OCSP API. 3198 [Steve Henson] 3199 3200 *) Delegate command line handling in s_client/s_server to SSL_CONF APIs. 3201 [Steve Henson] 3202 3203 *) SSL_CONF* functions. These provide a common framework for application 3204 configuration using configuration files or command lines. 3205 [Steve Henson] 3206 3207 *) SSL/TLS tracing code. This parses out SSL/TLS records using the 3208 message callback and prints the results. Needs compile time option 3209 "enable-ssl-trace". New options to s_client and s_server to enable 3210 tracing. 3211 [Steve Henson] 3212 3213 *) New ctrl and macro to retrieve supported points extensions. 3214 Print out extension in s_server and s_client. 3215 [Steve Henson] 3216 3217 *) New functions to retrieve certificate signature and signature 3218 OID NID. 3219 [Steve Henson] 3220 3221 *) Add functions to retrieve and manipulate the raw cipherlist sent by a 3222 client to OpenSSL. 3223 [Steve Henson] 3224 3225 *) New Suite B modes for TLS code. These use and enforce the requirements 3226 of RFC6460: restrict ciphersuites, only permit Suite B algorithms and 3227 only use Suite B curves. The Suite B modes can be set by using the 3228 strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring. 3229 [Steve Henson] 3230 3231 *) New chain verification flags for Suite B levels of security. Check 3232 algorithms are acceptable when flags are set in X509_verify_cert. 3233 [Steve Henson] 3234 3235 *) Make tls1_check_chain return a set of flags indicating checks passed 3236 by a certificate chain. Add additional tests to handle client 3237 certificates: checks for matching certificate type and issuer name 3238 comparison. 3239 [Steve Henson] 3240 3241 *) If an attempt is made to use a signature algorithm not in the peer 3242 preference list abort the handshake. If client has no suitable 3243 signature algorithms in response to a certificate request do not 3244 use the certificate. 3245 [Steve Henson] 3246 3247 *) If server EC tmp key is not in client preference list abort handshake. 3248 [Steve Henson] 3249 3250 *) Add support for certificate stores in CERT structure. This makes it 3251 possible to have different stores per SSL structure or one store in 3252 the parent SSL_CTX. Include distinct stores for certificate chain 3253 verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN 3254 to build and store a certificate chain in CERT structure: returning 3255 an error if the chain cannot be built: this will allow applications 3256 to test if a chain is correctly configured. 3257 3258 Note: if the CERT based stores are not set then the parent SSL_CTX 3259 store is used to retain compatibility with existing behaviour. 3260 3261 [Steve Henson] 3262 3263 *) New function ssl_set_client_disabled to set a ciphersuite disabled 3264 mask based on the current session, check mask when sending client 3265 hello and checking the requested ciphersuite. 3266 [Steve Henson] 3267 3268 *) New ctrls to retrieve and set certificate types in a certificate 3269 request message. Print out received values in s_client. If certificate 3270 types is not set with custom values set sensible values based on 3271 supported signature algorithms. 3272 [Steve Henson] 3273 3274 *) Support for distinct client and server supported signature algorithms. 3275 [Steve Henson] 3276 3277 *) Add certificate callback. If set this is called whenever a certificate 3278 is required by client or server. An application can decide which 3279 certificate chain to present based on arbitrary criteria: for example 3280 supported signature algorithms. Add very simple example to s_server. 3281 This fixes many of the problems and restrictions of the existing client 3282 certificate callback: for example you can now clear an existing 3283 certificate and specify the whole chain. 3284 [Steve Henson] 3285 3286 *) Add new "valid_flags" field to CERT_PKEY structure which determines what 3287 the certificate can be used for (if anything). Set valid_flags field 3288 in new tls1_check_chain function. Simplify ssl_set_cert_masks which used 3289 to have similar checks in it. 3290 3291 Add new "cert_flags" field to CERT structure and include a "strict mode". 3292 This enforces some TLS certificate requirements (such as only permitting 3293 certificate signature algorithms contained in the supported algorithms 3294 extension) which some implementations ignore: this option should be used 3295 with caution as it could cause interoperability issues. 3296 [Steve Henson] 3297 3298 *) Update and tidy signature algorithm extension processing. Work out 3299 shared signature algorithms based on preferences and peer algorithms 3300 and print them out in s_client and s_server. Abort handshake if no 3301 shared signature algorithms. 3302 [Steve Henson] 3303 3304 *) Add new functions to allow customised supported signature algorithms 3305 for SSL and SSL_CTX structures. Add options to s_client and s_server 3306 to support them. 3307 [Steve Henson] 3308 3309 *) New function SSL_certs_clear() to delete all references to certificates 3310 from an SSL structure. Before this once a certificate had been added 3311 it couldn't be removed. 3312 [Steve Henson] 3313 3314 *) Integrate hostname, email address and IP address checking with certificate 3315 verification. New verify options supporting checking in openssl utility. 3316 [Steve Henson] 3317 3318 *) Fixes and wildcard matching support to hostname and email checking 3319 functions. Add manual page. 3320 [Florian Weimer (Red Hat Product Security Team)] 3321 3322 *) New functions to check a hostname email or IP address against a 3323 certificate. Add options x509 utility to print results of checks against 3324 a certificate. 3325 [Steve Henson] 3326 3327 *) Fix OCSP checking. 3328 [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie] 3329 3330 *) Initial experimental support for explicitly trusted non-root CAs. 3331 OpenSSL still tries to build a complete chain to a root but if an 3332 intermediate CA has a trust setting included that is used. The first 3333 setting is used: whether to trust (e.g., -addtrust option to the x509 3334 utility) or reject. 3335 [Steve Henson] 3336 3337 *) Add -trusted_first option which attempts to find certificates in the 3338 trusted store even if an untrusted chain is also supplied. 3339 [Steve Henson] 3340 3341 *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE, 3342 platform support for Linux and Android. 3343 [Andy Polyakov] 3344 3345 *) Support for linux-x32, ILP32 environment in x86_64 framework. 3346 [Andy Polyakov] 3347 3348 *) Experimental multi-implementation support for FIPS capable OpenSSL. 3349 When in FIPS mode the approved implementations are used as normal, 3350 when not in FIPS mode the internal unapproved versions are used instead. 3351 This means that the FIPS capable OpenSSL isn't forced to use the 3352 (often lower performance) FIPS implementations outside FIPS mode. 3353 [Steve Henson] 3354 3355 *) Transparently support X9.42 DH parameters when calling 3356 PEM_read_bio_DHparameters. This means existing applications can handle 3357 the new parameter format automatically. 3358 [Steve Henson] 3359 3360 *) Initial experimental support for X9.42 DH parameter format: mainly 3361 to support use of 'q' parameter for RFC5114 parameters. 3362 [Steve Henson] 3363 3364 *) Add DH parameters from RFC5114 including test data to dhtest. 3365 [Steve Henson] 3366 3367 *) Support for automatic EC temporary key parameter selection. If enabled 3368 the most preferred EC parameters are automatically used instead of 3369 hardcoded fixed parameters. Now a server just has to call: 3370 SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically 3371 support ECDH and use the most appropriate parameters. 3372 [Steve Henson] 3373 3374 *) Enhance and tidy EC curve and point format TLS extension code. Use 3375 static structures instead of allocation if default values are used. 3376 New ctrls to set curves we wish to support and to retrieve shared curves. 3377 Print out shared curves in s_server. New options to s_server and s_client 3378 to set list of supported curves. 3379 [Steve Henson] 3380 3381 *) New ctrls to retrieve supported signature algorithms and 3382 supported curve values as an array of NIDs. Extend openssl utility 3383 to print out received values. 3384 [Steve Henson] 3385 3386 *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert 3387 between NIDs and the more common NIST names such as "P-256". Enhance 3388 ecparam utility and ECC method to recognise the NIST names for curves. 3389 [Steve Henson] 3390 3391 *) Enhance SSL/TLS certificate chain handling to support different 3392 chains for each certificate instead of one chain in the parent SSL_CTX. 3393 [Steve Henson] 3394 3395 *) Support for fixed DH ciphersuite client authentication: where both 3396 server and client use DH certificates with common parameters. 3397 [Steve Henson] 3398 3399 *) Support for fixed DH ciphersuites: those requiring DH server 3400 certificates. 3401 [Steve Henson] 3402 3403 *) New function i2d_re_X509_tbs for re-encoding the TBS portion of 3404 the certificate. 3405 Note: Related 1.0.2-beta specific macros X509_get_cert_info, 3406 X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and 3407 X509_CINF_get_signature were reverted post internal team review. 3408 3409 Changes between 1.0.1k and 1.0.1l [15 Jan 2015] 3410 3411 *) Build fixes for the Windows and OpenVMS platforms 3412 [Matt Caswell and Richard Levitte] 3413 3414 Changes between 1.0.1j and 1.0.1k [8 Jan 2015] 3415 3416 *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS 3417 message can cause a segmentation fault in OpenSSL due to a NULL pointer 3418 dereference. This could lead to a Denial Of Service attack. Thanks to 3419 Markus Stenberg of Cisco Systems, Inc. for reporting this issue. 3420 (CVE-2014-3571) 3421 [Steve Henson] 3422 3423 *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the 3424 dtls1_buffer_record function under certain conditions. In particular this 3425 could occur if an attacker sent repeated DTLS records with the same 3426 sequence number but for the next epoch. The memory leak could be exploited 3427 by an attacker in a Denial of Service attack through memory exhaustion. 3428 Thanks to Chris Mueller for reporting this issue. 3429 (CVE-2015-0206) 3430 [Matt Caswell] 3431 3432 *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is 3433 built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl 3434 method would be set to NULL which could later result in a NULL pointer 3435 dereference. Thanks to Frank Schmirler for reporting this issue. 3436 (CVE-2014-3569) 3437 [Kurt Roeckx] 3438 3439 *) Abort handshake if server key exchange message is omitted for ephemeral 3440 ECDH ciphersuites. 3441 3442 Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for 3443 reporting this issue. 3444 (CVE-2014-3572) 3445 [Steve Henson] 3446 3447 *) Remove non-export ephemeral RSA code on client and server. This code 3448 violated the TLS standard by allowing the use of temporary RSA keys in 3449 non-export ciphersuites and could be used by a server to effectively 3450 downgrade the RSA key length used to a value smaller than the server 3451 certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at 3452 INRIA or reporting this issue. 3453 (CVE-2015-0204) 3454 [Steve Henson] 3455 3456 *) Fixed issue where DH client certificates are accepted without verification. 3457 An OpenSSL server will accept a DH certificate for client authentication 3458 without the certificate verify message. This effectively allows a client to 3459 authenticate without the use of a private key. This only affects servers 3460 which trust a client certificate authority which issues certificates 3461 containing DH keys: these are extremely rare and hardly ever encountered. 3462 Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting 3463 this issue. 3464 (CVE-2015-0205) 3465 [Steve Henson] 3466 3467 *) Ensure that the session ID context of an SSL is updated when its 3468 SSL_CTX is updated via SSL_set_SSL_CTX. 3469 3470 The session ID context is typically set from the parent SSL_CTX, 3471 and can vary with the CTX. 3472 [Adam Langley] 3473 3474 *) Fix various certificate fingerprint issues. 3475 3476 By using non-DER or invalid encodings outside the signed portion of a 3477 certificate the fingerprint can be changed without breaking the signature. 3478 Although no details of the signed portion of the certificate can be changed 3479 this can cause problems with some applications: e.g. those using the 3480 certificate fingerprint for blacklists. 3481 3482 1. Reject signatures with non zero unused bits. 3483 3484 If the BIT STRING containing the signature has non zero unused bits reject 3485 the signature. All current signature algorithms require zero unused bits. 3486 3487 2. Check certificate algorithm consistency. 3488 3489 Check the AlgorithmIdentifier inside TBS matches the one in the 3490 certificate signature. NB: this will result in signature failure 3491 errors for some broken certificates. 3492 3493 Thanks to Konrad Kraszewski from Google for reporting this issue. 3494 3495 3. Check DSA/ECDSA signatures use DER. 3496 3497 Re-encode DSA/ECDSA signatures and compare with the original received 3498 signature. Return an error if there is a mismatch. 3499 3500 This will reject various cases including garbage after signature 3501 (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS 3502 program for discovering this case) and use of BER or invalid ASN.1 INTEGERs 3503 (negative or with leading zeroes). 3504 3505 Further analysis was conducted and fixes were developed by Stephen Henson 3506 of the OpenSSL core team. 3507 3508 (CVE-2014-8275) 3509 [Steve Henson] 3510 3511 *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect 3512 results on some platforms, including x86_64. This bug occurs at random 3513 with a very low probability, and is not known to be exploitable in any 3514 way, though its exact impact is difficult to determine. Thanks to Pieter 3515 Wuille (Blockstream) who reported this issue and also suggested an initial 3516 fix. Further analysis was conducted by the OpenSSL development team and 3517 Adam Langley of Google. The final fix was developed by Andy Polyakov of 3518 the OpenSSL core team. 3519 (CVE-2014-3570) 3520 [Andy Polyakov] 3521 3522 *) Do not resume sessions on the server if the negotiated protocol 3523 version does not match the session's version. Resuming with a different 3524 version, while not strictly forbidden by the RFC, is of questionable 3525 sanity and breaks all known clients. 3526 [David Benjamin, Emilia Käsper] 3527 3528 *) Tighten handling of the ChangeCipherSpec (CCS) message: reject 3529 early CCS messages during renegotiation. (Note that because 3530 renegotiation is encrypted, this early CCS was not exploitable.) 3531 [Emilia Käsper] 3532 3533 *) Tighten client-side session ticket handling during renegotiation: 3534 ensure that the client only accepts a session ticket if the server sends 3535 the extension anew in the ServerHello. Previously, a TLS client would 3536 reuse the old extension state and thus accept a session ticket if one was 3537 announced in the initial ServerHello. 3538 3539 Similarly, ensure that the client requires a session ticket if one 3540 was advertised in the ServerHello. Previously, a TLS client would 3541 ignore a missing NewSessionTicket message. 3542 [Emilia Käsper] 3543 3544 Changes between 1.0.1i and 1.0.1j [15 Oct 2014] 3545 3546 *) SRTP Memory Leak. 3547 3548 A flaw in the DTLS SRTP extension parsing code allows an attacker, who 3549 sends a carefully crafted handshake message, to cause OpenSSL to fail 3550 to free up to 64k of memory causing a memory leak. This could be 3551 exploited in a Denial Of Service attack. This issue affects OpenSSL 3552 1.0.1 server implementations for both SSL/TLS and DTLS regardless of 3553 whether SRTP is used or configured. Implementations of OpenSSL that 3554 have been compiled with OPENSSL_NO_SRTP defined are not affected. 3555 3556 The fix was developed by the OpenSSL team. 3557 (CVE-2014-3513) 3558 [OpenSSL team] 3559 3560 *) Session Ticket Memory Leak. 3561 3562 When an OpenSSL SSL/TLS/DTLS server receives a session ticket the 3563 integrity of that ticket is first verified. In the event of a session 3564 ticket integrity check failing, OpenSSL will fail to free memory 3565 causing a memory leak. By sending a large number of invalid session 3566 tickets an attacker could exploit this issue in a Denial Of Service 3567 attack. 3568 (CVE-2014-3567) 3569 [Steve Henson] 3570 3571 *) Build option no-ssl3 is incomplete. 3572 3573 When OpenSSL is configured with "no-ssl3" as a build option, servers 3574 could accept and complete a SSL 3.0 handshake, and clients could be 3575 configured to send them. 3576 (CVE-2014-3568) 3577 [Akamai and the OpenSSL team] 3578 3579 *) Add support for TLS_FALLBACK_SCSV. 3580 Client applications doing fallback retries should call 3581 SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV). 3582 (CVE-2014-3566) 3583 [Adam Langley, Bodo Moeller] 3584 3585 *) Add additional DigestInfo checks. 3586 3587 Re-encode DigestInto in DER and check against the original when 3588 verifying RSA signature: this will reject any improperly encoded 3589 DigestInfo structures. 3590 3591 Note: this is a precautionary measure and no attacks are currently known. 3592 3593 [Steve Henson] 3594 3595 Changes between 1.0.1h and 1.0.1i [6 Aug 2014] 3596 3597 *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the 3598 SRP code can be overrun an internal buffer. Add sanity check that 3599 g, A, B < N to SRP code. 3600 3601 Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC 3602 Group for discovering this issue. 3603 (CVE-2014-3512) 3604 [Steve Henson] 3605 3606 *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate 3607 TLS 1.0 instead of higher protocol versions when the ClientHello message 3608 is badly fragmented. This allows a man-in-the-middle attacker to force a 3609 downgrade to TLS 1.0 even if both the server and the client support a 3610 higher protocol version, by modifying the client's TLS records. 3611 3612 Thanks to David Benjamin and Adam Langley (Google) for discovering and 3613 researching this issue. 3614 (CVE-2014-3511) 3615 [David Benjamin] 3616 3617 *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject 3618 to a denial of service attack. A malicious server can crash the client 3619 with a null pointer dereference (read) by specifying an anonymous (EC)DH 3620 ciphersuite and sending carefully crafted handshake messages. 3621 3622 Thanks to Felix Gröbert (Google) for discovering and researching this 3623 issue. 3624 (CVE-2014-3510) 3625 [Emilia Käsper] 3626 3627 *) By sending carefully crafted DTLS packets an attacker could cause openssl 3628 to leak memory. This can be exploited through a Denial of Service attack. 3629 Thanks to Adam Langley for discovering and researching this issue. 3630 (CVE-2014-3507) 3631 [Adam Langley] 3632 3633 *) An attacker can force openssl to consume large amounts of memory whilst 3634 processing DTLS handshake messages. This can be exploited through a 3635 Denial of Service attack. 3636 Thanks to Adam Langley for discovering and researching this issue. 3637 (CVE-2014-3506) 3638 [Adam Langley] 3639 3640 *) An attacker can force an error condition which causes openssl to crash 3641 whilst processing DTLS packets due to memory being freed twice. This 3642 can be exploited through a Denial of Service attack. 3643 Thanks to Adam Langley and Wan-Teh Chang for discovering and researching 3644 this issue. 3645 (CVE-2014-3505) 3646 [Adam Langley] 3647 3648 *) If a multithreaded client connects to a malicious server using a resumed 3649 session and the server sends an ec point format extension it could write 3650 up to 255 bytes to freed memory. 3651 3652 Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this 3653 issue. 3654 (CVE-2014-3509) 3655 [Gabor Tyukasz] 3656 3657 *) A malicious server can crash an OpenSSL client with a null pointer 3658 dereference (read) by specifying an SRP ciphersuite even though it was not 3659 properly negotiated with the client. This can be exploited through a 3660 Denial of Service attack. 3661 3662 Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for 3663 discovering and researching this issue. 3664 (CVE-2014-5139) 3665 [Steve Henson] 3666 3667 *) A flaw in OBJ_obj2txt may cause pretty printing functions such as 3668 X509_name_oneline, X509_name_print_ex et al. to leak some information 3669 from the stack. Applications may be affected if they echo pretty printing 3670 output to the attacker. 3671 3672 Thanks to Ivan Fratric (Google) for discovering this issue. 3673 (CVE-2014-3508) 3674 [Emilia Käsper, and Steve Henson] 3675 3676 *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.) 3677 for corner cases. (Certain input points at infinity could lead to 3678 bogus results, with non-infinity inputs mapped to infinity too.) 3679 [Bodo Moeller] 3680 3681 Changes between 1.0.1g and 1.0.1h [5 Jun 2014] 3682 3683 *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted 3684 handshake can force the use of weak keying material in OpenSSL 3685 SSL/TLS clients and servers. 3686 3687 Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and 3688 researching this issue. (CVE-2014-0224) 3689 [KIKUCHI Masashi, Steve Henson] 3690 3691 *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an 3692 OpenSSL DTLS client the code can be made to recurse eventually crashing 3693 in a DoS attack. 3694 3695 Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue. 3696 (CVE-2014-0221) 3697 [Imre Rad, Steve Henson] 3698 3699 *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can 3700 be triggered by sending invalid DTLS fragments to an OpenSSL DTLS 3701 client or server. This is potentially exploitable to run arbitrary 3702 code on a vulnerable client or server. 3703 3704 Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195) 3705 [Jüri Aedla, Steve Henson] 3706 3707 *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites 3708 are subject to a denial of service attack. 3709 3710 Thanks to Felix Gröbert and Ivan Fratric at Google for discovering 3711 this issue. (CVE-2014-3470) 3712 [Felix Gröbert, Ivan Fratric, Steve Henson] 3713 3714 *) Harmonize version and its documentation. -f flag is used to display 3715 compilation flags. 3716 [mancha <mancha1@zoho.com>] 3717 3718 *) Fix eckey_priv_encode so it immediately returns an error upon a failure 3719 in i2d_ECPrivateKey. 3720 [mancha <mancha1@zoho.com>] 3721 3722 *) Fix some double frees. These are not thought to be exploitable. 3723 [mancha <mancha1@zoho.com>] 3724 3725 Changes between 1.0.1f and 1.0.1g [7 Apr 2014] 3726 3727 *) A missing bounds check in the handling of the TLS heartbeat extension 3728 can be used to reveal up to 64k of memory to a connected client or 3729 server. 3730 3731 Thanks for Neel Mehta of Google Security for discovering this bug and to 3732 Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for 3733 preparing the fix (CVE-2014-0160) 3734 [Adam Langley, Bodo Moeller] 3735 3736 *) Fix for the attack described in the paper "Recovering OpenSSL 3737 ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack" 3738 by Yuval Yarom and Naomi Benger. Details can be obtained from: 3739 http://eprint.iacr.org/2014/140 3740 3741 Thanks to Yuval Yarom and Naomi Benger for discovering this 3742 flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076) 3743 [Yuval Yarom and Naomi Benger] 3744 3745 *) TLS pad extension: draft-agl-tls-padding-03 3746 3747 Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the 3748 TLS client Hello record length value would otherwise be > 255 and 3749 less that 512 pad with a dummy extension containing zeroes so it 3750 is at least 512 bytes long. 3751 3752 [Adam Langley, Steve Henson] 3753 3754 Changes between 1.0.1e and 1.0.1f [6 Jan 2014] 3755 3756 *) Fix for TLS record tampering bug. A carefully crafted invalid 3757 handshake could crash OpenSSL with a NULL pointer exception. 3758 Thanks to Anton Johansson for reporting this issues. 3759 (CVE-2013-4353) 3760 3761 *) Keep original DTLS digest and encryption contexts in retransmission 3762 structures so we can use the previous session parameters if they need 3763 to be resent. (CVE-2013-6450) 3764 [Steve Henson] 3765 3766 *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which 3767 avoids preferring ECDHE-ECDSA ciphers when the client appears to be 3768 Safari on OS X. Safari on OS X 10.8..10.8.3 advertises support for 3769 several ECDHE-ECDSA ciphers, but fails to negotiate them. The bug 3770 is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing 3771 10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer. 3772 [Rob Stradling, Adam Langley] 3773 3774 Changes between 1.0.1d and 1.0.1e [11 Feb 2013] 3775 3776 *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI 3777 supporting platforms or when small records were transferred. 3778 [Andy Polyakov, Steve Henson] 3779 3780 Changes between 1.0.1c and 1.0.1d [5 Feb 2013] 3781 3782 *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time. 3783 3784 This addresses the flaw in CBC record processing discovered by 3785 Nadhem Alfardan and Kenny Paterson. Details of this attack can be found 3786 at: http://www.isg.rhul.ac.uk/tls/ 3787 3788 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 3789 Security Group at Royal Holloway, University of London 3790 (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and 3791 Emilia Käsper for the initial patch. 3792 (CVE-2013-0169) 3793 [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson] 3794 3795 *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode 3796 ciphersuites which can be exploited in a denial of service attack. 3797 Thanks go to and to Adam Langley <agl@chromium.org> for discovering 3798 and detecting this bug and to Wolfgang Ettlinger 3799 <wolfgang.ettlinger@gmail.com> for independently discovering this issue. 3800 (CVE-2012-2686) 3801 [Adam Langley] 3802 3803 *) Return an error when checking OCSP signatures when key is NULL. 3804 This fixes a DoS attack. (CVE-2013-0166) 3805 [Steve Henson] 3806 3807 *) Make openssl verify return errors. 3808 [Chris Palmer <palmer@google.com> and Ben Laurie] 3809 3810 *) Call OCSP Stapling callback after ciphersuite has been chosen, so 3811 the right response is stapled. Also change SSL_get_certificate() 3812 so it returns the certificate actually sent. 3813 See http://rt.openssl.org/Ticket/Display.html?id=2836. 3814 [Rob Stradling <rob.stradling@comodo.com>] 3815 3816 *) Fix possible deadlock when decoding public keys. 3817 [Steve Henson] 3818 3819 *) Don't use TLS 1.0 record version number in initial client hello 3820 if renegotiating. 3821 [Steve Henson] 3822 3823 Changes between 1.0.1b and 1.0.1c [10 May 2012] 3824 3825 *) Sanity check record length before skipping explicit IV in TLS 3826 1.2, 1.1 and DTLS to fix DoS attack. 3827 3828 Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic 3829 fuzzing as a service testing platform. 3830 (CVE-2012-2333) 3831 [Steve Henson] 3832 3833 *) Initialise tkeylen properly when encrypting CMS messages. 3834 Thanks to Solar Designer of Openwall for reporting this issue. 3835 [Steve Henson] 3836 3837 *) In FIPS mode don't try to use composite ciphers as they are not 3838 approved. 3839 [Steve Henson] 3840 3841 Changes between 1.0.1a and 1.0.1b [26 Apr 2012] 3842 3843 *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and 3844 1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately 3845 mean any application compiled against OpenSSL 1.0.0 headers setting 3846 SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling 3847 TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to 3848 0x10000000L Any application which was previously compiled against 3849 OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1 3850 will need to be recompiled as a result. Letting be results in 3851 inability to disable specifically TLS 1.1 and in client context, 3852 in unlike event, limit maximum offered version to TLS 1.0 [see below]. 3853 [Steve Henson] 3854 3855 *) In order to ensure interoperability SSL_OP_NO_protocolX does not 3856 disable just protocol X, but all protocols above X *if* there are 3857 protocols *below* X still enabled. In more practical terms it means 3858 that if application wants to disable TLS1.0 in favor of TLS1.1 and 3859 above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass 3860 SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to 3861 client side. 3862 [Andy Polyakov] 3863 3864 Changes between 1.0.1 and 1.0.1a [19 Apr 2012] 3865 3866 *) Check for potentially exploitable overflows in asn1_d2i_read_bio 3867 BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer 3868 in CRYPTO_realloc_clean. 3869 3870 Thanks to Tavis Ormandy, Google Security Team, for discovering this 3871 issue and to Adam Langley <agl@chromium.org> for fixing it. 3872 (CVE-2012-2110) 3873 [Adam Langley (Google), Tavis Ormandy, Google Security Team] 3874 3875 *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections. 3876 [Adam Langley] 3877 3878 *) Workarounds for some broken servers that "hang" if a client hello 3879 record length exceeds 255 bytes. 3880 3881 1. Do not use record version number > TLS 1.0 in initial client 3882 hello: some (but not all) hanging servers will now work. 3883 2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate 3884 the number of ciphers sent in the client hello. This should be 3885 set to an even number, such as 50, for example by passing: 3886 -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure. 3887 Most broken servers should now work. 3888 3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable 3889 TLS 1.2 client support entirely. 3890 [Steve Henson] 3891 3892 *) Fix SEGV in Vector Permutation AES module observed in OpenSSH. 3893 [Andy Polyakov] 3894 3895 Changes between 1.0.0h and 1.0.1 [14 Mar 2012] 3896 3897 *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET 3898 STRING form instead of a DigestInfo. 3899 [Steve Henson] 3900 3901 *) The format used for MDC2 RSA signatures is inconsistent between EVP 3902 and the RSA_sign/RSA_verify functions. This was made more apparent when 3903 OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular 3904 those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 3905 the correct format in RSA_verify so both forms transparently work. 3906 [Steve Henson] 3907 3908 *) Some servers which support TLS 1.0 can choke if we initially indicate 3909 support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA 3910 encrypted premaster secret. As a workaround use the maximum permitted 3911 client version in client hello, this should keep such servers happy 3912 and still work with previous versions of OpenSSL. 3913 [Steve Henson] 3914 3915 *) Add support for TLS/DTLS heartbeats. 3916 [Robin Seggelmann <seggelmann@fh-muenster.de>] 3917 3918 *) Add support for SCTP. 3919 [Robin Seggelmann <seggelmann@fh-muenster.de>] 3920 3921 *) Improved PRNG seeding for VOS. 3922 [Paul Green <Paul.Green@stratus.com>] 3923 3924 *) Extensive assembler packs updates, most notably: 3925 3926 - x86[_64]: AES-NI, PCLMULQDQ, RDRAND support; 3927 - x86[_64]: SSSE3 support (SHA1, vector-permutation AES); 3928 - x86_64: bit-sliced AES implementation; 3929 - ARM: NEON support, contemporary platforms optimizations; 3930 - s390x: z196 support; 3931 - *: GHASH and GF(2^m) multiplication implementations; 3932 3933 [Andy Polyakov] 3934 3935 *) Make TLS-SRP code conformant with RFC 5054 API cleanup 3936 (removal of unnecessary code) 3937 [Peter Sylvester <peter.sylvester@edelweb.fr>] 3938 3939 *) Add TLS key material exporter from RFC 5705. 3940 [Eric Rescorla] 3941 3942 *) Add DTLS-SRTP negotiation from RFC 5764. 3943 [Eric Rescorla] 3944 3945 *) Add Next Protocol Negotiation, 3946 http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be 3947 disabled with a no-npn flag to config or Configure. Code donated 3948 by Google. 3949 [Adam Langley <agl@google.com> and Ben Laurie] 3950 3951 *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224, 3952 NIST-P256, NIST-P521, with constant-time single point multiplication on 3953 typical inputs. Compiler support for the nonstandard type __uint128_t is 3954 required to use this (present in gcc 4.4 and later, for 64-bit builds). 3955 Code made available under Apache License version 2.0. 3956 3957 Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command 3958 line to include this in your build of OpenSSL, and run "make depend" (or 3959 "make update"). This enables the following EC_METHODs: 3960 3961 EC_GFp_nistp224_method() 3962 EC_GFp_nistp256_method() 3963 EC_GFp_nistp521_method() 3964 3965 EC_GROUP_new_by_curve_name() will automatically use these (while 3966 EC_GROUP_new_curve_GFp() currently prefers the more flexible 3967 implementations). 3968 [Emilia Käsper, Adam Langley, Bodo Moeller (Google)] 3969 3970 *) Use type ossl_ssize_t instead of ssize_t which isn't available on 3971 all platforms. Move ssize_t definition from e_os.h to the public 3972 header file e_os2.h as it now appears in public header file cms.h 3973 [Steve Henson] 3974 3975 *) New -sigopt option to the ca, req and x509 utilities. Additional 3976 signature parameters can be passed using this option and in 3977 particular PSS. 3978 [Steve Henson] 3979 3980 *) Add RSA PSS signing function. This will generate and set the 3981 appropriate AlgorithmIdentifiers for PSS based on those in the 3982 corresponding EVP_MD_CTX structure. No application support yet. 3983 [Steve Henson] 3984 3985 *) Support for companion algorithm specific ASN1 signing routines. 3986 New function ASN1_item_sign_ctx() signs a pre-initialised 3987 EVP_MD_CTX structure and sets AlgorithmIdentifiers based on 3988 the appropriate parameters. 3989 [Steve Henson] 3990 3991 *) Add new algorithm specific ASN1 verification initialisation function 3992 to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1 3993 handling will be the same no matter what EVP_PKEY_METHOD is used. 3994 Add a PSS handler to support verification of PSS signatures: checked 3995 against a number of sample certificates. 3996 [Steve Henson] 3997 3998 *) Add signature printing for PSS. Add PSS OIDs. 3999 [Steve Henson, Martin Kaiser <lists@kaiser.cx>] 4000 4001 *) Add algorithm specific signature printing. An individual ASN1 method 4002 can now print out signatures instead of the standard hex dump. 4003 4004 More complex signatures (e.g. PSS) can print out more meaningful 4005 information. Include DSA version that prints out the signature 4006 parameters r, s. 4007 [Steve Henson] 4008 4009 *) Password based recipient info support for CMS library: implementing 4010 RFC3211. 4011 [Steve Henson] 4012 4013 *) Split password based encryption into PBES2 and PBKDF2 functions. This 4014 neatly separates the code into cipher and PBE sections and is required 4015 for some algorithms that split PBES2 into separate pieces (such as 4016 password based CMS). 4017 [Steve Henson] 4018 4019 *) Session-handling fixes: 4020 - Fix handling of connections that are resuming with a session ID, 4021 but also support Session Tickets. 4022 - Fix a bug that suppressed issuing of a new ticket if the client 4023 presented a ticket with an expired session. 4024 - Try to set the ticket lifetime hint to something reasonable. 4025 - Make tickets shorter by excluding irrelevant information. 4026 - On the client side, don't ignore renewed tickets. 4027 [Adam Langley, Bodo Moeller (Google)] 4028 4029 *) Fix PSK session representation. 4030 [Bodo Moeller] 4031 4032 *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations. 4033 4034 This work was sponsored by Intel. 4035 [Andy Polyakov] 4036 4037 *) Add GCM support to TLS library. Some custom code is needed to split 4038 the IV between the fixed (from PRF) and explicit (from TLS record) 4039 portions. This adds all GCM ciphersuites supported by RFC5288 and 4040 RFC5289. Generalise some AES* cipherstrings to include GCM and 4041 add a special AESGCM string for GCM only. 4042 [Steve Henson] 4043 4044 *) Expand range of ctrls for AES GCM. Permit setting invocation 4045 field on decrypt and retrieval of invocation field only on encrypt. 4046 [Steve Henson] 4047 4048 *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support. 4049 As required by RFC5289 these ciphersuites cannot be used if for 4050 versions of TLS earlier than 1.2. 4051 [Steve Henson] 4052 4053 *) For FIPS capable OpenSSL interpret a NULL default public key method 4054 as unset and return the appropriate default but do *not* set the default. 4055 This means we can return the appropriate method in applications that 4056 switch between FIPS and non-FIPS modes. 4057 [Steve Henson] 4058 4059 *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an 4060 ENGINE is used then we cannot handle that in the FIPS module so we 4061 keep original code iff non-FIPS operations are allowed. 4062 [Steve Henson] 4063 4064 *) Add -attime option to openssl utilities. 4065 [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson] 4066 4067 *) Redirect DSA and DH operations to FIPS module in FIPS mode. 4068 [Steve Henson] 4069 4070 *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use 4071 FIPS EC methods unconditionally for now. 4072 [Steve Henson] 4073 4074 *) New build option no-ec2m to disable characteristic 2 code. 4075 [Steve Henson] 4076 4077 *) Backport libcrypto audit of return value checking from 1.1.0-dev; not 4078 all cases can be covered as some introduce binary incompatibilities. 4079 [Steve Henson] 4080 4081 *) Redirect RSA operations to FIPS module including keygen, 4082 encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods. 4083 [Steve Henson] 4084 4085 *) Add similar low level API blocking to ciphers. 4086 [Steve Henson] 4087 4088 *) Low level digest APIs are not approved in FIPS mode: any attempt 4089 to use these will cause a fatal error. Applications that *really* want 4090 to use them can use the private_* version instead. 4091 [Steve Henson] 4092 4093 *) Redirect cipher operations to FIPS module for FIPS builds. 4094 [Steve Henson] 4095 4096 *) Redirect digest operations to FIPS module for FIPS builds. 4097 [Steve Henson] 4098 4099 *) Update build system to add "fips" flag which will link in fipscanister.o 4100 for static and shared library builds embedding a signature if needed. 4101 [Steve Henson] 4102 4103 *) Output TLS supported curves in preference order instead of numerical 4104 order. This is currently hardcoded for the highest order curves first. 4105 This should be configurable so applications can judge speed vs strength. 4106 [Steve Henson] 4107 4108 *) Add TLS v1.2 server support for client authentication. 4109 [Steve Henson] 4110 4111 *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers 4112 and enable MD5. 4113 [Steve Henson] 4114 4115 *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying 4116 FIPS modules versions. 4117 [Steve Henson] 4118 4119 *) Add TLS v1.2 client side support for client authentication. Keep cache 4120 of handshake records longer as we don't know the hash algorithm to use 4121 until after the certificate request message is received. 4122 [Steve Henson] 4123 4124 *) Initial TLS v1.2 client support. Add a default signature algorithms 4125 extension including all the algorithms we support. Parse new signature 4126 format in client key exchange. Relax some ECC signing restrictions for 4127 TLS v1.2 as indicated in RFC5246. 4128 [Steve Henson] 4129 4130 *) Add server support for TLS v1.2 signature algorithms extension. Switch 4131 to new signature format when needed using client digest preference. 4132 All server ciphersuites should now work correctly in TLS v1.2. No client 4133 support yet and no support for client certificates. 4134 [Steve Henson] 4135 4136 *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch 4137 to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based 4138 ciphersuites. At present only RSA key exchange ciphersuites work with 4139 TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete 4140 SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods 4141 and version checking. 4142 [Steve Henson] 4143 4144 *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled 4145 with this defined it will not be affected by any changes to ssl internal 4146 structures. Add several utility functions to allow openssl application 4147 to work with OPENSSL_NO_SSL_INTERN defined. 4148 [Steve Henson] 4149 4150 *) A long standing patch to add support for SRP from EdelWeb (Peter 4151 Sylvester and Christophe Renou) was integrated. 4152 [Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester 4153 <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and 4154 Ben Laurie] 4155 4156 *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id. 4157 [Steve Henson] 4158 4159 *) Permit abbreviated handshakes when renegotiating using the function 4160 SSL_renegotiate_abbreviated(). 4161 [Robin Seggelmann <seggelmann@fh-muenster.de>] 4162 4163 *) Add call to ENGINE_register_all_complete() to 4164 ENGINE_load_builtin_engines(), so some implementations get used 4165 automatically instead of needing explicit application support. 4166 [Steve Henson] 4167 4168 *) Add support for TLS key exporter as described in RFC5705. 4169 [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson] 4170 4171 *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only 4172 a few changes are required: 4173 4174 Add SSL_OP_NO_TLSv1_1 flag. 4175 Add TLSv1_1 methods. 4176 Update version checking logic to handle version 1.1. 4177 Add explicit IV handling (ported from DTLS code). 4178 Add command line options to s_client/s_server. 4179 [Steve Henson] 4180 4181 Changes between 1.0.0g and 1.0.0h [12 Mar 2012] 4182 4183 *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness 4184 in CMS and PKCS7 code. When RSA decryption fails use a random key for 4185 content decryption and always return the same error. Note: this attack 4186 needs on average 2^20 messages so it only affects automated senders. The 4187 old behaviour can be re-enabled in the CMS code by setting the 4188 CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where 4189 an MMA defence is not necessary. 4190 Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering 4191 this issue. (CVE-2012-0884) 4192 [Steve Henson] 4193 4194 *) Fix CVE-2011-4619: make sure we really are receiving a 4195 client hello before rejecting multiple SGC restarts. Thanks to 4196 Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug. 4197 [Steve Henson] 4198 4199 Changes between 1.0.0f and 1.0.0g [18 Jan 2012] 4200 4201 *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109. 4202 Thanks to Antonio Martin, Enterprise Secure Access Research and 4203 Development, Cisco Systems, Inc. for discovering this bug and 4204 preparing a fix. (CVE-2012-0050) 4205 [Antonio Martin] 4206 4207 Changes between 1.0.0e and 1.0.0f [4 Jan 2012] 4208 4209 *) Nadhem Alfardan and Kenny Paterson have discovered an extension 4210 of the Vaudenay padding oracle attack on CBC mode encryption 4211 which enables an efficient plaintext recovery attack against 4212 the OpenSSL implementation of DTLS. Their attack exploits timing 4213 differences arising during decryption processing. A research 4214 paper describing this attack can be found at: 4215 http://www.isg.rhul.ac.uk/~kp/dtls.pdf 4216 Thanks go to Nadhem Alfardan and Kenny Paterson of the Information 4217 Security Group at Royal Holloway, University of London 4218 (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann 4219 <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de> 4220 for preparing the fix. (CVE-2011-4108) 4221 [Robin Seggelmann, Michael Tuexen] 4222 4223 *) Clear bytes used for block padding of SSL 3.0 records. 4224 (CVE-2011-4576) 4225 [Adam Langley (Google)] 4226 4227 *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George 4228 Kadianakis <desnacked@gmail.com> for discovering this issue and 4229 Adam Langley for preparing the fix. (CVE-2011-4619) 4230 [Adam Langley (Google)] 4231 4232 *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027) 4233 [Andrey Kulikov <amdeich@gmail.com>] 4234 4235 *) Prevent malformed RFC3779 data triggering an assertion failure. 4236 Thanks to Andrew Chi, BBN Technologies, for discovering the flaw 4237 and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577) 4238 [Rob Austein <sra@hactrn.net>] 4239 4240 *) Improved PRNG seeding for VOS. 4241 [Paul Green <Paul.Green@stratus.com>] 4242 4243 *) Fix ssl_ciph.c set-up race. 4244 [Adam Langley (Google)] 4245 4246 *) Fix spurious failures in ecdsatest.c. 4247 [Emilia Käsper (Google)] 4248 4249 *) Fix the BIO_f_buffer() implementation (which was mixing different 4250 interpretations of the '..._len' fields). 4251 [Adam Langley (Google)] 4252 4253 *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than 4254 BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent 4255 threads won't reuse the same blinding coefficients. 4256 4257 This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING 4258 lock to call BN_BLINDING_invert_ex, and avoids one use of 4259 BN_BLINDING_update for each BN_BLINDING structure (previously, 4260 the last update always remained unused). 4261 [Emilia Käsper (Google)] 4262 4263 *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf. 4264 [Bob Buckholz (Google)] 4265 4266 Changes between 1.0.0d and 1.0.0e [6 Sep 2011] 4267 4268 *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted 4269 by initialising X509_STORE_CTX properly. (CVE-2011-3207) 4270 [Kaspar Brand <ossl@velox.ch>] 4271 4272 *) Fix SSL memory handling for (EC)DH ciphersuites, in particular 4273 for multi-threaded use of ECDH. (CVE-2011-3210) 4274 [Adam Langley (Google)] 4275 4276 *) Fix x509_name_ex_d2i memory leak on bad inputs. 4277 [Bodo Moeller] 4278 4279 *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check 4280 signature public key algorithm by using OID xref utilities instead. 4281 Before this you could only use some ECC ciphersuites with SHA1 only. 4282 [Steve Henson] 4283 4284 *) Add protection against ECDSA timing attacks as mentioned in the paper 4285 by Billy Bob Brumley and Nicola Tuveri, see: 4286 4287 http://eprint.iacr.org/2011/232.pdf 4288 4289 [Billy Bob Brumley and Nicola Tuveri] 4290 4291 Changes between 1.0.0c and 1.0.0d [8 Feb 2011] 4292 4293 *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014 4294 [Neel Mehta, Adam Langley, Bodo Moeller (Google)] 4295 4296 *) Fix bug in string printing code: if *any* escaping is enabled we must 4297 escape the escape character (backslash) or the resulting string is 4298 ambiguous. 4299 [Steve Henson] 4300 4301 Changes between 1.0.0b and 1.0.0c [2 Dec 2010] 4302 4303 *) Disable code workaround for ancient and obsolete Netscape browsers 4304 and servers: an attacker can use it in a ciphersuite downgrade attack. 4305 Thanks to Martin Rex for discovering this bug. CVE-2010-4180 4306 [Steve Henson] 4307 4308 *) Fixed J-PAKE implementation error, originally discovered by 4309 Sebastien Martini, further info and confirmation from Stefan 4310 Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252 4311 [Ben Laurie] 4312 4313 Changes between 1.0.0a and 1.0.0b [16 Nov 2010] 4314 4315 *) Fix extension code to avoid race conditions which can result in a buffer 4316 overrun vulnerability: resumed sessions must not be modified as they can 4317 be shared by multiple threads. CVE-2010-3864 4318 [Steve Henson] 4319 4320 *) Fix WIN32 build system to correctly link an ENGINE directory into 4321 a DLL. 4322 [Steve Henson] 4323 4324 Changes between 1.0.0 and 1.0.0a [01 Jun 2010] 4325 4326 *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 4327 (CVE-2010-1633) 4328 [Steve Henson, Peter-Michael Hager <hager@dortmund.net>] 4329 4330 Changes between 0.9.8n and 1.0.0 [29 Mar 2010] 4331 4332 *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher 4333 context. The operation can be customised via the ctrl mechanism in 4334 case ENGINEs want to include additional functionality. 4335 [Steve Henson] 4336 4337 *) Tolerate yet another broken PKCS#8 key format: private key value negative. 4338 [Steve Henson] 4339 4340 *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to 4341 output hashes compatible with older versions of OpenSSL. 4342 [Willy Weisz <weisz@vcpc.univie.ac.at>] 4343 4344 *) Fix compression algorithm handling: if resuming a session use the 4345 compression algorithm of the resumed session instead of determining 4346 it from client hello again. Don't allow server to change algorithm. 4347 [Steve Henson] 4348 4349 *) Add load_crls() function to apps tidying load_certs() too. Add option 4350 to verify utility to allow additional CRLs to be included. 4351 [Steve Henson] 4352 4353 *) Update OCSP request code to permit adding custom headers to the request: 4354 some responders need this. 4355 [Steve Henson] 4356 4357 *) The function EVP_PKEY_sign() returns <=0 on error: check return code 4358 correctly. 4359 [Julia Lawall <julia@diku.dk>] 4360 4361 *) Update verify callback code in apps/s_cb.c and apps/verify.c, it 4362 needlessly dereferenced structures, used obsolete functions and 4363 didn't handle all updated verify codes correctly. 4364 [Steve Henson] 4365 4366 *) Disable MD2 in the default configuration. 4367 [Steve Henson] 4368 4369 *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to 4370 indicate the initial BIO being pushed or popped. This makes it possible 4371 to determine whether the BIO is the one explicitly called or as a result 4372 of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so 4373 it handles reference counts correctly and doesn't zero out the I/O bio 4374 when it is not being explicitly popped. WARNING: applications which 4375 included workarounds for the old buggy behaviour will need to be modified 4376 or they could free up already freed BIOs. 4377 [Steve Henson] 4378 4379 *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni 4380 renaming to all platforms (within the 0.9.8 branch, this was 4381 done conditionally on Netware platforms to avoid a name clash). 4382 [Guenter <lists@gknw.net>] 4383 4384 *) Add ECDHE and PSK support to DTLS. 4385 [Michael Tuexen <tuexen@fh-muenster.de>] 4386 4387 *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't 4388 be used on C++. 4389 [Steve Henson] 4390 4391 *) Add "missing" function EVP_MD_flags() (without this the only way to 4392 retrieve a digest flags is by accessing the structure directly. Update 4393 EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest 4394 or cipher is registered as in the "from" argument. Print out all 4395 registered digests in the dgst usage message instead of manually 4396 attempting to work them out. 4397 [Steve Henson] 4398 4399 *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello: 4400 this allows the use of compression and extensions. Change default cipher 4401 string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2 4402 by default unless an application cipher string requests it. 4403 [Steve Henson] 4404 4405 *) Alter match criteria in PKCS12_parse(). It used to try to use local 4406 key ids to find matching certificates and keys but some PKCS#12 files 4407 don't follow the (somewhat unwritten) rules and this strategy fails. 4408 Now just gather all certificates together and the first private key 4409 then look for the first certificate that matches the key. 4410 [Steve Henson] 4411 4412 *) Support use of registered digest and cipher names for dgst and cipher 4413 commands instead of having to add each one as a special case. So now 4414 you can do: 4415 4416 openssl sha256 foo 4417 4418 as well as: 4419 4420 openssl dgst -sha256 foo 4421 4422 and this works for ENGINE based algorithms too. 4423 4424 [Steve Henson] 4425 4426 *) Update Gost ENGINE to support parameter files. 4427 [Victor B. Wagner <vitus@cryptocom.ru>] 4428 4429 *) Support GeneralizedTime in ca utility. 4430 [Oliver Martin <oliver@volatilevoid.net>, Steve Henson] 4431 4432 *) Enhance the hash format used for certificate directory links. The new 4433 form uses the canonical encoding (meaning equivalent names will work 4434 even if they aren't identical) and uses SHA1 instead of MD5. This form 4435 is incompatible with the older format and as a result c_rehash should 4436 be used to rebuild symbolic links. 4437 [Steve Henson] 4438 4439 *) Make PKCS#8 the default write format for private keys, replacing the 4440 traditional format. This form is standardised, more secure and doesn't 4441 include an implicit MD5 dependency. 4442 [Steve Henson] 4443 4444 *) Add a $gcc_devteam_warn option to Configure. The idea is that any code 4445 committed to OpenSSL should pass this lot as a minimum. 4446 [Steve Henson] 4447 4448 *) Add session ticket override functionality for use by EAP-FAST. 4449 [Jouni Malinen <j@w1.fi>] 4450 4451 *) Modify HMAC functions to return a value. Since these can be implemented 4452 in an ENGINE errors can occur. 4453 [Steve Henson] 4454 4455 *) Type-checked OBJ_bsearch_ex. 4456 [Ben Laurie] 4457 4458 *) Type-checked OBJ_bsearch. Also some constification necessitated 4459 by type-checking. Still to come: TXT_DB, bsearch(?), 4460 OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING, 4461 CONF_VALUE. 4462 [Ben Laurie] 4463 4464 *) New function OPENSSL_gmtime_adj() to add a specific number of days and 4465 seconds to a tm structure directly, instead of going through OS 4466 specific date routines. This avoids any issues with OS routines such 4467 as the year 2038 bug. New *_adj() functions for ASN1 time structures 4468 and X509_time_adj_ex() to cover the extended range. The existing 4469 X509_time_adj() is still usable and will no longer have any date issues. 4470 [Steve Henson] 4471 4472 *) Delta CRL support. New use deltas option which will attempt to locate 4473 and search any appropriate delta CRLs available. 4474 4475 This work was sponsored by Google. 4476 [Steve Henson] 4477 4478 *) Support for CRLs partitioned by reason code. Reorganise CRL processing 4479 code and add additional score elements. Validate alternate CRL paths 4480 as part of the CRL checking and indicate a new error "CRL path validation 4481 error" in this case. Applications wanting additional details can use 4482 the verify callback and check the new "parent" field. If this is not 4483 NULL CRL path validation is taking place. Existing applications won't 4484 see this because it requires extended CRL support which is off by 4485 default. 4486 4487 This work was sponsored by Google. 4488 [Steve Henson] 4489 4490 *) Support for freshest CRL extension. 4491 4492 This work was sponsored by Google. 4493 [Steve Henson] 4494 4495 *) Initial indirect CRL support. Currently only supported in the CRLs 4496 passed directly and not via lookup. Process certificate issuer 4497 CRL entry extension and lookup CRL entries by bother issuer name 4498 and serial number. Check and process CRL issuer entry in IDP extension. 4499 4500 This work was sponsored by Google. 4501 [Steve Henson] 4502 4503 *) Add support for distinct certificate and CRL paths. The CRL issuer 4504 certificate is validated separately in this case. Only enabled if 4505 an extended CRL support flag is set: this flag will enable additional 4506 CRL functionality in future. 4507 4508 This work was sponsored by Google. 4509 [Steve Henson] 4510 4511 *) Add support for policy mappings extension. 4512 4513 This work was sponsored by Google. 4514 [Steve Henson] 4515 4516 *) Fixes to pathlength constraint, self issued certificate handling, 4517 policy processing to align with RFC3280 and PKITS tests. 4518 4519 This work was sponsored by Google. 4520 [Steve Henson] 4521 4522 *) Support for name constraints certificate extension. DN, email, DNS 4523 and URI types are currently supported. 4524 4525 This work was sponsored by Google. 4526 [Steve Henson] 4527 4528 *) To cater for systems that provide a pointer-based thread ID rather 4529 than numeric, deprecate the current numeric thread ID mechanism and 4530 replace it with a structure and associated callback type. This 4531 mechanism allows a numeric "hash" to be extracted from a thread ID in 4532 either case, and on platforms where pointers are larger than 'long', 4533 mixing is done to help ensure the numeric 'hash' is usable even if it 4534 can't be guaranteed unique. The default mechanism is to use "&errno" 4535 as a pointer-based thread ID to distinguish between threads. 4536 4537 Applications that want to provide their own thread IDs should now use 4538 CRYPTO_THREADID_set_callback() to register a callback that will call 4539 either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer(). 4540 4541 Note that ERR_remove_state() is now deprecated, because it is tied 4542 to the assumption that thread IDs are numeric. ERR_remove_state(0) 4543 to free the current thread's error state should be replaced by 4544 ERR_remove_thread_state(NULL). 4545 4546 (This new approach replaces the functions CRYPTO_set_idptr_callback(), 4547 CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in 4548 OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an 4549 application was previously providing a numeric thread callback that 4550 was inappropriate for distinguishing threads, then uniqueness might 4551 have been obtained with &errno that happened immediately in the 4552 intermediate development versions of OpenSSL; this is no longer the 4553 case, the numeric thread callback will now override the automatic use 4554 of &errno.) 4555 [Geoff Thorpe, with help from Bodo Moeller] 4556 4557 *) Initial support for different CRL issuing certificates. This covers a 4558 simple case where the self issued certificates in the chain exist and 4559 the real CRL issuer is higher in the existing chain. 4560 4561 This work was sponsored by Google. 4562 [Steve Henson] 4563 4564 *) Removed effectively defunct crypto/store from the build. 4565 [Ben Laurie] 4566 4567 *) Revamp of STACK to provide stronger type-checking. Still to come: 4568 TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE, 4569 ASN1_STRING, CONF_VALUE. 4570 [Ben Laurie] 4571 4572 *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer 4573 RAM on SSL connections. This option can save about 34k per idle SSL. 4574 [Nick Mathewson] 4575 4576 *) Revamp of LHASH to provide stronger type-checking. Still to come: 4577 STACK, TXT_DB, bsearch, qsort. 4578 [Ben Laurie] 4579 4580 *) Initial support for Cryptographic Message Syntax (aka CMS) based 4581 on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility, 4582 support for data, signedData, compressedData, digestedData and 4583 encryptedData, envelopedData types included. Scripts to check against 4584 RFC4134 examples draft and interop and consistency checks of many 4585 content types and variants. 4586 [Steve Henson] 4587 4588 *) Add options to enc utility to support use of zlib compression BIO. 4589 [Steve Henson] 4590 4591 *) Extend mk1mf to support importing of options and assembly language 4592 files from Configure script, currently only included in VC-WIN32. 4593 The assembly language rules can now optionally generate the source 4594 files from the associated perl scripts. 4595 [Steve Henson] 4596 4597 *) Implement remaining functionality needed to support GOST ciphersuites. 4598 Interop testing has been performed using CryptoPro implementations. 4599 [Victor B. Wagner <vitus@cryptocom.ru>] 4600 4601 *) s390x assembler pack. 4602 [Andy Polyakov] 4603 4604 *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU 4605 "family." 4606 [Andy Polyakov] 4607 4608 *) Implement Opaque PRF Input TLS extension as specified in 4609 draft-rescorla-tls-opaque-prf-input-00.txt. Since this is not an 4610 official specification yet and no extension type assignment by 4611 IANA exists, this extension (for now) will have to be explicitly 4612 enabled when building OpenSSL by providing the extension number 4613 to use. For example, specify an option 4614 4615 -DTLSEXT_TYPE_opaque_prf_input=0x9527 4616 4617 to the "config" or "Configure" script to enable the extension, 4618 assuming extension number 0x9527 (which is a completely arbitrary 4619 and unofficial assignment based on the MD5 hash of the Internet 4620 Draft). Note that by doing so, you potentially lose 4621 interoperability with other TLS implementations since these might 4622 be using the same extension number for other purposes. 4623 4624 SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the 4625 opaque PRF input value to use in the handshake. This will create 4626 an internal copy of the length-'len' string at 'src', and will 4627 return non-zero for success. 4628 4629 To get more control and flexibility, provide a callback function 4630 by using 4631 4632 SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb) 4633 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg) 4634 4635 where 4636 4637 int (*cb)(SSL *, void *peerinput, size_t len, void *arg); 4638 void *arg; 4639 4640 Callback function 'cb' will be called in handshakes, and is 4641 expected to use SSL_set_tlsext_opaque_prf_input() as appropriate. 4642 Argument 'arg' is for application purposes (the value as given to 4643 SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly 4644 be provided to the callback function). The callback function 4645 has to return non-zero to report success: usually 1 to use opaque 4646 PRF input just if possible, or 2 to enforce use of the opaque PRF 4647 input. In the latter case, the library will abort the handshake 4648 if opaque PRF input is not successfully negotiated. 4649 4650 Arguments 'peerinput' and 'len' given to the callback function 4651 will always be NULL and 0 in the case of a client. A server will 4652 see the client's opaque PRF input through these variables if 4653 available (NULL and 0 otherwise). Note that if the server 4654 provides an opaque PRF input, the length must be the same as the 4655 length of the client's opaque PRF input. 4656 4657 Note that the callback function will only be called when creating 4658 a new session (session resumption can resume whatever was 4659 previously negotiated), and will not be called in SSL 2.0 4660 handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or 4661 SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended 4662 for applications that need to enforce opaque PRF input. 4663 4664 [Bodo Moeller] 4665 4666 *) Update ssl code to support digests other than SHA1+MD5 for handshake 4667 MAC. 4668 4669 [Victor B. Wagner <vitus@cryptocom.ru>] 4670 4671 *) Add RFC4507 support to OpenSSL. This includes the corrections in 4672 RFC4507bis. The encrypted ticket format is an encrypted encoded 4673 SSL_SESSION structure, that way new session features are automatically 4674 supported. 4675 4676 If a client application caches session in an SSL_SESSION structure 4677 support is transparent because tickets are now stored in the encoded 4678 SSL_SESSION. 4679 4680 The SSL_CTX structure automatically generates keys for ticket 4681 protection in servers so again support should be possible 4682 with no application modification. 4683 4684 If a client or server wishes to disable RFC4507 support then the option 4685 SSL_OP_NO_TICKET can be set. 4686 4687 Add a TLS extension debugging callback to allow the contents of any client 4688 or server extensions to be examined. 4689 4690 This work was sponsored by Google. 4691 [Steve Henson] 4692 4693 *) Final changes to avoid use of pointer pointer casts in OpenSSL. 4694 OpenSSL should now compile cleanly on gcc 4.2 4695 [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson] 4696 4697 *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC 4698 support including streaming MAC support: this is required for GOST 4699 ciphersuite support. 4700 [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson] 4701 4702 *) Add option -stream to use PKCS#7 streaming in smime utility. New 4703 function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream() 4704 to output in BER and PEM format. 4705 [Steve Henson] 4706 4707 *) Experimental support for use of HMAC via EVP_PKEY interface. This 4708 allows HMAC to be handled via the EVP_DigestSign*() interface. The 4709 EVP_PKEY "key" in this case is the HMAC key, potentially allowing 4710 ENGINE support for HMAC keys which are unextractable. New -mac and 4711 -macopt options to dgst utility. 4712 [Steve Henson] 4713 4714 *) New option -sigopt to dgst utility. Update dgst to use 4715 EVP_Digest{Sign,Verify}*. These two changes make it possible to use 4716 alternative signing parameters such as X9.31 or PSS in the dgst 4717 utility. 4718 [Steve Henson] 4719 4720 *) Change ssl_cipher_apply_rule(), the internal function that does 4721 the work each time a ciphersuite string requests enabling 4722 ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or 4723 removing ("!foo+bar") a class of ciphersuites: Now it maintains 4724 the order of disabled ciphersuites such that those ciphersuites 4725 that most recently went from enabled to disabled not only stay 4726 in order with respect to each other, but also have higher priority 4727 than other disabled ciphersuites the next time ciphersuites are 4728 enabled again. 4729 4730 This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable 4731 the same ciphersuites as with "HIGH" alone, but in a specific 4732 order where the PSK ciphersuites come first (since they are the 4733 most recently disabled ciphersuites when "HIGH" is parsed). 4734 4735 Also, change ssl_create_cipher_list() (using this new 4736 functionality) such that between otherwise identical 4737 ciphersuites, ephemeral ECDH is preferred over ephemeral DH in 4738 the default order. 4739 [Bodo Moeller] 4740 4741 *) Change ssl_create_cipher_list() so that it automatically 4742 arranges the ciphersuites in reasonable order before starting 4743 to process the rule string. Thus, the definition for "DEFAULT" 4744 (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but 4745 remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH". 4746 This makes it much easier to arrive at a reasonable default order 4747 in applications for which anonymous ciphers are OK (meaning 4748 that you can't actually use DEFAULT). 4749 [Bodo Moeller; suggested by Victor Duchovni] 4750 4751 *) Split the SSL/TLS algorithm mask (as used for ciphersuite string 4752 processing) into multiple integers instead of setting 4753 "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK", 4754 "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer. 4755 (These masks as well as the individual bit definitions are hidden 4756 away into the non-exported interface ssl/ssl_locl.h, so this 4757 change to the definition of the SSL_CIPHER structure shouldn't 4758 affect applications.) This give us more bits for each of these 4759 categories, so there is no longer a need to coagulate AES128 and 4760 AES256 into a single algorithm bit, and to coagulate Camellia128 4761 and Camellia256 into a single algorithm bit, which has led to all 4762 kinds of kludges. 4763 4764 Thus, among other things, the kludge introduced in 0.9.7m and 4765 0.9.8e for masking out AES256 independently of AES128 or masking 4766 out Camellia256 independently of AES256 is not needed here in 0.9.9. 4767 4768 With the change, we also introduce new ciphersuite aliases that 4769 so far were missing: "AES128", "AES256", "CAMELLIA128", and 4770 "CAMELLIA256". 4771 [Bodo Moeller] 4772 4773 *) Add support for dsa-with-SHA224 and dsa-with-SHA256. 4774 Use the leftmost N bytes of the signature input if the input is 4775 larger than the prime q (with N being the size in bytes of q). 4776 [Nils Larsch] 4777 4778 *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses 4779 it yet and it is largely untested. 4780 [Steve Henson] 4781 4782 *) Add support for the ecdsa-with-SHA224/256/384/512 signature types. 4783 [Nils Larsch] 4784 4785 *) Initial incomplete changes to avoid need for function casts in OpenSSL 4786 some compilers (gcc 4.2 and later) reject their use. Safestack is 4787 reimplemented. Update ASN1 to avoid use of legacy functions. 4788 [Steve Henson] 4789 4790 *) Win32/64 targets are linked with Winsock2. 4791 [Andy Polyakov] 4792 4793 *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected 4794 to external functions. This can be used to increase CRL handling 4795 efficiency especially when CRLs are very large by (for example) storing 4796 the CRL revoked certificates in a database. 4797 [Steve Henson] 4798 4799 *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so 4800 new CRLs added to a directory can be used. New command line option 4801 -verify_return_error to s_client and s_server. This causes real errors 4802 to be returned by the verify callback instead of carrying on no matter 4803 what. This reflects the way a "real world" verify callback would behave. 4804 [Steve Henson] 4805 4806 *) GOST engine, supporting several GOST algorithms and public key formats. 4807 Kindly donated by Cryptocom. 4808 [Cryptocom] 4809 4810 *) Partial support for Issuing Distribution Point CRL extension. CRLs 4811 partitioned by DP are handled but no indirect CRL or reason partitioning 4812 (yet). Complete overhaul of CRL handling: now the most suitable CRL is 4813 selected via a scoring technique which handles IDP and AKID in CRLs. 4814 [Steve Henson] 4815 4816 *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which 4817 will ultimately be used for all verify operations: this will remove the 4818 X509_STORE dependency on certificate verification and allow alternative 4819 lookup methods. X509_STORE based implementations of these two callbacks. 4820 [Steve Henson] 4821 4822 *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names. 4823 Modify get_crl() to find a valid (unexpired) CRL if possible. 4824 [Steve Henson] 4825 4826 *) New function X509_CRL_match() to check if two CRLs are identical. Normally 4827 this would be called X509_CRL_cmp() but that name is already used by 4828 a function that just compares CRL issuer names. Cache several CRL 4829 extensions in X509_CRL structure and cache CRLDP in X509. 4830 [Steve Henson] 4831 4832 *) Store a "canonical" representation of X509_NAME structure (ASN1 Name) 4833 this maps equivalent X509_NAME structures into a consistent structure. 4834 Name comparison can then be performed rapidly using memcmp(). 4835 [Steve Henson] 4836 4837 *) Non-blocking OCSP request processing. Add -timeout option to ocsp 4838 utility. 4839 [Steve Henson] 4840 4841 *) Allow digests to supply their own micalg string for S/MIME type using 4842 the ctrl EVP_MD_CTRL_MICALG. 4843 [Steve Henson] 4844 4845 *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the 4846 EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN 4847 ctrl. It can then customise the structure before and/or after signing 4848 if necessary. 4849 [Steve Henson] 4850 4851 *) New function OBJ_add_sigid() to allow application defined signature OIDs 4852 to be added to OpenSSLs internal tables. New function OBJ_sigid_free() 4853 to free up any added signature OIDs. 4854 [Steve Henson] 4855 4856 *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(), 4857 EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal 4858 digest and cipher tables. New options added to openssl utility: 4859 list-message-digest-algorithms and list-cipher-algorithms. 4860 [Steve Henson] 4861 4862 *) Change the array representation of binary polynomials: the list 4863 of degrees of non-zero coefficients is now terminated with -1. 4864 Previously it was terminated with 0, which was also part of the 4865 value; thus, the array representation was not applicable to 4866 polynomials where t^0 has coefficient zero. This change makes 4867 the array representation useful in a more general context. 4868 [Douglas Stebila] 4869 4870 *) Various modifications and fixes to SSL/TLS cipher string 4871 handling. For ECC, the code now distinguishes between fixed ECDH 4872 with RSA certificates on the one hand and with ECDSA certificates 4873 on the other hand, since these are separate ciphersuites. The 4874 unused code for Fortezza ciphersuites has been removed. 4875 4876 For consistency with EDH, ephemeral ECDH is now called "EECDH" 4877 (not "ECDHE"). For consistency with the code for DH 4878 certificates, use of ECDH certificates is now considered ECDH 4879 authentication, not RSA or ECDSA authentication (the latter is 4880 merely the CA's signing algorithm and not actively used in the 4881 protocol). 4882 4883 The temporary ciphersuite alias "ECCdraft" is no longer 4884 available, and ECC ciphersuites are no longer excluded from "ALL" 4885 and "DEFAULT". The following aliases now exist for RFC 4492 4886 ciphersuites, most of these by analogy with the DH case: 4887 4888 kECDHr - ECDH cert, signed with RSA 4889 kECDHe - ECDH cert, signed with ECDSA 4890 kECDH - ECDH cert (signed with either RSA or ECDSA) 4891 kEECDH - ephemeral ECDH 4892 ECDH - ECDH cert or ephemeral ECDH 4893 4894 aECDH - ECDH cert 4895 aECDSA - ECDSA cert 4896 ECDSA - ECDSA cert 4897 4898 AECDH - anonymous ECDH 4899 EECDH - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH") 4900 4901 [Bodo Moeller] 4902 4903 *) Add additional S/MIME capabilities for AES and GOST ciphers if supported. 4904 Use correct micalg parameters depending on digest(s) in signed message. 4905 [Steve Henson] 4906 4907 *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process 4908 an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code. 4909 [Steve Henson] 4910 4911 *) Initial engine support for EVP_PKEY_METHOD. New functions to permit 4912 an engine to register a method. Add ENGINE lookups for methods and 4913 functional reference processing. 4914 [Steve Henson] 4915 4916 *) New functions EVP_Digest{Sign,Verify)*. These are enhanced versions of 4917 EVP_{Sign,Verify}* which allow an application to customise the signature 4918 process. 4919 [Steve Henson] 4920 4921 *) New -resign option to smime utility. This adds one or more signers 4922 to an existing PKCS#7 signedData structure. Also -md option to use an 4923 alternative message digest algorithm for signing. 4924 [Steve Henson] 4925 4926 *) Tidy up PKCS#7 routines and add new functions to make it easier to 4927 create PKCS7 structures containing multiple signers. Update smime 4928 application to support multiple signers. 4929 [Steve Henson] 4930 4931 *) New -macalg option to pkcs12 utility to allow setting of an alternative 4932 digest MAC. 4933 [Steve Henson] 4934 4935 *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC. 4936 Reorganize PBE internals to lookup from a static table using NIDs, 4937 add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl: 4938 EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative 4939 PRF which will be automatically used with PBES2. 4940 [Steve Henson] 4941 4942 *) Replace the algorithm specific calls to generate keys in "req" with the 4943 new API. 4944 [Steve Henson] 4945 4946 *) Update PKCS#7 enveloped data routines to use new API. This is now 4947 supported by any public key method supporting the encrypt operation. A 4948 ctrl is added to allow the public key algorithm to examine or modify 4949 the PKCS#7 RecipientInfo structure if it needs to: for RSA this is 4950 a no op. 4951 [Steve Henson] 4952 4953 *) Add a ctrl to asn1 method to allow a public key algorithm to express 4954 a default digest type to use. In most cases this will be SHA1 but some 4955 algorithms (such as GOST) need to specify an alternative digest. The 4956 return value indicates how strong the preference is 1 means optional and 4957 2 is mandatory (that is it is the only supported type). Modify 4958 ASN1_item_sign() to accept a NULL digest argument to indicate it should 4959 use the default md. Update openssl utilities to use the default digest 4960 type for signing if it is not explicitly indicated. 4961 [Steve Henson] 4962 4963 *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 4964 EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant 4965 signing method from the key type. This effectively removes the link 4966 between digests and public key types. 4967 [Steve Henson] 4968 4969 *) Add an OID cross reference table and utility functions. Its purpose is to 4970 translate between signature OIDs such as SHA1WithrsaEncryption and SHA1, 4971 rsaEncryption. This will allow some of the algorithm specific hackery 4972 needed to use the correct OID to be removed. 4973 [Steve Henson] 4974 4975 *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO 4976 structures for PKCS7_sign(). They are now set up by the relevant public 4977 key ASN1 method. 4978 [Steve Henson] 4979 4980 *) Add provisional EC pkey method with support for ECDSA and ECDH. 4981 [Steve Henson] 4982 4983 *) Add support for key derivation (agreement) in the API, DH method and 4984 pkeyutl. 4985 [Steve Henson] 4986 4987 *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support 4988 public and private key formats. As a side effect these add additional 4989 command line functionality not previously available: DSA signatures can be 4990 generated and verified using pkeyutl and DH key support and generation in 4991 pkey, genpkey. 4992 [Steve Henson] 4993 4994 *) BeOS support. 4995 [Oliver Tappe <zooey@hirschkaefer.de>] 4996 4997 *) New make target "install_html_docs" installs HTML renditions of the 4998 manual pages. 4999 [Oliver Tappe <zooey@hirschkaefer.de>] 5000 5001 *) New utility "genpkey" this is analogous to "genrsa" etc except it can 5002 generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to 5003 support key and parameter generation and add initial key generation 5004 functionality for RSA. 5005 [Steve Henson] 5006 5007 *) Add functions for main EVP_PKEY_method operations. The undocumented 5008 functions EVP_PKEY_{encrypt,decrypt} have been renamed to 5009 EVP_PKEY_{encrypt,decrypt}_old. 5010 [Steve Henson] 5011 5012 *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public 5013 key API, doesn't do much yet. 5014 [Steve Henson] 5015 5016 *) New function EVP_PKEY_asn1_get0_info() to retrieve information about 5017 public key algorithms. New option to openssl utility: 5018 "list-public-key-algorithms" to print out info. 5019 [Steve Henson] 5020 5021 *) Implement the Supported Elliptic Curves Extension for 5022 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 5023 [Douglas Stebila] 5024 5025 *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or 5026 EVP_CIPHER structures to avoid later problems in EVP_cleanup(). 5027 [Steve Henson] 5028 5029 *) New utilities pkey and pkeyparam. These are similar to algorithm specific 5030 utilities such as rsa, dsa, dsaparam etc except they process any key 5031 type. 5032 [Steve Henson] 5033 5034 *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 5035 functions EVP_PKEY_print_public(), EVP_PKEY_print_private(), 5036 EVP_PKEY_print_param() to print public key data from an EVP_PKEY 5037 structure. 5038 [Steve Henson] 5039 5040 *) Initial support for pluggable public key ASN1. 5041 De-spaghettify the public key ASN1 handling. Move public and private 5042 key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate 5043 algorithm specific handling to a single module within the relevant 5044 algorithm directory. Add functions to allow (near) opaque processing 5045 of public and private key structures. 5046 [Steve Henson] 5047 5048 *) Implement the Supported Point Formats Extension for 5049 ECC ciphersuites from draft-ietf-tls-ecc-12.txt. 5050 [Douglas Stebila] 5051 5052 *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members 5053 for the psk identity [hint] and the psk callback functions to the 5054 SSL_SESSION, SSL and SSL_CTX structure. 5055 5056 New ciphersuites: 5057 PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA, 5058 PSK-AES256-CBC-SHA 5059 5060 New functions: 5061 SSL_CTX_use_psk_identity_hint 5062 SSL_get_psk_identity_hint 5063 SSL_get_psk_identity 5064 SSL_use_psk_identity_hint 5065 5066 [Mika Kousa and Pasi Eronen of Nokia Corporation] 5067 5068 *) Add RFC 3161 compliant time stamp request creation, response generation 5069 and response verification functionality. 5070 [Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project] 5071 5072 *) Add initial support for TLS extensions, specifically for the server_name 5073 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 5074 have new members for a host name. The SSL data structure has an 5075 additional member SSL_CTX *initial_ctx so that new sessions can be 5076 stored in that context to allow for session resumption, even after the 5077 SSL has been switched to a new SSL_CTX in reaction to a client's 5078 server_name extension. 5079 5080 New functions (subject to change): 5081 5082 SSL_get_servername() 5083 SSL_get_servername_type() 5084 SSL_set_SSL_CTX() 5085 5086 New CTRL codes and macros (subject to change): 5087 5088 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 5089 - SSL_CTX_set_tlsext_servername_callback() 5090 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 5091 - SSL_CTX_set_tlsext_servername_arg() 5092 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 5093 5094 openssl s_client has a new '-servername ...' option. 5095 5096 openssl s_server has new options '-servername_host ...', '-cert2 ...', 5097 '-key2 ...', '-servername_fatal' (subject to change). This allows 5098 testing the HostName extension for a specific single host name ('-cert' 5099 and '-key' remain fallbacks for handshakes without HostName 5100 negotiation). If the unrecognized_name alert has to be sent, this by 5101 default is a warning; it becomes fatal with the '-servername_fatal' 5102 option. 5103 5104 [Peter Sylvester, Remy Allais, Christophe Renou] 5105 5106 *) Whirlpool hash implementation is added. 5107 [Andy Polyakov] 5108 5109 *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to 5110 bn(64,32). Because of instruction set limitations it doesn't have 5111 any negative impact on performance. This was done mostly in order 5112 to make it possible to share assembler modules, such as bn_mul_mont 5113 implementations, between 32- and 64-bit builds without hassle. 5114 [Andy Polyakov] 5115 5116 *) Move code previously exiled into file crypto/ec/ec2_smpt.c 5117 to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP 5118 macro. 5119 [Bodo Moeller] 5120 5121 *) New candidate for BIGNUM assembler implementation, bn_mul_mont, 5122 dedicated Montgomery multiplication procedure, is introduced. 5123 BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher 5124 "64-bit" performance on certain 32-bit targets. 5125 [Andy Polyakov] 5126 5127 *) New option SSL_OP_NO_COMP to disable use of compression selectively 5128 in SSL structures. New SSL ctrl to set maximum send fragment size. 5129 Save memory by setting the I/O buffer sizes dynamically instead of 5130 using the maximum available value. 5131 [Steve Henson] 5132 5133 *) New option -V for 'openssl ciphers'. This prints the ciphersuite code 5134 in addition to the text details. 5135 [Bodo Moeller] 5136 5137 *) Very, very preliminary EXPERIMENTAL support for printing of general 5138 ASN1 structures. This currently produces rather ugly output and doesn't 5139 handle several customised structures at all. 5140 [Steve Henson] 5141 5142 *) Integrated support for PVK file format and some related formats such 5143 as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support 5144 these in the 'rsa' and 'dsa' utilities. 5145 [Steve Henson] 5146 5147 *) Support for PKCS#1 RSAPublicKey format on rsa utility command line. 5148 [Steve Henson] 5149 5150 *) Remove the ancient ASN1_METHOD code. This was only ever used in one 5151 place for the (very old) "NETSCAPE" format certificates which are now 5152 handled using new ASN1 code equivalents. 5153 [Steve Henson] 5154 5155 *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD 5156 pointer and make the SSL_METHOD parameter in SSL_CTX_new, 5157 SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'. 5158 [Nils Larsch] 5159 5160 *) Modify CRL distribution points extension code to print out previously 5161 unsupported fields. Enhance extension setting code to allow setting of 5162 all fields. 5163 [Steve Henson] 5164 5165 *) Add print and set support for Issuing Distribution Point CRL extension. 5166 [Steve Henson] 5167 5168 *) Change 'Configure' script to enable Camellia by default. 5169 [NTT] 5170 5171 Changes between 0.9.8m and 0.9.8n [24 Mar 2010] 5172 5173 *) When rejecting SSL/TLS records due to an incorrect version number, never 5174 update s->server with a new major version number. As of 5175 - OpenSSL 0.9.8m if 'short' is a 16-bit type, 5176 - OpenSSL 0.9.8f if 'short' is longer than 16 bits, 5177 the previous behavior could result in a read attempt at NULL when 5178 receiving specific incorrect SSL/TLS records once record payload 5179 protection is active. (CVE-2010-0740) 5180 [Bodo Moeller, Adam Langley <agl@chromium.org>] 5181 5182 *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 5183 could be crashed if the relevant tables were not present (e.g. chrooted). 5184 [Tomas Hoger <thoger@redhat.com>] 5185 5186 Changes between 0.9.8l and 0.9.8m [25 Feb 2010] 5187 5188 *) Always check bn_wexpand() return values for failure. (CVE-2009-3245) 5189 [Martin Olsson, Neel Mehta] 5190 5191 *) Fix X509_STORE locking: Every 'objs' access requires a lock (to 5192 accommodate for stack sorting, always a write lock!). 5193 [Bodo Moeller] 5194 5195 *) On some versions of WIN32 Heap32Next is very slow. This can cause 5196 excessive delays in the RAND_poll(): over a minute. As a workaround 5197 include a time check in the inner Heap32Next loop too. 5198 [Steve Henson] 5199 5200 *) The code that handled flushing of data in SSL/TLS originally used the 5201 BIO_CTRL_INFO ctrl to see if any data was pending first. This caused 5202 the problem outlined in PR#1949. The fix suggested there however can 5203 trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions 5204 of Apache). So instead simplify the code to flush unconditionally. 5205 This should be fine since flushing with no data to flush is a no op. 5206 [Steve Henson] 5207 5208 *) Handle TLS versions 2.0 and later properly and correctly use the 5209 highest version of TLS/SSL supported. Although TLS >= 2.0 is some way 5210 off ancient servers have a habit of sticking around for a while... 5211 [Steve Henson] 5212 5213 *) Modify compression code so it frees up structures without using the 5214 ex_data callbacks. This works around a problem where some applications 5215 call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when 5216 restarting) then use compression (e.g. SSL with compression) later. 5217 This results in significant per-connection memory leaks and 5218 has caused some security issues including CVE-2008-1678 and 5219 CVE-2009-4355. 5220 [Steve Henson] 5221 5222 *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't 5223 change when encrypting or decrypting. 5224 [Bodo Moeller] 5225 5226 *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to 5227 connect and renegotiate with servers which do not support RI. 5228 Until RI is more widely deployed this option is enabled by default. 5229 [Steve Henson] 5230 5231 *) Add "missing" ssl ctrls to clear options and mode. 5232 [Steve Henson] 5233 5234 *) If client attempts to renegotiate and doesn't support RI respond with 5235 a no_renegotiation alert as required by RFC5746. Some renegotiating 5236 TLS clients will continue a connection gracefully when they receive 5237 the alert. Unfortunately OpenSSL mishandled this alert and would hang 5238 waiting for a server hello which it will never receive. Now we treat a 5239 received no_renegotiation alert as a fatal error. This is because 5240 applications requesting a renegotiation might well expect it to succeed 5241 and would have no code in place to handle the server denying it so the 5242 only safe thing to do is to terminate the connection. 5243 [Steve Henson] 5244 5245 *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if 5246 peer supports secure renegotiation and 0 otherwise. Print out peer 5247 renegotiation support in s_client/s_server. 5248 [Steve Henson] 5249 5250 *) Replace the highly broken and deprecated SPKAC certification method with 5251 the updated NID creation version. This should correctly handle UTF8. 5252 [Steve Henson] 5253 5254 *) Implement RFC5746. Re-enable renegotiation but require the extension 5255 as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION 5256 turns out to be a bad idea. It has been replaced by 5257 SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with 5258 SSL_CTX_set_options(). This is really not recommended unless you 5259 know what you are doing. 5260 [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson] 5261 5262 *) Fixes to stateless session resumption handling. Use initial_ctx when 5263 issuing and attempting to decrypt tickets in case it has changed during 5264 servername handling. Use a non-zero length session ID when attempting 5265 stateless session resumption: this makes it possible to determine if 5266 a resumption has occurred immediately after receiving server hello 5267 (several places in OpenSSL subtly assume this) instead of later in 5268 the handshake. 5269 [Steve Henson] 5270 5271 *) The functions ENGINE_ctrl(), OPENSSL_isservice(), 5272 CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error 5273 fixes for a few places where the return code is not checked 5274 correctly. 5275 [Julia Lawall <julia@diku.dk>] 5276 5277 *) Add --strict-warnings option to Configure script to include devteam 5278 warnings in other configurations. 5279 [Steve Henson] 5280 5281 *) Add support for --libdir option and LIBDIR variable in makefiles. This 5282 makes it possible to install openssl libraries in locations which 5283 have names other than "lib", for example "/usr/lib64" which some 5284 systems need. 5285 [Steve Henson, based on patch from Jeremy Utley] 5286 5287 *) Don't allow the use of leading 0x80 in OIDs. This is a violation of 5288 X690 8.9.12 and can produce some misleading textual output of OIDs. 5289 [Steve Henson, reported by Dan Kaminsky] 5290 5291 *) Delete MD2 from algorithm tables. This follows the recommendation in 5292 several standards that it is not used in new applications due to 5293 several cryptographic weaknesses. For binary compatibility reasons 5294 the MD2 API is still compiled in by default. 5295 [Steve Henson] 5296 5297 *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved 5298 and restored. 5299 [Steve Henson] 5300 5301 *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and 5302 OPENSSL_asc2uni conditionally on Netware platforms to avoid a name 5303 clash. 5304 [Guenter <lists@gknw.net>] 5305 5306 *) Fix the server certificate chain building code to use X509_verify_cert(), 5307 it used to have an ad-hoc builder which was unable to cope with anything 5308 other than a simple chain. 5309 [David Woodhouse <dwmw2@infradead.org>, Steve Henson] 5310 5311 *) Don't check self signed certificate signatures in X509_verify_cert() 5312 by default (a flag can override this): it just wastes time without 5313 adding any security. As a useful side effect self signed root CAs 5314 with non-FIPS digests are now usable in FIPS mode. 5315 [Steve Henson] 5316 5317 *) In dtls1_process_out_of_seq_message() the check if the current message 5318 is already buffered was missing. For every new message was memory 5319 allocated, allowing an attacker to perform an denial of service attack 5320 with sending out of seq handshake messages until there is no memory 5321 left. Additionally every future message was buffered, even if the 5322 sequence number made no sense and would be part of another handshake. 5323 So only messages with sequence numbers less than 10 in advance will be 5324 buffered. (CVE-2009-1378) 5325 [Robin Seggelmann, discovered by Daniel Mentz] 5326 5327 *) Records are buffered if they arrive with a future epoch to be 5328 processed after finishing the corresponding handshake. There is 5329 currently no limitation to this buffer allowing an attacker to perform 5330 a DOS attack with sending records with future epochs until there is no 5331 memory left. This patch adds the pqueue_size() function to determine 5332 the size of a buffer and limits the record buffer to 100 entries. 5333 (CVE-2009-1377) 5334 [Robin Seggelmann, discovered by Daniel Mentz] 5335 5336 *) Keep a copy of frag->msg_header.frag_len so it can be used after the 5337 parent structure is freed. (CVE-2009-1379) 5338 [Daniel Mentz] 5339 5340 *) Handle non-blocking I/O properly in SSL_shutdown() call. 5341 [Darryl Miles <darryl-mailinglists@netbauds.net>] 5342 5343 *) Add 2.5.4.* OIDs 5344 [Ilya O. <vrghost@gmail.com>] 5345 5346 Changes between 0.9.8k and 0.9.8l [5 Nov 2009] 5347 5348 *) Disable renegotiation completely - this fixes a severe security 5349 problem (CVE-2009-3555) at the cost of breaking all 5350 renegotiation. Renegotiation can be re-enabled by setting 5351 SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at 5352 run-time. This is really not recommended unless you know what 5353 you're doing. 5354 [Ben Laurie] 5355 5356 Changes between 0.9.8j and 0.9.8k [25 Mar 2009] 5357 5358 *) Don't set val to NULL when freeing up structures, it is freed up by 5359 underlying code. If sizeof(void *) > sizeof(long) this can result in 5360 zeroing past the valid field. (CVE-2009-0789) 5361 [Paolo Ganci <Paolo.Ganci@AdNovum.CH>] 5362 5363 *) Fix bug where return value of CMS_SignerInfo_verify_content() was not 5364 checked correctly. This would allow some invalid signed attributes to 5365 appear to verify correctly. (CVE-2009-0591) 5366 [Ivan Nestlerode <inestlerode@us.ibm.com>] 5367 5368 *) Reject UniversalString and BMPString types with invalid lengths. This 5369 prevents a crash in ASN1_STRING_print_ex() which assumes the strings have 5370 a legal length. (CVE-2009-0590) 5371 [Steve Henson] 5372 5373 *) Set S/MIME signing as the default purpose rather than setting it 5374 unconditionally. This allows applications to override it at the store 5375 level. 5376 [Steve Henson] 5377 5378 *) Permit restricted recursion of ASN1 strings. This is needed in practice 5379 to handle some structures. 5380 [Steve Henson] 5381 5382 *) Improve efficiency of mem_gets: don't search whole buffer each time 5383 for a '\n' 5384 [Jeremy Shapiro <jnshapir@us.ibm.com>] 5385 5386 *) New -hex option for openssl rand. 5387 [Matthieu Herrb] 5388 5389 *) Print out UTF8String and NumericString when parsing ASN1. 5390 [Steve Henson] 5391 5392 *) Support NumericString type for name components. 5393 [Steve Henson] 5394 5395 *) Allow CC in the environment to override the automatically chosen 5396 compiler. Note that nothing is done to ensure flags work with the 5397 chosen compiler. 5398 [Ben Laurie] 5399 5400 Changes between 0.9.8i and 0.9.8j [07 Jan 2009] 5401 5402 *) Properly check EVP_VerifyFinal() and similar return values 5403 (CVE-2008-5077). 5404 [Ben Laurie, Bodo Moeller, Google Security Team] 5405 5406 *) Enable TLS extensions by default. 5407 [Ben Laurie] 5408 5409 *) Allow the CHIL engine to be loaded, whether the application is 5410 multithreaded or not. (This does not release the developer from the 5411 obligation to set up the dynamic locking callbacks.) 5412 [Sander Temme <sander@temme.net>] 5413 5414 *) Use correct exit code if there is an error in dgst command. 5415 [Steve Henson; problem pointed out by Roland Dirlewanger] 5416 5417 *) Tweak Configure so that you need to say "experimental-jpake" to enable 5418 JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications. 5419 [Bodo Moeller] 5420 5421 *) Add experimental JPAKE support, including demo authentication in 5422 s_client and s_server. 5423 [Ben Laurie] 5424 5425 *) Set the comparison function in v3_addr_canonize(). 5426 [Rob Austein <sra@hactrn.net>] 5427 5428 *) Add support for XMPP STARTTLS in s_client. 5429 [Philip Paeps <philip@freebsd.org>] 5430 5431 *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior 5432 to ensure that even with this option, only ciphersuites in the 5433 server's preference list will be accepted. (Note that the option 5434 applies only when resuming a session, so the earlier behavior was 5435 just about the algorithm choice for symmetric cryptography.) 5436 [Bodo Moeller] 5437 5438 Changes between 0.9.8h and 0.9.8i [15 Sep 2008] 5439 5440 *) Fix NULL pointer dereference if a DTLS server received 5441 ChangeCipherSpec as first record (CVE-2009-1386). 5442 [PR #1679] 5443 5444 *) Fix a state transition in s3_srvr.c and d1_srvr.c 5445 (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...). 5446 [Nagendra Modadugu] 5447 5448 *) The fix in 0.9.8c that supposedly got rid of unsafe 5449 double-checked locking was incomplete for RSA blinding, 5450 addressing just one layer of what turns out to have been 5451 doubly unsafe triple-checked locking. 5452 5453 So now fix this for real by retiring the MONT_HELPER macro 5454 in crypto/rsa/rsa_eay.c. 5455 5456 [Bodo Moeller; problem pointed out by Marius Schilder] 5457 5458 *) Various precautionary measures: 5459 5460 - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h). 5461 5462 - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c). 5463 (NB: This would require knowledge of the secret session ticket key 5464 to exploit, in which case you'd be SOL either way.) 5465 5466 - Change bn_nist.c so that it will properly handle input BIGNUMs 5467 outside the expected range. 5468 5469 - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG 5470 builds. 5471 5472 [Neel Mehta, Bodo Moeller] 5473 5474 *) Allow engines to be "soft loaded" - i.e. optionally don't die if 5475 the load fails. Useful for distros. 5476 [Ben Laurie and the FreeBSD team] 5477 5478 *) Add support for Local Machine Keyset attribute in PKCS#12 files. 5479 [Steve Henson] 5480 5481 *) Fix BN_GF2m_mod_arr() top-bit cleanup code. 5482 [Huang Ying] 5483 5484 *) Expand ENGINE to support engine supplied SSL client certificate functions. 5485 5486 This work was sponsored by Logica. 5487 [Steve Henson] 5488 5489 *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows 5490 keystores. Support for SSL/TLS client authentication too. 5491 Not compiled unless enable-capieng specified to Configure. 5492 5493 This work was sponsored by Logica. 5494 [Steve Henson] 5495 5496 *) Fix bug in X509_ATTRIBUTE creation: don't set attribute using 5497 ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain 5498 attribute creation routines such as certificate requests and PKCS#12 5499 files. 5500 [Steve Henson] 5501 5502 Changes between 0.9.8g and 0.9.8h [28 May 2008] 5503 5504 *) Fix flaw if 'Server Key exchange message' is omitted from a TLS 5505 handshake which could lead to a client crash as found using the 5506 Codenomicon TLS test suite (CVE-2008-1672) 5507 [Steve Henson, Mark Cox] 5508 5509 *) Fix double free in TLS server name extensions which could lead to 5510 a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 5511 [Joe Orton] 5512 5513 *) Clear error queue in SSL_CTX_use_certificate_chain_file() 5514 5515 Clear the error queue to ensure that error entries left from 5516 older function calls do not interfere with the correct operation. 5517 [Lutz Jaenicke, Erik de Castro Lopo] 5518 5519 *) Remove root CA certificates of commercial CAs: 5520 5521 The OpenSSL project does not recommend any specific CA and does not 5522 have any policy with respect to including or excluding any CA. 5523 Therefore it does not make any sense to ship an arbitrary selection 5524 of root CA certificates with the OpenSSL software. 5525 [Lutz Jaenicke] 5526 5527 *) RSA OAEP patches to fix two separate invalid memory reads. 5528 The first one involves inputs when 'lzero' is greater than 5529 'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes 5530 before the beginning of from). The second one involves inputs where 5531 the 'db' section contains nothing but zeroes (there is a one-byte 5532 invalid read after the end of 'db'). 5533 [Ivan Nestlerode <inestlerode@us.ibm.com>] 5534 5535 *) Partial backport from 0.9.9-dev: 5536 5537 Introduce bn_mul_mont (dedicated Montgomery multiplication 5538 procedure) as a candidate for BIGNUM assembler implementation. 5539 While 0.9.9-dev uses assembler for various architectures, only 5540 x86_64 is available by default here in the 0.9.8 branch, and 5541 32-bit x86 is available through a compile-time setting. 5542 5543 To try the 32-bit x86 assembler implementation, use Configure 5544 option "enable-montasm" (which exists only for this backport). 5545 5546 As "enable-montasm" for 32-bit x86 disclaims code stability 5547 anyway, in this constellation we activate additional code 5548 backported from 0.9.9-dev for further performance improvements, 5549 namely BN_from_montgomery_word. (To enable this otherwise, 5550 e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".) 5551 5552 [Andy Polyakov (backport partially by Bodo Moeller)] 5553 5554 *) Add TLS session ticket callback. This allows an application to set 5555 TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed 5556 values. This is useful for key rollover for example where several key 5557 sets may exist with different names. 5558 [Steve Henson] 5559 5560 *) Reverse ENGINE-internal logic for caching default ENGINE handles. 5561 This was broken until now in 0.9.8 releases, such that the only way 5562 a registered ENGINE could be used (assuming it initialises 5563 successfully on the host) was to explicitly set it as the default 5564 for the relevant algorithms. This is in contradiction with 0.9.7 5565 behaviour and the documentation. With this fix, when an ENGINE is 5566 registered into a given algorithm's table of implementations, the 5567 'uptodate' flag is reset so that auto-discovery will be used next 5568 time a new context for that algorithm attempts to select an 5569 implementation. 5570 [Ian Lister (tweaked by Geoff Thorpe)] 5571 5572 *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9 5573 implementation in the following ways: 5574 5575 Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be 5576 hard coded. 5577 5578 Lack of BER streaming support means one pass streaming processing is 5579 only supported if data is detached: setting the streaming flag is 5580 ignored for embedded content. 5581 5582 CMS support is disabled by default and must be explicitly enabled 5583 with the enable-cms configuration option. 5584 [Steve Henson] 5585 5586 *) Update the GMP engine glue to do direct copies between BIGNUM and 5587 mpz_t when openssl and GMP use the same limb size. Otherwise the 5588 existing "conversion via a text string export" trick is still used. 5589 [Paul Sheer <paulsheer@gmail.com>] 5590 5591 *) Zlib compression BIO. This is a filter BIO which compressed and 5592 uncompresses any data passed through it. 5593 [Steve Henson] 5594 5595 *) Add AES_wrap_key() and AES_unwrap_key() functions to implement 5596 RFC3394 compatible AES key wrapping. 5597 [Steve Henson] 5598 5599 *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0(): 5600 sets string data without copying. X509_ALGOR_set0() and 5601 X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier) 5602 data. Attribute function X509at_get0_data_by_OBJ(): retrieves data 5603 from an X509_ATTRIBUTE structure optionally checking it occurs only 5604 once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied 5605 data. 5606 [Steve Henson] 5607 5608 *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set() 5609 to get the expected BN_FLG_CONSTTIME behavior. 5610 [Bodo Moeller (Google)] 5611 5612 *) Netware support: 5613 5614 - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets 5615 - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT) 5616 - added some more tests to do_tests.pl 5617 - fixed RunningProcess usage so that it works with newer LIBC NDKs too 5618 - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency 5619 - added new Configure targets netware-clib-bsdsock, netware-clib-gcc, 5620 netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc 5621 - various changes to netware.pl to enable gcc-cross builds on Win32 5622 platform 5623 - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD) 5624 - various changes to fix missing prototype warnings 5625 - fixed x86nasm.pl to create correct asm files for NASM COFF output 5626 - added AES, WHIRLPOOL and CPUID assembler code to build files 5627 - added missing AES assembler make rules to mk1mf.pl 5628 - fixed order of includes in apps/ocsp.c so that e_os.h settings apply 5629 [Guenter Knauf <eflash@gmx.net>] 5630 5631 *) Implement certificate status request TLS extension defined in RFC3546. 5632 A client can set the appropriate parameters and receive the encoded 5633 OCSP response via a callback. A server can query the supplied parameters 5634 and set the encoded OCSP response in the callback. Add simplified examples 5635 to s_client and s_server. 5636 [Steve Henson] 5637 5638 Changes between 0.9.8f and 0.9.8g [19 Oct 2007] 5639 5640 *) Fix various bugs: 5641 + Binary incompatibility of ssl_ctx_st structure 5642 + DTLS interoperation with non-compliant servers 5643 + Don't call get_session_cb() without proposed session 5644 + Fix ia64 assembler code 5645 [Andy Polyakov, Steve Henson] 5646 5647 Changes between 0.9.8e and 0.9.8f [11 Oct 2007] 5648 5649 *) DTLS Handshake overhaul. There were longstanding issues with 5650 OpenSSL DTLS implementation, which were making it impossible for 5651 RFC 4347 compliant client to communicate with OpenSSL server. 5652 Unfortunately just fixing these incompatibilities would "cut off" 5653 pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e 5654 server keeps tolerating non RFC compliant syntax. The opposite is 5655 not true, 0.9.8f client can not communicate with earlier server. 5656 This update even addresses CVE-2007-4995. 5657 [Andy Polyakov] 5658 5659 *) Changes to avoid need for function casts in OpenSSL: some compilers 5660 (gcc 4.2 and later) reject their use. 5661 [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>, 5662 Steve Henson] 5663 5664 *) Add RFC4507 support to OpenSSL. This includes the corrections in 5665 RFC4507bis. The encrypted ticket format is an encrypted encoded 5666 SSL_SESSION structure, that way new session features are automatically 5667 supported. 5668 5669 If a client application caches session in an SSL_SESSION structure 5670 support is transparent because tickets are now stored in the encoded 5671 SSL_SESSION. 5672 5673 The SSL_CTX structure automatically generates keys for ticket 5674 protection in servers so again support should be possible 5675 with no application modification. 5676 5677 If a client or server wishes to disable RFC4507 support then the option 5678 SSL_OP_NO_TICKET can be set. 5679 5680 Add a TLS extension debugging callback to allow the contents of any client 5681 or server extensions to be examined. 5682 5683 This work was sponsored by Google. 5684 [Steve Henson] 5685 5686 *) Add initial support for TLS extensions, specifically for the server_name 5687 extension so far. The SSL_SESSION, SSL_CTX, and SSL data structures now 5688 have new members for a host name. The SSL data structure has an 5689 additional member SSL_CTX *initial_ctx so that new sessions can be 5690 stored in that context to allow for session resumption, even after the 5691 SSL has been switched to a new SSL_CTX in reaction to a client's 5692 server_name extension. 5693 5694 New functions (subject to change): 5695 5696 SSL_get_servername() 5697 SSL_get_servername_type() 5698 SSL_set_SSL_CTX() 5699 5700 New CTRL codes and macros (subject to change): 5701 5702 SSL_CTRL_SET_TLSEXT_SERVERNAME_CB 5703 - SSL_CTX_set_tlsext_servername_callback() 5704 SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG 5705 - SSL_CTX_set_tlsext_servername_arg() 5706 SSL_CTRL_SET_TLSEXT_HOSTNAME - SSL_set_tlsext_host_name() 5707 5708 openssl s_client has a new '-servername ...' option. 5709 5710 openssl s_server has new options '-servername_host ...', '-cert2 ...', 5711 '-key2 ...', '-servername_fatal' (subject to change). This allows 5712 testing the HostName extension for a specific single host name ('-cert' 5713 and '-key' remain fallbacks for handshakes without HostName 5714 negotiation). If the unrecognized_name alert has to be sent, this by 5715 default is a warning; it becomes fatal with the '-servername_fatal' 5716 option. 5717 5718 [Peter Sylvester, Remy Allais, Christophe Renou, Steve Henson] 5719 5720 *) Add AES and SSE2 assembly language support to VC++ build. 5721 [Steve Henson] 5722 5723 *) Mitigate attack on final subtraction in Montgomery reduction. 5724 [Andy Polyakov] 5725 5726 *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0 5727 (which previously caused an internal error). 5728 [Bodo Moeller] 5729 5730 *) Squeeze another 10% out of IGE mode when in != out. 5731 [Ben Laurie] 5732 5733 *) AES IGE mode speedup. 5734 [Dean Gaudet (Google)] 5735 5736 *) Add the Korean symmetric 128-bit cipher SEED (see 5737 http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and 5738 add SEED ciphersuites from RFC 4162: 5739 5740 TLS_RSA_WITH_SEED_CBC_SHA = "SEED-SHA" 5741 TLS_DHE_DSS_WITH_SEED_CBC_SHA = "DHE-DSS-SEED-SHA" 5742 TLS_DHE_RSA_WITH_SEED_CBC_SHA = "DHE-RSA-SEED-SHA" 5743 TLS_DH_anon_WITH_SEED_CBC_SHA = "ADH-SEED-SHA" 5744 5745 To minimize changes between patchlevels in the OpenSSL 0.9.8 5746 series, SEED remains excluded from compilation unless OpenSSL 5747 is configured with 'enable-seed'. 5748 [KISA, Bodo Moeller] 5749 5750 *) Mitigate branch prediction attacks, which can be practical if a 5751 single processor is shared, allowing a spy process to extract 5752 information. For detailed background information, see 5753 http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron, 5754 J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL 5755 and Necessary Software Countermeasures"). The core of the change 5756 are new versions BN_div_no_branch() and 5757 BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(), 5758 respectively, which are slower, but avoid the security-relevant 5759 conditional branches. These are automatically called by BN_div() 5760 and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one 5761 of the input BIGNUMs. Also, BN_is_bit_set() has been changed to 5762 remove a conditional branch. 5763 5764 BN_FLG_CONSTTIME is the new name for the previous 5765 BN_FLG_EXP_CONSTTIME flag, since it now affects more than just 5766 modular exponentiation. (Since OpenSSL 0.9.7h, setting this flag 5767 in the exponent causes BN_mod_exp_mont() to use the alternative 5768 implementation in BN_mod_exp_mont_consttime().) The old name 5769 remains as a deprecated alias. 5770 5771 Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general 5772 RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses 5773 constant-time implementations for more than just exponentiation. 5774 Here too the old name is kept as a deprecated alias. 5775 5776 BN_BLINDING_new() will now use BN_dup() for the modulus so that 5777 the BN_BLINDING structure gets an independent copy of the 5778 modulus. This means that the previous "BIGNUM *m" argument to 5779 BN_BLINDING_new() and to BN_BLINDING_create_param() now 5780 essentially becomes "const BIGNUM *m", although we can't actually 5781 change this in the header file before 0.9.9. It allows 5782 RSA_setup_blinding() to use BN_with_flags() on the modulus to 5783 enable BN_FLG_CONSTTIME. 5784 5785 [Matthew D Wood (Intel Corp)] 5786 5787 *) In the SSL/TLS server implementation, be strict about session ID 5788 context matching (which matters if an application uses a single 5789 external cache for different purposes). Previously, 5790 out-of-context reuse was forbidden only if SSL_VERIFY_PEER was 5791 set. This did ensure strict client verification, but meant that, 5792 with applications using a single external cache for quite 5793 different requirements, clients could circumvent ciphersuite 5794 restrictions for a given session ID context by starting a session 5795 in a different context. 5796 [Bodo Moeller] 5797 5798 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 5799 a ciphersuite string such as "DEFAULT:RSA" cannot enable 5800 authentication-only ciphersuites. 5801 [Bodo Moeller] 5802 5803 *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was 5804 not complete and could lead to a possible single byte overflow 5805 (CVE-2007-5135) [Ben Laurie] 5806 5807 Changes between 0.9.8d and 0.9.8e [23 Feb 2007] 5808 5809 *) Since AES128 and AES256 (and similarly Camellia128 and 5810 Camellia256) share a single mask bit in the logic of 5811 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 5812 kludge to work properly if AES128 is available and AES256 isn't 5813 (or if Camellia128 is available and Camellia256 isn't). 5814 [Victor Duchovni] 5815 5816 *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c 5817 (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters): 5818 When a point or a seed is encoded in a BIT STRING, we need to 5819 prevent the removal of trailing zero bits to get the proper DER 5820 encoding. (By default, crypto/asn1/a_bitstr.c assumes the case 5821 of a NamedBitList, for which trailing 0 bits need to be removed.) 5822 [Bodo Moeller] 5823 5824 *) Have SSL/TLS server implementation tolerate "mismatched" record 5825 protocol version while receiving ClientHello even if the 5826 ClientHello is fragmented. (The server can't insist on the 5827 particular protocol version it has chosen before the ServerHello 5828 message has informed the client about his choice.) 5829 [Bodo Moeller] 5830 5831 *) Add RFC 3779 support. 5832 [Rob Austein for ARIN, Ben Laurie] 5833 5834 *) Load error codes if they are not already present instead of using a 5835 static variable. This allows them to be cleanly unloaded and reloaded. 5836 Improve header file function name parsing. 5837 [Steve Henson] 5838 5839 *) extend SMTP and IMAP protocol emulation in s_client to use EHLO 5840 or CAPABILITY handshake as required by RFCs. 5841 [Goetz Babin-Ebell] 5842 5843 Changes between 0.9.8c and 0.9.8d [28 Sep 2006] 5844 5845 *) Introduce limits to prevent malicious keys being able to 5846 cause a denial of service. (CVE-2006-2940) 5847 [Steve Henson, Bodo Moeller] 5848 5849 *) Fix ASN.1 parsing of certain invalid structures that can result 5850 in a denial of service. (CVE-2006-2937) [Steve Henson] 5851 5852 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 5853 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 5854 5855 *) Fix SSL client code which could crash if connecting to a 5856 malicious SSLv2 server. (CVE-2006-4343) 5857 [Tavis Ormandy and Will Drewry, Google Security Team] 5858 5859 *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites 5860 match only those. Before that, "AES256-SHA" would be interpreted 5861 as a pattern and match "AES128-SHA" too (since AES128-SHA got 5862 the same strength classification in 0.9.7h) as we currently only 5863 have a single AES bit in the ciphersuite description bitmap. 5864 That change, however, also applied to ciphersuite strings such as 5865 "RC4-MD5" that intentionally matched multiple ciphersuites -- 5866 namely, SSL 2.0 ciphersuites in addition to the more common ones 5867 from SSL 3.0/TLS 1.0. 5868 5869 So we change the selection algorithm again: Naming an explicit 5870 ciphersuite selects this one ciphersuite, and any other similar 5871 ciphersuite (same bitmap) from *other* protocol versions. 5872 Thus, "RC4-MD5" again will properly select both the SSL 2.0 5873 ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite. 5874 5875 Since SSL 2.0 does not have any ciphersuites for which the 5876 128/256 bit distinction would be relevant, this works for now. 5877 The proper fix will be to use different bits for AES128 and 5878 AES256, which would have avoided the problems from the beginning; 5879 however, bits are scarce, so we can only do this in a new release 5880 (not just a patchlevel) when we can change the SSL_CIPHER 5881 definition to split the single 'unsigned long mask' bitmap into 5882 multiple values to extend the available space. 5883 5884 [Bodo Moeller] 5885 5886 Changes between 0.9.8b and 0.9.8c [05 Sep 2006] 5887 5888 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 5889 (CVE-2006-4339) [Ben Laurie and Google Security Team] 5890 5891 *) Add AES IGE and biIGE modes. 5892 [Ben Laurie] 5893 5894 *) Change the Unix randomness entropy gathering to use poll() when 5895 possible instead of select(), since the latter has some 5896 undesirable limitations. 5897 [Darryl Miles via Richard Levitte and Bodo Moeller] 5898 5899 *) Disable "ECCdraft" ciphersuites more thoroughly. Now special 5900 treatment in ssl/ssl_ciph.s makes sure that these ciphersuites 5901 cannot be implicitly activated as part of, e.g., the "AES" alias. 5902 However, please upgrade to OpenSSL 0.9.9[-dev] for 5903 non-experimental use of the ECC ciphersuites to get TLS extension 5904 support, which is required for curve and point format negotiation 5905 to avoid potential handshake problems. 5906 [Bodo Moeller] 5907 5908 *) Disable rogue ciphersuites: 5909 5910 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 5911 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 5912 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 5913 5914 The latter two were purportedly from 5915 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 5916 appear there. 5917 5918 Also deactivate the remaining ciphersuites from 5919 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 5920 unofficial, and the ID has long expired. 5921 [Bodo Moeller] 5922 5923 *) Fix RSA blinding Heisenbug (problems sometimes occurred on 5924 dual-core machines) and other potential thread-safety issues. 5925 [Bodo Moeller] 5926 5927 *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key 5928 versions), which is now available for royalty-free use 5929 (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html). 5930 Also, add Camellia TLS ciphersuites from RFC 4132. 5931 5932 To minimize changes between patchlevels in the OpenSSL 0.9.8 5933 series, Camellia remains excluded from compilation unless OpenSSL 5934 is configured with 'enable-camellia'. 5935 [NTT] 5936 5937 *) Disable the padding bug check when compression is in use. The padding 5938 bug check assumes the first packet is of even length, this is not 5939 necessarily true if compression is enabled and can result in false 5940 positives causing handshake failure. The actual bug test is ancient 5941 code so it is hoped that implementations will either have fixed it by 5942 now or any which still have the bug do not support compression. 5943 [Steve Henson] 5944 5945 Changes between 0.9.8a and 0.9.8b [04 May 2006] 5946 5947 *) When applying a cipher rule check to see if string match is an explicit 5948 cipher suite and only match that one cipher suite if it is. 5949 [Steve Henson] 5950 5951 *) Link in manifests for VC++ if needed. 5952 [Austin Ziegler <halostatue@gmail.com>] 5953 5954 *) Update support for ECC-based TLS ciphersuites according to 5955 draft-ietf-tls-ecc-12.txt with proposed changes (but without 5956 TLS extensions, which are supported starting with the 0.9.9 5957 branch, not in the OpenSSL 0.9.8 branch). 5958 [Douglas Stebila] 5959 5960 *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support 5961 opaque EVP_CIPHER_CTX handling. 5962 [Steve Henson] 5963 5964 *) Fixes and enhancements to zlib compression code. We now only use 5965 "zlib1.dll" and use the default __cdecl calling convention on Win32 5966 to conform with the standards mentioned here: 5967 http://www.zlib.net/DLL_FAQ.txt 5968 Static zlib linking now works on Windows and the new --with-zlib-include 5969 --with-zlib-lib options to Configure can be used to supply the location 5970 of the headers and library. Gracefully handle case where zlib library 5971 can't be loaded. 5972 [Steve Henson] 5973 5974 *) Several fixes and enhancements to the OID generation code. The old code 5975 sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't 5976 handle numbers larger than ULONG_MAX, truncated printing and had a 5977 non standard OBJ_obj2txt() behaviour. 5978 [Steve Henson] 5979 5980 *) Add support for building of engines under engine/ as shared libraries 5981 under VC++ build system. 5982 [Steve Henson] 5983 5984 *) Corrected the numerous bugs in the Win32 path splitter in DSO. 5985 Hopefully, we will not see any false combination of paths any more. 5986 [Richard Levitte] 5987 5988 Changes between 0.9.8 and 0.9.8a [11 Oct 2005] 5989 5990 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 5991 (part of SSL_OP_ALL). This option used to disable the 5992 countermeasure against man-in-the-middle protocol-version 5993 rollback in the SSL 2.0 server implementation, which is a bad 5994 idea. (CVE-2005-2969) 5995 5996 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 5997 for Information Security, National Institute of Advanced Industrial 5998 Science and Technology [AIST], Japan)] 5999 6000 *) Add two function to clear and return the verify parameter flags. 6001 [Steve Henson] 6002 6003 *) Keep cipherlists sorted in the source instead of sorting them at 6004 runtime, thus removing the need for a lock. 6005 [Nils Larsch] 6006 6007 *) Avoid some small subgroup attacks in Diffie-Hellman. 6008 [Nick Mathewson and Ben Laurie] 6009 6010 *) Add functions for well-known primes. 6011 [Nick Mathewson] 6012 6013 *) Extended Windows CE support. 6014 [Satoshi Nakamura and Andy Polyakov] 6015 6016 *) Initialize SSL_METHOD structures at compile time instead of during 6017 runtime, thus removing the need for a lock. 6018 [Steve Henson] 6019 6020 *) Make PKCS7_decrypt() work even if no certificate is supplied by 6021 attempting to decrypt each encrypted key in turn. Add support to 6022 smime utility. 6023 [Steve Henson] 6024 6025 Changes between 0.9.7h and 0.9.8 [05 Jul 2005] 6026 6027 [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after 6028 OpenSSL 0.9.8.] 6029 6030 *) Add libcrypto.pc and libssl.pc for those who feel they need them. 6031 [Richard Levitte] 6032 6033 *) Change CA.sh and CA.pl so they don't bundle the CSR and the private 6034 key into the same file any more. 6035 [Richard Levitte] 6036 6037 *) Add initial support for Win64, both IA64 and AMD64/x64 flavors. 6038 [Andy Polyakov] 6039 6040 *) Add -utf8 command line and config file option to 'ca'. 6041 [Stefan <stf@udoma.org] 6042 6043 *) Removed the macro des_crypt(), as it seems to conflict with some 6044 libraries. Use DES_crypt(). 6045 [Richard Levitte] 6046 6047 *) Correct naming of the 'chil' and '4758cca' ENGINEs. This 6048 involves renaming the source and generated shared-libs for 6049 both. The engines will accept the corrected or legacy ids 6050 ('ncipher' and '4758_cca' respectively) when binding. NB, 6051 this only applies when building 'shared'. 6052 [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe] 6053 6054 *) Add attribute functions to EVP_PKEY structure. Modify 6055 PKCS12_create() to recognize a CSP name attribute and 6056 use it. Make -CSP option work again in pkcs12 utility. 6057 [Steve Henson] 6058 6059 *) Add new functionality to the bn blinding code: 6060 - automatic re-creation of the BN_BLINDING parameters after 6061 a fixed number of uses (currently 32) 6062 - add new function for parameter creation 6063 - introduce flags to control the update behaviour of the 6064 BN_BLINDING parameters 6065 - hide BN_BLINDING structure 6066 Add a second BN_BLINDING slot to the RSA structure to improve 6067 performance when a single RSA object is shared among several 6068 threads. 6069 [Nils Larsch] 6070 6071 *) Add support for DTLS. 6072 [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie] 6073 6074 *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1) 6075 to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file() 6076 [Walter Goulet] 6077 6078 *) Remove buggy and incomplete DH cert support from 6079 ssl/ssl_rsa.c and ssl/s3_both.c 6080 [Nils Larsch] 6081 6082 *) Use SHA-1 instead of MD5 as the default digest algorithm for 6083 the apps/openssl applications. 6084 [Nils Larsch] 6085 6086 *) Compile clean with "-Wall -Wmissing-prototypes 6087 -Wstrict-prototypes -Wmissing-declarations -Werror". Currently 6088 DEBUG_SAFESTACK must also be set. 6089 [Ben Laurie] 6090 6091 *) Change ./Configure so that certain algorithms can be disabled by default. 6092 The new counterpiece to "no-xxx" is "enable-xxx". 6093 6094 The patented RC5 and MDC2 algorithms will now be disabled unless 6095 "enable-rc5" and "enable-mdc2", respectively, are specified. 6096 6097 (IDEA remains enabled despite being patented. This is because IDEA 6098 is frequently required for interoperability, and there is no license 6099 fee for non-commercial use. As before, "no-idea" can be used to 6100 avoid this algorithm.) 6101 6102 [Bodo Moeller] 6103 6104 *) Add processing of proxy certificates (see RFC 3820). This work was 6105 sponsored by KTH (The Royal Institute of Technology in Stockholm) and 6106 EGEE (Enabling Grids for E-science in Europe). 6107 [Richard Levitte] 6108 6109 *) RC4 performance overhaul on modern architectures/implementations, such 6110 as Intel P4, IA-64 and AMD64. 6111 [Andy Polyakov] 6112 6113 *) New utility extract-section.pl. This can be used specify an alternative 6114 section number in a pod file instead of having to treat each file as 6115 a separate case in Makefile. This can be done by adding two lines to the 6116 pod file: 6117 6118 =for comment openssl_section:XXX 6119 6120 The blank line is mandatory. 6121 6122 [Steve Henson] 6123 6124 *) New arguments -certform, -keyform and -pass for s_client and s_server 6125 to allow alternative format key and certificate files and passphrase 6126 sources. 6127 [Steve Henson] 6128 6129 *) New structure X509_VERIFY_PARAM which combines current verify parameters, 6130 update associated structures and add various utility functions. 6131 6132 Add new policy related verify parameters, include policy checking in 6133 standard verify code. Enhance 'smime' application with extra parameters 6134 to support policy checking and print out. 6135 [Steve Henson] 6136 6137 *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3 6138 Nehemiah processors. These extensions support AES encryption in hardware 6139 as well as RNG (though RNG support is currently disabled). 6140 [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov] 6141 6142 *) Deprecate BN_[get|set]_params() functions (they were ignored internally). 6143 [Geoff Thorpe] 6144 6145 *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented. 6146 [Andy Polyakov and a number of other people] 6147 6148 *) Improved PowerPC platform support. Most notably BIGNUM assembler 6149 implementation contributed by IBM. 6150 [Suresh Chari, Peter Waltenberg, Andy Polyakov] 6151 6152 *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public 6153 exponent rather than 'unsigned long'. There is a corresponding change to 6154 the new 'rsa_keygen' element of the RSA_METHOD structure. 6155 [Jelte Jansen, Geoff Thorpe] 6156 6157 *) Functionality for creating the initial serial number file is now 6158 moved from CA.pl to the 'ca' utility with a new option -create_serial. 6159 6160 (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial 6161 number file to 1, which is bound to cause problems. To avoid 6162 the problems while respecting compatibility between different 0.9.7 6163 patchlevels, 0.9.7e employed 'openssl x509 -next_serial' in 6164 CA.pl for serial number initialization. With the new release 0.9.8, 6165 we can fix the problem directly in the 'ca' utility.) 6166 [Steve Henson] 6167 6168 *) Reduced header interdependencies by declaring more opaque objects in 6169 ossl_typ.h. As a consequence, including some headers (eg. engine.h) will 6170 give fewer recursive includes, which could break lazy source code - so 6171 this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always, 6172 developers should define this symbol when building and using openssl to 6173 ensure they track the recommended behaviour, interfaces, [etc], but 6174 backwards-compatible behaviour prevails when this isn't defined. 6175 [Geoff Thorpe] 6176 6177 *) New function X509_POLICY_NODE_print() which prints out policy nodes. 6178 [Steve Henson] 6179 6180 *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality. 6181 This will generate a random key of the appropriate length based on the 6182 cipher context. The EVP_CIPHER can provide its own random key generation 6183 routine to support keys of a specific form. This is used in the des and 6184 3des routines to generate a key of the correct parity. Update S/MIME 6185 code to use new functions and hence generate correct parity DES keys. 6186 Add EVP_CHECK_DES_KEY #define to return an error if the key is not 6187 valid (weak or incorrect parity). 6188 [Steve Henson] 6189 6190 *) Add a local set of CRLs that can be used by X509_verify_cert() as well 6191 as looking them up. This is useful when the verified structure may contain 6192 CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs 6193 present unless the new PKCS7_NO_CRL flag is asserted. 6194 [Steve Henson] 6195 6196 *) Extend ASN1 oid configuration module. It now additionally accepts the 6197 syntax: 6198 6199 shortName = some long name, 1.2.3.4 6200 [Steve Henson] 6201 6202 *) Reimplemented the BN_CTX implementation. There is now no more static 6203 limitation on the number of variables it can handle nor the depth of the 6204 "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack 6205 information can now expand as required, and rather than having a single 6206 static array of bignums, BN_CTX now uses a linked-list of such arrays 6207 allowing it to expand on demand whilst maintaining the usefulness of 6208 BN_CTX's "bundling". 6209 [Geoff Thorpe] 6210 6211 *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD 6212 to allow all RSA operations to function using a single BN_CTX. 6213 [Geoff Thorpe] 6214 6215 *) Preliminary support for certificate policy evaluation and checking. This 6216 is initially intended to pass the tests outlined in "Conformance Testing 6217 of Relying Party Client Certificate Path Processing Logic" v1.07. 6218 [Steve Henson] 6219 6220 *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and 6221 remained unused and not that useful. A variety of other little bignum 6222 tweaks and fixes have also been made continuing on from the audit (see 6223 below). 6224 [Geoff Thorpe] 6225 6226 *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with 6227 associated ASN1, EVP and SSL functions and old ASN1 macros. 6228 [Richard Levitte] 6229 6230 *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results, 6231 and this should never fail. So the return value from the use of 6232 BN_set_word() (which can fail due to needless expansion) is now deprecated; 6233 if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro. 6234 [Geoff Thorpe] 6235 6236 *) BN_CTX_get() should return zero-valued bignums, providing the same 6237 initialised value as BN_new(). 6238 [Geoff Thorpe, suggested by Ulf Möller] 6239 6240 *) Support for inhibitAnyPolicy certificate extension. 6241 [Steve Henson] 6242 6243 *) An audit of the BIGNUM code is underway, for which debugging code is 6244 enabled when BN_DEBUG is defined. This makes stricter enforcements on what 6245 is considered valid when processing BIGNUMs, and causes execution to 6246 assert() when a problem is discovered. If BN_DEBUG_RAND is defined, 6247 further steps are taken to deliberately pollute unused data in BIGNUM 6248 structures to try and expose faulty code further on. For now, openssl will 6249 (in its default mode of operation) continue to tolerate the inconsistent 6250 forms that it has tolerated in the past, but authors and packagers should 6251 consider trying openssl and their own applications when compiled with 6252 these debugging symbols defined. It will help highlight potential bugs in 6253 their own code, and will improve the test coverage for OpenSSL itself. At 6254 some point, these tighter rules will become openssl's default to improve 6255 maintainability, though the assert()s and other overheads will remain only 6256 in debugging configurations. See bn.h for more details. 6257 [Geoff Thorpe, Nils Larsch, Ulf Möller] 6258 6259 *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure 6260 that can only be obtained through BN_CTX_new() (which implicitly 6261 initialises it). The presence of this function only made it possible 6262 to overwrite an existing structure (and cause memory leaks). 6263 [Geoff Thorpe] 6264 6265 *) Because of the callback-based approach for implementing LHASH as a 6266 template type, lh_insert() adds opaque objects to hash-tables and 6267 lh_doall() or lh_doall_arg() are typically used with a destructor callback 6268 to clean up those corresponding objects before destroying the hash table 6269 (and losing the object pointers). So some over-zealous constifications in 6270 LHASH have been relaxed so that lh_insert() does not take (nor store) the 6271 objects as "const" and the lh_doall[_arg] callback wrappers are not 6272 prototyped to have "const" restrictions on the object pointers they are 6273 given (and so aren't required to cast them away any more). 6274 [Geoff Thorpe] 6275 6276 *) The tmdiff.h API was so ugly and minimal that our own timing utility 6277 (speed) prefers to use its own implementation. The two implementations 6278 haven't been consolidated as yet (volunteers?) but the tmdiff API has had 6279 its object type properly exposed (MS_TM) instead of casting to/from "char 6280 *". This may still change yet if someone realises MS_TM and "ms_time_***" 6281 aren't necessarily the greatest nomenclatures - but this is what was used 6282 internally to the implementation so I've used that for now. 6283 [Geoff Thorpe] 6284 6285 *) Ensure that deprecated functions do not get compiled when 6286 OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of 6287 the self-tests were still using deprecated key-generation functions so 6288 these have been updated also. 6289 [Geoff Thorpe] 6290 6291 *) Reorganise PKCS#7 code to separate the digest location functionality 6292 into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest(). 6293 New function PKCS7_set_digest() to set the digest type for PKCS#7 6294 digestedData type. Add additional code to correctly generate the 6295 digestedData type and add support for this type in PKCS7 initialization 6296 functions. 6297 [Steve Henson] 6298 6299 *) New function PKCS7_set0_type_other() this initializes a PKCS7 6300 structure of type "other". 6301 [Steve Henson] 6302 6303 *) Fix prime generation loop in crypto/bn/bn_prime.pl by making 6304 sure the loop does correctly stop and breaking ("division by zero") 6305 modulus operations are not performed. The (pre-generated) prime 6306 table crypto/bn/bn_prime.h was already correct, but it could not be 6307 re-generated on some platforms because of the "division by zero" 6308 situation in the script. 6309 [Ralf S. Engelschall] 6310 6311 *) Update support for ECC-based TLS ciphersuites according to 6312 draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with 6313 SHA-1 now is only used for "small" curves (where the 6314 representation of a field element takes up to 24 bytes); for 6315 larger curves, the field element resulting from ECDH is directly 6316 used as premaster secret. 6317 [Douglas Stebila (Sun Microsystems Laboratories)] 6318 6319 *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2 6320 curve secp160r1 to the tests. 6321 [Douglas Stebila (Sun Microsystems Laboratories)] 6322 6323 *) Add the possibility to load symbols globally with DSO. 6324 [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte] 6325 6326 *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better 6327 control of the error stack. 6328 [Richard Levitte] 6329 6330 *) Add support for STORE in ENGINE. 6331 [Richard Levitte] 6332 6333 *) Add the STORE type. The intention is to provide a common interface 6334 to certificate and key stores, be they simple file-based stores, or 6335 HSM-type store, or LDAP stores, or... 6336 NOTE: The code is currently UNTESTED and isn't really used anywhere. 6337 [Richard Levitte] 6338 6339 *) Add a generic structure called OPENSSL_ITEM. This can be used to 6340 pass a list of arguments to any function as well as provide a way 6341 for a function to pass data back to the caller. 6342 [Richard Levitte] 6343 6344 *) Add the functions BUF_strndup() and BUF_memdup(). BUF_strndup() 6345 works like BUF_strdup() but can be used to duplicate a portion of 6346 a string. The copy gets NUL-terminated. BUF_memdup() duplicates 6347 a memory area. 6348 [Richard Levitte] 6349 6350 *) Add the function sk_find_ex() which works like sk_find(), but will 6351 return an index to an element even if an exact match couldn't be 6352 found. The index is guaranteed to point at the element where the 6353 searched-for key would be inserted to preserve sorting order. 6354 [Richard Levitte] 6355 6356 *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but 6357 takes an extra flags argument for optional functionality. Currently, 6358 the following flags are defined: 6359 6360 OBJ_BSEARCH_VALUE_ON_NOMATCH 6361 This one gets OBJ_bsearch_ex() to return a pointer to the first 6362 element where the comparing function returns a negative or zero 6363 number. 6364 6365 OBJ_BSEARCH_FIRST_VALUE_ON_MATCH 6366 This one gets OBJ_bsearch_ex() to return a pointer to the first 6367 element where the comparing function returns zero. This is useful 6368 if there are more than one element where the comparing function 6369 returns zero. 6370 [Richard Levitte] 6371 6372 *) Make it possible to create self-signed certificates with 'openssl ca' 6373 in such a way that the self-signed certificate becomes part of the 6374 CA database and uses the same mechanisms for serial number generation 6375 as all other certificate signing. The new flag '-selfsign' enables 6376 this functionality. Adapt CA.sh and CA.pl.in. 6377 [Richard Levitte] 6378 6379 *) Add functionality to check the public key of a certificate request 6380 against a given private. This is useful to check that a certificate 6381 request can be signed by that key (self-signing). 6382 [Richard Levitte] 6383 6384 *) Make it possible to have multiple active certificates with the same 6385 subject in the CA index file. This is done only if the keyword 6386 'unique_subject' is set to 'no' in the main CA section (default 6387 if 'CA_default') of the configuration file. The value is saved 6388 with the database itself in a separate index attribute file, 6389 named like the index file with '.attr' appended to the name. 6390 [Richard Levitte] 6391 6392 *) Generate multi-valued AVAs using '+' notation in config files for 6393 req and dirName. 6394 [Steve Henson] 6395 6396 *) Support for nameConstraints certificate extension. 6397 [Steve Henson] 6398 6399 *) Support for policyConstraints certificate extension. 6400 [Steve Henson] 6401 6402 *) Support for policyMappings certificate extension. 6403 [Steve Henson] 6404 6405 *) Make sure the default DSA_METHOD implementation only uses its 6406 dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL, 6407 and change its own handlers to be NULL so as to remove unnecessary 6408 indirection. This lets alternative implementations fallback to the 6409 default implementation more easily. 6410 [Geoff Thorpe] 6411 6412 *) Support for directoryName in GeneralName related extensions 6413 in config files. 6414 [Steve Henson] 6415 6416 *) Make it possible to link applications using Makefile.shared. 6417 Make that possible even when linking against static libraries! 6418 [Richard Levitte] 6419 6420 *) Support for single pass processing for S/MIME signing. This now 6421 means that S/MIME signing can be done from a pipe, in addition 6422 cleartext signing (multipart/signed type) is effectively streaming 6423 and the signed data does not need to be all held in memory. 6424 6425 This is done with a new flag PKCS7_STREAM. When this flag is set 6426 PKCS7_sign() only initializes the PKCS7 structure and the actual signing 6427 is done after the data is output (and digests calculated) in 6428 SMIME_write_PKCS7(). 6429 [Steve Henson] 6430 6431 *) Add full support for -rpath/-R, both in shared libraries and 6432 applications, at least on the platforms where it's known how 6433 to do it. 6434 [Richard Levitte] 6435 6436 *) In crypto/ec/ec_mult.c, implement fast point multiplication with 6437 precomputation, based on wNAF splitting: EC_GROUP_precompute_mult() 6438 will now compute a table of multiples of the generator that 6439 makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul() 6440 faster (notably in the case of a single point multiplication, 6441 scalar * generator). 6442 [Nils Larsch, Bodo Moeller] 6443 6444 *) IPv6 support for certificate extensions. The various extensions 6445 which use the IP:a.b.c.d can now take IPv6 addresses using the 6446 formats of RFC1884 2.2 . IPv6 addresses are now also displayed 6447 correctly. 6448 [Steve Henson] 6449 6450 *) Added an ENGINE that implements RSA by performing private key 6451 exponentiations with the GMP library. The conversions to and from 6452 GMP's mpz_t format aren't optimised nor are any montgomery forms 6453 cached, and on x86 it appears OpenSSL's own performance has caught up. 6454 However there are likely to be other architectures where GMP could 6455 provide a boost. This ENGINE is not built in by default, but it can be 6456 specified at Configure time and should be accompanied by the necessary 6457 linker additions, eg; 6458 ./config -DOPENSSL_USE_GMP -lgmp 6459 [Geoff Thorpe] 6460 6461 *) "openssl engine" will not display ENGINE/DSO load failure errors when 6462 testing availability of engines with "-t" - the old behaviour is 6463 produced by increasing the feature's verbosity with "-tt". 6464 [Geoff Thorpe] 6465 6466 *) ECDSA routines: under certain error conditions uninitialized BN objects 6467 could be freed. Solution: make sure initialization is performed early 6468 enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de> 6469 via PR#459) 6470 [Lutz Jaenicke] 6471 6472 *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD 6473 and DH_METHOD (eg. by ENGINE implementations) to override the normal 6474 software implementations. For DSA and DH, parameter generation can 6475 also be overridden by providing the appropriate method callbacks. 6476 [Geoff Thorpe] 6477 6478 *) Change the "progress" mechanism used in key-generation and 6479 primality testing to functions that take a new BN_GENCB pointer in 6480 place of callback/argument pairs. The new API functions have "_ex" 6481 postfixes and the older functions are reimplemented as wrappers for 6482 the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide 6483 declarations of the old functions to help (graceful) attempts to 6484 migrate to the new functions. Also, the new key-generation API 6485 functions operate on a caller-supplied key-structure and return 6486 success/failure rather than returning a key or NULL - this is to 6487 help make "keygen" another member function of RSA_METHOD etc. 6488 6489 Example for using the new callback interface: 6490 6491 int (*my_callback)(int a, int b, BN_GENCB *cb) = ...; 6492 void *my_arg = ...; 6493 BN_GENCB my_cb; 6494 6495 BN_GENCB_set(&my_cb, my_callback, my_arg); 6496 6497 return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb); 6498 /* For the meaning of a, b in calls to my_callback(), see the 6499 * documentation of the function that calls the callback. 6500 * cb will point to my_cb; my_arg can be retrieved as cb->arg. 6501 * my_callback should return 1 if it wants BN_is_prime_ex() 6502 * to continue, or 0 to stop. 6503 */ 6504 6505 [Geoff Thorpe] 6506 6507 *) Change the ZLIB compression method to be stateful, and make it 6508 available to TLS with the number defined in 6509 draft-ietf-tls-compression-04.txt. 6510 [Richard Levitte] 6511 6512 *) Add the ASN.1 structures and functions for CertificatePair, which 6513 is defined as follows (according to X.509_4thEditionDraftV6.pdf): 6514 6515 CertificatePair ::= SEQUENCE { 6516 forward [0] Certificate OPTIONAL, 6517 reverse [1] Certificate OPTIONAL, 6518 -- at least one of the pair shall be present -- } 6519 6520 Also implement the PEM functions to read and write certificate 6521 pairs, and defined the PEM tag as "CERTIFICATE PAIR". 6522 6523 This needed to be defined, mostly for the sake of the LDAP 6524 attribute crossCertificatePair, but may prove useful elsewhere as 6525 well. 6526 [Richard Levitte] 6527 6528 *) Make it possible to inhibit symlinking of shared libraries in 6529 Makefile.shared, for Cygwin's sake. 6530 [Richard Levitte] 6531 6532 *) Extend the BIGNUM API by creating a function 6533 void BN_set_negative(BIGNUM *a, int neg); 6534 and a macro that behave like 6535 int BN_is_negative(const BIGNUM *a); 6536 6537 to avoid the need to access 'a->neg' directly in applications. 6538 [Nils Larsch] 6539 6540 *) Implement fast modular reduction for pseudo-Mersenne primes 6541 used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c). 6542 EC_GROUP_new_curve_GFp() will now automatically use this 6543 if applicable. 6544 [Nils Larsch <nla@trustcenter.de>] 6545 6546 *) Add new lock type (CRYPTO_LOCK_BN). 6547 [Bodo Moeller] 6548 6549 *) Change the ENGINE framework to automatically load engines 6550 dynamically from specific directories unless they could be 6551 found to already be built in or loaded. Move all the 6552 current engines except for the cryptodev one to a new 6553 directory engines/. 6554 The engines in engines/ are built as shared libraries if 6555 the "shared" options was given to ./Configure or ./config. 6556 Otherwise, they are inserted in libcrypto.a. 6557 /usr/local/ssl/engines is the default directory for dynamic 6558 engines, but that can be overridden at configure time through 6559 the usual use of --prefix and/or --openssldir, and at run 6560 time with the environment variable OPENSSL_ENGINES. 6561 [Geoff Thorpe and Richard Levitte] 6562 6563 *) Add Makefile.shared, a helper makefile to build shared 6564 libraries. Adapt Makefile.org. 6565 [Richard Levitte] 6566 6567 *) Add version info to Win32 DLLs. 6568 [Peter 'Luna' Runestig" <peter@runestig.com>] 6569 6570 *) Add new 'medium level' PKCS#12 API. Certificates and keys 6571 can be added using this API to created arbitrary PKCS#12 6572 files while avoiding the low level API. 6573 6574 New options to PKCS12_create(), key or cert can be NULL and 6575 will then be omitted from the output file. The encryption 6576 algorithm NIDs can be set to -1 for no encryption, the mac 6577 iteration count can be set to 0 to omit the mac. 6578 6579 Enhance pkcs12 utility by making the -nokeys and -nocerts 6580 options work when creating a PKCS#12 file. New option -nomac 6581 to omit the mac, NONE can be set for an encryption algorithm. 6582 New code is modified to use the enhanced PKCS12_create() 6583 instead of the low level API. 6584 [Steve Henson] 6585 6586 *) Extend ASN1 encoder to support indefinite length constructed 6587 encoding. This can output sequences tags and octet strings in 6588 this form. Modify pk7_asn1.c to support indefinite length 6589 encoding. This is experimental and needs additional code to 6590 be useful, such as an ASN1 bio and some enhanced streaming 6591 PKCS#7 code. 6592 6593 Extend template encode functionality so that tagging is passed 6594 down to the template encoder. 6595 [Steve Henson] 6596 6597 *) Let 'openssl req' fail if an argument to '-newkey' is not 6598 recognized instead of using RSA as a default. 6599 [Bodo Moeller] 6600 6601 *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt. 6602 As these are not official, they are not included in "ALL"; 6603 the "ECCdraft" ciphersuite group alias can be used to select them. 6604 [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)] 6605 6606 *) Add ECDH engine support. 6607 [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)] 6608 6609 *) Add ECDH in new directory crypto/ecdh/. 6610 [Douglas Stebila (Sun Microsystems Laboratories)] 6611 6612 *) Let BN_rand_range() abort with an error after 100 iterations 6613 without success (which indicates a broken PRNG). 6614 [Bodo Moeller] 6615 6616 *) Change BN_mod_sqrt() so that it verifies that the input value 6617 is really the square of the return value. (Previously, 6618 BN_mod_sqrt would show GIGO behaviour.) 6619 [Bodo Moeller] 6620 6621 *) Add named elliptic curves over binary fields from X9.62, SECG, 6622 and WAP/WTLS; add OIDs that were still missing. 6623 6624 [Sheueling Chang Shantz and Douglas Stebila 6625 (Sun Microsystems Laboratories)] 6626 6627 *) Extend the EC library for elliptic curves over binary fields 6628 (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/). 6629 New EC_METHOD: 6630 6631 EC_GF2m_simple_method 6632 6633 New API functions: 6634 6635 EC_GROUP_new_curve_GF2m 6636 EC_GROUP_set_curve_GF2m 6637 EC_GROUP_get_curve_GF2m 6638 EC_POINT_set_affine_coordinates_GF2m 6639 EC_POINT_get_affine_coordinates_GF2m 6640 EC_POINT_set_compressed_coordinates_GF2m 6641 6642 Point compression for binary fields is disabled by default for 6643 patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to 6644 enable it). 6645 6646 As binary polynomials are represented as BIGNUMs, various members 6647 of the EC_GROUP and EC_POINT data structures can be shared 6648 between the implementations for prime fields and binary fields; 6649 the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m) 6650 are essentially identical to their ..._GFp counterparts. 6651 (For simplicity, the '..._GFp' prefix has been dropped from 6652 various internal method names.) 6653 6654 An internal 'field_div' method (similar to 'field_mul' and 6655 'field_sqr') has been added; this is used only for binary fields. 6656 6657 [Sheueling Chang Shantz and Douglas Stebila 6658 (Sun Microsystems Laboratories)] 6659 6660 *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult() 6661 through methods ('mul', 'precompute_mult'). 6662 6663 The generic implementations (now internally called 'ec_wNAF_mul' 6664 and 'ec_wNAF_precomputed_mult') remain the default if these 6665 methods are undefined. 6666 6667 [Sheueling Chang Shantz and Douglas Stebila 6668 (Sun Microsystems Laboratories)] 6669 6670 *) New function EC_GROUP_get_degree, which is defined through 6671 EC_METHOD. For curves over prime fields, this returns the bit 6672 length of the modulus. 6673 6674 [Sheueling Chang Shantz and Douglas Stebila 6675 (Sun Microsystems Laboratories)] 6676 6677 *) New functions EC_GROUP_dup, EC_POINT_dup. 6678 (These simply call ..._new and ..._copy). 6679 6680 [Sheueling Chang Shantz and Douglas Stebila 6681 (Sun Microsystems Laboratories)] 6682 6683 *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c. 6684 Polynomials are represented as BIGNUMs (where the sign bit is not 6685 used) in the following functions [macros]: 6686 6687 BN_GF2m_add 6688 BN_GF2m_sub [= BN_GF2m_add] 6689 BN_GF2m_mod [wrapper for BN_GF2m_mod_arr] 6690 BN_GF2m_mod_mul [wrapper for BN_GF2m_mod_mul_arr] 6691 BN_GF2m_mod_sqr [wrapper for BN_GF2m_mod_sqr_arr] 6692 BN_GF2m_mod_inv 6693 BN_GF2m_mod_exp [wrapper for BN_GF2m_mod_exp_arr] 6694 BN_GF2m_mod_sqrt [wrapper for BN_GF2m_mod_sqrt_arr] 6695 BN_GF2m_mod_solve_quad [wrapper for BN_GF2m_mod_solve_quad_arr] 6696 BN_GF2m_cmp [= BN_ucmp] 6697 6698 (Note that only the 'mod' functions are actually for fields GF(2^m). 6699 BN_GF2m_add() is misnomer, but this is for the sake of consistency.) 6700 6701 For some functions, an the irreducible polynomial defining a 6702 field can be given as an 'unsigned int[]' with strictly 6703 decreasing elements giving the indices of those bits that are set; 6704 i.e., p[] represents the polynomial 6705 f(t) = t^p[0] + t^p[1] + ... + t^p[k] 6706 where 6707 p[0] > p[1] > ... > p[k] = 0. 6708 This applies to the following functions: 6709 6710 BN_GF2m_mod_arr 6711 BN_GF2m_mod_mul_arr 6712 BN_GF2m_mod_sqr_arr 6713 BN_GF2m_mod_inv_arr [wrapper for BN_GF2m_mod_inv] 6714 BN_GF2m_mod_div_arr [wrapper for BN_GF2m_mod_div] 6715 BN_GF2m_mod_exp_arr 6716 BN_GF2m_mod_sqrt_arr 6717 BN_GF2m_mod_solve_quad_arr 6718 BN_GF2m_poly2arr 6719 BN_GF2m_arr2poly 6720 6721 Conversion can be performed by the following functions: 6722 6723 BN_GF2m_poly2arr 6724 BN_GF2m_arr2poly 6725 6726 bntest.c has additional tests for binary polynomial arithmetic. 6727 6728 Two implementations for BN_GF2m_mod_div() are available. 6729 The default algorithm simply uses BN_GF2m_mod_inv() and 6730 BN_GF2m_mod_mul(). The alternative algorithm is compiled in only 6731 if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the 6732 copyright notice in crypto/bn/bn_gf2m.c before enabling it). 6733 6734 [Sheueling Chang Shantz and Douglas Stebila 6735 (Sun Microsystems Laboratories)] 6736 6737 *) Add new error code 'ERR_R_DISABLED' that can be used when some 6738 functionality is disabled at compile-time. 6739 [Douglas Stebila <douglas.stebila@sun.com>] 6740 6741 *) Change default behaviour of 'openssl asn1parse' so that more 6742 information is visible when viewing, e.g., a certificate: 6743 6744 Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump' 6745 mode the content of non-printable OCTET STRINGs is output in a 6746 style similar to INTEGERs, but with '[HEX DUMP]' prepended to 6747 avoid the appearance of a printable string. 6748 [Nils Larsch <nla@trustcenter.de>] 6749 6750 *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access 6751 functions 6752 EC_GROUP_set_asn1_flag() 6753 EC_GROUP_get_asn1_flag() 6754 EC_GROUP_set_point_conversion_form() 6755 EC_GROUP_get_point_conversion_form() 6756 These control ASN1 encoding details: 6757 - Curves (i.e., groups) are encoded explicitly unless asn1_flag 6758 has been set to OPENSSL_EC_NAMED_CURVE. 6759 - Points are encoded in uncompressed form by default; options for 6760 asn1_for are as for point2oct, namely 6761 POINT_CONVERSION_COMPRESSED 6762 POINT_CONVERSION_UNCOMPRESSED 6763 POINT_CONVERSION_HYBRID 6764 6765 Also add 'seed' and 'seed_len' members to EC_GROUP with access 6766 functions 6767 EC_GROUP_set_seed() 6768 EC_GROUP_get0_seed() 6769 EC_GROUP_get_seed_len() 6770 This is used only for ASN1 purposes (so far). 6771 [Nils Larsch <nla@trustcenter.de>] 6772 6773 *) Add 'field_type' member to EC_METHOD, which holds the NID 6774 of the appropriate field type OID. The new function 6775 EC_METHOD_get_field_type() returns this value. 6776 [Nils Larsch <nla@trustcenter.de>] 6777 6778 *) Add functions 6779 EC_POINT_point2bn() 6780 EC_POINT_bn2point() 6781 EC_POINT_point2hex() 6782 EC_POINT_hex2point() 6783 providing useful interfaces to EC_POINT_point2oct() and 6784 EC_POINT_oct2point(). 6785 [Nils Larsch <nla@trustcenter.de>] 6786 6787 *) Change internals of the EC library so that the functions 6788 EC_GROUP_set_generator() 6789 EC_GROUP_get_generator() 6790 EC_GROUP_get_order() 6791 EC_GROUP_get_cofactor() 6792 are implemented directly in crypto/ec/ec_lib.c and not dispatched 6793 to methods, which would lead to unnecessary code duplication when 6794 adding different types of curves. 6795 [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller] 6796 6797 *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM 6798 arithmetic, and such that modified wNAFs are generated 6799 (which avoid length expansion in many cases). 6800 [Bodo Moeller] 6801 6802 *) Add a function EC_GROUP_check_discriminant() (defined via 6803 EC_METHOD) that verifies that the curve discriminant is non-zero. 6804 6805 Add a function EC_GROUP_check() that makes some sanity tests 6806 on a EC_GROUP, its generator and order. This includes 6807 EC_GROUP_check_discriminant(). 6808 [Nils Larsch <nla@trustcenter.de>] 6809 6810 *) Add ECDSA in new directory crypto/ecdsa/. 6811 6812 Add applications 'openssl ecparam' and 'openssl ecdsa' 6813 (these are based on 'openssl dsaparam' and 'openssl dsa'). 6814 6815 ECDSA support is also included in various other files across the 6816 library. Most notably, 6817 - 'openssl req' now has a '-newkey ecdsa:file' option; 6818 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA; 6819 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and 6820 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make 6821 them suitable for ECDSA where domain parameters must be 6822 extracted before the specific public key; 6823 - ECDSA engine support has been added. 6824 [Nils Larsch <nla@trustcenter.de>] 6825 6826 *) Include some named elliptic curves, and add OIDs from X9.62, 6827 SECG, and WAP/WTLS. Each curve can be obtained from the new 6828 function 6829 EC_GROUP_new_by_curve_name(), 6830 and the list of available named curves can be obtained with 6831 EC_get_builtin_curves(). 6832 Also add a 'curve_name' member to EC_GROUP objects, which can be 6833 accessed via 6834 EC_GROUP_set_curve_name() 6835 EC_GROUP_get_curve_name() 6836 [Nils Larsch <larsch@trustcenter.de, Bodo Moeller] 6837 6838 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 6839 was actually never needed) and in BN_mul(). The removal in BN_mul() 6840 required a small change in bn_mul_part_recursive() and the addition 6841 of the functions bn_cmp_part_words(), bn_sub_part_words() and 6842 bn_add_part_words(), which do the same thing as bn_cmp_words(), 6843 bn_sub_words() and bn_add_words() except they take arrays with 6844 differing sizes. 6845 [Richard Levitte] 6846 6847 Changes between 0.9.7l and 0.9.7m [23 Feb 2007] 6848 6849 *) Cleanse PEM buffers before freeing them since they may contain 6850 sensitive data. 6851 [Benjamin Bennett <ben@psc.edu>] 6852 6853 *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that 6854 a ciphersuite string such as "DEFAULT:RSA" cannot enable 6855 authentication-only ciphersuites. 6856 [Bodo Moeller] 6857 6858 *) Since AES128 and AES256 share a single mask bit in the logic of 6859 ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a 6860 kludge to work properly if AES128 is available and AES256 isn't. 6861 [Victor Duchovni] 6862 6863 *) Expand security boundary to match 1.1.1 module. 6864 [Steve Henson] 6865 6866 *) Remove redundant features: hash file source, editing of test vectors 6867 modify fipsld to use external fips_premain.c signature. 6868 [Steve Henson] 6869 6870 *) New perl script mkfipsscr.pl to create shell scripts or batch files to 6871 run algorithm test programs. 6872 [Steve Henson] 6873 6874 *) Make algorithm test programs more tolerant of whitespace. 6875 [Steve Henson] 6876 6877 *) Have SSL/TLS server implementation tolerate "mismatched" record 6878 protocol version while receiving ClientHello even if the 6879 ClientHello is fragmented. (The server can't insist on the 6880 particular protocol version it has chosen before the ServerHello 6881 message has informed the client about his choice.) 6882 [Bodo Moeller] 6883 6884 *) Load error codes if they are not already present instead of using a 6885 static variable. This allows them to be cleanly unloaded and reloaded. 6886 [Steve Henson] 6887 6888 Changes between 0.9.7k and 0.9.7l [28 Sep 2006] 6889 6890 *) Introduce limits to prevent malicious keys being able to 6891 cause a denial of service. (CVE-2006-2940) 6892 [Steve Henson, Bodo Moeller] 6893 6894 *) Fix ASN.1 parsing of certain invalid structures that can result 6895 in a denial of service. (CVE-2006-2937) [Steve Henson] 6896 6897 *) Fix buffer overflow in SSL_get_shared_ciphers() function. 6898 (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team] 6899 6900 *) Fix SSL client code which could crash if connecting to a 6901 malicious SSLv2 server. (CVE-2006-4343) 6902 [Tavis Ormandy and Will Drewry, Google Security Team] 6903 6904 *) Change ciphersuite string processing so that an explicit 6905 ciphersuite selects this one ciphersuite (so that "AES256-SHA" 6906 will no longer include "AES128-SHA"), and any other similar 6907 ciphersuite (same bitmap) from *other* protocol versions (so that 6908 "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the 6909 SSL 3.0/TLS 1.0 ciphersuite). This is a backport combining 6910 changes from 0.9.8b and 0.9.8d. 6911 [Bodo Moeller] 6912 6913 Changes between 0.9.7j and 0.9.7k [05 Sep 2006] 6914 6915 *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher 6916 (CVE-2006-4339) [Ben Laurie and Google Security Team] 6917 6918 *) Change the Unix randomness entropy gathering to use poll() when 6919 possible instead of select(), since the latter has some 6920 undesirable limitations. 6921 [Darryl Miles via Richard Levitte and Bodo Moeller] 6922 6923 *) Disable rogue ciphersuites: 6924 6925 - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5") 6926 - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5") 6927 - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5") 6928 6929 The latter two were purportedly from 6930 draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really 6931 appear there. 6932 6933 Also deactivate the remaining ciphersuites from 6934 draft-ietf-tls-56-bit-ciphersuites-01.txt. These are just as 6935 unofficial, and the ID has long expired. 6936 [Bodo Moeller] 6937 6938 *) Fix RSA blinding Heisenbug (problems sometimes occurred on 6939 dual-core machines) and other potential thread-safety issues. 6940 [Bodo Moeller] 6941 6942 Changes between 0.9.7i and 0.9.7j [04 May 2006] 6943 6944 *) Adapt fipsld and the build system to link against the validated FIPS 6945 module in FIPS mode. 6946 [Steve Henson] 6947 6948 *) Fixes for VC++ 2005 build under Windows. 6949 [Steve Henson] 6950 6951 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 6952 from a Windows bash shell such as MSYS. It is autodetected from the 6953 "config" script when run from a VC++ environment. Modify standard VC++ 6954 build to use fipscanister.o from the GNU make build. 6955 [Steve Henson] 6956 6957 Changes between 0.9.7h and 0.9.7i [14 Oct 2005] 6958 6959 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS. 6960 The value now differs depending on if you build for FIPS or not. 6961 BEWARE! A program linked with a shared FIPSed libcrypto can't be 6962 safely run with a non-FIPSed libcrypto, as it may crash because of 6963 the difference induced by this change. 6964 [Andy Polyakov] 6965 6966 Changes between 0.9.7g and 0.9.7h [11 Oct 2005] 6967 6968 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING 6969 (part of SSL_OP_ALL). This option used to disable the 6970 countermeasure against man-in-the-middle protocol-version 6971 rollback in the SSL 2.0 server implementation, which is a bad 6972 idea. (CVE-2005-2969) 6973 6974 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center 6975 for Information Security, National Institute of Advanced Industrial 6976 Science and Technology [AIST], Japan)] 6977 6978 *) Minimal support for X9.31 signatures and PSS padding modes. This is 6979 mainly for FIPS compliance and not fully integrated at this stage. 6980 [Steve Henson] 6981 6982 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform 6983 the exponentiation using a fixed-length exponent. (Otherwise, 6984 the information leaked through timing could expose the secret key 6985 after many signatures; cf. Bleichenbacher's attack on DSA with 6986 biased k.) 6987 [Bodo Moeller] 6988 6989 *) Make a new fixed-window mod_exp implementation the default for 6990 RSA, DSA, and DH private-key operations so that the sequence of 6991 squares and multiplies and the memory access pattern are 6992 independent of the particular secret key. This will mitigate 6993 cache-timing and potential related attacks. 6994 6995 BN_mod_exp_mont_consttime() is the new exponentiation implementation, 6996 and this is automatically used by BN_mod_exp_mont() if the new flag 6997 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH 6998 will use this BN flag for private exponents unless the flag 6999 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or 7000 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set. 7001 7002 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller] 7003 7004 *) Change the client implementation for SSLv23_method() and 7005 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0 7006 Client Hello message format if the SSL_OP_NO_SSLv2 option is set. 7007 (Previously, the SSL 2.0 backwards compatible Client Hello 7008 message format would be used even with SSL_OP_NO_SSLv2.) 7009 [Bodo Moeller] 7010 7011 *) Add support for smime-type MIME parameter in S/MIME messages which some 7012 clients need. 7013 [Steve Henson] 7014 7015 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in 7016 a threadsafe manner. Modify rsa code to use new function and add calls 7017 to dsa and dh code (which had race conditions before). 7018 [Steve Henson] 7019 7020 *) Include the fixed error library code in the C error file definitions 7021 instead of fixing them up at runtime. This keeps the error code 7022 structures constant. 7023 [Steve Henson] 7024 7025 Changes between 0.9.7f and 0.9.7g [11 Apr 2005] 7026 7027 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after 7028 OpenSSL 0.9.8.] 7029 7030 *) Fixes for newer kerberos headers. NB: the casts are needed because 7031 the 'length' field is signed on one version and unsigned on another 7032 with no (?) obvious way to tell the difference, without these VC++ 7033 complains. Also the "definition" of FAR (blank) is no longer included 7034 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up 7035 some needed definitions. 7036 [Steve Henson] 7037 7038 *) Undo Cygwin change. 7039 [Ulf Möller] 7040 7041 *) Added support for proxy certificates according to RFC 3820. 7042 Because they may be a security thread to unaware applications, 7043 they must be explicitly allowed in run-time. See 7044 docs/HOWTO/proxy_certificates.txt for further information. 7045 [Richard Levitte] 7046 7047 Changes between 0.9.7e and 0.9.7f [22 Mar 2005] 7048 7049 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating 7050 server and client random values. Previously 7051 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in 7052 less random data when sizeof(time_t) > 4 (some 64 bit platforms). 7053 7054 This change has negligible security impact because: 7055 7056 1. Server and client random values still have 24 bytes of pseudo random 7057 data. 7058 7059 2. Server and client random values are sent in the clear in the initial 7060 handshake. 7061 7062 3. The master secret is derived using the premaster secret (48 bytes in 7063 size for static RSA ciphersuites) as well as client server and random 7064 values. 7065 7066 The OpenSSL team would like to thank the UK NISCC for bringing this issue 7067 to our attention. 7068 7069 [Stephen Henson, reported by UK NISCC] 7070 7071 *) Use Windows randomness collection on Cygwin. 7072 [Ulf Möller] 7073 7074 *) Fix hang in EGD/PRNGD query when communication socket is closed 7075 prematurely by EGD/PRNGD. 7076 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014] 7077 7078 *) Prompt for pass phrases when appropriate for PKCS12 input format. 7079 [Steve Henson] 7080 7081 *) Back-port of selected performance improvements from development 7082 branch, as well as improved support for PowerPC platforms. 7083 [Andy Polyakov] 7084 7085 *) Add lots of checks for memory allocation failure, error codes to indicate 7086 failure and freeing up memory if a failure occurs. 7087 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson] 7088 7089 *) Add new -passin argument to dgst. 7090 [Steve Henson] 7091 7092 *) Perform some character comparisons of different types in X509_NAME_cmp: 7093 this is needed for some certificates that re-encode DNs into UTF8Strings 7094 (in violation of RFC3280) and can't or won't issue name rollover 7095 certificates. 7096 [Steve Henson] 7097 7098 *) Make an explicit check during certificate validation to see that 7099 the CA setting in each certificate on the chain is correct. As a 7100 side effect always do the following basic checks on extensions, 7101 not just when there's an associated purpose to the check: 7102 7103 - if there is an unhandled critical extension (unless the user 7104 has chosen to ignore this fault) 7105 - if the path length has been exceeded (if one is set at all) 7106 - that certain extensions fit the associated purpose (if one has 7107 been given) 7108 [Richard Levitte] 7109 7110 Changes between 0.9.7d and 0.9.7e [25 Oct 2004] 7111 7112 *) Avoid a race condition when CRLs are checked in a multi threaded 7113 environment. This would happen due to the reordering of the revoked 7114 entries during signature checking and serial number lookup. Now the 7115 encoding is cached and the serial number sort performed under a lock. 7116 Add new STACK function sk_is_sorted(). 7117 [Steve Henson] 7118 7119 *) Add Delta CRL to the extension code. 7120 [Steve Henson] 7121 7122 *) Various fixes to s3_pkt.c so alerts are sent properly. 7123 [David Holmes <d.holmes@f5.com>] 7124 7125 *) Reduce the chances of duplicate issuer name and serial numbers (in 7126 violation of RFC3280) using the OpenSSL certificate creation utilities. 7127 This is done by creating a random 64 bit value for the initial serial 7128 number when a serial number file is created or when a self signed 7129 certificate is created using 'openssl req -x509'. The initial serial 7130 number file is created using 'openssl x509 -next_serial' in CA.pl 7131 rather than being initialized to 1. 7132 [Steve Henson] 7133 7134 Changes between 0.9.7c and 0.9.7d [17 Mar 2004] 7135 7136 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 7137 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 7138 [Joe Orton, Steve Henson] 7139 7140 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites 7141 (CVE-2004-0112) 7142 [Joe Orton, Steve Henson] 7143 7144 *) Make it possible to have multiple active certificates with the same 7145 subject in the CA index file. This is done only if the keyword 7146 'unique_subject' is set to 'no' in the main CA section (default 7147 if 'CA_default') of the configuration file. The value is saved 7148 with the database itself in a separate index attribute file, 7149 named like the index file with '.attr' appended to the name. 7150 [Richard Levitte] 7151 7152 *) X509 verify fixes. Disable broken certificate workarounds when 7153 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if 7154 keyUsage extension present. Don't accept CRLs with unhandled critical 7155 extensions: since verify currently doesn't process CRL extensions this 7156 rejects a CRL with *any* critical extensions. Add new verify error codes 7157 for these cases. 7158 [Steve Henson] 7159 7160 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue. 7161 A clarification of RFC2560 will require the use of OCTET STRINGs and 7162 some implementations cannot handle the current raw format. Since OpenSSL 7163 copies and compares OCSP nonces as opaque blobs without any attempt at 7164 parsing them this should not create any compatibility issues. 7165 [Steve Henson] 7166 7167 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when 7168 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without 7169 this HMAC (and other) operations are several times slower than OpenSSL 7170 < 0.9.7. 7171 [Steve Henson] 7172 7173 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex(). 7174 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 7175 7176 *) Use the correct content when signing type "other". 7177 [Steve Henson] 7178 7179 Changes between 0.9.7b and 0.9.7c [30 Sep 2003] 7180 7181 *) Fix various bugs revealed by running the NISCC test suite: 7182 7183 Stop out of bounds reads in the ASN1 code when presented with 7184 invalid tags (CVE-2003-0543 and CVE-2003-0544). 7185 7186 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545). 7187 7188 If verify callback ignores invalid public key errors don't try to check 7189 certificate signature with the NULL public key. 7190 7191 [Steve Henson] 7192 7193 *) New -ignore_err option in ocsp application to stop the server 7194 exiting on the first error in a request. 7195 [Steve Henson] 7196 7197 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 7198 if the server requested one: as stated in TLS 1.0 and SSL 3.0 7199 specifications. 7200 [Steve Henson] 7201 7202 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 7203 extra data after the compression methods not only for TLS 1.0 7204 but also for SSL 3.0 (as required by the specification). 7205 [Bodo Moeller; problem pointed out by Matthias Loepfe] 7206 7207 *) Change X509_certificate_type() to mark the key as exported/exportable 7208 when it's 512 *bits* long, not 512 bytes. 7209 [Richard Levitte] 7210 7211 *) Change AES_cbc_encrypt() so it outputs exact multiple of 7212 blocks during encryption. 7213 [Richard Levitte] 7214 7215 *) Various fixes to base64 BIO and non blocking I/O. On write 7216 flushes were not handled properly if the BIO retried. On read 7217 data was not being buffered properly and had various logic bugs. 7218 This also affects blocking I/O when the data being decoded is a 7219 certain size. 7220 [Steve Henson] 7221 7222 *) Various S/MIME bugfixes and compatibility changes: 7223 output correct application/pkcs7 MIME type if 7224 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures. 7225 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening 7226 of files as .eml work). Correctly handle very long lines in MIME 7227 parser. 7228 [Steve Henson] 7229 7230 Changes between 0.9.7a and 0.9.7b [10 Apr 2003] 7231 7232 *) Countermeasure against the Klima-Pokorny-Rosa extension of 7233 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 7234 a protocol version number mismatch like a decryption error 7235 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 7236 [Bodo Moeller] 7237 7238 *) Turn on RSA blinding by default in the default implementation 7239 to avoid a timing attack. Applications that don't want it can call 7240 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 7241 They would be ill-advised to do so in most cases. 7242 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 7243 7244 *) Change RSA blinding code so that it works when the PRNG is not 7245 seeded (in this case, the secret RSA exponent is abused as 7246 an unpredictable seed -- if it is not unpredictable, there 7247 is no point in blinding anyway). Make RSA blinding thread-safe 7248 by remembering the creator's thread ID in rsa->blinding and 7249 having all other threads use local one-time blinding factors 7250 (this requires more computation than sharing rsa->blinding, but 7251 avoids excessive locking; and if an RSA object is not shared 7252 between threads, blinding will still be very fast). 7253 [Bodo Moeller] 7254 7255 *) Fixed a typo bug that would cause ENGINE_set_default() to set an 7256 ENGINE as defaults for all supported algorithms irrespective of 7257 the 'flags' parameter. 'flags' is now honoured, so applications 7258 should make sure they are passing it correctly. 7259 [Geoff Thorpe] 7260 7261 *) Target "mingw" now allows native Windows code to be generated in 7262 the Cygwin environment as well as with the MinGW compiler. 7263 [Ulf Moeller] 7264 7265 Changes between 0.9.7 and 0.9.7a [19 Feb 2003] 7266 7267 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 7268 via timing by performing a MAC computation even if incorrect 7269 block cipher padding has been found. This is a countermeasure 7270 against active attacks where the attacker has to distinguish 7271 between bad padding and a MAC verification error. (CVE-2003-0078) 7272 7273 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 7274 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 7275 Martin Vuagnoux (EPFL, Ilion)] 7276 7277 *) Make the no-err option work as intended. The intention with no-err 7278 is not to have the whole error stack handling routines removed from 7279 libcrypto, it's only intended to remove all the function name and 7280 reason texts, thereby removing some of the footprint that may not 7281 be interesting if those errors aren't displayed anyway. 7282 7283 NOTE: it's still possible for any application or module to have its 7284 own set of error texts inserted. The routines are there, just not 7285 used by default when no-err is given. 7286 [Richard Levitte] 7287 7288 *) Add support for FreeBSD on IA64. 7289 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454] 7290 7291 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT 7292 Kerberos function mit_des_cbc_cksum(). Before this change, 7293 the value returned by DES_cbc_cksum() was like the one from 7294 mit_des_cbc_cksum(), except the bytes were swapped. 7295 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte] 7296 7297 *) Allow an application to disable the automatic SSL chain building. 7298 Before this a rather primitive chain build was always performed in 7299 ssl3_output_cert_chain(): an application had no way to send the 7300 correct chain if the automatic operation produced an incorrect result. 7301 7302 Now the chain builder is disabled if either: 7303 7304 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert(). 7305 7306 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set. 7307 7308 The reasoning behind this is that an application would not want the 7309 auto chain building to take place if extra chain certificates are 7310 present and it might also want a means of sending no additional 7311 certificates (for example the chain has two certificates and the 7312 root is omitted). 7313 [Steve Henson] 7314 7315 *) Add the possibility to build without the ENGINE framework. 7316 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 7317 7318 *) Under Win32 gmtime() can return NULL: check return value in 7319 OPENSSL_gmtime(). Add error code for case where gmtime() fails. 7320 [Steve Henson] 7321 7322 *) DSA routines: under certain error conditions uninitialized BN objects 7323 could be freed. Solution: make sure initialization is performed early 7324 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>, 7325 Nils Larsch <nla@trustcenter.de> via PR#459) 7326 [Lutz Jaenicke] 7327 7328 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly 7329 checked on reconnect on the client side, therefore session resumption 7330 could still fail with a "ssl session id is different" error. This 7331 behaviour is masked when SSL_OP_ALL is used due to 7332 SSL_OP_MICROSOFT_SESS_ID_BUG being set. 7333 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 7334 followup to PR #377. 7335 [Lutz Jaenicke] 7336 7337 *) IA-32 assembler support enhancements: unified ELF targets, support 7338 for SCO/Caldera platforms, fix for Cygwin shared build. 7339 [Andy Polyakov] 7340 7341 *) Add support for FreeBSD on sparc64. As a consequence, support for 7342 FreeBSD on non-x86 processors is separate from x86 processors on 7343 the config script, much like the NetBSD support. 7344 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>] 7345 7346 Changes between 0.9.6h and 0.9.7 [31 Dec 2002] 7347 7348 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after 7349 OpenSSL 0.9.7.] 7350 7351 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED 7352 code (06) was taken as the first octet of the session ID and the last 7353 octet was ignored consequently. As a result SSLv2 client side session 7354 caching could not have worked due to the session ID mismatch between 7355 client and server. 7356 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as 7357 PR #377. 7358 [Lutz Jaenicke] 7359 7360 *) Change the declaration of needed Kerberos libraries to use EX_LIBS 7361 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is 7362 removed entirely. 7363 [Richard Levitte] 7364 7365 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it 7366 seems that in spite of existing for more than a year, many application 7367 author have done nothing to provide the necessary callbacks, which 7368 means that this particular engine will not work properly anywhere. 7369 This is a very unfortunate situation which forces us, in the name 7370 of usability, to give the hw_ncipher.c a static lock, which is part 7371 of libcrypto. 7372 NOTE: This is for the 0.9.7 series ONLY. This hack will never 7373 appear in 0.9.8 or later. We EXPECT application authors to have 7374 dealt properly with this when 0.9.8 is released (unless we actually 7375 make such changes in the libcrypto locking code that changes will 7376 have to be made anyway). 7377 [Richard Levitte] 7378 7379 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content 7380 octets have been read, EOF or an error occurs. Without this change 7381 some truncated ASN1 structures will not produce an error. 7382 [Steve Henson] 7383 7384 *) Disable Heimdal support, since it hasn't been fully implemented. 7385 Still give the possibility to force the use of Heimdal, but with 7386 warnings and a request that patches get sent to openssl-dev. 7387 [Richard Levitte] 7388 7389 *) Add the VC-CE target, introduce the WINCE sysname, and add 7390 INSTALL.WCE and appropriate conditionals to make it build. 7391 [Steven Reddie <smr@essemer.com.au> via Richard Levitte] 7392 7393 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and 7394 cygssl-x.y.z.dll, where x, y and z are the major, minor and 7395 edit numbers of the version. 7396 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 7397 7398 *) Introduce safe string copy and catenation functions 7399 (BUF_strlcpy() and BUF_strlcat()). 7400 [Ben Laurie (CHATS) and Richard Levitte] 7401 7402 *) Avoid using fixed-size buffers for one-line DNs. 7403 [Ben Laurie (CHATS)] 7404 7405 *) Add BUF_MEM_grow_clean() to avoid information leakage when 7406 resizing buffers containing secrets, and use where appropriate. 7407 [Ben Laurie (CHATS)] 7408 7409 *) Avoid using fixed size buffers for configuration file location. 7410 [Ben Laurie (CHATS)] 7411 7412 *) Avoid filename truncation for various CA files. 7413 [Ben Laurie (CHATS)] 7414 7415 *) Use sizeof in preference to magic numbers. 7416 [Ben Laurie (CHATS)] 7417 7418 *) Avoid filename truncation in cert requests. 7419 [Ben Laurie (CHATS)] 7420 7421 *) Add assertions to check for (supposedly impossible) buffer 7422 overflows. 7423 [Ben Laurie (CHATS)] 7424 7425 *) Don't cache truncated DNS entries in the local cache (this could 7426 potentially lead to a spoofing attack). 7427 [Ben Laurie (CHATS)] 7428 7429 *) Fix various buffers to be large enough for hex/decimal 7430 representations in a platform independent manner. 7431 [Ben Laurie (CHATS)] 7432 7433 *) Add CRYPTO_realloc_clean() to avoid information leakage when 7434 resizing buffers containing secrets, and use where appropriate. 7435 [Ben Laurie (CHATS)] 7436 7437 *) Add BIO_indent() to avoid much slightly worrying code to do 7438 indents. 7439 [Ben Laurie (CHATS)] 7440 7441 *) Convert sprintf()/BIO_puts() to BIO_printf(). 7442 [Ben Laurie (CHATS)] 7443 7444 *) buffer_gets() could terminate with the buffer only half 7445 full. Fixed. 7446 [Ben Laurie (CHATS)] 7447 7448 *) Add assertions to prevent user-supplied crypto functions from 7449 overflowing internal buffers by having large block sizes, etc. 7450 [Ben Laurie (CHATS)] 7451 7452 *) New OPENSSL_assert() macro (similar to assert(), but enabled 7453 unconditionally). 7454 [Ben Laurie (CHATS)] 7455 7456 *) Eliminate unused copy of key in RC4. 7457 [Ben Laurie (CHATS)] 7458 7459 *) Eliminate unused and incorrectly sized buffers for IV in pem.h. 7460 [Ben Laurie (CHATS)] 7461 7462 *) Fix off-by-one error in EGD path. 7463 [Ben Laurie (CHATS)] 7464 7465 *) If RANDFILE path is too long, ignore instead of truncating. 7466 [Ben Laurie (CHATS)] 7467 7468 *) Eliminate unused and incorrectly sized X.509 structure 7469 CBCParameter. 7470 [Ben Laurie (CHATS)] 7471 7472 *) Eliminate unused and dangerous function knumber(). 7473 [Ben Laurie (CHATS)] 7474 7475 *) Eliminate unused and dangerous structure, KSSL_ERR. 7476 [Ben Laurie (CHATS)] 7477 7478 *) Protect against overlong session ID context length in an encoded 7479 session object. Since these are local, this does not appear to be 7480 exploitable. 7481 [Ben Laurie (CHATS)] 7482 7483 *) Change from security patch (see 0.9.6e below) that did not affect 7484 the 0.9.6 release series: 7485 7486 Remote buffer overflow in SSL3 protocol - an attacker could 7487 supply an oversized master key in Kerberos-enabled versions. 7488 (CVE-2002-0657) 7489 [Ben Laurie (CHATS)] 7490 7491 *) Change the SSL kerb5 codes to match RFC 2712. 7492 [Richard Levitte] 7493 7494 *) Make -nameopt work fully for req and add -reqopt switch. 7495 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson] 7496 7497 *) The "block size" for block ciphers in CFB and OFB mode should be 1. 7498 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>] 7499 7500 *) Make sure tests can be performed even if the corresponding algorithms 7501 have been removed entirely. This was also the last step to make 7502 OpenSSL compilable with DJGPP under all reasonable conditions. 7503 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>] 7504 7505 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT 7506 to allow version independent disabling of normally unselected ciphers, 7507 which may be activated as a side-effect of selecting a single cipher. 7508 7509 (E.g., cipher list string "RSA" enables ciphersuites that are left 7510 out of "ALL" because they do not provide symmetric encryption. 7511 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.) 7512 [Lutz Jaenicke, Bodo Moeller] 7513 7514 *) Add appropriate support for separate platform-dependent build 7515 directories. The recommended way to make a platform-dependent 7516 build directory is the following (tested on Linux), maybe with 7517 some local tweaks: 7518 7519 # Place yourself outside of the OpenSSL source tree. In 7520 # this example, the environment variable OPENSSL_SOURCE 7521 # is assumed to contain the absolute OpenSSL source directory. 7522 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`" 7523 cd objtree/"`uname -s`-`uname -r`-`uname -m`" 7524 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do 7525 mkdir -p `dirname $F` 7526 ln -s $OPENSSL_SOURCE/$F $F 7527 done 7528 7529 To be absolutely sure not to disturb the source tree, a "make clean" 7530 is a good thing. If it isn't successful, don't worry about it, 7531 it probably means the source directory is very clean. 7532 [Richard Levitte] 7533 7534 *) Make sure any ENGINE control commands make local copies of string 7535 pointers passed to them whenever necessary. Otherwise it is possible 7536 the caller may have overwritten (or deallocated) the original string 7537 data when a later ENGINE operation tries to use the stored values. 7538 [Götz Babin-Ebell <babinebell@trustcenter.de>] 7539 7540 *) Improve diagnostics in file reading and command-line digests. 7541 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>] 7542 7543 *) Add AES modes CFB and OFB to the object database. Correct an 7544 error in AES-CFB decryption. 7545 [Richard Levitte] 7546 7547 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 7548 allows existing EVP_CIPHER_CTX structures to be reused after 7549 calling EVP_*Final(). This behaviour is used by encryption 7550 BIOs and some applications. This has the side effect that 7551 applications must explicitly clean up cipher contexts with 7552 EVP_CIPHER_CTX_cleanup() or they will leak memory. 7553 [Steve Henson] 7554 7555 *) Check the values of dna and dnb in bn_mul_recursive before calling 7556 bn_mul_comba (a non zero value means the a or b arrays do not contain 7557 n2 elements) and fallback to bn_mul_normal if either is not zero. 7558 [Steve Henson] 7559 7560 *) Fix escaping of non-ASCII characters when using the -subj option 7561 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>) 7562 [Lutz Jaenicke] 7563 7564 *) Make object definitions compliant to LDAP (RFC2256): SN is the short 7565 form for "surname", serialNumber has no short form. 7566 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798; 7567 therefore remove "mail" short name for "internet 7". 7568 The OID for unique identifiers in X509 certificates is 7569 x500UniqueIdentifier, not uniqueIdentifier. 7570 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>) 7571 [Lutz Jaenicke] 7572 7573 *) Add an "init" command to the ENGINE config module and auto initialize 7574 ENGINEs. Without any "init" command the ENGINE will be initialized 7575 after all ctrl commands have been executed on it. If init=1 the 7576 ENGINE is initialized at that point (ctrls before that point are run 7577 on the uninitialized ENGINE and after on the initialized one). If 7578 init=0 then the ENGINE will not be initialized at all. 7579 [Steve Henson] 7580 7581 *) Fix the 'app_verify_callback' interface so that the user-defined 7582 argument is actually passed to the callback: In the 7583 SSL_CTX_set_cert_verify_callback() prototype, the callback 7584 declaration has been changed from 7585 int (*cb)() 7586 into 7587 int (*cb)(X509_STORE_CTX *,void *); 7588 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call 7589 i=s->ctx->app_verify_callback(&ctx) 7590 has been changed into 7591 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg). 7592 7593 To update applications using SSL_CTX_set_cert_verify_callback(), 7594 a dummy argument can be added to their callback functions. 7595 [D. K. Smetters <smetters@parc.xerox.com>] 7596 7597 *) Added the '4758cca' ENGINE to support IBM 4758 cards. 7598 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe] 7599 7600 *) Add and OPENSSL_LOAD_CONF define which will cause 7601 OpenSSL_add_all_algorithms() to load the openssl.cnf config file. 7602 This allows older applications to transparently support certain 7603 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading. 7604 Two new functions OPENSSL_add_all_algorithms_noconf() which will never 7605 load the config file and OPENSSL_add_all_algorithms_conf() which will 7606 always load it have also been added. 7607 [Steve Henson] 7608 7609 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES. 7610 Adjust NIDs and EVP layer. 7611 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 7612 7613 *) Config modules support in openssl utility. 7614 7615 Most commands now load modules from the config file, 7616 though in a few (such as version) this isn't done 7617 because it couldn't be used for anything. 7618 7619 In the case of ca and req the config file used is 7620 the same as the utility itself: that is the -config 7621 command line option can be used to specify an 7622 alternative file. 7623 [Steve Henson] 7624 7625 *) Move default behaviour from OPENSSL_config(). If appname is NULL 7626 use "openssl_conf" if filename is NULL use default openssl config file. 7627 [Steve Henson] 7628 7629 *) Add an argument to OPENSSL_config() to allow the use of an alternative 7630 config section name. Add a new flag to tolerate a missing config file 7631 and move code to CONF_modules_load_file(). 7632 [Steve Henson] 7633 7634 *) Support for crypto accelerator cards from Accelerated Encryption 7635 Processing, www.aep.ie. (Use engine 'aep') 7636 The support was copied from 0.9.6c [engine] and adapted/corrected 7637 to work with the new engine framework. 7638 [AEP Inc. and Richard Levitte] 7639 7640 *) Support for SureWare crypto accelerator cards from Baltimore 7641 Technologies. (Use engine 'sureware') 7642 The support was copied from 0.9.6c [engine] and adapted 7643 to work with the new engine framework. 7644 [Richard Levitte] 7645 7646 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually 7647 make the newer ENGINE framework commands for the CHIL engine work. 7648 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte] 7649 7650 *) Make it possible to produce shared libraries on ReliantUNIX. 7651 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte] 7652 7653 *) Add the configuration target debug-linux-ppro. 7654 Make 'openssl rsa' use the general key loading routines 7655 implemented in apps.c, and make those routines able to 7656 handle the key format FORMAT_NETSCAPE and the variant 7657 FORMAT_IISSGC. 7658 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7659 7660 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 7661 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 7662 7663 *) Add -keyform to rsautl, and document -engine. 7664 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>] 7665 7666 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new 7667 BIO_R_NO_SUCH_FILE error code rather than the generic 7668 ERR_R_SYS_LIB error code if fopen() fails with ENOENT. 7669 [Ben Laurie] 7670 7671 *) Add new functions 7672 ERR_peek_last_error 7673 ERR_peek_last_error_line 7674 ERR_peek_last_error_line_data. 7675 These are similar to 7676 ERR_peek_error 7677 ERR_peek_error_line 7678 ERR_peek_error_line_data, 7679 but report on the latest error recorded rather than the first one 7680 still in the error queue. 7681 [Ben Laurie, Bodo Moeller] 7682 7683 *) default_algorithms option in ENGINE config module. This allows things 7684 like: 7685 default_algorithms = ALL 7686 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS 7687 [Steve Henson] 7688 7689 *) Preliminary ENGINE config module. 7690 [Steve Henson] 7691 7692 *) New experimental application configuration code. 7693 [Steve Henson] 7694 7695 *) Change the AES code to follow the same name structure as all other 7696 symmetric ciphers, and behave the same way. Move everything to 7697 the directory crypto/aes, thereby obsoleting crypto/rijndael. 7698 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte] 7699 7700 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c. 7701 [Ben Laurie and Theo de Raadt] 7702 7703 *) Add option to output public keys in req command. 7704 [Massimiliano Pala madwolf@openca.org] 7705 7706 *) Use wNAFs in EC_POINTs_mul() for improved efficiency 7707 (up to about 10% better than before for P-192 and P-224). 7708 [Bodo Moeller] 7709 7710 *) New functions/macros 7711 7712 SSL_CTX_set_msg_callback(ctx, cb) 7713 SSL_CTX_set_msg_callback_arg(ctx, arg) 7714 SSL_set_msg_callback(ssl, cb) 7715 SSL_set_msg_callback_arg(ssl, arg) 7716 7717 to request calling a callback function 7718 7719 void cb(int write_p, int version, int content_type, 7720 const void *buf, size_t len, SSL *ssl, void *arg) 7721 7722 whenever a protocol message has been completely received 7723 (write_p == 0) or sent (write_p == 1). Here 'version' is the 7724 protocol version according to which the SSL library interprets 7725 the current protocol message (SSL2_VERSION, SSL3_VERSION, or 7726 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or 7727 the content type as defined in the SSL 3.0/TLS 1.0 protocol 7728 specification (change_cipher_spec(20), alert(21), handshake(22)). 7729 'buf' and 'len' point to the actual message, 'ssl' to the 7730 SSL object, and 'arg' is the application-defined value set by 7731 SSL[_CTX]_set_msg_callback_arg(). 7732 7733 'openssl s_client' and 'openssl s_server' have new '-msg' options 7734 to enable a callback that displays all protocol messages. 7735 [Bodo Moeller] 7736 7737 *) Change the shared library support so shared libraries are built as 7738 soon as the corresponding static library is finished, and thereby get 7739 openssl and the test programs linked against the shared library. 7740 This still only happens when the keyword "shard" has been given to 7741 the configuration scripts. 7742 7743 NOTE: shared library support is still an experimental thing, and 7744 backward binary compatibility is still not guaranteed. 7745 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte] 7746 7747 *) Add support for Subject Information Access extension. 7748 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>] 7749 7750 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero 7751 additional bytes when new memory had to be allocated, not just 7752 when reusing an existing buffer. 7753 [Bodo Moeller] 7754 7755 *) New command line and configuration option 'utf8' for the req command. 7756 This allows field values to be specified as UTF8 strings. 7757 [Steve Henson] 7758 7759 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel 7760 runs for the former and machine-readable output for the latter. 7761 [Ben Laurie] 7762 7763 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion 7764 of the e-mail address in the DN (i.e., it will go into a certificate 7765 extension only). The new configuration file option 'email_in_dn = no' 7766 has the same effect. 7767 [Massimiliano Pala madwolf@openca.org] 7768 7769 *) Change all functions with names starting with des_ to be starting 7770 with DES_ instead. Add wrappers that are compatible with libdes, 7771 but are named _ossl_old_des_*. Finally, add macros that map the 7772 des_* symbols to the corresponding _ossl_old_des_* if libdes 7773 compatibility is desired. If OpenSSL 0.9.6c compatibility is 7774 desired, the des_* symbols will be mapped to DES_*, with one 7775 exception. 7776 7777 Since we provide two compatibility mappings, the user needs to 7778 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes 7779 compatibility is desired. The default (i.e., when that macro 7780 isn't defined) is OpenSSL 0.9.6c compatibility. 7781 7782 There are also macros that enable and disable the support of old 7783 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT 7784 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those 7785 are defined, the default will apply: to support the old des routines. 7786 7787 In either case, one must include openssl/des.h to get the correct 7788 definitions. Do not try to just include openssl/des_old.h, that 7789 won't work. 7790 7791 NOTE: This is a major break of an old API into a new one. Software 7792 authors are encouraged to switch to the DES_ style functions. Some 7793 time in the future, des_old.h and the libdes compatibility functions 7794 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the 7795 default), and then completely removed. 7796 [Richard Levitte] 7797 7798 *) Test for certificates which contain unsupported critical extensions. 7799 If such a certificate is found during a verify operation it is 7800 rejected by default: this behaviour can be overridden by either 7801 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or 7802 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function 7803 X509_supported_extension() has also been added which returns 1 if a 7804 particular extension is supported. 7805 [Steve Henson] 7806 7807 *) Modify the behaviour of EVP cipher functions in similar way to digests 7808 to retain compatibility with existing code. 7809 [Steve Henson] 7810 7811 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain 7812 compatibility with existing code. In particular the 'ctx' parameter does 7813 not have to be to be initialized before the call to EVP_DigestInit() and 7814 it is tidied up after a call to EVP_DigestFinal(). New function 7815 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function 7816 EVP_MD_CTX_copy() changed to not require the destination to be 7817 initialized valid and new function EVP_MD_CTX_copy_ex() added which 7818 requires the destination to be valid. 7819 7820 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(), 7821 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex(). 7822 [Steve Henson] 7823 7824 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it 7825 so that complete 'Handshake' protocol structures are kept in memory 7826 instead of overwriting 'msg_type' and 'length' with 'body' data. 7827 [Bodo Moeller] 7828 7829 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32. 7830 [Massimo Santin via Richard Levitte] 7831 7832 *) Major restructuring to the underlying ENGINE code. This includes 7833 reduction of linker bloat, separation of pure "ENGINE" manipulation 7834 (initialisation, etc) from functionality dealing with implementations 7835 of specific crypto interfaces. This change also introduces integrated 7836 support for symmetric ciphers and digest implementations - so ENGINEs 7837 can now accelerate these by providing EVP_CIPHER and EVP_MD 7838 implementations of their own. This is detailed in crypto/engine/README 7839 as it couldn't be adequately described here. However, there are a few 7840 API changes worth noting - some RSA, DSA, DH, and RAND functions that 7841 were changed in the original introduction of ENGINE code have now 7842 reverted back - the hooking from this code to ENGINE is now a good 7843 deal more passive and at run-time, operations deal directly with 7844 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than 7845 dereferencing through an ENGINE pointer any more. Also, the ENGINE 7846 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed - 7847 they were not being used by the framework as there is no concept of a 7848 BIGNUM_METHOD and they could not be generalised to the new 7849 'ENGINE_TABLE' mechanism that underlies the new code. Similarly, 7850 ENGINE_cpy() has been removed as it cannot be consistently defined in 7851 the new code. 7852 [Geoff Thorpe] 7853 7854 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds. 7855 [Steve Henson] 7856 7857 *) Change mkdef.pl to sort symbols that get the same entry number, 7858 and make sure the automatically generated functions ERR_load_* 7859 become part of libeay.num as well. 7860 [Richard Levitte] 7861 7862 *) New function SSL_renegotiate_pending(). This returns true once 7863 renegotiation has been requested (either SSL_renegotiate() call 7864 or HelloRequest/ClientHello received from the peer) and becomes 7865 false once a handshake has been completed. 7866 (For servers, SSL_renegotiate() followed by SSL_do_handshake() 7867 sends a HelloRequest, but does not ensure that a handshake takes 7868 place. SSL_renegotiate_pending() is useful for checking if the 7869 client has followed the request.) 7870 [Bodo Moeller] 7871 7872 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION. 7873 By default, clients may request session resumption even during 7874 renegotiation (if session ID contexts permit); with this option, 7875 session resumption is possible only in the first handshake. 7876 7877 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes 7878 more bits available for options that should not be part of 7879 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION). 7880 [Bodo Moeller] 7881 7882 *) Add some demos for certificate and certificate request creation. 7883 [Steve Henson] 7884 7885 *) Make maximum certificate chain size accepted from the peer application 7886 settable (SSL*_get/set_max_cert_list()), as proposed by 7887 "Douglas E. Engert" <deengert@anl.gov>. 7888 [Lutz Jaenicke] 7889 7890 *) Add support for shared libraries for Unixware-7 7891 (Boyd Lynn Gerber <gerberb@zenez.com>). 7892 [Lutz Jaenicke] 7893 7894 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to 7895 be done prior to destruction. Use this to unload error strings from 7896 ENGINEs that load their own error strings. NB: This adds two new API 7897 functions to "get" and "set" this destroy handler in an ENGINE. 7898 [Geoff Thorpe] 7899 7900 *) Alter all existing ENGINE implementations (except "openssl" and 7901 "openbsd") to dynamically instantiate their own error strings. This 7902 makes them more flexible to be built both as statically-linked ENGINEs 7903 and self-contained shared-libraries loadable via the "dynamic" ENGINE. 7904 Also, add stub code to each that makes building them as self-contained 7905 shared-libraries easier (see README.ENGINE). 7906 [Geoff Thorpe] 7907 7908 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE 7909 implementations into applications that are completely implemented in 7910 self-contained shared-libraries. The "dynamic" ENGINE exposes control 7911 commands that can be used to configure what shared-library to load and 7912 to control aspects of the way it is handled. Also, made an update to 7913 the README.ENGINE file that brings its information up-to-date and 7914 provides some information and instructions on the "dynamic" ENGINE 7915 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc). 7916 [Geoff Thorpe] 7917 7918 *) Make it possible to unload ranges of ERR strings with a new 7919 "ERR_unload_strings" function. 7920 [Geoff Thorpe] 7921 7922 *) Add a copy() function to EVP_MD. 7923 [Ben Laurie] 7924 7925 *) Make EVP_MD routines take a context pointer instead of just the 7926 md_data void pointer. 7927 [Ben Laurie] 7928 7929 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates 7930 that the digest can only process a single chunk of data 7931 (typically because it is provided by a piece of 7932 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application 7933 is only going to provide a single chunk of data, and hence the 7934 framework needn't accumulate the data for oneshot drivers. 7935 [Ben Laurie] 7936 7937 *) As with "ERR", make it possible to replace the underlying "ex_data" 7938 functions. This change also alters the storage and management of global 7939 ex_data state - it's now all inside ex_data.c and all "class" code (eg. 7940 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class 7941 index counters. The API functions that use this state have been changed 7942 to take a "class_index" rather than pointers to the class's local STACK 7943 and counter, and there is now an API function to dynamically create new 7944 classes. This centralisation allows us to (a) plug a lot of the 7945 thread-safety problems that existed, and (b) makes it possible to clean 7946 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b) 7947 such data would previously have always leaked in application code and 7948 workarounds were in place to make the memory debugging turn a blind eye 7949 to it. Application code that doesn't use this new function will still 7950 leak as before, but their memory debugging output will announce it now 7951 rather than letting it slide. 7952 7953 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change 7954 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now 7955 has a return value to indicate success or failure. 7956 [Geoff Thorpe] 7957 7958 *) Make it possible to replace the underlying "ERR" functions such that the 7959 global state (2 LHASH tables and 2 locks) is only used by the "default" 7960 implementation. This change also adds two functions to "get" and "set" 7961 the implementation prior to it being automatically set the first time 7962 any other ERR function takes place. Ie. an application can call "get", 7963 pass the return value to a module it has just loaded, and that module 7964 can call its own "set" function using that value. This means the 7965 module's "ERR" operations will use (and modify) the error state in the 7966 application and not in its own statically linked copy of OpenSSL code. 7967 [Geoff Thorpe] 7968 7969 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment 7970 reference counts. This performs normal REF_PRINT/REF_CHECK macros on 7971 the operation, and provides a more encapsulated way for external code 7972 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code 7973 to use these functions rather than manually incrementing the counts. 7974 7975 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()". 7976 [Geoff Thorpe] 7977 7978 *) Add EVP test program. 7979 [Ben Laurie] 7980 7981 *) Add symmetric cipher support to ENGINE. Expect the API to change! 7982 [Ben Laurie] 7983 7984 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name() 7985 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(), 7986 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate(). 7987 These allow a CRL to be built without having to access X509_CRL fields 7988 directly. Modify 'ca' application to use new functions. 7989 [Steve Henson] 7990 7991 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended 7992 bug workarounds. Rollback attack detection is a security feature. 7993 The problem will only arise on OpenSSL servers when TLSv1 is not 7994 available (sslv3_server_method() or SSL_OP_NO_TLSv1). 7995 Software authors not wanting to support TLSv1 will have special reasons 7996 for their choice and can explicitly enable this option. 7997 [Bodo Moeller, Lutz Jaenicke] 7998 7999 *) Rationalise EVP so it can be extended: don't include a union of 8000 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX 8001 (similar to those existing for EVP_CIPHER_CTX). 8002 Usage example: 8003 8004 EVP_MD_CTX md; 8005 8006 EVP_MD_CTX_init(&md); /* new function call */ 8007 EVP_DigestInit(&md, EVP_sha1()); 8008 EVP_DigestUpdate(&md, in, len); 8009 EVP_DigestFinal(&md, out, NULL); 8010 EVP_MD_CTX_cleanup(&md); /* new function call */ 8011 8012 [Ben Laurie] 8013 8014 *) Make DES key schedule conform to the usual scheme, as well as 8015 correcting its structure. This means that calls to DES functions 8016 now have to pass a pointer to a des_key_schedule instead of a 8017 plain des_key_schedule (which was actually always a pointer 8018 anyway): E.g., 8019 8020 des_key_schedule ks; 8021 8022 des_set_key_checked(..., &ks); 8023 des_ncbc_encrypt(..., &ks, ...); 8024 8025 (Note that a later change renames 'des_...' into 'DES_...'.) 8026 [Ben Laurie] 8027 8028 *) Initial reduction of linker bloat: the use of some functions, such as 8029 PEM causes large amounts of unused functions to be linked in due to 8030 poor organisation. For example pem_all.c contains every PEM function 8031 which has a knock on effect of linking in large amounts of (unused) 8032 ASN1 code. Grouping together similar functions and splitting unrelated 8033 functions prevents this. 8034 [Steve Henson] 8035 8036 *) Cleanup of EVP macros. 8037 [Ben Laurie] 8038 8039 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the 8040 correct _ecb suffix. 8041 [Ben Laurie] 8042 8043 *) Add initial OCSP responder support to ocsp application. The 8044 revocation information is handled using the text based index 8045 use by the ca application. The responder can either handle 8046 requests generated internally, supplied in files (for example 8047 via a CGI script) or using an internal minimal server. 8048 [Steve Henson] 8049 8050 *) Add configuration choices to get zlib compression for TLS. 8051 [Richard Levitte] 8052 8053 *) Changes to Kerberos SSL for RFC 2712 compliance: 8054 1. Implemented real KerberosWrapper, instead of just using 8055 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>] 8056 2. Implemented optional authenticator field of KerberosWrapper. 8057 8058 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req, 8059 and authenticator structs; see crypto/krb5/. 8060 8061 Generalized Kerberos calls to support multiple Kerberos libraries. 8062 [Vern Staats <staatsvr@asc.hpc.mil>, 8063 Jeffrey Altman <jaltman@columbia.edu> 8064 via Richard Levitte] 8065 8066 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it 8067 already does with RSA. testdsa.h now has 'priv_key/pub_key' 8068 values for each of the key sizes rather than having just 8069 parameters (and 'speed' generating keys each time). 8070 [Geoff Thorpe] 8071 8072 *) Speed up EVP routines. 8073 Before: 8074encrypt 8075type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 8076des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k 8077des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k 8078des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k 8079decrypt 8080des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k 8081des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k 8082des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k 8083 After: 8084encrypt 8085des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k 8086decrypt 8087des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k 8088 [Ben Laurie] 8089 8090 *) Added the OS2-EMX target. 8091 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte] 8092 8093 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions 8094 to support NCONF routines in extension code. New function CONF_set_nconf() 8095 to allow functions which take an NCONF to also handle the old LHASH 8096 structure: this means that the old CONF compatible routines can be 8097 retained (in particular wrt extensions) without having to duplicate the 8098 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack. 8099 [Steve Henson] 8100 8101 *) Enhance the general user interface with mechanisms for inner control 8102 and with possibilities to have yes/no kind of prompts. 8103 [Richard Levitte] 8104 8105 *) Change all calls to low level digest routines in the library and 8106 applications to use EVP. Add missing calls to HMAC_cleanup() and 8107 don't assume HMAC_CTX can be copied using memcpy(). 8108 [Verdon Walker <VWalker@novell.com>, Steve Henson] 8109 8110 *) Add the possibility to control engines through control names but with 8111 arbitrary arguments instead of just a string. 8112 Change the key loaders to take a UI_METHOD instead of a callback 8113 function pointer. NOTE: this breaks binary compatibility with earlier 8114 versions of OpenSSL [engine]. 8115 Adapt the nCipher code for these new conditions and add a card insertion 8116 callback. 8117 [Richard Levitte] 8118 8119 *) Enhance the general user interface with mechanisms to better support 8120 dialog box interfaces, application-defined prompts, the possibility 8121 to use defaults (for example default passwords from somewhere else) 8122 and interrupts/cancellations. 8123 [Richard Levitte] 8124 8125 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name 8126 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility. 8127 [Steve Henson] 8128 8129 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also 8130 tidy up some unnecessarily weird code in 'sk_new()'). 8131 [Geoff, reported by Diego Tartara <dtartara@novamens.com>] 8132 8133 *) Change the key loading routines for ENGINEs to use the same kind 8134 callback (pem_password_cb) as all other routines that need this 8135 kind of callback. 8136 [Richard Levitte] 8137 8138 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with 8139 256 bit (=32 byte) keys. Of course seeding with more entropy bytes 8140 than this minimum value is recommended. 8141 [Lutz Jaenicke] 8142 8143 *) New random seeder for OpenVMS, using the system process statistics 8144 that are easily reachable. 8145 [Richard Levitte] 8146 8147 *) Windows apparently can't transparently handle global 8148 variables defined in DLLs. Initialisations such as: 8149 8150 const ASN1_ITEM *it = &ASN1_INTEGER_it; 8151 8152 won't compile. This is used by the any applications that need to 8153 declare their own ASN1 modules. This was fixed by adding the option 8154 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly 8155 needed for static libraries under Win32. 8156 [Steve Henson] 8157 8158 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle 8159 setting of purpose and trust fields. New X509_STORE trust and 8160 purpose functions and tidy up setting in other SSL functions. 8161 [Steve Henson] 8162 8163 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE 8164 structure. These are inherited by X509_STORE_CTX when it is 8165 initialised. This allows various defaults to be set in the 8166 X509_STORE structure (such as flags for CRL checking and custom 8167 purpose or trust settings) for functions which only use X509_STORE_CTX 8168 internally such as S/MIME. 8169 8170 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and 8171 trust settings if they are not set in X509_STORE. This allows X509_STORE 8172 purposes and trust (in S/MIME for example) to override any set by default. 8173 8174 Add command line options for CRL checking to smime, s_client and s_server 8175 applications. 8176 [Steve Henson] 8177 8178 *) Initial CRL based revocation checking. If the CRL checking flag(s) 8179 are set then the CRL is looked up in the X509_STORE structure and 8180 its validity and signature checked, then if the certificate is found 8181 in the CRL the verify fails with a revoked error. 8182 8183 Various new CRL related callbacks added to X509_STORE_CTX structure. 8184 8185 Command line options added to 'verify' application to support this. 8186 8187 This needs some additional work, such as being able to handle multiple 8188 CRLs with different times, extension based lookup (rather than just 8189 by subject name) and ultimately more complete V2 CRL extension 8190 handling. 8191 [Steve Henson] 8192 8193 *) Add a general user interface API (crypto/ui/). This is designed 8194 to replace things like des_read_password and friends (backward 8195 compatibility functions using this new API are provided). 8196 The purpose is to remove prompting functions from the DES code 8197 section as well as provide for prompting through dialog boxes in 8198 a window system and the like. 8199 [Richard Levitte] 8200 8201 *) Add "ex_data" support to ENGINE so implementations can add state at a 8202 per-structure level rather than having to store it globally. 8203 [Geoff] 8204 8205 *) Make it possible for ENGINE structures to be copied when retrieved by 8206 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY. 8207 This causes the "original" ENGINE structure to act like a template, 8208 analogous to the RSA vs. RSA_METHOD type of separation. Because of this 8209 operational state can be localised to each ENGINE structure, despite the 8210 fact they all share the same "methods". New ENGINE structures returned in 8211 this case have no functional references and the return value is the single 8212 structural reference. This matches the single structural reference returned 8213 by ENGINE_by_id() normally, when it is incremented on the pre-existing 8214 ENGINE structure. 8215 [Geoff] 8216 8217 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this 8218 needs to match any other type at all we need to manually clear the 8219 tag cache. 8220 [Steve Henson] 8221 8222 *) Changes to the "openssl engine" utility to include; 8223 - verbosity levels ('-v', '-vv', and '-vvv') that provide information 8224 about an ENGINE's available control commands. 8225 - executing control commands from command line arguments using the 8226 '-pre' and '-post' switches. '-post' is only used if '-t' is 8227 specified and the ENGINE is successfully initialised. The syntax for 8228 the individual commands are colon-separated, for example; 8229 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so 8230 [Geoff] 8231 8232 *) New dynamic control command support for ENGINEs. ENGINEs can now 8233 declare their own commands (numbers), names (strings), descriptions, 8234 and input types for run-time discovery by calling applications. A 8235 subset of these commands are implicitly classed as "executable" 8236 depending on their input type, and only these can be invoked through 8237 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this 8238 can be based on user input, config files, etc). The distinction is 8239 that "executable" commands cannot return anything other than a boolean 8240 result and can only support numeric or string input, whereas some 8241 discoverable commands may only be for direct use through 8242 ENGINE_ctrl(), eg. supporting the exchange of binary data, function 8243 pointers, or other custom uses. The "executable" commands are to 8244 support parameterisations of ENGINE behaviour that can be 8245 unambiguously defined by ENGINEs and used consistently across any 8246 OpenSSL-based application. Commands have been added to all the 8247 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow 8248 control over shared-library paths without source code alterations. 8249 [Geoff] 8250 8251 *) Changed all ENGINE implementations to dynamically allocate their 8252 ENGINEs rather than declaring them statically. Apart from this being 8253 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction, 8254 this also allows the implementations to compile without using the 8255 internal engine_int.h header. 8256 [Geoff] 8257 8258 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a 8259 'const' value. Any code that should be able to modify a RAND_METHOD 8260 should already have non-const pointers to it (ie. they should only 8261 modify their own ones). 8262 [Geoff] 8263 8264 *) Made a variety of little tweaks to the ENGINE code. 8265 - "atalla" and "ubsec" string definitions were moved from header files 8266 to C code. "nuron" string definitions were placed in variables 8267 rather than hard-coded - allowing parameterisation of these values 8268 later on via ctrl() commands. 8269 - Removed unused "#if 0"'d code. 8270 - Fixed engine list iteration code so it uses ENGINE_free() to release 8271 structural references. 8272 - Constified the RAND_METHOD element of ENGINE structures. 8273 - Constified various get/set functions as appropriate and added 8274 missing functions (including a catch-all ENGINE_cpy that duplicates 8275 all ENGINE values onto a new ENGINE except reference counts/state). 8276 - Removed NULL parameter checks in get/set functions. Setting a method 8277 or function to NULL is a way of cancelling out a previously set 8278 value. Passing a NULL ENGINE parameter is just plain stupid anyway 8279 and doesn't justify the extra error symbols and code. 8280 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for 8281 flags from engine_int.h to engine.h. 8282 - Changed prototypes for ENGINE handler functions (init(), finish(), 8283 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter. 8284 [Geoff] 8285 8286 *) Implement binary inversion algorithm for BN_mod_inverse in addition 8287 to the algorithm using long division. The binary algorithm can be 8288 used only if the modulus is odd. On 32-bit systems, it is faster 8289 only for relatively small moduli (roughly 20-30% for 128-bit moduli, 8290 roughly 5-15% for 256-bit moduli), so we use it only for moduli 8291 up to 450 bits. In 64-bit environments, the binary algorithm 8292 appears to be advantageous for much longer moduli; here we use it 8293 for moduli up to 2048 bits. 8294 [Bodo Moeller] 8295 8296 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code 8297 could not support the combine flag in choice fields. 8298 [Steve Henson] 8299 8300 *) Add a 'copy_extensions' option to the 'ca' utility. This copies 8301 extensions from a certificate request to the certificate. 8302 [Steve Henson] 8303 8304 *) Allow multiple 'certopt' and 'nameopt' options to be separated 8305 by commas. Add 'namopt' and 'certopt' options to the 'ca' config 8306 file: this allows the display of the certificate about to be 8307 signed to be customised, to allow certain fields to be included 8308 or excluded and extension details. The old system didn't display 8309 multicharacter strings properly, omitted fields not in the policy 8310 and couldn't display additional details such as extensions. 8311 [Steve Henson] 8312 8313 *) Function EC_POINTs_mul for multiple scalar multiplication 8314 of an arbitrary number of elliptic curve points 8315 \sum scalars[i]*points[i], 8316 optionally including the generator defined for the EC_GROUP: 8317 scalar*generator + \sum scalars[i]*points[i]. 8318 8319 EC_POINT_mul is a simple wrapper function for the typical case 8320 that the point list has just one item (besides the optional 8321 generator). 8322 [Bodo Moeller] 8323 8324 *) First EC_METHODs for curves over GF(p): 8325 8326 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr 8327 operations and provides various method functions that can also 8328 operate with faster implementations of modular arithmetic. 8329 8330 EC_GFp_mont_method() reuses most functions that are part of 8331 EC_GFp_simple_method, but uses Montgomery arithmetic. 8332 8333 [Bodo Moeller; point addition and point doubling 8334 implementation directly derived from source code provided by 8335 Lenka Fibikova <fibikova@exp-math.uni-essen.de>] 8336 8337 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h, 8338 crypto/ec/ec_lib.c): 8339 8340 Curves are EC_GROUP objects (with an optional group generator) 8341 based on EC_METHODs that are built into the library. 8342 8343 Points are EC_POINT objects based on EC_GROUP objects. 8344 8345 Most of the framework would be able to handle curves over arbitrary 8346 finite fields, but as there are no obvious types for fields other 8347 than GF(p), some functions are limited to that for now. 8348 [Bodo Moeller] 8349 8350 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires 8351 that the file contains a complete HTTP response. 8352 [Richard Levitte] 8353 8354 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl 8355 change the def and num file printf format specifier from "%-40sXXX" 8356 to "%-39s XXX". The latter will always guarantee a space after the 8357 field while the former will cause them to run together if the field 8358 is 40 of more characters long. 8359 [Steve Henson] 8360 8361 *) Constify the cipher and digest 'method' functions and structures 8362 and modify related functions to take constant EVP_MD and EVP_CIPHER 8363 pointers. 8364 [Steve Henson] 8365 8366 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them 8367 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32. 8368 [Bodo Moeller] 8369 8370 *) Modify EVP_Digest*() routines so they now return values. Although the 8371 internal software routines can never fail additional hardware versions 8372 might. 8373 [Steve Henson] 8374 8375 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts: 8376 8377 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7 8378 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32. 8379 8380 ASN1 error codes 8381 ERR_R_NESTED_ASN1_ERROR 8382 ... 8383 ERR_R_MISSING_ASN1_EOS 8384 were 4 .. 9, conflicting with 8385 ERR_LIB_RSA (= ERR_R_RSA_LIB) 8386 ... 8387 ERR_LIB_PEM (= ERR_R_PEM_LIB). 8388 They are now 58 .. 63 (i.e., just below ERR_R_FATAL). 8389 8390 Add new error code 'ERR_R_INTERNAL_ERROR'. 8391 [Bodo Moeller] 8392 8393 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock 8394 suffices. 8395 [Bodo Moeller] 8396 8397 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This 8398 sets the subject name for a new request or supersedes the 8399 subject name in a given request. Formats that can be parsed are 8400 'CN=Some Name, OU=myOU, C=IT' 8401 and 8402 'CN=Some Name/OU=myOU/C=IT'. 8403 8404 Add options '-batch' and '-verbose' to 'openssl req'. 8405 [Massimiliano Pala <madwolf@hackmasters.net>] 8406 8407 *) Introduce the possibility to access global variables through 8408 functions on platform were that's the best way to handle exporting 8409 global variables in shared libraries. To enable this functionality, 8410 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro 8411 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter 8412 is normally done by Configure or something similar). 8413 8414 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL 8415 in the source file (foo.c) like this: 8416 8417 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1; 8418 OPENSSL_IMPLEMENT_GLOBAL(double,bar); 8419 8420 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL 8421 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this: 8422 8423 OPENSSL_DECLARE_GLOBAL(int,foo); 8424 #define foo OPENSSL_GLOBAL_REF(foo) 8425 OPENSSL_DECLARE_GLOBAL(double,bar); 8426 #define bar OPENSSL_GLOBAL_REF(bar) 8427 8428 The #defines are very important, and therefore so is including the 8429 header file everywhere where the defined globals are used. 8430 8431 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition 8432 of ASN.1 items, but that structure is a bit different. 8433 8434 The largest change is in util/mkdef.pl which has been enhanced with 8435 better and easier to understand logic to choose which symbols should 8436 go into the Windows .def files as well as a number of fixes and code 8437 cleanup (among others, algorithm keywords are now sorted 8438 lexicographically to avoid constant rewrites). 8439 [Richard Levitte] 8440 8441 *) In BN_div() keep a copy of the sign of 'num' before writing the 8442 result to 'rm' because if rm==num the value will be overwritten 8443 and produce the wrong result if 'num' is negative: this caused 8444 problems with BN_mod() and BN_nnmod(). 8445 [Steve Henson] 8446 8447 *) Function OCSP_request_verify(). This checks the signature on an 8448 OCSP request and verifies the signer certificate. The signer 8449 certificate is just checked for a generic purpose and OCSP request 8450 trust settings. 8451 [Steve Henson] 8452 8453 *) Add OCSP_check_validity() function to check the validity of OCSP 8454 responses. OCSP responses are prepared in real time and may only 8455 be a few seconds old. Simply checking that the current time lies 8456 between thisUpdate and nextUpdate max reject otherwise valid responses 8457 caused by either OCSP responder or client clock inaccuracy. Instead 8458 we allow thisUpdate and nextUpdate to fall within a certain period of 8459 the current time. The age of the response can also optionally be 8460 checked. Two new options -validity_period and -status_age added to 8461 ocsp utility. 8462 [Steve Henson] 8463 8464 *) If signature or public key algorithm is unrecognized print out its 8465 OID rather that just UNKNOWN. 8466 [Steve Henson] 8467 8468 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and 8469 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate 8470 ID to be generated from the issuer certificate alone which can then be 8471 passed to OCSP_id_issuer_cmp(). 8472 [Steve Henson] 8473 8474 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new 8475 ASN1 modules to export functions returning ASN1_ITEM pointers 8476 instead of the ASN1_ITEM structures themselves. This adds several 8477 new macros which allow the underlying ASN1 function/structure to 8478 be accessed transparently. As a result code should not use ASN1_ITEM 8479 references directly (such as &X509_it) but instead use the relevant 8480 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow 8481 use of the new ASN1 code on platforms where exporting structures 8482 is problematical (for example in shared libraries) but exporting 8483 functions returning pointers to structures is not. 8484 [Steve Henson] 8485 8486 *) Add support for overriding the generation of SSL/TLS session IDs. 8487 These callbacks can be registered either in an SSL_CTX or per SSL. 8488 The purpose of this is to allow applications to control, if they wish, 8489 the arbitrary values chosen for use as session IDs, particularly as it 8490 can be useful for session caching in multiple-server environments. A 8491 command-line switch for testing this (and any client code that wishes 8492 to use such a feature) has been added to "s_server". 8493 [Geoff Thorpe, Lutz Jaenicke] 8494 8495 *) Modify mkdef.pl to recognise and parse preprocessor conditionals 8496 of the form '#if defined(...) || defined(...) || ...' and 8497 '#if !defined(...) && !defined(...) && ...'. This also avoids 8498 the growing number of special cases it was previously handling. 8499 [Richard Levitte] 8500 8501 *) Make all configuration macros available for application by making 8502 sure they are available in opensslconf.h, by giving them names starting 8503 with "OPENSSL_" to avoid conflicts with other packages and by making 8504 sure e_os2.h will cover all platform-specific cases together with 8505 opensslconf.h. 8506 Additionally, it is now possible to define configuration/platform- 8507 specific names (called "system identities"). In the C code, these 8508 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another 8509 macro with the name beginning with "OPENSSL_SYS_", which is determined 8510 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on 8511 what is available. 8512 [Richard Levitte] 8513 8514 *) New option -set_serial to 'req' and 'x509' this allows the serial 8515 number to use to be specified on the command line. Previously self 8516 signed certificates were hard coded with serial number 0 and the 8517 CA options of 'x509' had to use a serial number in a file which was 8518 auto incremented. 8519 [Steve Henson] 8520 8521 *) New options to 'ca' utility to support V2 CRL entry extensions. 8522 Currently CRL reason, invalidity date and hold instruction are 8523 supported. Add new CRL extensions to V3 code and some new objects. 8524 [Steve Henson] 8525 8526 *) New function EVP_CIPHER_CTX_set_padding() this is used to 8527 disable standard block padding (aka PKCS#5 padding) in the EVP 8528 API, which was previously mandatory. This means that the data is 8529 not padded in any way and so the total length much be a multiple 8530 of the block size, otherwise an error occurs. 8531 [Steve Henson] 8532 8533 *) Initial (incomplete) OCSP SSL support. 8534 [Steve Henson] 8535 8536 *) New function OCSP_parse_url(). This splits up a URL into its host, 8537 port and path components: primarily to parse OCSP URLs. New -url 8538 option to ocsp utility. 8539 [Steve Henson] 8540 8541 *) New nonce behavior. The return value of OCSP_check_nonce() now 8542 reflects the various checks performed. Applications can decide 8543 whether to tolerate certain situations such as an absent nonce 8544 in a response when one was present in a request: the ocsp application 8545 just prints out a warning. New function OCSP_add1_basic_nonce() 8546 this is to allow responders to include a nonce in a response even if 8547 the request is nonce-less. 8548 [Steve Henson] 8549 8550 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are 8551 skipped when using openssl x509 multiple times on a single input file, 8552 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs". 8553 [Bodo Moeller] 8554 8555 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string() 8556 set string type: to handle setting ASN1_TIME structures. Fix ca 8557 utility to correctly initialize revocation date of CRLs. 8558 [Steve Henson] 8559 8560 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override 8561 the clients preferred ciphersuites and rather use its own preferences. 8562 Should help to work around M$ SGC (Server Gated Cryptography) bug in 8563 Internet Explorer by ensuring unchanged hash method during stepup. 8564 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.) 8565 [Lutz Jaenicke] 8566 8567 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael 8568 to aes and add a new 'exist' option to print out symbols that don't 8569 appear to exist. 8570 [Steve Henson] 8571 8572 *) Additional options to ocsp utility to allow flags to be set and 8573 additional certificates supplied. 8574 [Steve Henson] 8575 8576 *) Add the option -VAfile to 'openssl ocsp', so the user can give the 8577 OCSP client a number of certificate to only verify the response 8578 signature against. 8579 [Richard Levitte] 8580 8581 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to 8582 handle the new API. Currently only ECB, CBC modes supported. Add new 8583 AES OIDs. 8584 8585 Add TLS AES ciphersuites as described in RFC3268, "Advanced 8586 Encryption Standard (AES) Ciphersuites for Transport Layer 8587 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were 8588 not enabled by default and were not part of the "ALL" ciphersuite 8589 alias because they were not yet official; they could be 8590 explicitly requested by specifying the "AESdraft" ciphersuite 8591 group alias. In the final release of OpenSSL 0.9.7, the group 8592 alias is called "AES" and is part of "ALL".) 8593 [Ben Laurie, Steve Henson, Bodo Moeller] 8594 8595 *) New function OCSP_copy_nonce() to copy nonce value (if present) from 8596 request to response. 8597 [Steve Henson] 8598 8599 *) Functions for OCSP responders. OCSP_request_onereq_count(), 8600 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info() 8601 extract information from a certificate request. OCSP_response_create() 8602 creates a response and optionally adds a basic response structure. 8603 OCSP_basic_add1_status() adds a complete single response to a basic 8604 response and returns the OCSP_SINGLERESP structure just added (to allow 8605 extensions to be included for example). OCSP_basic_add1_cert() adds a 8606 certificate to a basic response and OCSP_basic_sign() signs a basic 8607 response with various flags. New helper functions ASN1_TIME_check() 8608 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime() 8609 (converts ASN1_TIME to GeneralizedTime). 8610 [Steve Henson] 8611 8612 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}() 8613 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key 8614 structure from a certificate. X509_pubkey_digest() digests the public_key 8615 contents: this is used in various key identifiers. 8616 [Steve Henson] 8617 8618 *) Make sk_sort() tolerate a NULL argument. 8619 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>] 8620 8621 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates 8622 passed by the function are trusted implicitly. If any of them signed the 8623 response then it is assumed to be valid and is not verified. 8624 [Steve Henson] 8625 8626 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT 8627 to data. This was previously part of the PKCS7 ASN1 code. This 8628 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures. 8629 [Steve Henson, reported by Kenneth R. Robinette 8630 <support@securenetterm.com>] 8631 8632 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1 8633 routines: without these tracing memory leaks is very painful. 8634 Fix leaks in PKCS12 and PKCS7 routines. 8635 [Steve Henson] 8636 8637 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new(). 8638 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which 8639 effectively meant GeneralizedTime would never be used. Now it 8640 is initialised to -1 but X509_time_adj() now has to check the value 8641 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or 8642 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime. 8643 [Steve Henson, reported by Kenneth R. Robinette 8644 <support@securenetterm.com>] 8645 8646 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously 8647 result in a zero length in the ASN1_INTEGER structure which was 8648 not consistent with the structure when d2i_ASN1_INTEGER() was used 8649 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER() 8650 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER() 8651 where it did not print out a minus for negative ASN1_INTEGER. 8652 [Steve Henson] 8653 8654 *) Add summary printout to ocsp utility. The various functions which 8655 convert status values to strings have been renamed to: 8656 OCSP_response_status_str(), OCSP_cert_status_str() and 8657 OCSP_crl_reason_str() and are no longer static. New options 8658 to verify nonce values and to disable verification. OCSP response 8659 printout format cleaned up. 8660 [Steve Henson] 8661 8662 *) Add additional OCSP certificate checks. These are those specified 8663 in RFC2560. This consists of two separate checks: the CA of the 8664 certificate being checked must either be the OCSP signer certificate 8665 or the issuer of the OCSP signer certificate. In the latter case the 8666 OCSP signer certificate must contain the OCSP signing extended key 8667 usage. This check is performed by attempting to match the OCSP 8668 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash 8669 in the OCSP_CERTID structures of the response. 8670 [Steve Henson] 8671 8672 *) Initial OCSP certificate verification added to OCSP_basic_verify() 8673 and related routines. This uses the standard OpenSSL certificate 8674 verify routines to perform initial checks (just CA validity) and 8675 to obtain the certificate chain. Then additional checks will be 8676 performed on the chain. Currently the root CA is checked to see 8677 if it is explicitly trusted for OCSP signing. This is used to set 8678 a root CA as a global signing root: that is any certificate that 8679 chains to that CA is an acceptable OCSP signing certificate. 8680 [Steve Henson] 8681 8682 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3 8683 extensions from a separate configuration file. 8684 As when reading extensions from the main configuration file, 8685 the '-extensions ...' option may be used for specifying the 8686 section to use. 8687 [Massimiliano Pala <madwolf@comune.modena.it>] 8688 8689 *) New OCSP utility. Allows OCSP requests to be generated or 8690 read. The request can be sent to a responder and the output 8691 parsed, outputted or printed in text form. Not complete yet: 8692 still needs to check the OCSP response validity. 8693 [Steve Henson] 8694 8695 *) New subcommands for 'openssl ca': 8696 'openssl ca -status <serial>' prints the status of the cert with 8697 the given serial number (according to the index file). 8698 'openssl ca -updatedb' updates the expiry status of certificates 8699 in the index file. 8700 [Massimiliano Pala <madwolf@comune.modena.it>] 8701 8702 *) New '-newreq-nodes' command option to CA.pl. This is like 8703 '-newreq', but calls 'openssl req' with the '-nodes' option 8704 so that the resulting key is not encrypted. 8705 [Damien Miller <djm@mindrot.org>] 8706 8707 *) New configuration for the GNU Hurd. 8708 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte] 8709 8710 *) Initial code to implement OCSP basic response verify. This 8711 is currently incomplete. Currently just finds the signer's 8712 certificate and verifies the signature on the response. 8713 [Steve Henson] 8714 8715 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in 8716 value of OPENSSLDIR. This is available via the new '-d' option 8717 to 'openssl version', and is also included in 'openssl version -a'. 8718 [Bodo Moeller] 8719 8720 *) Allowing defining memory allocation callbacks that will be given 8721 file name and line number information in additional arguments 8722 (a const char* and an int). The basic functionality remains, as 8723 well as the original possibility to just replace malloc(), 8724 realloc() and free() by functions that do not know about these 8725 additional arguments. To register and find out the current 8726 settings for extended allocation functions, the following 8727 functions are provided: 8728 8729 CRYPTO_set_mem_ex_functions 8730 CRYPTO_set_locked_mem_ex_functions 8731 CRYPTO_get_mem_ex_functions 8732 CRYPTO_get_locked_mem_ex_functions 8733 8734 These work the same way as CRYPTO_set_mem_functions and friends. 8735 CRYPTO_get_[locked_]mem_functions now writes 0 where such an 8736 extended allocation function is enabled. 8737 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where 8738 a conventional allocation function is enabled. 8739 [Richard Levitte, Bodo Moeller] 8740 8741 *) Finish off removing the remaining LHASH function pointer casts. 8742 There should no longer be any prototype-casting required when using 8743 the LHASH abstraction, and any casts that remain are "bugs". See 8744 the callback types and macros at the head of lhash.h for details 8745 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example). 8746 [Geoff Thorpe] 8747 8748 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant. 8749 If /dev/[u]random devices are not available or do not return enough 8750 entropy, EGD style sockets (served by EGD or PRNGD) will automatically 8751 be queried. 8752 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and 8753 /etc/entropy will be queried once each in this sequence, querying stops 8754 when enough entropy was collected without querying more sockets. 8755 [Lutz Jaenicke] 8756 8757 *) Change the Unix RAND_poll() variant to be able to poll several 8758 random devices, as specified by DEVRANDOM, until a sufficient amount 8759 of data has been collected. We spend at most 10 ms on each file 8760 (select timeout) and read in non-blocking mode. DEVRANDOM now 8761 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom" 8762 (previously it was just the string "/dev/urandom"), so on typical 8763 platforms the 10 ms delay will never occur. 8764 Also separate out the Unix variant to its own file, rand_unix.c. 8765 For VMS, there's a currently-empty rand_vms.c. 8766 [Richard Levitte] 8767 8768 *) Move OCSP client related routines to ocsp_cl.c. These 8769 provide utility functions which an application needing 8770 to issue a request to an OCSP responder and analyse the 8771 response will typically need: as opposed to those which an 8772 OCSP responder itself would need which will be added later. 8773 8774 OCSP_request_sign() signs an OCSP request with an API similar 8775 to PKCS7_sign(). OCSP_response_status() returns status of OCSP 8776 response. OCSP_response_get1_basic() extracts basic response 8777 from response. OCSP_resp_find_status(): finds and extracts status 8778 information from an OCSP_CERTID structure (which will be created 8779 when the request structure is built). These are built from lower 8780 level functions which work on OCSP_SINGLERESP structures but 8781 won't normally be used unless the application wishes to examine 8782 extensions in the OCSP response for example. 8783 8784 Replace nonce routines with a pair of functions. 8785 OCSP_request_add1_nonce() adds a nonce value and optionally 8786 generates a random value. OCSP_check_nonce() checks the 8787 validity of the nonce in an OCSP response. 8788 [Steve Henson] 8789 8790 *) Change function OCSP_request_add() to OCSP_request_add0_id(). 8791 This doesn't copy the supplied OCSP_CERTID and avoids the 8792 need to free up the newly created id. Change return type 8793 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure. 8794 This can then be used to add extensions to the request. 8795 Deleted OCSP_request_new(), since most of its functionality 8796 is now in OCSP_REQUEST_new() (and the case insensitive name 8797 clash) apart from the ability to set the request name which 8798 will be added elsewhere. 8799 [Steve Henson] 8800 8801 *) Update OCSP API. Remove obsolete extensions argument from 8802 various functions. Extensions are now handled using the new 8803 OCSP extension code. New simple OCSP HTTP function which 8804 can be used to send requests and parse the response. 8805 [Steve Henson] 8806 8807 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new 8808 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN 8809 uses the special reorder version of SET OF to sort the attributes 8810 and reorder them to match the encoded order. This resolves a long 8811 standing problem: a verify on a PKCS7 structure just after signing 8812 it used to fail because the attribute order did not match the 8813 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes: 8814 it uses the received order. This is necessary to tolerate some broken 8815 software that does not order SET OF. This is handled by encoding 8816 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class) 8817 to produce the required SET OF. 8818 [Steve Henson] 8819 8820 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and 8821 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header 8822 files to get correct declarations of the ASN.1 item variables. 8823 [Richard Levitte] 8824 8825 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many 8826 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs: 8827 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was 8828 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i(). 8829 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant 8830 ASN1_ITEM and no wrapper functions. 8831 [Steve Henson] 8832 8833 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These 8834 replace the old function pointer based I/O routines. Change most of 8835 the *_d2i_bio() and *_d2i_fp() functions to use these. 8836 [Steve Henson] 8837 8838 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor 8839 lines, recognize more "algorithms" that can be deselected, and make 8840 it complain about algorithm deselection that isn't recognised. 8841 [Richard Levitte] 8842 8843 *) New ASN1 functions to handle dup, sign, verify, digest, pack and 8844 unpack operations in terms of ASN1_ITEM. Modify existing wrappers 8845 to use new functions. Add NO_ASN1_OLD which can be set to remove 8846 some old style ASN1 functions: this can be used to determine if old 8847 code will still work when these eventually go away. 8848 [Steve Henson] 8849 8850 *) New extension functions for OCSP structures, these follow the 8851 same conventions as certificates and CRLs. 8852 [Steve Henson] 8853 8854 *) New function X509V3_add1_i2d(). This automatically encodes and 8855 adds an extension. Its behaviour can be customised with various 8856 flags to append, replace or delete. Various wrappers added for 8857 certificates and CRLs. 8858 [Steve Henson] 8859 8860 *) Fix to avoid calling the underlying ASN1 print routine when 8861 an extension cannot be parsed. Correct a typo in the 8862 OCSP_SERVICELOC extension. Tidy up print OCSP format. 8863 [Steve Henson] 8864 8865 *) Make mkdef.pl parse some of the ASN1 macros and add appropriate 8866 entries for variables. 8867 [Steve Henson] 8868 8869 *) Add functionality to apps/openssl.c for detecting locking 8870 problems: As the program is single-threaded, all we have 8871 to do is register a locking callback using an array for 8872 storing which locks are currently held by the program. 8873 [Bodo Moeller] 8874 8875 *) Use a lock around the call to CRYPTO_get_ex_new_index() in 8876 SSL_get_ex_data_X509_STORE_idx(), which is used in 8877 ssl_verify_cert_chain() and thus can be called at any time 8878 during TLS/SSL handshakes so that thread-safety is essential. 8879 Unfortunately, the ex_data design is not at all suited 8880 for multi-threaded use, so it probably should be abolished. 8881 [Bodo Moeller] 8882 8883 *) Added Broadcom "ubsec" ENGINE to OpenSSL. 8884 [Broadcom, tweaked and integrated by Geoff Thorpe] 8885 8886 *) Move common extension printing code to new function 8887 X509V3_print_extensions(). Reorganise OCSP print routines and 8888 implement some needed OCSP ASN1 functions. Add OCSP extensions. 8889 [Steve Henson] 8890 8891 *) New function X509_signature_print() to remove duplication in some 8892 print routines. 8893 [Steve Henson] 8894 8895 *) Add a special meaning when SET OF and SEQUENCE OF flags are both 8896 set (this was treated exactly the same as SET OF previously). This 8897 is used to reorder the STACK representing the structure to match the 8898 encoding. This will be used to get round a problem where a PKCS7 8899 structure which was signed could not be verified because the STACK 8900 order did not reflect the encoded order. 8901 [Steve Henson] 8902 8903 *) Reimplement the OCSP ASN1 module using the new code. 8904 [Steve Henson] 8905 8906 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure 8907 for its ASN1 operations. The old style function pointers still exist 8908 for now but they will eventually go away. 8909 [Steve Henson] 8910 8911 *) Merge in replacement ASN1 code from the ASN1 branch. This almost 8912 completely replaces the old ASN1 functionality with a table driven 8913 encoder and decoder which interprets an ASN1_ITEM structure describing 8914 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is 8915 largely maintained. Almost all of the old asn1_mac.h macro based ASN1 8916 has also been converted to the new form. 8917 [Steve Henson] 8918 8919 *) Change BN_mod_exp_recp so that negative moduli are tolerated 8920 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set 8921 so that BN_mod_exp_mont and BN_mod_exp_mont_word work 8922 for negative moduli. 8923 [Bodo Moeller] 8924 8925 *) Fix BN_uadd and BN_usub: Always return non-negative results instead 8926 of not touching the result's sign bit. 8927 [Bodo Moeller] 8928 8929 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be 8930 set. 8931 [Bodo Moeller] 8932 8933 *) Changed the LHASH code to use prototypes for callbacks, and created 8934 macros to declare and implement thin (optionally static) functions 8935 that provide type-safety and avoid function pointer casting for the 8936 type-specific callbacks. 8937 [Geoff Thorpe] 8938 8939 *) Added Kerberos Cipher Suites to be used with TLS, as written in 8940 RFC 2712. 8941 [Veers Staats <staatsvr@asc.hpc.mil>, 8942 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte] 8943 8944 *) Reformat the FAQ so the different questions and answers can be divided 8945 in sections depending on the subject. 8946 [Richard Levitte] 8947 8948 *) Have the zlib compression code load ZLIB.DLL dynamically under 8949 Windows. 8950 [Richard Levitte] 8951 8952 *) New function BN_mod_sqrt for computing square roots modulo a prime 8953 (using the probabilistic Tonelli-Shanks algorithm unless 8954 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can 8955 be handled deterministically). 8956 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 8957 8958 *) Make BN_mod_inverse faster by explicitly handling small quotients 8959 in the Euclid loop. (Speed gain about 20% for small moduli [256 or 8960 512 bits], about 30% for larger ones [1024 or 2048 bits].) 8961 [Bodo Moeller] 8962 8963 *) New function BN_kronecker. 8964 [Bodo Moeller] 8965 8966 *) Fix BN_gcd so that it works on negative inputs; the result is 8967 positive unless both parameters are zero. 8968 Previously something reasonably close to an infinite loop was 8969 possible because numbers could be growing instead of shrinking 8970 in the implementation of Euclid's algorithm. 8971 [Bodo Moeller] 8972 8973 *) Fix BN_is_word() and BN_is_one() macros to take into account the 8974 sign of the number in question. 8975 8976 Fix BN_is_word(a,w) to work correctly for w == 0. 8977 8978 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w) 8979 because its test if the absolute value of 'a' equals 'w'. 8980 Note that BN_abs_is_word does *not* handle w == 0 reliably; 8981 it exists mostly for use in the implementations of BN_is_zero(), 8982 BN_is_one(), and BN_is_word(). 8983 [Bodo Moeller] 8984 8985 *) New function BN_swap. 8986 [Bodo Moeller] 8987 8988 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that 8989 the exponentiation functions are more likely to produce reasonable 8990 results on negative inputs. 8991 [Bodo Moeller] 8992 8993 *) Change BN_mod_mul so that the result is always non-negative. 8994 Previously, it could be negative if one of the factors was negative; 8995 I don't think anyone really wanted that behaviour. 8996 [Bodo Moeller] 8997 8998 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c 8999 (except for exponentiation, which stays in crypto/bn/bn_exp.c, 9000 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c) 9001 and add new functions: 9002 9003 BN_nnmod 9004 BN_mod_sqr 9005 BN_mod_add 9006 BN_mod_add_quick 9007 BN_mod_sub 9008 BN_mod_sub_quick 9009 BN_mod_lshift1 9010 BN_mod_lshift1_quick 9011 BN_mod_lshift 9012 BN_mod_lshift_quick 9013 9014 These functions always generate non-negative results. 9015 9016 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r 9017 such that |m| < r < 0, BN_nnmod will output rem + |m| instead). 9018 9019 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as 9020 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b] 9021 be reduced modulo m. 9022 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller] 9023 9024#if 0 9025 The following entry accidentally appeared in the CHANGES file 9026 distributed with OpenSSL 0.9.7. The modifications described in 9027 it do *not* apply to OpenSSL 0.9.7. 9028 9029 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there 9030 was actually never needed) and in BN_mul(). The removal in BN_mul() 9031 required a small change in bn_mul_part_recursive() and the addition 9032 of the functions bn_cmp_part_words(), bn_sub_part_words() and 9033 bn_add_part_words(), which do the same thing as bn_cmp_words(), 9034 bn_sub_words() and bn_add_words() except they take arrays with 9035 differing sizes. 9036 [Richard Levitte] 9037#endif 9038 9039 *) In 'openssl passwd', verify passwords read from the terminal 9040 unless the '-salt' option is used (which usually means that 9041 verification would just waste user's time since the resulting 9042 hash is going to be compared with some given password hash) 9043 or the new '-noverify' option is used. 9044 9045 This is an incompatible change, but it does not affect 9046 non-interactive use of 'openssl passwd' (passwords on the command 9047 line, '-stdin' option, '-in ...' option) and thus should not 9048 cause any problems. 9049 [Bodo Moeller] 9050 9051 *) Remove all references to RSAref, since there's no more need for it. 9052 [Richard Levitte] 9053 9054 *) Make DSO load along a path given through an environment variable 9055 (SHLIB_PATH) with shl_load(). 9056 [Richard Levitte] 9057 9058 *) Constify the ENGINE code as a result of BIGNUM constification. 9059 Also constify the RSA code and most things related to it. In a 9060 few places, most notable in the depth of the ASN.1 code, ugly 9061 casts back to non-const were required (to be solved at a later 9062 time) 9063 [Richard Levitte] 9064 9065 *) Make it so the openssl application has all engines loaded by default. 9066 [Richard Levitte] 9067 9068 *) Constify the BIGNUM routines a little more. 9069 [Richard Levitte] 9070 9071 *) Add the following functions: 9072 9073 ENGINE_load_cswift() 9074 ENGINE_load_chil() 9075 ENGINE_load_atalla() 9076 ENGINE_load_nuron() 9077 ENGINE_load_builtin_engines() 9078 9079 That way, an application can itself choose if external engines that 9080 are built-in in OpenSSL shall ever be used or not. The benefit is 9081 that applications won't have to be linked with libdl or other dso 9082 libraries unless it's really needed. 9083 9084 Changed 'openssl engine' to load all engines on demand. 9085 Changed the engine header files to avoid the duplication of some 9086 declarations (they differed!). 9087 [Richard Levitte] 9088 9089 *) 'openssl engine' can now list capabilities. 9090 [Richard Levitte] 9091 9092 *) Better error reporting in 'openssl engine'. 9093 [Richard Levitte] 9094 9095 *) Never call load_dh_param(NULL) in s_server. 9096 [Bodo Moeller] 9097 9098 *) Add engine application. It can currently list engines by name and 9099 identity, and test if they are actually available. 9100 [Richard Levitte] 9101 9102 *) Improve RPM specification file by forcing symbolic linking and making 9103 sure the installed documentation is also owned by root.root. 9104 [Damien Miller <djm@mindrot.org>] 9105 9106 *) Give the OpenSSL applications more possibilities to make use of 9107 keys (public as well as private) handled by engines. 9108 [Richard Levitte] 9109 9110 *) Add OCSP code that comes from CertCo. 9111 [Richard Levitte] 9112 9113 *) Add VMS support for the Rijndael code. 9114 [Richard Levitte] 9115 9116 *) Added untested support for Nuron crypto accelerator. 9117 [Ben Laurie] 9118 9119 *) Add support for external cryptographic devices. This code was 9120 previously distributed separately as the "engine" branch. 9121 [Geoff Thorpe, Richard Levitte] 9122 9123 *) Rework the filename-translation in the DSO code. It is now possible to 9124 have far greater control over how a "name" is turned into a filename 9125 depending on the operating environment and any oddities about the 9126 different shared library filenames on each system. 9127 [Geoff Thorpe] 9128 9129 *) Support threads on FreeBSD-elf in Configure. 9130 [Richard Levitte] 9131 9132 *) Fix for SHA1 assembly problem with MASM: it produces 9133 warnings about corrupt line number information when assembling 9134 with debugging information. This is caused by the overlapping 9135 of two sections. 9136 [Bernd Matthes <mainbug@celocom.de>, Steve Henson] 9137 9138 *) NCONF changes. 9139 NCONF_get_number() has no error checking at all. As a replacement, 9140 NCONF_get_number_e() is defined (_e for "error checking") and is 9141 promoted strongly. The old NCONF_get_number is kept around for 9142 binary backward compatibility. 9143 Make it possible for methods to load from something other than a BIO, 9144 by providing a function pointer that is given a name instead of a BIO. 9145 For example, this could be used to load configuration data from an 9146 LDAP server. 9147 [Richard Levitte] 9148 9149 *) Fix for non blocking accept BIOs. Added new I/O special reason 9150 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs 9151 with non blocking I/O was not possible because no retry code was 9152 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover 9153 this case. 9154 [Steve Henson] 9155 9156 *) Added the beginnings of Rijndael support. 9157 [Ben Laurie] 9158 9159 *) Fix for bug in DirectoryString mask setting. Add support for 9160 X509_NAME_print_ex() in 'req' and X509_print_ex() function 9161 to allow certificate printing to more controllable, additional 9162 'certopt' option to 'x509' to allow new printing options to be 9163 set. 9164 [Steve Henson] 9165 9166 *) Clean old EAY MD5 hack from e_os.h. 9167 [Richard Levitte] 9168 9169 Changes between 0.9.6l and 0.9.6m [17 Mar 2004] 9170 9171 *) Fix null-pointer assignment in do_change_cipher_spec() revealed 9172 by using the Codenomicon TLS Test Tool (CVE-2004-0079) 9173 [Joe Orton, Steve Henson] 9174 9175 Changes between 0.9.6k and 0.9.6l [04 Nov 2003] 9176 9177 *) Fix additional bug revealed by the NISCC test suite: 9178 9179 Stop bug triggering large recursion when presented with 9180 certain ASN.1 tags (CVE-2003-0851) 9181 [Steve Henson] 9182 9183 Changes between 0.9.6j and 0.9.6k [30 Sep 2003] 9184 9185 *) Fix various bugs revealed by running the NISCC test suite: 9186 9187 Stop out of bounds reads in the ASN1 code when presented with 9188 invalid tags (CVE-2003-0543 and CVE-2003-0544). 9189 9190 If verify callback ignores invalid public key errors don't try to check 9191 certificate signature with the NULL public key. 9192 9193 [Steve Henson] 9194 9195 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate 9196 if the server requested one: as stated in TLS 1.0 and SSL 3.0 9197 specifications. 9198 [Steve Henson] 9199 9200 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional 9201 extra data after the compression methods not only for TLS 1.0 9202 but also for SSL 3.0 (as required by the specification). 9203 [Bodo Moeller; problem pointed out by Matthias Loepfe] 9204 9205 *) Change X509_certificate_type() to mark the key as exported/exportable 9206 when it's 512 *bits* long, not 512 bytes. 9207 [Richard Levitte] 9208 9209 Changes between 0.9.6i and 0.9.6j [10 Apr 2003] 9210 9211 *) Countermeasure against the Klima-Pokorny-Rosa extension of 9212 Bleichbacher's attack on PKCS #1 v1.5 padding: treat 9213 a protocol version number mismatch like a decryption error 9214 in ssl3_get_client_key_exchange (ssl/s3_srvr.c). 9215 [Bodo Moeller] 9216 9217 *) Turn on RSA blinding by default in the default implementation 9218 to avoid a timing attack. Applications that don't want it can call 9219 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING. 9220 They would be ill-advised to do so in most cases. 9221 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller] 9222 9223 *) Change RSA blinding code so that it works when the PRNG is not 9224 seeded (in this case, the secret RSA exponent is abused as 9225 an unpredictable seed -- if it is not unpredictable, there 9226 is no point in blinding anyway). Make RSA blinding thread-safe 9227 by remembering the creator's thread ID in rsa->blinding and 9228 having all other threads use local one-time blinding factors 9229 (this requires more computation than sharing rsa->blinding, but 9230 avoids excessive locking; and if an RSA object is not shared 9231 between threads, blinding will still be very fast). 9232 [Bodo Moeller] 9233 9234 Changes between 0.9.6h and 0.9.6i [19 Feb 2003] 9235 9236 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked 9237 via timing by performing a MAC computation even if incorrect 9238 block cipher padding has been found. This is a countermeasure 9239 against active attacks where the attacker has to distinguish 9240 between bad padding and a MAC verification error. (CVE-2003-0078) 9241 9242 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL), 9243 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and 9244 Martin Vuagnoux (EPFL, Ilion)] 9245 9246 Changes between 0.9.6g and 0.9.6h [5 Dec 2002] 9247 9248 *) New function OPENSSL_cleanse(), which is used to cleanse a section of 9249 memory from its contents. This is done with a counter that will 9250 place alternating values in each byte. This can be used to solve 9251 two issues: 1) the removal of calls to memset() by highly optimizing 9252 compilers, and 2) cleansing with other values than 0, since those can 9253 be read through on certain media, for example a swap space on disk. 9254 [Geoff Thorpe] 9255 9256 *) Bugfix: client side session caching did not work with external caching, 9257 because the session->cipher setting was not restored when reloading 9258 from the external cache. This problem was masked, when 9259 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set. 9260 (Found by Steve Haslam <steve@araqnid.ddts.net>.) 9261 [Lutz Jaenicke] 9262 9263 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total 9264 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33. 9265 [Zeev Lieber <zeev-l@yahoo.com>] 9266 9267 *) Undo an undocumented change introduced in 0.9.6e which caused 9268 repeated calls to OpenSSL_add_all_ciphers() and 9269 OpenSSL_add_all_digests() to be ignored, even after calling 9270 EVP_cleanup(). 9271 [Richard Levitte] 9272 9273 *) Change the default configuration reader to deal with last line not 9274 being properly terminated. 9275 [Richard Levitte] 9276 9277 *) Change X509_NAME_cmp() so it applies the special rules on handling 9278 DN values that are of type PrintableString, as well as RDNs of type 9279 emailAddress where the value has the type ia5String. 9280 [stefank@valicert.com via Richard Levitte] 9281 9282 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half 9283 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently 9284 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be 9285 the bitwise-OR of the two for use by the majority of applications 9286 wanting this behaviour, and update the docs. The documented 9287 behaviour and actual behaviour were inconsistent and had been 9288 changing anyway, so this is more a bug-fix than a behavioural 9289 change. 9290 [Geoff Thorpe, diagnosed by Nadav Har'El] 9291 9292 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c 9293 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes). 9294 [Bodo Moeller] 9295 9296 *) Fix initialization code race conditions in 9297 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(), 9298 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(), 9299 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(), 9300 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(), 9301 ssl2_get_cipher_by_char(), 9302 ssl3_get_cipher_by_char(). 9303 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller] 9304 9305 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after 9306 the cached sessions are flushed, as the remove_cb() might use ex_data 9307 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com> 9308 (see [openssl.org #212]). 9309 [Geoff Thorpe, Lutz Jaenicke] 9310 9311 *) Fix typo in OBJ_txt2obj which incorrectly passed the content 9312 length, instead of the encoding length to d2i_ASN1_OBJECT. 9313 [Steve Henson] 9314 9315 Changes between 0.9.6f and 0.9.6g [9 Aug 2002] 9316 9317 *) [In 0.9.6g-engine release:] 9318 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall'). 9319 [Lynn Gazis <lgazis@rainbow.com>] 9320 9321 Changes between 0.9.6e and 0.9.6f [8 Aug 2002] 9322 9323 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX 9324 and get fix the header length calculation. 9325 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>, 9326 Alon Kantor <alonk@checkpoint.com> (and others), 9327 Steve Henson] 9328 9329 *) Use proper error handling instead of 'assertions' in buffer 9330 overflow checks added in 0.9.6e. This prevents DoS (the 9331 assertions could call abort()). 9332 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller] 9333 9334 Changes between 0.9.6d and 0.9.6e [30 Jul 2002] 9335 9336 *) Add various sanity checks to asn1_get_length() to reject 9337 the ASN1 length bytes if they exceed sizeof(long), will appear 9338 negative or the content length exceeds the length of the 9339 supplied buffer. 9340 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 9341 9342 *) Fix cipher selection routines: ciphers without encryption had no flags 9343 for the cipher strength set and where therefore not handled correctly 9344 by the selection routines (PR #130). 9345 [Lutz Jaenicke] 9346 9347 *) Fix EVP_dsa_sha macro. 9348 [Nils Larsch] 9349 9350 *) New option 9351 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS 9352 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure 9353 that was added in OpenSSL 0.9.6d. 9354 9355 As the countermeasure turned out to be incompatible with some 9356 broken SSL implementations, the new option is part of SSL_OP_ALL. 9357 SSL_OP_ALL is usually employed when compatibility with weird SSL 9358 implementations is desired (e.g. '-bugs' option to 's_client' and 9359 's_server'), so the new option is automatically set in many 9360 applications. 9361 [Bodo Moeller] 9362 9363 *) Changes in security patch: 9364 9365 Changes marked "(CHATS)" were sponsored by the Defense Advanced 9366 Research Projects Agency (DARPA) and Air Force Research Laboratory, 9367 Air Force Materiel Command, USAF, under agreement number 9368 F30602-01-2-0537. 9369 9370 *) Add various sanity checks to asn1_get_length() to reject 9371 the ASN1 length bytes if they exceed sizeof(long), will appear 9372 negative or the content length exceeds the length of the 9373 supplied buffer. (CVE-2002-0659) 9374 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>] 9375 9376 *) Assertions for various potential buffer overflows, not known to 9377 happen in practice. 9378 [Ben Laurie (CHATS)] 9379 9380 *) Various temporary buffers to hold ASCII versions of integers were 9381 too small for 64 bit platforms. (CVE-2002-0655) 9382 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)> 9383 9384 *) Remote buffer overflow in SSL3 protocol - an attacker could 9385 supply an oversized session ID to a client. (CVE-2002-0656) 9386 [Ben Laurie (CHATS)] 9387 9388 *) Remote buffer overflow in SSL2 protocol - an attacker could 9389 supply an oversized client master key. (CVE-2002-0656) 9390 [Ben Laurie (CHATS)] 9391 9392 Changes between 0.9.6c and 0.9.6d [9 May 2002] 9393 9394 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not 9395 encoded as NULL) with id-dsa-with-sha1. 9396 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller] 9397 9398 *) Check various X509_...() return values in apps/req.c. 9399 [Nils Larsch <nla@trustcenter.de>] 9400 9401 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines: 9402 an end-of-file condition would erroneously be flagged, when the CRLF 9403 was just at the end of a processed block. The bug was discovered when 9404 processing data through a buffering memory BIO handing the data to a 9405 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov 9406 <ptsekov@syntrex.com> and Nedelcho Stanev. 9407 [Lutz Jaenicke] 9408 9409 *) Implement a countermeasure against a vulnerability recently found 9410 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment 9411 before application data chunks to avoid the use of known IVs 9412 with data potentially chosen by the attacker. 9413 [Bodo Moeller] 9414 9415 *) Fix length checks in ssl3_get_client_hello(). 9416 [Bodo Moeller] 9417 9418 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently 9419 to prevent ssl3_read_internal() from incorrectly assuming that 9420 ssl3_read_bytes() found application data while handshake 9421 processing was enabled when in fact s->s3->in_read_app_data was 9422 merely automatically cleared during the initial handshake. 9423 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>] 9424 9425 *) Fix object definitions for Private and Enterprise: they were not 9426 recognized in their shortname (=lowercase) representation. Extend 9427 obj_dat.pl to issue an error when using undefined keywords instead 9428 of silently ignoring the problem (Svenning Sorensen 9429 <sss@sss.dnsalias.net>). 9430 [Lutz Jaenicke] 9431 9432 *) Fix DH_generate_parameters() so that it works for 'non-standard' 9433 generators, i.e. generators other than 2 and 5. (Previously, the 9434 code did not properly initialise the 'add' and 'rem' values to 9435 BN_generate_prime().) 9436 9437 In the new general case, we do not insist that 'generator' is 9438 actually a primitive root: This requirement is rather pointless; 9439 a generator of the order-q subgroup is just as good, if not 9440 better. 9441 [Bodo Moeller] 9442 9443 *) Map new X509 verification errors to alerts. Discovered and submitted by 9444 Tom Wu <tom@arcot.com>. 9445 [Lutz Jaenicke] 9446 9447 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from 9448 returning non-zero before the data has been completely received 9449 when using non-blocking I/O. 9450 [Bodo Moeller; problem pointed out by John Hughes] 9451 9452 *) Some of the ciphers missed the strength entry (SSL_LOW etc). 9453 [Ben Laurie, Lutz Jaenicke] 9454 9455 *) Fix bug in SSL_clear(): bad sessions were not removed (found by 9456 Yoram Zahavi <YoramZ@gilian.com>). 9457 [Lutz Jaenicke] 9458 9459 *) Add information about CygWin 1.3 and on, and preserve proper 9460 configuration for the versions before that. 9461 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte] 9462 9463 *) Make removal from session cache (SSL_CTX_remove_session()) more robust: 9464 check whether we deal with a copy of a session and do not delete from 9465 the cache in this case. Problem reported by "Izhar Shoshani Levi" 9466 <izhar@checkpoint.com>. 9467 [Lutz Jaenicke] 9468 9469 *) Do not store session data into the internal session cache, if it 9470 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP 9471 flag is set). Proposed by Aslam <aslam@funk.com>. 9472 [Lutz Jaenicke] 9473 9474 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested 9475 value is 0. 9476 [Richard Levitte] 9477 9478 *) [In 0.9.6d-engine release:] 9479 Fix a crashbug and a logic bug in hwcrhk_load_pubkey(). 9480 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte] 9481 9482 *) Add the configuration target linux-s390x. 9483 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte] 9484 9485 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of 9486 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag 9487 variable as an indication that a ClientHello message has been 9488 received. As the flag value will be lost between multiple 9489 invocations of ssl3_accept when using non-blocking I/O, the 9490 function may not be aware that a handshake has actually taken 9491 place, thus preventing a new session from being added to the 9492 session cache. 9493 9494 To avoid this problem, we now set s->new_session to 2 instead of 9495 using a local variable. 9496 [Lutz Jaenicke, Bodo Moeller] 9497 9498 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c) 9499 if the SSL_R_LENGTH_MISMATCH error is detected. 9500 [Geoff Thorpe, Bodo Moeller] 9501 9502 *) New 'shared_ldflag' column in Configure platform table. 9503 [Richard Levitte] 9504 9505 *) Fix EVP_CIPHER_mode macro. 9506 ["Dan S. Camper" <dan@bti.net>] 9507 9508 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown 9509 type, we must throw them away by setting rr->length to 0. 9510 [D P Chang <dpc@qualys.com>] 9511 9512 Changes between 0.9.6b and 0.9.6c [21 dec 2001] 9513 9514 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl 9515 <Dominikus.Scherkl@biodata.com>. (The previous implementation 9516 worked incorrectly for those cases where range = 10..._2 and 9517 3*range is two bits longer than range.) 9518 [Bodo Moeller] 9519 9520 *) Only add signing time to PKCS7 structures if it is not already 9521 present. 9522 [Steve Henson] 9523 9524 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce", 9525 OBJ_ld_ce should be OBJ_id_ce. 9526 Also some ip-pda OIDs in crypto/objects/objects.txt were 9527 incorrect (cf. RFC 3039). 9528 [Matt Cooper, Frederic Giudicelli, Bodo Moeller] 9529 9530 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid() 9531 returns early because it has nothing to do. 9532 [Andy Schneider <andy.schneider@bjss.co.uk>] 9533 9534 *) [In 0.9.6c-engine release:] 9535 Fix mutex callback return values in crypto/engine/hw_ncipher.c. 9536 [Andy Schneider <andy.schneider@bjss.co.uk>] 9537 9538 *) [In 0.9.6c-engine release:] 9539 Add support for Cryptographic Appliance's keyserver technology. 9540 (Use engine 'keyclient') 9541 [Cryptographic Appliances and Geoff Thorpe] 9542 9543 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89' 9544 is called via tools/c89.sh because arguments have to be 9545 rearranged (all '-L' options must appear before the first object 9546 modules). 9547 [Richard Shapiro <rshapiro@abinitio.com>] 9548 9549 *) [In 0.9.6c-engine release:] 9550 Add support for Broadcom crypto accelerator cards, backported 9551 from 0.9.7. 9552 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox] 9553 9554 *) [In 0.9.6c-engine release:] 9555 Add support for SureWare crypto accelerator cards from 9556 Baltimore Technologies. (Use engine 'sureware') 9557 [Baltimore Technologies and Mark Cox] 9558 9559 *) [In 0.9.6c-engine release:] 9560 Add support for crypto accelerator cards from Accelerated 9561 Encryption Processing, www.aep.ie. (Use engine 'aep') 9562 [AEP Inc. and Mark Cox] 9563 9564 *) Add a configuration entry for gcc on UnixWare. 9565 [Gary Benson <gbenson@redhat.com>] 9566 9567 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake 9568 messages are stored in a single piece (fixed-length part and 9569 variable-length part combined) and fix various bugs found on the way. 9570 [Bodo Moeller] 9571 9572 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname() 9573 instead. BIO_gethostbyname() does not know what timeouts are 9574 appropriate, so entries would stay in cache even when they have 9575 become invalid. 9576 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com> 9577 9578 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when 9579 faced with a pathologically small ClientHello fragment that does 9580 not contain client_version: Instead of aborting with an error, 9581 simply choose the highest available protocol version (i.e., 9582 TLS 1.0 unless it is disabled). In practice, ClientHello 9583 messages are never sent like this, but this change gives us 9584 strictly correct behaviour at least for TLS. 9585 [Bodo Moeller] 9586 9587 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear() 9588 never resets s->method to s->ctx->method when called from within 9589 one of the SSL handshake functions. 9590 [Bodo Moeller; problem pointed out by Niko Baric] 9591 9592 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert 9593 (sent using the client's version number) if client_version is 9594 smaller than the protocol version in use. Also change 9595 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if 9596 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then 9597 the client will at least see that alert. 9598 [Bodo Moeller] 9599 9600 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation 9601 correctly. 9602 [Bodo Moeller] 9603 9604 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a 9605 client receives HelloRequest while in a handshake. 9606 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>] 9607 9608 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C 9609 should end in 'break', not 'goto end' which circumvents various 9610 cleanups done in state SSL_ST_OK. But session related stuff 9611 must be disabled for SSL_ST_OK in the case that we just sent a 9612 HelloRequest. 9613 9614 Also avoid some overhead by not calling ssl_init_wbio_buffer() 9615 before just sending a HelloRequest. 9616 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>] 9617 9618 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't 9619 reveal whether illegal block cipher padding was found or a MAC 9620 verification error occurred. (Neither SSLerr() codes nor alerts 9621 are directly visible to potential attackers, but the information 9622 may leak via logfiles.) 9623 9624 Similar changes are not required for the SSL 2.0 implementation 9625 because the number of padding bytes is sent in clear for SSL 2.0, 9626 and the extra bytes are just ignored. However ssl/s2_pkt.c 9627 failed to verify that the purported number of padding bytes is in 9628 the legal range. 9629 [Bodo Moeller] 9630 9631 *) Add OpenUNIX-8 support including shared libraries 9632 (Boyd Lynn Gerber <gerberb@zenez.com>). 9633 [Lutz Jaenicke] 9634 9635 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid 9636 'wristwatch attack' using huge encoding parameters (cf. 9637 James H. Manger's CRYPTO 2001 paper). Note that the 9638 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use 9639 encoding parameters and hence was not vulnerable. 9640 [Bodo Moeller] 9641 9642 *) BN_sqr() bug fix. 9643 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>] 9644 9645 *) Rabin-Miller test analyses assume uniformly distributed witnesses, 9646 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand() 9647 followed by modular reduction. 9648 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>] 9649 9650 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range() 9651 equivalent based on BN_pseudo_rand() instead of BN_rand(). 9652 [Bodo Moeller] 9653 9654 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB). 9655 This function was broken, as the check for a new client hello message 9656 to handle SGC did not allow these large messages. 9657 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.) 9658 [Lutz Jaenicke] 9659 9660 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long](). 9661 [Lutz Jaenicke] 9662 9663 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl() 9664 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>). 9665 [Lutz Jaenicke] 9666 9667 *) Rework the configuration and shared library support for Tru64 Unix. 9668 The configuration part makes use of modern compiler features and 9669 still retains old compiler behavior for those that run older versions 9670 of the OS. The shared library support part includes a variant that 9671 uses the RPATH feature, and is available through the special 9672 configuration target "alpha-cc-rpath", which will never be selected 9673 automatically. 9674 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte] 9675 9676 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message() 9677 with the same message size as in ssl3_get_certificate_request(). 9678 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest 9679 messages might inadvertently be reject as too long. 9680 [Petr Lampa <lampa@fee.vutbr.cz>] 9681 9682 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX). 9683 [Andy Polyakov] 9684 9685 *) Modified SSL library such that the verify_callback that has been set 9686 specifically for an SSL object with SSL_set_verify() is actually being 9687 used. Before the change, a verify_callback set with this function was 9688 ignored and the verify_callback() set in the SSL_CTX at the time of 9689 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced 9690 to allow the necessary settings. 9691 [Lutz Jaenicke] 9692 9693 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c 9694 explicitly to NULL, as at least on Solaris 8 this seems not always to be 9695 done automatically (in contradiction to the requirements of the C 9696 standard). This made problems when used from OpenSSH. 9697 [Lutz Jaenicke] 9698 9699 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored 9700 dh->length and always used 9701 9702 BN_rand_range(priv_key, dh->p). 9703 9704 BN_rand_range() is not necessary for Diffie-Hellman, and this 9705 specific range makes Diffie-Hellman unnecessarily inefficient if 9706 dh->length (recommended exponent length) is much smaller than the 9707 length of dh->p. We could use BN_rand_range() if the order of 9708 the subgroup was stored in the DH structure, but we only have 9709 dh->length. 9710 9711 So switch back to 9712 9713 BN_rand(priv_key, l, ...) 9714 9715 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1 9716 otherwise. 9717 [Bodo Moeller] 9718 9719 *) In 9720 9721 RSA_eay_public_encrypt 9722 RSA_eay_private_decrypt 9723 RSA_eay_private_encrypt (signing) 9724 RSA_eay_public_decrypt (signature verification) 9725 9726 (default implementations for RSA_public_encrypt, 9727 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt), 9728 always reject numbers >= n. 9729 [Bodo Moeller] 9730 9731 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2 9732 to synchronize access to 'locking_thread'. This is necessary on 9733 systems where access to 'locking_thread' (an 'unsigned long' 9734 variable) is not atomic. 9735 [Bodo Moeller] 9736 9737 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID 9738 *before* setting the 'crypto_lock_rand' flag. The previous code had 9739 a race condition if 0 is a valid thread ID. 9740 [Travis Vitek <vitek@roguewave.com>] 9741 9742 *) Add support for shared libraries under Irix. 9743 [Albert Chin-A-Young <china@thewrittenword.com>] 9744 9745 *) Add configuration option to build on Linux on both big-endian and 9746 little-endian MIPS. 9747 [Ralf Baechle <ralf@uni-koblenz.de>] 9748 9749 *) Add the possibility to create shared libraries on HP-UX. 9750 [Richard Levitte] 9751 9752 Changes between 0.9.6a and 0.9.6b [9 Jul 2001] 9753 9754 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c) 9755 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by 9756 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>: 9757 PRNG state recovery was possible based on the output of 9758 one PRNG request appropriately sized to gain knowledge on 9759 'md' followed by enough consecutive 1-byte PRNG requests 9760 to traverse all of 'state'. 9761 9762 1. When updating 'md_local' (the current thread's copy of 'md') 9763 during PRNG output generation, hash all of the previous 9764 'md_local' value, not just the half used for PRNG output. 9765 9766 2. Make the number of bytes from 'state' included into the hash 9767 independent from the number of PRNG bytes requested. 9768 9769 The first measure alone would be sufficient to avoid 9770 Markku-Juhani's attack. (Actually it had never occurred 9771 to me that the half of 'md_local' used for chaining was the 9772 half from which PRNG output bytes were taken -- I had always 9773 assumed that the secret half would be used.) The second 9774 measure makes sure that additional data from 'state' is never 9775 mixed into 'md_local' in small portions; this heuristically 9776 further strengthens the PRNG. 9777 [Bodo Moeller] 9778 9779 *) Fix crypto/bn/asm/mips3.s. 9780 [Andy Polyakov] 9781 9782 *) When only the key is given to "enc", the IV is undefined. Print out 9783 an error message in this case. 9784 [Lutz Jaenicke] 9785 9786 *) Handle special case when X509_NAME is empty in X509 printing routines. 9787 [Steve Henson] 9788 9789 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are 9790 positive and less than q. 9791 [Bodo Moeller] 9792 9793 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is 9794 used: it isn't thread safe and the add_lock_callback should handle 9795 that itself. 9796 [Paul Rose <Paul.Rose@bridge.com>] 9797 9798 *) Verify that incoming data obeys the block size in 9799 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c). 9800 [Bodo Moeller] 9801 9802 *) Fix OAEP check. 9803 [Ulf Möller, Bodo Möller] 9804 9805 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5 9806 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5 9807 when fixing the server behaviour for backwards-compatible 'client 9808 hello' messages. (Note that the attack is impractical against 9809 SSL 3.0 and TLS 1.0 anyway because length and version checking 9810 means that the probability of guessing a valid ciphertext is 9811 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98 9812 paper.) 9813 9814 Before 0.9.5, the countermeasure (hide the error by generating a 9815 random 'decryption result') did not work properly because 9816 ERR_clear_error() was missing, meaning that SSL_get_error() would 9817 detect the supposedly ignored error. 9818 9819 Both problems are now fixed. 9820 [Bodo Moeller] 9821 9822 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096 9823 (previously it was 1024). 9824 [Bodo Moeller] 9825 9826 *) Fix for compatibility mode trust settings: ignore trust settings 9827 unless some valid trust or reject settings are present. 9828 [Steve Henson] 9829 9830 *) Fix for blowfish EVP: its a variable length cipher. 9831 [Steve Henson] 9832 9833 *) Fix various bugs related to DSA S/MIME verification. Handle missing 9834 parameters in DSA public key structures and return an error in the 9835 DSA routines if parameters are absent. 9836 [Steve Henson] 9837 9838 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd" 9839 in the current directory if neither $RANDFILE nor $HOME was set. 9840 RAND_file_name() in 0.9.6a returned NULL in this case. This has 9841 caused some confusion to Windows users who haven't defined $HOME. 9842 Thus RAND_file_name() is changed again: e_os.h can define a 9843 DEFAULT_HOME, which will be used if $HOME is not set. 9844 For Windows, we use "C:"; on other platforms, we still require 9845 environment variables. 9846 9847 *) Move 'if (!initialized) RAND_poll()' into regions protected by 9848 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids 9849 having multiple threads call RAND_poll() concurrently. 9850 [Bodo Moeller] 9851 9852 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a 9853 combination of a flag and a thread ID variable. 9854 Otherwise while one thread is in ssleay_rand_bytes (which sets the 9855 flag), *other* threads can enter ssleay_add_bytes without obeying 9856 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock 9857 that they do not hold after the first thread unsets add_do_not_lock). 9858 [Bodo Moeller] 9859 9860 *) Change bctest again: '-x' expressions are not available in all 9861 versions of 'test'. 9862 [Bodo Moeller] 9863 9864 Changes between 0.9.6 and 0.9.6a [5 Apr 2001] 9865 9866 *) Fix a couple of memory leaks in PKCS7_dataDecode() 9867 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>] 9868 9869 *) Change Configure and Makefiles to provide EXE_EXT, which will contain 9870 the default extension for executables, if any. Also, make the perl 9871 scripts that use symlink() to test if it really exists and use "cp" 9872 if it doesn't. All this made OpenSSL compilable and installable in 9873 CygWin. 9874 [Richard Levitte] 9875 9876 *) Fix for asn1_GetSequence() for indefinite length constructed data. 9877 If SEQUENCE is length is indefinite just set c->slen to the total 9878 amount of data available. 9879 [Steve Henson, reported by shige@FreeBSD.org] 9880 [This change does not apply to 0.9.7.] 9881 9882 *) Change bctest to avoid here-documents inside command substitution 9883 (workaround for FreeBSD /bin/sh bug). 9884 For compatibility with Ultrix, avoid shell functions (introduced 9885 in the bctest version that searches along $PATH). 9886 [Bodo Moeller] 9887 9888 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes 9889 with des_encrypt() defined on some operating systems, like Solaris 9890 and UnixWare. 9891 [Richard Levitte] 9892 9893 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton: 9894 On the Importance of Eliminating Errors in Cryptographic 9895 Computations, J. Cryptology 14 (2001) 2, 101-119, 9896 http://theory.stanford.edu/~dabo/papers/faults.ps.gz). 9897 [Ulf Moeller] 9898 9899 *) MIPS assembler BIGNUM division bug fix. 9900 [Andy Polyakov] 9901 9902 *) Disabled incorrect Alpha assembler code. 9903 [Richard Levitte] 9904 9905 *) Fix PKCS#7 decode routines so they correctly update the length 9906 after reading an EOC for the EXPLICIT tag. 9907 [Steve Henson] 9908 [This change does not apply to 0.9.7.] 9909 9910 *) Fix bug in PKCS#12 key generation routines. This was triggered 9911 if a 3DES key was generated with a 0 initial byte. Include 9912 PKCS12_BROKEN_KEYGEN compilation option to retain the old 9913 (but broken) behaviour. 9914 [Steve Henson] 9915 9916 *) Enhance bctest to search for a working bc along $PATH and print 9917 it when found. 9918 [Tim Rice <tim@multitalents.net> via Richard Levitte] 9919 9920 *) Fix memory leaks in err.c: free err_data string if necessary; 9921 don't write to the wrong index in ERR_set_error_data. 9922 [Bodo Moeller] 9923 9924 *) Implement ssl23_peek (analogous to ssl23_read), which previously 9925 did not exist. 9926 [Bodo Moeller] 9927 9928 *) Replace rdtsc with _emit statements for VC++ version 5. 9929 [Jeremy Cooper <jeremy@baymoo.org>] 9930 9931 *) Make it possible to reuse SSLv2 sessions. 9932 [Richard Levitte] 9933 9934 *) In copy_email() check for >= 0 as a return value for 9935 X509_NAME_get_index_by_NID() since 0 is a valid index. 9936 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>] 9937 9938 *) Avoid coredump with unsupported or invalid public keys by checking if 9939 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when 9940 PKCS7_verify() fails with non detached data. 9941 [Steve Henson] 9942 9943 *) Don't use getenv in library functions when run as setuid/setgid. 9944 New function OPENSSL_issetugid(). 9945 [Ulf Moeller] 9946 9947 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c) 9948 due to incorrect handling of multi-threading: 9949 9950 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl(). 9951 9952 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on(). 9953 9954 3. Count how many times MemCheck_off() has been called so that 9955 nested use can be treated correctly. This also avoids 9956 inband-signalling in the previous code (which relied on the 9957 assumption that thread ID 0 is impossible). 9958 [Bodo Moeller] 9959 9960 *) Add "-rand" option also to s_client and s_server. 9961 [Lutz Jaenicke] 9962 9963 *) Fix CPU detection on Irix 6.x. 9964 [Kurt Hockenbury <khockenb@stevens-tech.edu> and 9965 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 9966 9967 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME 9968 was empty. 9969 [Steve Henson] 9970 [This change does not apply to 0.9.7.] 9971 9972 *) Use the cached encoding of an X509_NAME structure rather than 9973 copying it. This is apparently the reason for the libsafe "errors" 9974 but the code is actually correct. 9975 [Steve Henson] 9976 9977 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent 9978 Bleichenbacher's DSA attack. 9979 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits 9980 to be set and top=0 forces the highest bit to be set; top=-1 is new 9981 and leaves the highest bit random. 9982 [Ulf Moeller, Bodo Moeller] 9983 9984 *) In the NCONF_...-based implementations for CONF_... queries 9985 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using 9986 a temporary CONF structure with the data component set to NULL 9987 (which gives segmentation faults in lh_retrieve). 9988 Instead, use NULL for the CONF pointer in CONF_get_string and 9989 CONF_get_number (which may use environment variables) and directly 9990 return NULL from CONF_get_section. 9991 [Bodo Moeller] 9992 9993 *) Fix potential buffer overrun for EBCDIC. 9994 [Ulf Moeller] 9995 9996 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign 9997 keyUsage if basicConstraints absent for a CA. 9998 [Steve Henson] 9999 10000 *) Make SMIME_write_PKCS7() write mail header values with a format that 10001 is more generally accepted (no spaces before the semicolon), since 10002 some programs can't parse those values properly otherwise. Also make 10003 sure BIO's that break lines after each write do not create invalid 10004 headers. 10005 [Richard Levitte] 10006 10007 *) Make the CRL encoding routines work with empty SEQUENCE OF. The 10008 macros previously used would not encode an empty SEQUENCE OF 10009 and break the signature. 10010 [Steve Henson] 10011 [This change does not apply to 0.9.7.] 10012 10013 *) Zero the premaster secret after deriving the master secret in 10014 DH ciphersuites. 10015 [Steve Henson] 10016 10017 *) Add some EVP_add_digest_alias registrations (as found in 10018 OpenSSL_add_all_digests()) to SSL_library_init() 10019 aka OpenSSL_add_ssl_algorithms(). This provides improved 10020 compatibility with peers using X.509 certificates 10021 with unconventional AlgorithmIdentifier OIDs. 10022 [Bodo Moeller] 10023 10024 *) Fix for Irix with NO_ASM. 10025 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>] 10026 10027 *) ./config script fixes. 10028 [Ulf Moeller, Richard Levitte] 10029 10030 *) Fix 'openssl passwd -1'. 10031 [Bodo Moeller] 10032 10033 *) Change PKCS12_key_gen_asc() so it can cope with non null 10034 terminated strings whose length is passed in the passlen 10035 parameter, for example from PEM callbacks. This was done 10036 by adding an extra length parameter to asc2uni(). 10037 [Steve Henson, reported by <oddissey@samsung.co.kr>] 10038 10039 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn 10040 call failed, free the DSA structure. 10041 [Bodo Moeller] 10042 10043 *) Fix to uni2asc() to cope with zero length Unicode strings. 10044 These are present in some PKCS#12 files. 10045 [Steve Henson] 10046 10047 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c). 10048 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits 10049 when writing a 32767 byte record. 10050 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>] 10051 10052 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c), 10053 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}. 10054 10055 (RSA objects have a reference count access to which is protected 10056 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c], 10057 so they are meant to be shared between threads.) 10058 [Bodo Moeller, Geoff Thorpe; original patch submitted by 10059 "Reddie, Steven" <Steven.Reddie@ca.com>] 10060 10061 *) Fix a deadlock in CRYPTO_mem_leaks(). 10062 [Bodo Moeller] 10063 10064 *) Use better test patterns in bntest. 10065 [Ulf Möller] 10066 10067 *) rand_win.c fix for Borland C. 10068 [Ulf Möller] 10069 10070 *) BN_rshift bugfix for n == 0. 10071 [Bodo Moeller] 10072 10073 *) Add a 'bctest' script that checks for some known 'bc' bugs 10074 so that 'make test' does not abort just because 'bc' is broken. 10075 [Bodo Moeller] 10076 10077 *) Store verify_result within SSL_SESSION also for client side to 10078 avoid potential security hole. (Re-used sessions on the client side 10079 always resulted in verify_result==X509_V_OK, not using the original 10080 result of the server certificate verification.) 10081 [Lutz Jaenicke] 10082 10083 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type 10084 SSL3_RT_APPLICATION_DATA, return 0. 10085 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true. 10086 [Bodo Moeller] 10087 10088 *) Fix SSL_peek: 10089 Both ssl2_peek and ssl3_peek, which were totally broken in earlier 10090 releases, have been re-implemented by renaming the previous 10091 implementations of ssl2_read and ssl3_read to ssl2_read_internal 10092 and ssl3_read_internal, respectively, and adding 'peek' parameters 10093 to them. The new ssl[23]_{read,peek} functions are calls to 10094 ssl[23]_read_internal with the 'peek' flag set appropriately. 10095 A 'peek' parameter has also been added to ssl3_read_bytes, which 10096 does the actual work for ssl3_read_internal. 10097 [Bodo Moeller] 10098 10099 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling 10100 the method-specific "init()" handler. Also clean up ex_data after 10101 calling the method-specific "finish()" handler. Previously, this was 10102 happening the other way round. 10103 [Geoff Thorpe] 10104 10105 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16. 10106 The previous value, 12, was not always sufficient for BN_mod_exp(). 10107 [Bodo Moeller] 10108 10109 *) Make sure that shared libraries get the internal name engine with 10110 the full version number and not just 0. This should mark the 10111 shared libraries as not backward compatible. Of course, this should 10112 be changed again when we can guarantee backward binary compatibility. 10113 [Richard Levitte] 10114 10115 *) Fix typo in get_cert_by_subject() in by_dir.c 10116 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>] 10117 10118 *) Rework the system to generate shared libraries: 10119 10120 - Make note of the expected extension for the shared libraries and 10121 if there is a need for symbolic links from for example libcrypto.so.0 10122 to libcrypto.so.0.9.7. There is extended info in Configure for 10123 that. 10124 10125 - Make as few rebuilds of the shared libraries as possible. 10126 10127 - Still avoid linking the OpenSSL programs with the shared libraries. 10128 10129 - When installing, install the shared libraries separately from the 10130 static ones. 10131 [Richard Levitte] 10132 10133 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument. 10134 10135 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new 10136 and not in SSL_clear because the latter is also used by the 10137 accept/connect functions; previously, the settings made by 10138 SSL_set_read_ahead would be lost during the handshake. 10139 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>] 10140 10141 *) Correct util/mkdef.pl to be selective about disabled algorithms. 10142 Previously, it would create entries for disabled algorithms no 10143 matter what. 10144 [Richard Levitte] 10145 10146 *) Added several new manual pages for SSL_* function. 10147 [Lutz Jaenicke] 10148 10149 Changes between 0.9.5a and 0.9.6 [24 Sep 2000] 10150 10151 *) In ssl23_get_client_hello, generate an error message when faced 10152 with an initial SSL 3.0/TLS record that is too small to contain the 10153 first two bytes of the ClientHello message, i.e. client_version. 10154 (Note that this is a pathologic case that probably has never happened 10155 in real life.) The previous approach was to use the version number 10156 from the record header as a substitute; but our protocol choice 10157 should not depend on that one because it is not authenticated 10158 by the Finished messages. 10159 [Bodo Moeller] 10160 10161 *) More robust randomness gathering functions for Windows. 10162 [Jeffrey Altman <jaltman@columbia.edu>] 10163 10164 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is 10165 not set then we don't setup the error code for issuer check errors 10166 to avoid possibly overwriting other errors which the callback does 10167 handle. If an application does set the flag then we assume it knows 10168 what it is doing and can handle the new informational codes 10169 appropriately. 10170 [Steve Henson] 10171 10172 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for 10173 a general "ANY" type, as such it should be able to decode anything 10174 including tagged types. However it didn't check the class so it would 10175 wrongly interpret tagged types in the same way as their universal 10176 counterpart and unknown types were just rejected. Changed so that the 10177 tagged and unknown types are handled in the same way as a SEQUENCE: 10178 that is the encoding is stored intact. There is also a new type 10179 "V_ASN1_OTHER" which is used when the class is not universal, in this 10180 case we have no idea what the actual type is so we just lump them all 10181 together. 10182 [Steve Henson] 10183 10184 *) On VMS, stdout may very well lead to a file that is written to 10185 in a record-oriented fashion. That means that every write() will 10186 write a separate record, which will be read separately by the 10187 programs trying to read from it. This can be very confusing. 10188 10189 The solution is to put a BIO filter in the way that will buffer 10190 text until a linefeed is reached, and then write everything a 10191 line at a time, so every record written will be an actual line, 10192 not chunks of lines and not (usually doesn't happen, but I've 10193 seen it once) several lines in one record. BIO_f_linebuffer() is 10194 the answer. 10195 10196 Currently, it's a VMS-only method, because that's where it has 10197 been tested well enough. 10198 [Richard Levitte] 10199 10200 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery, 10201 it can return incorrect results. 10202 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a, 10203 but it was in 0.9.6-beta[12].) 10204 [Bodo Moeller] 10205 10206 *) Disable the check for content being present when verifying detached 10207 signatures in pk7_smime.c. Some versions of Netscape (wrongly) 10208 include zero length content when signing messages. 10209 [Steve Henson] 10210 10211 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR 10212 BIO_ctrl (for BIO pairs). 10213 [Bodo Möller] 10214 10215 *) Add DSO method for VMS. 10216 [Richard Levitte] 10217 10218 *) Bug fix: Montgomery multiplication could produce results with the 10219 wrong sign. 10220 [Ulf Möller] 10221 10222 *) Add RPM specification openssl.spec and modify it to build three 10223 packages. The default package contains applications, application 10224 documentation and run-time libraries. The devel package contains 10225 include files, static libraries and function documentation. The 10226 doc package contains the contents of the doc directory. The original 10227 openssl.spec was provided by Damien Miller <djm@mindrot.org>. 10228 [Richard Levitte] 10229 10230 *) Add a large number of documentation files for many SSL routines. 10231 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 10232 10233 *) Add a configuration entry for Sony News 4. 10234 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>] 10235 10236 *) Don't set the two most significant bits to one when generating a 10237 random number < q in the DSA library. 10238 [Ulf Möller] 10239 10240 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default 10241 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if 10242 the underlying transport is blocking) if a handshake took place. 10243 (The default behaviour is needed by applications such as s_client 10244 and s_server that use select() to determine when to use SSL_read; 10245 but for applications that know in advance when to expect data, it 10246 just makes things more complicated.) 10247 [Bodo Moeller] 10248 10249 *) Add RAND_egd_bytes(), which gives control over the number of bytes read 10250 from EGD. 10251 [Ben Laurie] 10252 10253 *) Add a few more EBCDIC conditionals that make `req' and `x509' 10254 work better on such systems. 10255 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 10256 10257 *) Add two demo programs for PKCS12_parse() and PKCS12_create(). 10258 Update PKCS12_parse() so it copies the friendlyName and the 10259 keyid to the certificates aux info. 10260 [Steve Henson] 10261 10262 *) Fix bug in PKCS7_verify() which caused an infinite loop 10263 if there was more than one signature. 10264 [Sven Uszpelkat <su@celocom.de>] 10265 10266 *) Major change in util/mkdef.pl to include extra information 10267 about each symbol, as well as presenting variables as well 10268 as functions. This change means that there's n more need 10269 to rebuild the .num files when some algorithms are excluded. 10270 [Richard Levitte] 10271 10272 *) Allow the verify time to be set by an application, 10273 rather than always using the current time. 10274 [Steve Henson] 10275 10276 *) Phase 2 verify code reorganisation. The certificate 10277 verify code now looks up an issuer certificate by a 10278 number of criteria: subject name, authority key id 10279 and key usage. It also verifies self signed certificates 10280 by the same criteria. The main comparison function is 10281 X509_check_issued() which performs these checks. 10282 10283 Lot of changes were necessary in order to support this 10284 without completely rewriting the lookup code. 10285 10286 Authority and subject key identifier are now cached. 10287 10288 The LHASH 'certs' is X509_STORE has now been replaced 10289 by a STACK_OF(X509_OBJECT). This is mainly because an 10290 LHASH can't store or retrieve multiple objects with 10291 the same hash value. 10292 10293 As a result various functions (which were all internal 10294 use only) have changed to handle the new X509_STORE 10295 structure. This will break anything that messed round 10296 with X509_STORE internally. 10297 10298 The functions X509_STORE_add_cert() now checks for an 10299 exact match, rather than just subject name. 10300 10301 The X509_STORE API doesn't directly support the retrieval 10302 of multiple certificates matching a given criteria, however 10303 this can be worked round by performing a lookup first 10304 (which will fill the cache with candidate certificates) 10305 and then examining the cache for matches. This is probably 10306 the best we can do without throwing out X509_LOOKUP 10307 entirely (maybe later...). 10308 10309 The X509_VERIFY_CTX structure has been enhanced considerably. 10310 10311 All certificate lookup operations now go via a get_issuer() 10312 callback. Although this currently uses an X509_STORE it 10313 can be replaced by custom lookups. This is a simple way 10314 to bypass the X509_STORE hackery necessary to make this 10315 work and makes it possible to use more efficient techniques 10316 in future. A very simple version which uses a simple 10317 STACK for its trusted certificate store is also provided 10318 using X509_STORE_CTX_trusted_stack(). 10319 10320 The verify_cb() and verify() callbacks now have equivalents 10321 in the X509_STORE_CTX structure. 10322 10323 X509_STORE_CTX also has a 'flags' field which can be used 10324 to customise the verify behaviour. 10325 [Steve Henson] 10326 10327 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 10328 excludes S/MIME capabilities. 10329 [Steve Henson] 10330 10331 *) When a certificate request is read in keep a copy of the 10332 original encoding of the signed data and use it when outputting 10333 again. Signatures then use the original encoding rather than 10334 a decoded, encoded version which may cause problems if the 10335 request is improperly encoded. 10336 [Steve Henson] 10337 10338 *) For consistency with other BIO_puts implementations, call 10339 buffer_write(b, ...) directly in buffer_puts instead of calling 10340 BIO_write(b, ...). 10341 10342 In BIO_puts, increment b->num_write as in BIO_write. 10343 [Peter.Sylvester@EdelWeb.fr] 10344 10345 *) Fix BN_mul_word for the case where the word is 0. (We have to use 10346 BN_zero, we may not return a BIGNUM with an array consisting of 10347 words set to zero.) 10348 [Bodo Moeller] 10349 10350 *) Avoid calling abort() from within the library when problems are 10351 detected, except if preprocessor symbols have been defined 10352 (such as REF_CHECK, BN_DEBUG etc.). 10353 [Bodo Moeller] 10354 10355 *) New openssl application 'rsautl'. This utility can be 10356 used for low level RSA operations. DER public key 10357 BIO/fp routines also added. 10358 [Steve Henson] 10359 10360 *) New Configure entry and patches for compiling on QNX 4. 10361 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>] 10362 10363 *) A demo state-machine implementation was sponsored by 10364 Nuron (http://www.nuron.com/) and is now available in 10365 demos/state_machine. 10366 [Ben Laurie] 10367 10368 *) New options added to the 'dgst' utility for signature 10369 generation and verification. 10370 [Steve Henson] 10371 10372 *) Unrecognized PKCS#7 content types are now handled via a 10373 catch all ASN1_TYPE structure. This allows unsupported 10374 types to be stored as a "blob" and an application can 10375 encode and decode it manually. 10376 [Steve Henson] 10377 10378 *) Fix various signed/unsigned issues to make a_strex.c 10379 compile under VC++. 10380 [Oscar Jacobsson <oscar.jacobsson@celocom.com>] 10381 10382 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct 10383 length if passed a buffer. ASN1_INTEGER_to_BN failed 10384 if passed a NULL BN and its argument was negative. 10385 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>] 10386 10387 *) Modification to PKCS#7 encoding routines to output definite 10388 length encoding. Since currently the whole structures are in 10389 memory there's not real point in using indefinite length 10390 constructed encoding. However if OpenSSL is compiled with 10391 the flag PKCS7_INDEFINITE_ENCODING the old form is used. 10392 [Steve Henson] 10393 10394 *) Added BIO_vprintf() and BIO_vsnprintf(). 10395 [Richard Levitte] 10396 10397 *) Added more prefixes to parse for in the strings written 10398 through a logging bio, to cover all the levels that are available 10399 through syslog. The prefixes are now: 10400 10401 PANIC, EMERG, EMR => LOG_EMERG 10402 ALERT, ALR => LOG_ALERT 10403 CRIT, CRI => LOG_CRIT 10404 ERROR, ERR => LOG_ERR 10405 WARNING, WARN, WAR => LOG_WARNING 10406 NOTICE, NOTE, NOT => LOG_NOTICE 10407 INFO, INF => LOG_INFO 10408 DEBUG, DBG => LOG_DEBUG 10409 10410 and as before, if none of those prefixes are present at the 10411 beginning of the string, LOG_ERR is chosen. 10412 10413 On Win32, the LOG_* levels are mapped according to this: 10414 10415 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE 10416 LOG_WARNING => EVENTLOG_WARNING_TYPE 10417 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE 10418 10419 [Richard Levitte] 10420 10421 *) Made it possible to reconfigure with just the configuration 10422 argument "reconf" or "reconfigure". The command line arguments 10423 are stored in Makefile.ssl in the variable CONFIGURE_ARGS, 10424 and are retrieved from there when reconfiguring. 10425 [Richard Levitte] 10426 10427 *) MD4 implemented. 10428 [Assar Westerlund <assar@sics.se>, Richard Levitte] 10429 10430 *) Add the arguments -CAfile and -CApath to the pkcs12 utility. 10431 [Richard Levitte] 10432 10433 *) The obj_dat.pl script was messing up the sorting of object 10434 names. The reason was that it compared the quoted version 10435 of strings as a result "OCSP" > "OCSP Signing" because 10436 " > SPACE. Changed script to store unquoted versions of 10437 names and add quotes on output. It was also omitting some 10438 names from the lookup table if they were given a default 10439 value (that is if SN is missing it is given the same 10440 value as LN and vice versa), these are now added on the 10441 grounds that if an object has a name we should be able to 10442 look it up. Finally added warning output when duplicate 10443 short or long names are found. 10444 [Steve Henson] 10445 10446 *) Changes needed for Tandem NSK. 10447 [Scott Uroff <scott@xypro.com>] 10448 10449 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in 10450 RSA_padding_check_SSLv23(), special padding was never detected 10451 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol 10452 version rollback attacks was not effective. 10453 10454 In s23_clnt.c, don't use special rollback-attack detection padding 10455 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the 10456 client; similarly, in s23_srvr.c, don't do the rollback check if 10457 SSL 2.0 is the only protocol enabled in the server. 10458 [Bodo Moeller] 10459 10460 *) Make it possible to get hexdumps of unprintable data with 'openssl 10461 asn1parse'. By implication, the functions ASN1_parse_dump() and 10462 BIO_dump_indent() are added. 10463 [Richard Levitte] 10464 10465 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex() 10466 these print out strings and name structures based on various 10467 flags including RFC2253 support and proper handling of 10468 multibyte characters. Added options to the 'x509' utility 10469 to allow the various flags to be set. 10470 [Steve Henson] 10471 10472 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME. 10473 Also change the functions X509_cmp_current_time() and 10474 X509_gmtime_adj() work with an ASN1_TIME structure, 10475 this will enable certificates using GeneralizedTime in validity 10476 dates to be checked. 10477 [Steve Henson] 10478 10479 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid 10480 negative public key encodings) on by default, 10481 NO_NEG_PUBKEY_BUG can be set to disable it. 10482 [Steve Henson] 10483 10484 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT 10485 content octets. An i2c_ASN1_OBJECT is unnecessary because 10486 the encoding can be trivially obtained from the structure. 10487 [Steve Henson] 10488 10489 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock), 10490 not read locks (CRYPTO_r_[un]lock). 10491 [Bodo Moeller] 10492 10493 *) A first attempt at creating official support for shared 10494 libraries through configuration. I've kept it so the 10495 default is static libraries only, and the OpenSSL programs 10496 are always statically linked for now, but there are 10497 preparations for dynamic linking in place. 10498 This has been tested on Linux and Tru64. 10499 [Richard Levitte] 10500 10501 *) Randomness polling function for Win9x, as described in: 10502 Peter Gutmann, Software Generation of Practically Strong 10503 Random Numbers. 10504 [Ulf Möller] 10505 10506 *) Fix so PRNG is seeded in req if using an already existing 10507 DSA key. 10508 [Steve Henson] 10509 10510 *) New options to smime application. -inform and -outform 10511 allow alternative formats for the S/MIME message including 10512 PEM and DER. The -content option allows the content to be 10513 specified separately. This should allow things like Netscape 10514 form signing output easier to verify. 10515 [Steve Henson] 10516 10517 *) Fix the ASN1 encoding of tags using the 'long form'. 10518 [Steve Henson] 10519 10520 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT 10521 STRING types. These convert content octets to and from the 10522 underlying type. The actual tag and length octets are 10523 already assumed to have been read in and checked. These 10524 are needed because all other string types have virtually 10525 identical handling apart from the tag. By having versions 10526 of the ASN1 functions that just operate on content octets 10527 IMPLICIT tagging can be handled properly. It also allows 10528 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED 10529 and ASN1_INTEGER are identical apart from the tag. 10530 [Steve Henson] 10531 10532 *) Change the handling of OID objects as follows: 10533 10534 - New object identifiers are inserted in objects.txt, following 10535 the syntax given in objects.README. 10536 - objects.pl is used to process obj_mac.num and create a new 10537 obj_mac.h. 10538 - obj_dat.pl is used to create a new obj_dat.h, using the data in 10539 obj_mac.h. 10540 10541 This is currently kind of a hack, and the perl code in objects.pl 10542 isn't very elegant, but it works as I intended. The simplest way 10543 to check that it worked correctly is to look in obj_dat.h and 10544 check the array nid_objs and make sure the objects haven't moved 10545 around (this is important!). Additions are OK, as well as 10546 consistent name changes. 10547 [Richard Levitte] 10548 10549 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1'). 10550 [Bodo Moeller] 10551 10552 *) Addition of the command line parameter '-rand file' to 'openssl req'. 10553 The given file adds to whatever has already been seeded into the 10554 random pool through the RANDFILE configuration file option or 10555 environment variable, or the default random state file. 10556 [Richard Levitte] 10557 10558 *) mkstack.pl now sorts each macro group into lexical order. 10559 Previously the output order depended on the order the files 10560 appeared in the directory, resulting in needless rewriting 10561 of safestack.h . 10562 [Steve Henson] 10563 10564 *) Patches to make OpenSSL compile under Win32 again. Mostly 10565 work arounds for the VC++ problem that it treats func() as 10566 func(void). Also stripped out the parts of mkdef.pl that 10567 added extra typesafe functions: these no longer exist. 10568 [Steve Henson] 10569 10570 *) Reorganisation of the stack code. The macros are now all 10571 collected in safestack.h . Each macro is defined in terms of 10572 a "stack macro" of the form SKM_<name>(type, a, b). The 10573 DEBUG_SAFESTACK is now handled in terms of function casts, 10574 this has the advantage of retaining type safety without the 10575 use of additional functions. If DEBUG_SAFESTACK is not defined 10576 then the non typesafe macros are used instead. Also modified the 10577 mkstack.pl script to handle the new form. Needs testing to see 10578 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK 10579 the default if no major problems. Similar behaviour for ASN1_SET_OF 10580 and PKCS12_STACK_OF. 10581 [Steve Henson] 10582 10583 *) When some versions of IIS use the 'NET' form of private key the 10584 key derivation algorithm is different. Normally MD5(password) is 10585 used as a 128 bit RC4 key. In the modified case 10586 MD5(MD5(password) + "SGCKEYSALT") is used instead. Added some 10587 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same 10588 as the old Netscape_RSA functions except they have an additional 10589 'sgckey' parameter which uses the modified algorithm. Also added 10590 an -sgckey command line option to the rsa utility. Thanks to 10591 Adrian Peck <bertie@ncipher.com> for posting details of the modified 10592 algorithm to openssl-dev. 10593 [Steve Henson] 10594 10595 *) The evp_local.h macros were using 'c.##kname' which resulted in 10596 invalid expansion on some systems (SCO 5.0.5 for example). 10597 Corrected to 'c.kname'. 10598 [Phillip Porch <root@theporch.com>] 10599 10600 *) New X509_get1_email() and X509_REQ_get1_email() functions that return 10601 a STACK of email addresses from a certificate or request, these look 10602 in the subject name and the subject alternative name extensions and 10603 omit any duplicate addresses. 10604 [Steve Henson] 10605 10606 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows. 10607 This makes DSA verification about 2 % faster. 10608 [Bodo Moeller] 10609 10610 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5 10611 (meaning that now 2^5 values will be precomputed, which is only 4 KB 10612 plus overhead for 1024 bit moduli). 10613 This makes exponentiations about 0.5 % faster for 1024 bit 10614 exponents (as measured by "openssl speed rsa2048"). 10615 [Bodo Moeller] 10616 10617 *) Rename memory handling macros to avoid conflicts with other 10618 software: 10619 Malloc => OPENSSL_malloc 10620 Malloc_locked => OPENSSL_malloc_locked 10621 Realloc => OPENSSL_realloc 10622 Free => OPENSSL_free 10623 [Richard Levitte] 10624 10625 *) New function BN_mod_exp_mont_word for small bases (roughly 15% 10626 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange). 10627 [Bodo Moeller] 10628 10629 *) CygWin32 support. 10630 [John Jarvie <jjarvie@newsguy.com>] 10631 10632 *) The type-safe stack code has been rejigged. It is now only compiled 10633 in when OpenSSL is configured with the DEBUG_SAFESTACK option and 10634 by default all type-specific stack functions are "#define"d back to 10635 standard stack functions. This results in more streamlined output 10636 but retains the type-safety checking possibilities of the original 10637 approach. 10638 [Geoff Thorpe] 10639 10640 *) The STACK code has been cleaned up, and certain type declarations 10641 that didn't make a lot of sense have been brought in line. This has 10642 also involved a cleanup of sorts in safestack.h to more correctly 10643 map type-safe stack functions onto their plain stack counterparts. 10644 This work has also resulted in a variety of "const"ifications of 10645 lots of the code, especially "_cmp" operations which should normally 10646 be prototyped with "const" parameters anyway. 10647 [Geoff Thorpe] 10648 10649 *) When generating bytes for the first time in md_rand.c, 'stir the pool' 10650 by seeding with STATE_SIZE dummy bytes (with zero entropy count). 10651 (The PRNG state consists of two parts, the large pool 'state' and 'md', 10652 where all of 'md' is used each time the PRNG is used, but 'state' 10653 is used only indexed by a cyclic counter. As entropy may not be 10654 well distributed from the beginning, 'md' is important as a 10655 chaining variable. However, the output function chains only half 10656 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains 10657 all of 'md', and seeding with STATE_SIZE dummy bytes will result 10658 in all of 'state' being rewritten, with the new values depending 10659 on virtually all of 'md'. This overcomes the 80 bit limitation.) 10660 [Bodo Moeller] 10661 10662 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when 10663 the handshake is continued after ssl_verify_cert_chain(); 10664 otherwise, if SSL_VERIFY_NONE is set, remaining error codes 10665 can lead to 'unexplainable' connection aborts later. 10666 [Bodo Moeller; problem tracked down by Lutz Jaenicke] 10667 10668 *) Major EVP API cipher revision. 10669 Add hooks for extra EVP features. This allows various cipher 10670 parameters to be set in the EVP interface. Support added for variable 10671 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and 10672 setting of RC2 and RC5 parameters. 10673 10674 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length 10675 ciphers. 10676 10677 Remove lots of duplicated code from the EVP library. For example *every* 10678 cipher init() function handles the 'iv' in the same way according to the 10679 cipher mode. They also all do nothing if the 'key' parameter is NULL and 10680 for CFB and OFB modes they zero ctx->num. 10681 10682 New functionality allows removal of S/MIME code RC2 hack. 10683 10684 Most of the routines have the same form and so can be declared in terms 10685 of macros. 10686 10687 By shifting this to the top level EVP_CipherInit() it can be removed from 10688 all individual ciphers. If the cipher wants to handle IVs or keys 10689 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT 10690 flags. 10691 10692 Change lots of functions like EVP_EncryptUpdate() to now return a 10693 value: although software versions of the algorithms cannot fail 10694 any installed hardware versions can. 10695 [Steve Henson] 10696 10697 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if 10698 this option is set, tolerate broken clients that send the negotiated 10699 protocol version number instead of the requested protocol version 10700 number. 10701 [Bodo Moeller] 10702 10703 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag; 10704 i.e. non-zero for export ciphersuites, zero otherwise. 10705 Previous versions had this flag inverted, inconsistent with 10706 rsa_tmp_cb (..._TMP_RSA_CB). 10707 [Bodo Moeller; problem reported by Amit Chopra] 10708 10709 *) Add missing DSA library text string. Work around for some IIS 10710 key files with invalid SEQUENCE encoding. 10711 [Steve Henson] 10712 10713 *) Add a document (doc/standards.txt) that list all kinds of standards 10714 and so on that are implemented in OpenSSL. 10715 [Richard Levitte] 10716 10717 *) Enhance c_rehash script. Old version would mishandle certificates 10718 with the same subject name hash and wouldn't handle CRLs at all. 10719 Added -fingerprint option to crl utility, to support new c_rehash 10720 features. 10721 [Steve Henson] 10722 10723 *) Eliminate non-ANSI declarations in crypto.h and stack.h. 10724 [Ulf Möller] 10725 10726 *) Fix for SSL server purpose checking. Server checking was 10727 rejecting certificates which had extended key usage present 10728 but no ssl client purpose. 10729 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>] 10730 10731 *) Make PKCS#12 code work with no password. The PKCS#12 spec 10732 is a little unclear about how a blank password is handled. 10733 Since the password in encoded as a BMPString with terminating 10734 double NULL a zero length password would end up as just the 10735 double NULL. However no password at all is different and is 10736 handled differently in the PKCS#12 key generation code. NS 10737 treats a blank password as zero length. MSIE treats it as no 10738 password on export: but it will try both on import. We now do 10739 the same: PKCS12_parse() tries zero length and no password if 10740 the password is set to "" or NULL (NULL is now a valid password: 10741 it wasn't before) as does the pkcs12 application. 10742 [Steve Henson] 10743 10744 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use 10745 perror when PEM_read_bio_X509_REQ fails, the error message must 10746 be obtained from the error queue. 10747 [Bodo Moeller] 10748 10749 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing 10750 it in ERR_remove_state if appropriate, and change ERR_get_state 10751 accordingly to avoid race conditions (this is necessary because 10752 thread_hash is no longer constant once set). 10753 [Bodo Moeller] 10754 10755 *) Bugfix for linux-elf makefile.one. 10756 [Ulf Möller] 10757 10758 *) RSA_get_default_method() will now cause a default 10759 RSA_METHOD to be chosen if one doesn't exist already. 10760 Previously this was only set during a call to RSA_new() 10761 or RSA_new_method(NULL) meaning it was possible for 10762 RSA_get_default_method() to return NULL. 10763 [Geoff Thorpe] 10764 10765 *) Added native name translation to the existing DSO code 10766 that will convert (if the flag to do so is set) filenames 10767 that are sufficiently small and have no path information 10768 into a canonical native form. Eg. "blah" converted to 10769 "libblah.so" or "blah.dll" etc. 10770 [Geoff Thorpe] 10771 10772 *) New function ERR_error_string_n(e, buf, len) which is like 10773 ERR_error_string(e, buf), but writes at most 'len' bytes 10774 including the 0 terminator. For ERR_error_string_n, 'buf' 10775 may not be NULL. 10776 [Damien Miller <djm@mindrot.org>, Bodo Moeller] 10777 10778 *) CONF library reworked to become more general. A new CONF 10779 configuration file reader "class" is implemented as well as a 10780 new functions (NCONF_*, for "New CONF") to handle it. The now 10781 old CONF_* functions are still there, but are reimplemented to 10782 work in terms of the new functions. Also, a set of functions 10783 to handle the internal storage of the configuration data is 10784 provided to make it easier to write new configuration file 10785 reader "classes" (I can definitely see something reading a 10786 configuration file in XML format, for example), called _CONF_*, 10787 or "the configuration storage API"... 10788 10789 The new configuration file reading functions are: 10790 10791 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio, 10792 NCONF_get_section, NCONF_get_string, NCONF_get_numbre 10793 10794 NCONF_default, NCONF_WIN32 10795 10796 NCONF_dump_fp, NCONF_dump_bio 10797 10798 NCONF_default and NCONF_WIN32 are method (or "class") choosers, 10799 NCONF_new creates a new CONF object. This works in the same way 10800 as other interfaces in OpenSSL, like the BIO interface. 10801 NCONF_dump_* dump the internal storage of the configuration file, 10802 which is useful for debugging. All other functions take the same 10803 arguments as the old CONF_* functions with the exception of the 10804 first that must be a `CONF *' instead of a `LHASH *'. 10805 10806 To make it easier to use the new classes with the old CONF_* functions, 10807 the function CONF_set_default_method is provided. 10808 [Richard Levitte] 10809 10810 *) Add '-tls1' option to 'openssl ciphers', which was already 10811 mentioned in the documentation but had not been implemented. 10812 (This option is not yet really useful because even the additional 10813 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.) 10814 [Bodo Moeller] 10815 10816 *) Initial DSO code added into libcrypto for letting OpenSSL (and 10817 OpenSSL-based applications) load shared libraries and bind to 10818 them in a portable way. 10819 [Geoff Thorpe, with contributions from Richard Levitte] 10820 10821 Changes between 0.9.5 and 0.9.5a [1 Apr 2000] 10822 10823 *) Make sure _lrotl and _lrotr are only used with MSVC. 10824 10825 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status 10826 (the default implementation of RAND_status). 10827 10828 *) Rename openssl x509 option '-crlext', which was added in 0.9.5, 10829 to '-clrext' (= clear extensions), as intended and documented. 10830 [Bodo Moeller; inconsistency pointed out by Michael Attili 10831 <attili@amaxo.com>] 10832 10833 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length 10834 was larger than the MD block size. 10835 [Steve Henson, pointed out by Yost William <YostW@tce.com>] 10836 10837 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument 10838 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set() 10839 using the passed key: if the passed key was a private key the result 10840 of X509_print(), for example, would be to print out all the private key 10841 components. 10842 [Steve Henson] 10843 10844 *) des_quad_cksum() byte order bug fix. 10845 [Ulf Möller, using the problem description in krb4-0.9.7, where 10846 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>] 10847 10848 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly 10849 discouraged. 10850 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>] 10851 10852 *) For easily testing in shell scripts whether some command 10853 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX' 10854 returns with exit code 0 iff no command of the given name is available. 10855 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases, 10856 the output goes to stdout and nothing is printed to stderr. 10857 Additional arguments are always ignored. 10858 10859 Since for each cipher there is a command of the same name, 10860 the 'no-cipher' compilation switches can be tested this way. 10861 10862 ('openssl no-XXX' is not able to detect pseudo-commands such 10863 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.) 10864 [Bodo Moeller] 10865 10866 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration. 10867 [Bodo Moeller] 10868 10869 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE 10870 is set; it will be thrown away anyway because each handshake creates 10871 its own key. 10872 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition 10873 to parameters -- in previous versions (since OpenSSL 0.9.3) the 10874 'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning 10875 you effectively got SSL_OP_SINGLE_DH_USE when using this macro. 10876 [Bodo Moeller] 10877 10878 *) New s_client option -ign_eof: EOF at stdin is ignored, and 10879 'Q' and 'R' lose their special meanings (quit/renegotiate). 10880 This is part of what -quiet does; unlike -quiet, -ign_eof 10881 does not suppress any output. 10882 [Richard Levitte] 10883 10884 *) Add compatibility options to the purpose and trust code. The 10885 purpose X509_PURPOSE_ANY is "any purpose" which automatically 10886 accepts a certificate or CA, this was the previous behaviour, 10887 with all the associated security issues. 10888 10889 X509_TRUST_COMPAT is the old trust behaviour: only and 10890 automatically trust self signed roots in certificate store. A 10891 new trust setting X509_TRUST_DEFAULT is used to specify that 10892 a purpose has no associated trust setting and it should instead 10893 use the value in the default purpose. 10894 [Steve Henson] 10895 10896 *) Fix the PKCS#8 DSA private key code so it decodes keys again 10897 and fix a memory leak. 10898 [Steve Henson] 10899 10900 *) In util/mkerr.pl (which implements 'make errors'), preserve 10901 reason strings from the previous version of the .c file, as 10902 the default to have only downcase letters (and digits) in 10903 automatically generated reasons codes is not always appropriate. 10904 [Bodo Moeller] 10905 10906 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table 10907 using strerror. Previously, ERR_reason_error_string() returned 10908 library names as reason strings for SYSerr; but SYSerr is a special 10909 case where small numbers are errno values, not library numbers. 10910 [Bodo Moeller] 10911 10912 *) Add '-dsaparam' option to 'openssl dhparam' application. This 10913 converts DSA parameters into DH parameters. (When creating parameters, 10914 DSA_generate_parameters is used.) 10915 [Bodo Moeller] 10916 10917 *) Include 'length' (recommended exponent length) in C code generated 10918 by 'openssl dhparam -C'. 10919 [Bodo Moeller] 10920 10921 *) The second argument to set_label in perlasm was already being used 10922 so couldn't be used as a "file scope" flag. Moved to third argument 10923 which was free. 10924 [Steve Henson] 10925 10926 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes 10927 instead of RAND_bytes for encryption IVs and salts. 10928 [Bodo Moeller] 10929 10930 *) Include RAND_status() into RAND_METHOD instead of implementing 10931 it only for md_rand.c Otherwise replacing the PRNG by calling 10932 RAND_set_rand_method would be impossible. 10933 [Bodo Moeller] 10934 10935 *) Don't let DSA_generate_key() enter an infinite loop if the random 10936 number generation fails. 10937 [Bodo Moeller] 10938 10939 *) New 'rand' application for creating pseudo-random output. 10940 [Bodo Moeller] 10941 10942 *) Added configuration support for Linux/IA64 10943 [Rolf Haberrecker <rolf@suse.de>] 10944 10945 *) Assembler module support for Mingw32. 10946 [Ulf Möller] 10947 10948 *) Shared library support for HPUX (in shlib/). 10949 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous] 10950 10951 *) Shared library support for Solaris gcc. 10952 [Lutz Behnke <behnke@trustcenter.de>] 10953 10954 Changes between 0.9.4 and 0.9.5 [28 Feb 2000] 10955 10956 *) PKCS7_encrypt() was adding text MIME headers twice because they 10957 were added manually and by SMIME_crlf_copy(). 10958 [Steve Henson] 10959 10960 *) In bntest.c don't call BN_rand with zero bits argument. 10961 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>] 10962 10963 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n] 10964 case was implemented. This caused BN_div_recp() to fail occasionally. 10965 [Ulf Möller] 10966 10967 *) Add an optional second argument to the set_label() in the perl 10968 assembly language builder. If this argument exists and is set 10969 to 1 it signals that the assembler should use a symbol whose 10970 scope is the entire file, not just the current function. This 10971 is needed with MASM which uses the format label:: for this scope. 10972 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>] 10973 10974 *) Change the ASN1 types so they are typedefs by default. Before 10975 almost all types were #define'd to ASN1_STRING which was causing 10976 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING) 10977 for example. 10978 [Steve Henson] 10979 10980 *) Change names of new functions to the new get1/get0 naming 10981 convention: After 'get1', the caller owns a reference count 10982 and has to call ..._free; 'get0' returns a pointer to some 10983 data structure without incrementing reference counters. 10984 (Some of the existing 'get' functions increment a reference 10985 counter, some don't.) 10986 Similarly, 'set1' and 'add1' functions increase reference 10987 counters or duplicate objects. 10988 [Steve Henson] 10989 10990 *) Allow for the possibility of temp RSA key generation failure: 10991 the code used to assume it always worked and crashed on failure. 10992 [Steve Henson] 10993 10994 *) Fix potential buffer overrun problem in BIO_printf(). 10995 [Ulf Möller, using public domain code by Patrick Powell; problem 10996 pointed out by David Sacerdote <das33@cornell.edu>] 10997 10998 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions 10999 RAND_egd() and RAND_status(). In the command line application, 11000 the EGD socket can be specified like a seed file using RANDFILE 11001 or -rand. 11002 [Ulf Möller] 11003 11004 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures. 11005 Some CAs (e.g. Verisign) distribute certificates in this form. 11006 [Steve Henson] 11007 11008 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher 11009 list to exclude them. This means that no special compilation option 11010 is needed to use anonymous DH: it just needs to be included in the 11011 cipher list. 11012 [Steve Henson] 11013 11014 *) Change the EVP_MD_CTX_type macro so its meaning consistent with 11015 EVP_MD_type. The old functionality is available in a new macro called 11016 EVP_MD_md(). Change code that uses it and update docs. 11017 [Steve Henson] 11018 11019 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions 11020 where the 'void *' argument is replaced by a function pointer argument. 11021 Previously 'void *' was abused to point to functions, which works on 11022 many platforms, but is not correct. As these functions are usually 11023 called by macros defined in OpenSSL header files, most source code 11024 should work without changes. 11025 [Richard Levitte] 11026 11027 *) <openssl/opensslconf.h> (which is created by Configure) now contains 11028 sections with information on -D... compiler switches used for 11029 compiling the library so that applications can see them. To enable 11030 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES 11031 must be defined. E.g., 11032 #define OPENSSL_ALGORITHM_DEFINES 11033 #include <openssl/opensslconf.h> 11034 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc. 11035 [Richard Levitte, Ulf and Bodo Möller] 11036 11037 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS 11038 record layer. 11039 [Bodo Moeller] 11040 11041 *) Change the 'other' type in certificate aux info to a STACK_OF 11042 X509_ALGOR. Although not an AlgorithmIdentifier as such it has 11043 the required ASN1 format: arbitrary types determined by an OID. 11044 [Steve Henson] 11045 11046 *) Add some PEM_write_X509_REQ_NEW() functions and a command line 11047 argument to 'req'. This is not because the function is newer or 11048 better than others it just uses the work 'NEW' in the certificate 11049 request header lines. Some software needs this. 11050 [Steve Henson] 11051 11052 *) Reorganise password command line arguments: now passwords can be 11053 obtained from various sources. Delete the PEM_cb function and make 11054 it the default behaviour: i.e. if the callback is NULL and the 11055 usrdata argument is not NULL interpret it as a null terminated pass 11056 phrase. If usrdata and the callback are NULL then the pass phrase 11057 is prompted for as usual. 11058 [Steve Henson] 11059 11060 *) Add support for the Compaq Atalla crypto accelerator. If it is installed, 11061 the support is automatically enabled. The resulting binaries will 11062 autodetect the card and use it if present. 11063 [Ben Laurie and Compaq Inc.] 11064 11065 *) Work around for Netscape hang bug. This sends certificate request 11066 and server done in one record. Since this is perfectly legal in the 11067 SSL/TLS protocol it isn't a "bug" option and is on by default. See 11068 the bugs/SSLv3 entry for more info. 11069 [Steve Henson] 11070 11071 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround. 11072 [Andy Polyakov] 11073 11074 *) Add -rand argument to smime and pkcs12 applications and read/write 11075 of seed file. 11076 [Steve Henson] 11077 11078 *) New 'passwd' tool for crypt(3) and apr1 password hashes. 11079 [Bodo Moeller] 11080 11081 *) Add command line password options to the remaining applications. 11082 [Steve Henson] 11083 11084 *) Bug fix for BN_div_recp() for numerators with an even number of 11085 bits. 11086 [Ulf Möller] 11087 11088 *) More tests in bntest.c, and changed test_bn output. 11089 [Ulf Möller] 11090 11091 *) ./config recognizes MacOS X now. 11092 [Andy Polyakov] 11093 11094 *) Bug fix for BN_div() when the first words of num and divisor are 11095 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0). 11096 [Ulf Möller] 11097 11098 *) Add support for various broken PKCS#8 formats, and command line 11099 options to produce them. 11100 [Steve Henson] 11101 11102 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to 11103 get temporary BIGNUMs from a BN_CTX. 11104 [Ulf Möller] 11105 11106 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont() 11107 for p == 0. 11108 [Ulf Möller] 11109 11110 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and 11111 include a #define from the old name to the new. The original intent 11112 was that statically linked binaries could for example just call 11113 SSLeay_add_all_ciphers() to just add ciphers to the table and not 11114 link with digests. This never worked because SSLeay_add_all_digests() 11115 and SSLeay_add_all_ciphers() were in the same source file so calling 11116 one would link with the other. They are now in separate source files. 11117 [Steve Henson] 11118 11119 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'. 11120 [Steve Henson] 11121 11122 *) Use a less unusual form of the Miller-Rabin primality test (it used 11123 a binary algorithm for exponentiation integrated into the Miller-Rabin 11124 loop, our standard modexp algorithms are faster). 11125 [Bodo Moeller] 11126 11127 *) Support for the EBCDIC character set completed. 11128 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>] 11129 11130 *) Source code cleanups: use const where appropriate, eliminate casts, 11131 use void * instead of char * in lhash. 11132 [Ulf Möller] 11133 11134 *) Bugfix: ssl3_send_server_key_exchange was not restartable 11135 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of 11136 this the server could overwrite ephemeral keys that the client 11137 has already seen). 11138 [Bodo Moeller] 11139 11140 *) Turn DSA_is_prime into a macro that calls BN_is_prime, 11141 using 50 iterations of the Rabin-Miller test. 11142 11143 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50 11144 iterations of the Rabin-Miller test as required by the appendix 11145 to FIPS PUB 186[-1]) instead of DSA_is_prime. 11146 As BN_is_prime_fasttest includes trial division, DSA parameter 11147 generation becomes much faster. 11148 11149 This implies a change for the callback functions in DSA_is_prime 11150 and DSA_generate_parameters: The callback function is called once 11151 for each positive witness in the Rabin-Miller test, not just 11152 occasionally in the inner loop; and the parameters to the 11153 callback function now provide an iteration count for the outer 11154 loop rather than for the current invocation of the inner loop. 11155 DSA_generate_parameters additionally can call the callback 11156 function with an 'iteration count' of -1, meaning that a 11157 candidate has passed the trial division test (when q is generated 11158 from an application-provided seed, trial division is skipped). 11159 [Bodo Moeller] 11160 11161 *) New function BN_is_prime_fasttest that optionally does trial 11162 division before starting the Rabin-Miller test and has 11163 an additional BN_CTX * argument (whereas BN_is_prime always 11164 has to allocate at least one BN_CTX). 11165 'callback(1, -1, cb_arg)' is called when a number has passed the 11166 trial division stage. 11167 [Bodo Moeller] 11168 11169 *) Fix for bug in CRL encoding. The validity dates weren't being handled 11170 as ASN1_TIME. 11171 [Steve Henson] 11172 11173 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file. 11174 [Steve Henson] 11175 11176 *) New function BN_pseudo_rand(). 11177 [Ulf Möller] 11178 11179 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable) 11180 bignum version of BN_from_montgomery() with the working code from 11181 SSLeay 0.9.0 (the word based version is faster anyway), and clean up 11182 the comments. 11183 [Ulf Möller] 11184 11185 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that 11186 made it impossible to use the same SSL_SESSION data structure in 11187 SSL2 clients in multiple threads. 11188 [Bodo Moeller] 11189 11190 *) The return value of RAND_load_file() no longer counts bytes obtained 11191 by stat(). RAND_load_file(..., -1) is new and uses the complete file 11192 to seed the PRNG (previously an explicit byte count was required). 11193 [Ulf Möller, Bodo Möller] 11194 11195 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes 11196 used (char *) instead of (void *) and had casts all over the place. 11197 [Steve Henson] 11198 11199 *) Make BN_generate_prime() return NULL on error if ret!=NULL. 11200 [Ulf Möller] 11201 11202 *) Retain source code compatibility for BN_prime_checks macro: 11203 BN_is_prime(..., BN_prime_checks, ...) now uses 11204 BN_prime_checks_for_size to determine the appropriate number of 11205 Rabin-Miller iterations. 11206 [Ulf Möller] 11207 11208 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to 11209 DH_CHECK_P_NOT_SAFE_PRIME. 11210 (Check if this is true? OpenPGP calls them "strong".) 11211 [Ulf Möller] 11212 11213 *) Merge the functionality of "dh" and "gendh" programs into a new program 11214 "dhparam". The old programs are retained for now but will handle DH keys 11215 (instead of parameters) in future. 11216 [Steve Henson] 11217 11218 *) Make the ciphers, s_server and s_client programs check the return values 11219 when a new cipher list is set. 11220 [Steve Henson] 11221 11222 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit 11223 ciphers. Before when the 56bit ciphers were enabled the sorting was 11224 wrong. 11225 11226 The syntax for the cipher sorting has been extended to support sorting by 11227 cipher-strength (using the strength_bits hard coded in the tables). 11228 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod). 11229 11230 Fix a bug in the cipher-command parser: when supplying a cipher command 11231 string with an "undefined" symbol (neither command nor alphanumeric 11232 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now 11233 an error is flagged. 11234 11235 Due to the strength-sorting extension, the code of the 11236 ssl_create_cipher_list() function was completely rearranged. I hope that 11237 the readability was also increased :-) 11238 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>] 11239 11240 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1 11241 for the first serial number and places 2 in the serial number file. This 11242 avoids problems when the root CA is created with serial number zero and 11243 the first user certificate has the same issuer name and serial number 11244 as the root CA. 11245 [Steve Henson] 11246 11247 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses 11248 the new code. Add documentation for this stuff. 11249 [Steve Henson] 11250 11251 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from 11252 X509_*() to X509at_*() on the grounds that they don't handle X509 11253 structures and behave in an analogous way to the X509v3 functions: 11254 they shouldn't be called directly but wrapper functions should be used 11255 instead. 11256 11257 So we also now have some wrapper functions that call the X509at functions 11258 when passed certificate requests. (TO DO: similar things can be done with 11259 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other 11260 things. Some of these need some d2i or i2d and print functionality 11261 because they handle more complex structures.) 11262 [Steve Henson] 11263 11264 *) Add missing #ifndefs that caused missing symbols when building libssl 11265 as a shared library without RSA. Use #ifndef NO_SSL2 instead of 11266 NO_RSA in ssl/s2*.c. 11267 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller] 11268 11269 *) Precautions against using the PRNG uninitialized: RAND_bytes() now 11270 has a return value which indicates the quality of the random data 11271 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's 11272 error queue. New function RAND_pseudo_bytes() generates output that is 11273 guaranteed to be unique but not unpredictable. RAND_add is like 11274 RAND_seed, but takes an extra argument for an entropy estimate 11275 (RAND_seed always assumes full entropy). 11276 [Ulf Möller] 11277 11278 *) Do more iterations of Rabin-Miller probable prime test (specifically, 11279 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes 11280 instead of only 2 for all lengths; see BN_prime_checks_for_size definition 11281 in crypto/bn/bn_prime.c for the complete table). This guarantees a 11282 false-positive rate of at most 2^-80 for random input. 11283 [Bodo Moeller] 11284 11285 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs. 11286 [Bodo Moeller] 11287 11288 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain 11289 in the 0.9.5 release), this returns the chain 11290 from an X509_CTX structure with a dup of the stack and all 11291 the X509 reference counts upped: so the stack will exist 11292 after X509_CTX_cleanup() has been called. Modify pkcs12.c 11293 to use this. 11294 11295 Also make SSL_SESSION_print() print out the verify return 11296 code. 11297 [Steve Henson] 11298 11299 *) Add manpage for the pkcs12 command. Also change the default 11300 behaviour so MAC iteration counts are used unless the new 11301 -nomaciter option is used. This improves file security and 11302 only older versions of MSIE (4.0 for example) need it. 11303 [Steve Henson] 11304 11305 *) Honor the no-xxx Configure options when creating .DEF files. 11306 [Ulf Möller] 11307 11308 *) Add PKCS#10 attributes to field table: challengePassword, 11309 unstructuredName and unstructuredAddress. These are taken from 11310 draft PKCS#9 v2.0 but are compatible with v1.2 provided no 11311 international characters are used. 11312 11313 More changes to X509_ATTRIBUTE code: allow the setting of types 11314 based on strings. Remove the 'loc' parameter when adding 11315 attributes because these will be a SET OF encoding which is sorted 11316 in ASN1 order. 11317 [Steve Henson] 11318 11319 *) Initial changes to the 'req' utility to allow request generation 11320 automation. This will allow an application to just generate a template 11321 file containing all the field values and have req construct the 11322 request. 11323 11324 Initial support for X509_ATTRIBUTE handling. Stacks of these are 11325 used all over the place including certificate requests and PKCS#7 11326 structures. They are currently handled manually where necessary with 11327 some primitive wrappers for PKCS#7. The new functions behave in a 11328 manner analogous to the X509 extension functions: they allow 11329 attributes to be looked up by NID and added. 11330 11331 Later something similar to the X509V3 code would be desirable to 11332 automatically handle the encoding, decoding and printing of the 11333 more complex types. The string types like challengePassword can 11334 be handled by the string table functions. 11335 11336 Also modified the multi byte string table handling. Now there is 11337 a 'global mask' which masks out certain types. The table itself 11338 can use the flag STABLE_NO_MASK to ignore the mask setting: this 11339 is useful when for example there is only one permissible type 11340 (as in countryName) and using the mask might result in no valid 11341 types at all. 11342 [Steve Henson] 11343 11344 *) Clean up 'Finished' handling, and add functions SSL_get_finished and 11345 SSL_get_peer_finished to allow applications to obtain the latest 11346 Finished messages sent to the peer or expected from the peer, 11347 respectively. (SSL_get_peer_finished is usually the Finished message 11348 actually received from the peer, otherwise the protocol will be aborted.) 11349 11350 As the Finished message are message digests of the complete handshake 11351 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can 11352 be used for external authentication procedures when the authentication 11353 provided by SSL/TLS is not desired or is not enough. 11354 [Bodo Moeller] 11355 11356 *) Enhanced support for Alpha Linux is added. Now ./config checks if 11357 the host supports BWX extension and if Compaq C is present on the 11358 $PATH. Just exploiting of the BWX extension results in 20-30% 11359 performance kick for some algorithms, e.g. DES and RC4 to mention 11360 a couple. Compaq C in turn generates ~20% faster code for MD5 and 11361 SHA1. 11362 [Andy Polyakov] 11363 11364 *) Add support for MS "fast SGC". This is arguably a violation of the 11365 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with 11366 weak crypto and after checking the certificate is SGC a second one 11367 with strong crypto. MS SGC stops the first handshake after receiving 11368 the server certificate message and sends a second client hello. Since 11369 a server will typically do all the time consuming operations before 11370 expecting any further messages from the client (server key exchange 11371 is the most expensive) there is little difference between the two. 11372 11373 To get OpenSSL to support MS SGC we have to permit a second client 11374 hello message after we have sent server done. In addition we have to 11375 reset the MAC if we do get this second client hello. 11376 [Steve Henson] 11377 11378 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide 11379 if a DER encoded private key is RSA or DSA traditional format. Changed 11380 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional" 11381 format DER encoded private key. Newer code should use PKCS#8 format which 11382 has the key type encoded in the ASN1 structure. Added DER private key 11383 support to pkcs8 application. 11384 [Steve Henson] 11385 11386 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous 11387 ciphersuites has been selected (as required by the SSL 3/TLS 1 11388 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT 11389 is set, we interpret this as a request to violate the specification 11390 (the worst that can happen is a handshake failure, and 'correct' 11391 behaviour would result in a handshake failure anyway). 11392 [Bodo Moeller] 11393 11394 *) In SSL_CTX_add_session, take into account that there might be multiple 11395 SSL_SESSION structures with the same session ID (e.g. when two threads 11396 concurrently obtain them from an external cache). 11397 The internal cache can handle only one SSL_SESSION with a given ID, 11398 so if there's a conflict, we now throw out the old one to achieve 11399 consistency. 11400 [Bodo Moeller] 11401 11402 *) Add OIDs for idea and blowfish in CBC mode. This will allow both 11403 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to 11404 some routines that use cipher OIDs: some ciphers do not have OIDs 11405 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for 11406 example. 11407 [Steve Henson] 11408 11409 *) Simplify the trust setting structure and code. Now we just have 11410 two sequences of OIDs for trusted and rejected settings. These will 11411 typically have values the same as the extended key usage extension 11412 and any application specific purposes. 11413 11414 The trust checking code now has a default behaviour: it will just 11415 check for an object with the same NID as the passed id. Functions can 11416 be provided to override either the default behaviour or the behaviour 11417 for a given id. SSL client, server and email already have functions 11418 in place for compatibility: they check the NID and also return "trusted" 11419 if the certificate is self signed. 11420 [Steve Henson] 11421 11422 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the 11423 traditional format into an EVP_PKEY structure. 11424 [Steve Henson] 11425 11426 *) Add a password callback function PEM_cb() which either prompts for 11427 a password if usr_data is NULL or otherwise assumes it is a null 11428 terminated password. Allow passwords to be passed on command line 11429 environment or config files in a few more utilities. 11430 [Steve Henson] 11431 11432 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private 11433 keys. Add some short names for PKCS#8 PBE algorithms and allow them 11434 to be specified on the command line for the pkcs8 and pkcs12 utilities. 11435 Update documentation. 11436 [Steve Henson] 11437 11438 *) Support for ASN1 "NULL" type. This could be handled before by using 11439 ASN1_TYPE but there wasn't any function that would try to read a NULL 11440 and produce an error if it couldn't. For compatibility we also have 11441 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and 11442 don't allocate anything because they don't need to. 11443 [Steve Henson] 11444 11445 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS 11446 for details. 11447 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>] 11448 11449 *) Rebuild of the memory allocation routines used by OpenSSL code and 11450 possibly others as well. The purpose is to make an interface that 11451 provide hooks so anyone can build a separate set of allocation and 11452 deallocation routines to be used by OpenSSL, for example memory 11453 pool implementations, or something else, which was previously hard 11454 since Malloc(), Realloc() and Free() were defined as macros having 11455 the values malloc, realloc and free, respectively (except for Win32 11456 compilations). The same is provided for memory debugging code. 11457 OpenSSL already comes with functionality to find memory leaks, but 11458 this gives people a chance to debug other memory problems. 11459 11460 With these changes, a new set of functions and macros have appeared: 11461 11462 CRYPTO_set_mem_debug_functions() [F] 11463 CRYPTO_get_mem_debug_functions() [F] 11464 CRYPTO_dbg_set_options() [F] 11465 CRYPTO_dbg_get_options() [F] 11466 CRYPTO_malloc_debug_init() [M] 11467 11468 The memory debug functions are NULL by default, unless the library 11469 is compiled with CRYPTO_MDEBUG or friends is defined. If someone 11470 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which 11471 gives the standard debugging functions that come with OpenSSL) or 11472 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions 11473 provided by the library user) must be used. When the standard 11474 debugging functions are used, CRYPTO_dbg_set_options can be used to 11475 request additional information: 11476 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting 11477 the CRYPTO_MDEBUG_xxx macro when compiling the library. 11478 11479 Also, things like CRYPTO_set_mem_functions will always give the 11480 expected result (the new set of functions is used for allocation 11481 and deallocation) at all times, regardless of platform and compiler 11482 options. 11483 11484 To finish it up, some functions that were never use in any other 11485 way than through macros have a new API and new semantic: 11486 11487 CRYPTO_dbg_malloc() 11488 CRYPTO_dbg_realloc() 11489 CRYPTO_dbg_free() 11490 11491 All macros of value have retained their old syntax. 11492 [Richard Levitte and Bodo Moeller] 11493 11494 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the 11495 ordering of SMIMECapabilities wasn't in "strength order" and there 11496 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature 11497 algorithm. 11498 [Steve Henson] 11499 11500 *) Some ASN1 types with illegal zero length encoding (INTEGER, 11501 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines. 11502 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson] 11503 11504 *) Merge in my S/MIME library for OpenSSL. This provides a simple 11505 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough 11506 functionality to handle multipart/signed properly) and a utility 11507 called 'smime' to call all this stuff. This is based on code I 11508 originally wrote for Celo who have kindly allowed it to be 11509 included in OpenSSL. 11510 [Steve Henson] 11511 11512 *) Add variants des_set_key_checked and des_set_key_unchecked of 11513 des_set_key (aka des_key_sched). Global variable des_check_key 11514 decides which of these is called by des_set_key; this way 11515 des_check_key behaves as it always did, but applications and 11516 the library itself, which was buggy for des_check_key == 1, 11517 have a cleaner way to pick the version they need. 11518 [Bodo Moeller] 11519 11520 *) New function PKCS12_newpass() which changes the password of a 11521 PKCS12 structure. 11522 [Steve Henson] 11523 11524 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and 11525 dynamic mix. In both cases the ids can be used as an index into the 11526 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add() 11527 functions so they accept a list of the field values and the 11528 application doesn't need to directly manipulate the X509_TRUST 11529 structure. 11530 [Steve Henson] 11531 11532 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't 11533 need initialising. 11534 [Steve Henson] 11535 11536 *) Modify the way the V3 extension code looks up extensions. This now 11537 works in a similar way to the object code: we have some "standard" 11538 extensions in a static table which is searched with OBJ_bsearch() 11539 and the application can add dynamic ones if needed. The file 11540 crypto/x509v3/ext_dat.h now has the info: this file needs to be 11541 updated whenever a new extension is added to the core code and kept 11542 in ext_nid order. There is a simple program 'tabtest.c' which checks 11543 this. New extensions are not added too often so this file can readily 11544 be maintained manually. 11545 11546 There are two big advantages in doing things this way. The extensions 11547 can be looked up immediately and no longer need to be "added" using 11548 X509V3_add_standard_extensions(): this function now does nothing. 11549 [Side note: I get *lots* of email saying the extension code doesn't 11550 work because people forget to call this function] 11551 Also no dynamic allocation is done unless new extensions are added: 11552 so if we don't add custom extensions there is no need to call 11553 X509V3_EXT_cleanup(). 11554 [Steve Henson] 11555 11556 *) Modify enc utility's salting as follows: make salting the default. Add a 11557 magic header, so unsalted files fail gracefully instead of just decrypting 11558 to garbage. This is because not salting is a big security hole, so people 11559 should be discouraged from doing it. 11560 [Ben Laurie] 11561 11562 *) Fixes and enhancements to the 'x509' utility. It allowed a message 11563 digest to be passed on the command line but it only used this 11564 parameter when signing a certificate. Modified so all relevant 11565 operations are affected by the digest parameter including the 11566 -fingerprint and -x509toreq options. Also -x509toreq choked if a 11567 DSA key was used because it didn't fix the digest. 11568 [Steve Henson] 11569 11570 *) Initial certificate chain verify code. Currently tests the untrusted 11571 certificates for consistency with the verify purpose (which is set 11572 when the X509_STORE_CTX structure is set up) and checks the pathlength. 11573 11574 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour: 11575 this is because it will reject chains with invalid extensions whereas 11576 every previous version of OpenSSL and SSLeay made no checks at all. 11577 11578 Trust code: checks the root CA for the relevant trust settings. Trust 11579 settings have an initial value consistent with the verify purpose: e.g. 11580 if the verify purpose is for SSL client use it expects the CA to be 11581 trusted for SSL client use. However the default value can be changed to 11582 permit custom trust settings: one example of this would be to only trust 11583 certificates from a specific "secure" set of CAs. 11584 11585 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions 11586 which should be used for version portability: especially since the 11587 verify structure is likely to change more often now. 11588 11589 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions 11590 to set them. If not set then assume SSL clients will verify SSL servers 11591 and vice versa. 11592 11593 Two new options to the verify program: -untrusted allows a set of 11594 untrusted certificates to be passed in and -purpose which sets the 11595 intended purpose of the certificate. If a purpose is set then the 11596 new chain verify code is used to check extension consistency. 11597 [Steve Henson] 11598 11599 *) Support for the authority information access extension. 11600 [Steve Henson] 11601 11602 *) Modify RSA and DSA PEM read routines to transparently handle 11603 PKCS#8 format private keys. New *_PUBKEY_* functions that handle 11604 public keys in a format compatible with certificate 11605 SubjectPublicKeyInfo structures. Unfortunately there were already 11606 functions called *_PublicKey_* which used various odd formats so 11607 these are retained for compatibility: however the DSA variants were 11608 never in a public release so they have been deleted. Changed dsa/rsa 11609 utilities to handle the new format: note no releases ever handled public 11610 keys so we should be OK. 11611 11612 The primary motivation for this change is to avoid the same fiasco 11613 that dogs private keys: there are several incompatible private key 11614 formats some of which are standard and some OpenSSL specific and 11615 require various evil hacks to allow partial transparent handling and 11616 even then it doesn't work with DER formats. Given the option anything 11617 other than PKCS#8 should be dumped: but the other formats have to 11618 stay in the name of compatibility. 11619 11620 With public keys and the benefit of hindsight one standard format 11621 is used which works with EVP_PKEY, RSA or DSA structures: though 11622 it clearly returns an error if you try to read the wrong kind of key. 11623 11624 Added a -pubkey option to the 'x509' utility to output the public key. 11625 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*() 11626 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add 11627 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*()) 11628 that do the same as the EVP_PKEY_assign_*() except they up the 11629 reference count of the added key (they don't "swallow" the 11630 supplied key). 11631 [Steve Henson] 11632 11633 *) Fixes to crypto/x509/by_file.c the code to read in certificates and 11634 CRLs would fail if the file contained no certificates or no CRLs: 11635 added a new function to read in both types and return the number 11636 read: this means that if none are read it will be an error. The 11637 DER versions of the certificate and CRL reader would always fail 11638 because it isn't possible to mix certificates and CRLs in DER format 11639 without choking one or the other routine. Changed this to just read 11640 a certificate: this is the best we can do. Also modified the code 11641 in apps/verify.c to take notice of return codes: it was previously 11642 attempting to read in certificates from NULL pointers and ignoring 11643 any errors: this is one reason why the cert and CRL reader seemed 11644 to work. It doesn't check return codes from the default certificate 11645 routines: these may well fail if the certificates aren't installed. 11646 [Steve Henson] 11647 11648 *) Code to support otherName option in GeneralName. 11649 [Steve Henson] 11650 11651 *) First update to verify code. Change the verify utility 11652 so it warns if it is passed a self signed certificate: 11653 for consistency with the normal behaviour. X509_verify 11654 has been modified to it will now verify a self signed 11655 certificate if *exactly* the same certificate appears 11656 in the store: it was previously impossible to trust a 11657 single self signed certificate. This means that: 11658 openssl verify ss.pem 11659 now gives a warning about a self signed certificate but 11660 openssl verify -CAfile ss.pem ss.pem 11661 is OK. 11662 [Steve Henson] 11663 11664 *) For servers, store verify_result in SSL_SESSION data structure 11665 (and add it to external session representation). 11666 This is needed when client certificate verifications fails, 11667 but an application-provided verification callback (set by 11668 SSL_CTX_set_cert_verify_callback) allows accepting the session 11669 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK 11670 but returns 1): When the session is reused, we have to set 11671 ssl->verify_result to the appropriate error code to avoid 11672 security holes. 11673 [Bodo Moeller, problem pointed out by Lutz Jaenicke] 11674 11675 *) Fix a bug in the new PKCS#7 code: it didn't consider the 11676 case in PKCS7_dataInit() where the signed PKCS7 structure 11677 didn't contain any existing data because it was being created. 11678 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson] 11679 11680 *) Add a salt to the key derivation routines in enc.c. This 11681 forms the first 8 bytes of the encrypted file. Also add a 11682 -S option to allow a salt to be input on the command line. 11683 [Steve Henson] 11684 11685 *) New function X509_cmp(). Oddly enough there wasn't a function 11686 to compare two certificates. We do this by working out the SHA1 11687 hash and comparing that. X509_cmp() will be needed by the trust 11688 code. 11689 [Steve Henson] 11690 11691 *) SSL_get1_session() is like SSL_get_session(), but increments 11692 the reference count in the SSL_SESSION returned. 11693 [Geoff Thorpe <geoff@eu.c2.net>] 11694 11695 *) Fix for 'req': it was adding a null to request attributes. 11696 Also change the X509_LOOKUP and X509_INFO code to handle 11697 certificate auxiliary information. 11698 [Steve Henson] 11699 11700 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document 11701 the 'enc' command. 11702 [Steve Henson] 11703 11704 *) Add the possibility to add extra information to the memory leak 11705 detecting output, to form tracebacks, showing from where each 11706 allocation was originated: CRYPTO_push_info("constant string") adds 11707 the string plus current file name and line number to a per-thread 11708 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info() 11709 is like calling CYRPTO_pop_info() until the stack is empty. 11710 Also updated memory leak detection code to be multi-thread-safe. 11711 [Richard Levitte] 11712 11713 *) Add options -text and -noout to pkcs7 utility and delete the 11714 encryption options which never did anything. Update docs. 11715 [Steve Henson] 11716 11717 *) Add options to some of the utilities to allow the pass phrase 11718 to be included on either the command line (not recommended on 11719 OSes like Unix) or read from the environment. Update the 11720 manpages and fix a few bugs. 11721 [Steve Henson] 11722 11723 *) Add a few manpages for some of the openssl commands. 11724 [Steve Henson] 11725 11726 *) Fix the -revoke option in ca. It was freeing up memory twice, 11727 leaking and not finding already revoked certificates. 11728 [Steve Henson] 11729 11730 *) Extensive changes to support certificate auxiliary information. 11731 This involves the use of X509_CERT_AUX structure and X509_AUX 11732 functions. An X509_AUX function such as PEM_read_X509_AUX() 11733 can still read in a certificate file in the usual way but it 11734 will also read in any additional "auxiliary information". By 11735 doing things this way a fair degree of compatibility can be 11736 retained: existing certificates can have this information added 11737 using the new 'x509' options. 11738 11739 Current auxiliary information includes an "alias" and some trust 11740 settings. The trust settings will ultimately be used in enhanced 11741 certificate chain verification routines: currently a certificate 11742 can only be trusted if it is self signed and then it is trusted 11743 for all purposes. 11744 [Steve Henson] 11745 11746 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD). 11747 The problem was that one of the replacement routines had not been working 11748 since SSLeay releases. For now the offending routine has been replaced 11749 with non-optimised assembler. Even so, this now gives around 95% 11750 performance improvement for 1024 bit RSA signs. 11751 [Mark Cox] 11752 11753 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 11754 handling. Most clients have the effective key size in bits equal to 11755 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key. 11756 A few however don't do this and instead use the size of the decrypted key 11757 to determine the RC2 key length and the AlgorithmIdentifier to determine 11758 the effective key length. In this case the effective key length can still 11759 be 40 bits but the key length can be 168 bits for example. This is fixed 11760 by manually forcing an RC2 key into the EVP_PKEY structure because the 11761 EVP code can't currently handle unusual RC2 key sizes: it always assumes 11762 the key length and effective key length are equal. 11763 [Steve Henson] 11764 11765 *) Add a bunch of functions that should simplify the creation of 11766 X509_NAME structures. Now you should be able to do: 11767 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0); 11768 and have it automatically work out the correct field type and fill in 11769 the structures. The more adventurous can try: 11770 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0); 11771 and it will (hopefully) work out the correct multibyte encoding. 11772 [Steve Henson] 11773 11774 *) Change the 'req' utility to use the new field handling and multibyte 11775 copy routines. Before the DN field creation was handled in an ad hoc 11776 way in req, ca, and x509 which was rather broken and didn't support 11777 BMPStrings or UTF8Strings. Since some software doesn't implement 11778 BMPStrings or UTF8Strings yet, they can be enabled using the config file 11779 using the dirstring_type option. See the new comment in the default 11780 openssl.cnf for more info. 11781 [Steve Henson] 11782 11783 *) Make crypto/rand/md_rand.c more robust: 11784 - Assure unique random numbers after fork(). 11785 - Make sure that concurrent threads access the global counter and 11786 md serializably so that we never lose entropy in them 11787 or use exactly the same state in multiple threads. 11788 Access to the large state is not always serializable because 11789 the additional locking could be a performance killer, and 11790 md should be large enough anyway. 11791 [Bodo Moeller] 11792 11793 *) New file apps/app_rand.c with commonly needed functionality 11794 for handling the random seed file. 11795 11796 Use the random seed file in some applications that previously did not: 11797 ca, 11798 dsaparam -genkey (which also ignored its '-rand' option), 11799 s_client, 11800 s_server, 11801 x509 (when signing). 11802 Except on systems with /dev/urandom, it is crucial to have a random 11803 seed file at least for key creation, DSA signing, and for DH exchanges; 11804 for RSA signatures we could do without one. 11805 11806 gendh and gendsa (unlike genrsa) used to read only the first byte 11807 of each file listed in the '-rand' option. The function as previously 11808 found in genrsa is now in app_rand.c and is used by all programs 11809 that support '-rand'. 11810 [Bodo Moeller] 11811 11812 *) In RAND_write_file, use mode 0600 for creating files; 11813 don't just chmod when it may be too late. 11814 [Bodo Moeller] 11815 11816 *) Report an error from X509_STORE_load_locations 11817 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed. 11818 [Bill Perry] 11819 11820 *) New function ASN1_mbstring_copy() this copies a string in either 11821 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format 11822 into an ASN1_STRING type. A mask of permissible types is passed 11823 and it chooses the "minimal" type to use or an error if not type 11824 is suitable. 11825 [Steve Henson] 11826 11827 *) Add function equivalents to the various macros in asn1.h. The old 11828 macros are retained with an M_ prefix. Code inside the library can 11829 use the M_ macros. External code (including the openssl utility) 11830 should *NOT* in order to be "shared library friendly". 11831 [Steve Henson] 11832 11833 *) Add various functions that can check a certificate's extensions 11834 to see if it usable for various purposes such as SSL client, 11835 server or S/MIME and CAs of these types. This is currently 11836 VERY EXPERIMENTAL but will ultimately be used for certificate chain 11837 verification. Also added a -purpose flag to x509 utility to 11838 print out all the purposes. 11839 [Steve Henson] 11840 11841 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated 11842 functions. 11843 [Steve Henson] 11844 11845 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search 11846 for, obtain and decode and extension and obtain its critical flag. 11847 This allows all the necessary extension code to be handled in a 11848 single function call. 11849 [Steve Henson] 11850 11851 *) RC4 tune-up featuring 30-40% performance improvement on most RISC 11852 platforms. See crypto/rc4/rc4_enc.c for further details. 11853 [Andy Polyakov] 11854 11855 *) New -noout option to asn1parse. This causes no output to be produced 11856 its main use is when combined with -strparse and -out to extract data 11857 from a file (which may not be in ASN.1 format). 11858 [Steve Henson] 11859 11860 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer 11861 when producing the local key id. 11862 [Richard Levitte <levitte@stacken.kth.se>] 11863 11864 *) New option -dhparam in s_server. This allows a DH parameter file to be 11865 stated explicitly. If it is not stated then it tries the first server 11866 certificate file. The previous behaviour hard coded the filename 11867 "server.pem". 11868 [Steve Henson] 11869 11870 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow 11871 a public key to be input or output. For example: 11872 openssl rsa -in key.pem -pubout -out pubkey.pem 11873 Also added necessary DSA public key functions to handle this. 11874 [Steve Henson] 11875 11876 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained 11877 in the message. This was handled by allowing 11878 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it. 11879 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>] 11880 11881 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null 11882 to the end of the strings whereas this didn't. This would cause problems 11883 if strings read with d2i_ASN1_bytes() were later modified. 11884 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>] 11885 11886 *) Fix for base64 decode bug. When a base64 bio reads only one line of 11887 data and it contains EOF it will end up returning an error. This is 11888 caused by input 46 bytes long. The cause is due to the way base64 11889 BIOs find the start of base64 encoded data. They do this by trying a 11890 trial decode on each line until they find one that works. When they 11891 do a flag is set and it starts again knowing it can pass all the 11892 data directly through the decoder. Unfortunately it doesn't reset 11893 the context it uses. This means that if EOF is reached an attempt 11894 is made to pass two EOFs through the context and this causes the 11895 resulting error. This can also cause other problems as well. As is 11896 usual with these problems it takes *ages* to find and the fix is 11897 trivial: move one line. 11898 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ] 11899 11900 *) Ugly workaround to get s_client and s_server working under Windows. The 11901 old code wouldn't work because it needed to select() on sockets and the 11902 tty (for keypresses and to see if data could be written). Win32 only 11903 supports select() on sockets so we select() with a 1s timeout on the 11904 sockets and then see if any characters are waiting to be read, if none 11905 are present then we retry, we also assume we can always write data to 11906 the tty. This isn't nice because the code then blocks until we've 11907 received a complete line of data and it is effectively polling the 11908 keyboard at 1s intervals: however it's quite a bit better than not 11909 working at all :-) A dedicated Windows application might handle this 11910 with an event loop for example. 11911 [Steve Henson] 11912 11913 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign 11914 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions 11915 will be called when RSA_sign() and RSA_verify() are used. This is useful 11916 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available. 11917 For this to work properly RSA_public_decrypt() and RSA_private_encrypt() 11918 should *not* be used: RSA_sign() and RSA_verify() must be used instead. 11919 This necessitated the support of an extra signature type NID_md5_sha1 11920 for SSL signatures and modifications to the SSL library to use it instead 11921 of calling RSA_public_decrypt() and RSA_private_encrypt(). 11922 [Steve Henson] 11923 11924 *) Add new -verify -CAfile and -CApath options to the crl program, these 11925 will lookup a CRL issuers certificate and verify the signature in a 11926 similar way to the verify program. Tidy up the crl program so it 11927 no longer accesses structures directly. Make the ASN1 CRL parsing a bit 11928 less strict. It will now permit CRL extensions even if it is not 11929 a V2 CRL: this will allow it to tolerate some broken CRLs. 11930 [Steve Henson] 11931 11932 *) Initialize all non-automatic variables each time one of the openssl 11933 sub-programs is started (this is necessary as they may be started 11934 multiple times from the "OpenSSL>" prompt). 11935 [Lennart Bang, Bodo Moeller] 11936 11937 *) Preliminary compilation option RSA_NULL which disables RSA crypto without 11938 removing all other RSA functionality (this is what NO_RSA does). This 11939 is so (for example) those in the US can disable those operations covered 11940 by the RSA patent while allowing storage and parsing of RSA keys and RSA 11941 key generation. 11942 [Steve Henson] 11943 11944 *) Non-copying interface to BIO pairs. 11945 (still largely untested) 11946 [Bodo Moeller] 11947 11948 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive 11949 ASCII string. This was handled independently in various places before. 11950 [Steve Henson] 11951 11952 *) New functions UTF8_getc() and UTF8_putc() that parse and generate 11953 UTF8 strings a character at a time. 11954 [Steve Henson] 11955 11956 *) Use client_version from client hello to select the protocol 11957 (s23_srvr.c) and for RSA client key exchange verification 11958 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications. 11959 [Bodo Moeller] 11960 11961 *) Add various utility functions to handle SPKACs, these were previously 11962 handled by poking round in the structure internals. Added new function 11963 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to 11964 print, verify and generate SPKACs. Based on an original idea from 11965 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified. 11966 [Steve Henson] 11967 11968 *) RIPEMD160 is operational on all platforms and is back in 'make test'. 11969 [Andy Polyakov] 11970 11971 *) Allow the config file extension section to be overwritten on the 11972 command line. Based on an original idea from Massimiliano Pala 11973 <madwolf@comune.modena.it>. The new option is called -extensions 11974 and can be applied to ca, req and x509. Also -reqexts to override 11975 the request extensions in req and -crlexts to override the crl extensions 11976 in ca. 11977 [Steve Henson] 11978 11979 *) Add new feature to the SPKAC handling in ca. Now you can include 11980 the same field multiple times by preceding it by "XXXX." for example: 11981 1.OU="Unit name 1" 11982 2.OU="Unit name 2" 11983 this is the same syntax as used in the req config file. 11984 [Steve Henson] 11985 11986 *) Allow certificate extensions to be added to certificate requests. These 11987 are specified in a 'req_extensions' option of the req section of the 11988 config file. They can be printed out with the -text option to req but 11989 are otherwise ignored at present. 11990 [Steve Henson] 11991 11992 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first 11993 data read consists of only the final block it would not decrypted because 11994 EVP_CipherUpdate() would correctly report zero bytes had been decrypted. 11995 A misplaced 'break' also meant the decrypted final block might not be 11996 copied until the next read. 11997 [Steve Henson] 11998 11999 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added 12000 a few extra parameters to the DH structure: these will be useful if 12001 for example we want the value of 'q' or implement X9.42 DH. 12002 [Steve Henson] 12003 12004 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and 12005 provides hooks that allow the default DSA functions or functions on a 12006 "per key" basis to be replaced. This allows hardware acceleration and 12007 hardware key storage to be handled without major modification to the 12008 library. Also added low level modexp hooks and CRYPTO_EX structure and 12009 associated functions. 12010 [Steve Henson] 12011 12012 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO 12013 as "read only": it can't be written to and the buffer it points to will 12014 not be freed. Reading from a read only BIO is much more efficient than 12015 a normal memory BIO. This was added because there are several times when 12016 an area of memory needs to be read from a BIO. The previous method was 12017 to create a memory BIO and write the data to it, this results in two 12018 copies of the data and an O(n^2) reading algorithm. There is a new 12019 function BIO_new_mem_buf() which creates a read only memory BIO from 12020 an area of memory. Also modified the PKCS#7 routines to use read only 12021 memory BIOs. 12022 [Steve Henson] 12023 12024 *) Bugfix: ssl23_get_client_hello did not work properly when called in 12025 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of 12026 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read, 12027 but a retry condition occurred while trying to read the rest. 12028 [Bodo Moeller] 12029 12030 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as 12031 NID_pkcs7_encrypted by default: this was wrong since this should almost 12032 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle 12033 the encrypted data type: this is a more sensible place to put it and it 12034 allows the PKCS#12 code to be tidied up that duplicated this 12035 functionality. 12036 [Steve Henson] 12037 12038 *) Changed obj_dat.pl script so it takes its input and output files on 12039 the command line. This should avoid shell escape redirection problems 12040 under Win32. 12041 [Steve Henson] 12042 12043 *) Initial support for certificate extension requests, these are included 12044 in things like Xenroll certificate requests. Included functions to allow 12045 extensions to be obtained and added. 12046 [Steve Henson] 12047 12048 *) -crlf option to s_client and s_server for sending newlines as 12049 CRLF (as required by many protocols). 12050 [Bodo Moeller] 12051 12052 Changes between 0.9.3a and 0.9.4 [09 Aug 1999] 12053 12054 *) Install libRSAglue.a when OpenSSL is built with RSAref. 12055 [Ralf S. Engelschall] 12056 12057 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency. 12058 [Andrija Antonijevic <TheAntony2@bigfoot.com>] 12059 12060 *) Fix -startdate and -enddate (which was missing) arguments to 'ca' 12061 program. 12062 [Steve Henson] 12063 12064 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as 12065 DH parameters/keys (q is lost during that conversion, but the resulting 12066 DH parameters contain its length). 12067 12068 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is 12069 much faster than DH_generate_parameters (which creates parameters 12070 where p = 2*q + 1), and also the smaller q makes DH computations 12071 much more efficient (160-bit exponentiation instead of 1024-bit 12072 exponentiation); so this provides a convenient way to support DHE 12073 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of 12074 utter importance to use 12075 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 12076 or 12077 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE); 12078 when such DH parameters are used, because otherwise small subgroup 12079 attacks may become possible! 12080 [Bodo Moeller] 12081 12082 *) Avoid memory leak in i2d_DHparams. 12083 [Bodo Moeller] 12084 12085 *) Allow the -k option to be used more than once in the enc program: 12086 this allows the same encrypted message to be read by multiple recipients. 12087 [Steve Henson] 12088 12089 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts 12090 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then 12091 it will always use the numerical form of the OID, even if it has a short 12092 or long name. 12093 [Steve Henson] 12094 12095 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp 12096 method only got called if p,q,dmp1,dmq1,iqmp components were present, 12097 otherwise bn_mod_exp was called. In the case of hardware keys for example 12098 no private key components need be present and it might store extra data 12099 in the RSA structure, which cannot be accessed from bn_mod_exp. 12100 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for 12101 private key operations. 12102 [Steve Henson] 12103 12104 *) Added support for SPARC Linux. 12105 [Andy Polyakov] 12106 12107 *) pem_password_cb function type incompatibly changed from 12108 typedef int pem_password_cb(char *buf, int size, int rwflag); 12109 to 12110 ....(char *buf, int size, int rwflag, void *userdata); 12111 so that applications can pass data to their callbacks: 12112 The PEM[_ASN1]_{read,write}... functions and macros now take an 12113 additional void * argument, which is just handed through whenever 12114 the password callback is called. 12115 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller] 12116 12117 New function SSL_CTX_set_default_passwd_cb_userdata. 12118 12119 Compatibility note: As many C implementations push function arguments 12120 onto the stack in reverse order, the new library version is likely to 12121 interoperate with programs that have been compiled with the old 12122 pem_password_cb definition (PEM_whatever takes some data that 12123 happens to be on the stack as its last argument, and the callback 12124 just ignores this garbage); but there is no guarantee whatsoever that 12125 this will work. 12126 12127 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=... 12128 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused 12129 problems not only on Windows, but also on some Unix platforms. 12130 To avoid problematic command lines, these definitions are now in an 12131 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl 12132 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds). 12133 [Bodo Moeller] 12134 12135 *) MIPS III/IV assembler module is reimplemented. 12136 [Andy Polyakov] 12137 12138 *) More DES library cleanups: remove references to srand/rand and 12139 delete an unused file. 12140 [Ulf Möller] 12141 12142 *) Add support for the free Netwide assembler (NASM) under Win32, 12143 since not many people have MASM (ml) and it can be hard to obtain. 12144 This is currently experimental but it seems to work OK and pass all 12145 the tests. Check out INSTALL.W32 for info. 12146 [Steve Henson] 12147 12148 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections 12149 without temporary keys kept an extra copy of the server key, 12150 and connections with temporary keys did not free everything in case 12151 of an error. 12152 [Bodo Moeller] 12153 12154 *) New function RSA_check_key and new openssl rsa option -check 12155 for verifying the consistency of RSA keys. 12156 [Ulf Moeller, Bodo Moeller] 12157 12158 *) Various changes to make Win32 compile work: 12159 1. Casts to avoid "loss of data" warnings in p5_crpt2.c 12160 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned 12161 comparison" warnings. 12162 3. Add sk_<TYPE>_sort to DEF file generator and do make update. 12163 [Steve Henson] 12164 12165 *) Add a debugging option to PKCS#5 v2 key generation function: when 12166 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and 12167 derived keys are printed to stderr. 12168 [Steve Henson] 12169 12170 *) Copy the flags in ASN1_STRING_dup(). 12171 [Roman E. Pavlov <pre@mo.msk.ru>] 12172 12173 *) The x509 application mishandled signing requests containing DSA 12174 keys when the signing key was also DSA and the parameters didn't match. 12175 12176 It was supposed to omit the parameters when they matched the signing key: 12177 the verifying software was then supposed to automatically use the CA's 12178 parameters if they were absent from the end user certificate. 12179 12180 Omitting parameters is no longer recommended. The test was also 12181 the wrong way round! This was probably due to unusual behaviour in 12182 EVP_cmp_parameters() which returns 1 if the parameters match. 12183 This meant that parameters were omitted when they *didn't* match and 12184 the certificate was useless. Certificates signed with 'ca' didn't have 12185 this bug. 12186 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>] 12187 12188 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems. 12189 The interface is as follows: 12190 Applications can use 12191 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(), 12192 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop(); 12193 "off" is now the default. 12194 The library internally uses 12195 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(), 12196 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on() 12197 to disable memory-checking temporarily. 12198 12199 Some inconsistent states that previously were possible (and were 12200 even the default) are now avoided. 12201 12202 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time 12203 with each memory chunk allocated; this is occasionally more helpful 12204 than just having a counter. 12205 12206 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID. 12207 12208 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future 12209 extensions. 12210 [Bodo Moeller] 12211 12212 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX), 12213 which largely parallels "options", but is for changing API behaviour, 12214 whereas "options" are about protocol behaviour. 12215 Initial "mode" flags are: 12216 12217 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when 12218 a single record has been written. 12219 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write 12220 retries use the same buffer location. 12221 (But all of the contents must be 12222 copied!) 12223 [Bodo Moeller] 12224 12225 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options 12226 worked. 12227 12228 *) Fix problems with no-hmac etc. 12229 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>] 12230 12231 *) New functions RSA_get_default_method(), RSA_set_method() and 12232 RSA_get_method(). These allows replacement of RSA_METHODs without having 12233 to mess around with the internals of an RSA structure. 12234 [Steve Henson] 12235 12236 *) Fix memory leaks in DSA_do_sign and DSA_is_prime. 12237 Also really enable memory leak checks in openssl.c and in some 12238 test programs. 12239 [Chad C. Mulligan, Bodo Moeller] 12240 12241 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess 12242 up the length of negative integers. This has now been simplified to just 12243 store the length when it is first determined and use it later, rather 12244 than trying to keep track of where data is copied and updating it to 12245 point to the end. 12246 [Steve Henson, reported by Brien Wheeler 12247 <bwheeler@authentica-security.com>] 12248 12249 *) Add a new function PKCS7_signatureVerify. This allows the verification 12250 of a PKCS#7 signature but with the signing certificate passed to the 12251 function itself. This contrasts with PKCS7_dataVerify which assumes the 12252 certificate is present in the PKCS#7 structure. This isn't always the 12253 case: certificates can be omitted from a PKCS#7 structure and be 12254 distributed by "out of band" means (such as a certificate database). 12255 [Steve Henson] 12256 12257 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the 12258 function prototypes in pem.h, also change util/mkdef.pl to add the 12259 necessary function names. 12260 [Steve Henson] 12261 12262 *) mk1mf.pl (used by Windows builds) did not properly read the 12263 options set by Configure in the top level Makefile, and Configure 12264 was not even able to write more than one option correctly. 12265 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended. 12266 [Bodo Moeller] 12267 12268 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config 12269 file to be loaded from a BIO or FILE pointer. The BIO version will 12270 for example allow memory BIOs to contain config info. 12271 [Steve Henson] 12272 12273 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS. 12274 Whoever hopes to achieve shared-library compatibility across versions 12275 must use this, not the compile-time macro. 12276 (Exercise 0.9.4: Which is the minimum library version required by 12277 such programs?) 12278 Note: All this applies only to multi-threaded programs, others don't 12279 need locks. 12280 [Bodo Moeller] 12281 12282 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests 12283 through a BIO pair triggered the default case, i.e. 12284 SSLerr(...,SSL_R_UNKNOWN_STATE). 12285 [Bodo Moeller] 12286 12287 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications 12288 can use the SSL library even if none of the specific BIOs is 12289 appropriate. 12290 [Bodo Moeller] 12291 12292 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value 12293 for the encoded length. 12294 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>] 12295 12296 *) Add initial documentation of the X509V3 functions. 12297 [Steve Henson] 12298 12299 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 12300 PEM_write_bio_PKCS8PrivateKey() that are equivalent to 12301 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more 12302 secure PKCS#8 private key format with a high iteration count. 12303 [Steve Henson] 12304 12305 *) Fix determination of Perl interpreter: A perl or perl5 12306 _directory_ in $PATH was also accepted as the interpreter. 12307 [Ralf S. Engelschall] 12308 12309 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking 12310 wrong with it but it was very old and did things like calling 12311 PEM_ASN1_read() directly and used MD5 for the hash not to mention some 12312 unusual formatting. 12313 [Steve Henson] 12314 12315 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed 12316 to use the new extension code. 12317 [Steve Henson] 12318 12319 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c 12320 with macros. This should make it easier to change their form, add extra 12321 arguments etc. Fix a few PEM prototypes which didn't have cipher as a 12322 constant. 12323 [Steve Henson] 12324 12325 *) Add to configuration table a new entry that can specify an alternative 12326 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep, 12327 according to Mark Crispin <MRC@Panda.COM>. 12328 [Bodo Moeller] 12329 12330#if 0 12331 *) DES CBC did not update the IV. Weird. 12332 [Ben Laurie] 12333#else 12334 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does. 12335 Changing the behaviour of the former might break existing programs -- 12336 where IV updating is needed, des_ncbc_encrypt can be used. 12337#endif 12338 12339 *) When bntest is run from "make test" it drives bc to check its 12340 calculations, as well as internally checking them. If an internal check 12341 fails, it needs to cause bc to give a non-zero result or make test carries 12342 on without noticing the failure. Fixed. 12343 [Ben Laurie] 12344 12345 *) DES library cleanups. 12346 [Ulf Möller] 12347 12348 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be 12349 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit 12350 ciphers. NOTE: although the key derivation function has been verified 12351 against some published test vectors it has not been extensively tested 12352 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use 12353 of v2.0. 12354 [Steve Henson] 12355 12356 *) Instead of "mkdir -p", which is not fully portable, use new 12357 Perl script "util/mkdir-p.pl". 12358 [Bodo Moeller] 12359 12360 *) Rewrite the way password based encryption (PBE) is handled. It used to 12361 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter 12362 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms 12363 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now 12364 the 'parameter' field of the AlgorithmIdentifier is passed to the 12365 underlying key generation function so it must do its own ASN1 parsing. 12366 This has also changed the EVP_PBE_CipherInit() function which now has a 12367 'parameter' argument instead of literal salt and iteration count values 12368 and the function EVP_PBE_ALGOR_CipherInit() has been deleted. 12369 [Steve Henson] 12370 12371 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms 12372 and PKCS#8 functionality. New 'pkcs8' application linked to openssl. 12373 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE 12374 KEY" because this clashed with PKCS#8 unencrypted string. Since this 12375 value was just used as a "magic string" and not used directly its 12376 value doesn't matter. 12377 [Steve Henson] 12378 12379 *) Introduce some semblance of const correctness to BN. Shame C doesn't 12380 support mutable. 12381 [Ben Laurie] 12382 12383 *) "linux-sparc64" configuration (ultrapenguin). 12384 [Ray Miller <ray.miller@oucs.ox.ac.uk>] 12385 "linux-sparc" configuration. 12386 [Christian Forster <fo@hawo.stw.uni-erlangen.de>] 12387 12388 *) config now generates no-xxx options for missing ciphers. 12389 [Ulf Möller] 12390 12391 *) Support the EBCDIC character set (work in progress). 12392 File ebcdic.c not yet included because it has a different license. 12393 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 12394 12395 *) Support BS2000/OSD-POSIX. 12396 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>] 12397 12398 *) Make callbacks for key generation use void * instead of char *. 12399 [Ben Laurie] 12400 12401 *) Make S/MIME samples compile (not yet tested). 12402 [Ben Laurie] 12403 12404 *) Additional typesafe stacks. 12405 [Ben Laurie] 12406 12407 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x). 12408 [Bodo Moeller] 12409 12410 12411 Changes between 0.9.3 and 0.9.3a [29 May 1999] 12412 12413 *) New configuration variant "sco5-gcc". 12414 12415 *) Updated some demos. 12416 [Sean O Riordain, Wade Scholine] 12417 12418 *) Add missing BIO_free at exit of pkcs12 application. 12419 [Wu Zhigang] 12420 12421 *) Fix memory leak in conf.c. 12422 [Steve Henson] 12423 12424 *) Updates for Win32 to assembler version of MD5. 12425 [Steve Henson] 12426 12427 *) Set #! path to perl in apps/der_chop to where we found it 12428 instead of using a fixed path. 12429 [Bodo Moeller] 12430 12431 *) SHA library changes for irix64-mips4-cc. 12432 [Andy Polyakov] 12433 12434 *) Improvements for VMS support. 12435 [Richard Levitte] 12436 12437 12438 Changes between 0.9.2b and 0.9.3 [24 May 1999] 12439 12440 *) Bignum library bug fix. IRIX 6 passes "make test" now! 12441 This also avoids the problems with SC4.2 and unpatched SC5. 12442 [Andy Polyakov <appro@fy.chalmers.se>] 12443 12444 *) New functions sk_num, sk_value and sk_set to replace the previous macros. 12445 These are required because of the typesafe stack would otherwise break 12446 existing code. If old code used a structure member which used to be STACK 12447 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with 12448 sk_num or sk_value it would produce an error because the num, data members 12449 are not present in STACK_OF. Now it just produces a warning. sk_set 12450 replaces the old method of assigning a value to sk_value 12451 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code 12452 that does this will no longer work (and should use sk_set instead) but 12453 this could be regarded as a "questionable" behaviour anyway. 12454 [Steve Henson] 12455 12456 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now 12457 correctly handle encrypted S/MIME data. 12458 [Steve Henson] 12459 12460 *) Change type of various DES function arguments from des_cblock 12461 (which means, in function argument declarations, pointer to char) 12462 to des_cblock * (meaning pointer to array with 8 char elements), 12463 which allows the compiler to do more typechecking; it was like 12464 that back in SSLeay, but with lots of ugly casts. 12465 12466 Introduce new type const_des_cblock. 12467 [Bodo Moeller] 12468 12469 *) Reorganise the PKCS#7 library and get rid of some of the more obvious 12470 problems: find RecipientInfo structure that matches recipient certificate 12471 and initialise the ASN1 structures properly based on passed cipher. 12472 [Steve Henson] 12473 12474 *) Belatedly make the BN tests actually check the results. 12475 [Ben Laurie] 12476 12477 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion 12478 to and from BNs: it was completely broken. New compilation option 12479 NEG_PUBKEY_BUG to allow for some broken certificates that encode public 12480 key elements as negative integers. 12481 [Steve Henson] 12482 12483 *) Reorganize and speed up MD5. 12484 [Andy Polyakov <appro@fy.chalmers.se>] 12485 12486 *) VMS support. 12487 [Richard Levitte <richard@levitte.org>] 12488 12489 *) New option -out to asn1parse to allow the parsed structure to be 12490 output to a file. This is most useful when combined with the -strparse 12491 option to examine the output of things like OCTET STRINGS. 12492 [Steve Henson] 12493 12494 *) Make SSL library a little more fool-proof by not requiring any longer 12495 that SSL_set_{accept,connect}_state be called before 12496 SSL_{accept,connect} may be used (SSL_set_..._state is omitted 12497 in many applications because usually everything *appeared* to work as 12498 intended anyway -- now it really works as intended). 12499 [Bodo Moeller] 12500 12501 *) Move openssl.cnf out of lib/. 12502 [Ulf Möller] 12503 12504 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall 12505 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes 12506 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 12507 [Ralf S. Engelschall] 12508 12509 *) Various fixes to the EVP and PKCS#7 code. It may now be able to 12510 handle PKCS#7 enveloped data properly. 12511 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve] 12512 12513 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of 12514 copying pointers. The cert_st handling is changed by this in 12515 various ways (and thus what used to be known as ctx->default_cert 12516 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert 12517 any longer when s->cert does not give us what we need). 12518 ssl_cert_instantiate becomes obsolete by this change. 12519 As soon as we've got the new code right (possibly it already is?), 12520 we have solved a couple of bugs of the earlier code where s->cert 12521 was used as if it could not have been shared with other SSL structures. 12522 12523 Note that using the SSL API in certain dirty ways now will result 12524 in different behaviour than observed with earlier library versions: 12525 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx) 12526 does not influence s as it used to. 12527 12528 In order to clean up things more thoroughly, inside SSL_SESSION 12529 we don't use CERT any longer, but a new structure SESS_CERT 12530 that holds per-session data (if available); currently, this is 12531 the peer's certificate chain and, for clients, the server's certificate 12532 and temporary key. CERT holds only those values that can have 12533 meaningful defaults in an SSL_CTX. 12534 [Bodo Moeller] 12535 12536 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure 12537 from the internal representation. Various PKCS#7 fixes: remove some 12538 evil casts and set the enc_dig_alg field properly based on the signing 12539 key type. 12540 [Steve Henson] 12541 12542 *) Allow PKCS#12 password to be set from the command line or the 12543 environment. Let 'ca' get its config file name from the environment 12544 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req' 12545 and 'x509'). 12546 [Steve Henson] 12547 12548 *) Allow certificate policies extension to use an IA5STRING for the 12549 organization field. This is contrary to the PKIX definition but 12550 VeriSign uses it and IE5 only recognises this form. Document 'x509' 12551 extension option. 12552 [Steve Henson] 12553 12554 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic, 12555 without disallowing inline assembler and the like for non-pedantic builds. 12556 [Ben Laurie] 12557 12558 *) Support Borland C++ builder. 12559 [Janez Jere <jj@void.si>, modified by Ulf Möller] 12560 12561 *) Support Mingw32. 12562 [Ulf Möller] 12563 12564 *) SHA-1 cleanups and performance enhancements. 12565 [Andy Polyakov <appro@fy.chalmers.se>] 12566 12567 *) Sparc v8plus assembler for the bignum library. 12568 [Andy Polyakov <appro@fy.chalmers.se>] 12569 12570 *) Accept any -xxx and +xxx compiler options in Configure. 12571 [Ulf Möller] 12572 12573 *) Update HPUX configuration. 12574 [Anonymous] 12575 12576 *) Add missing sk_<type>_unshift() function to safestack.h 12577 [Ralf S. Engelschall] 12578 12579 *) New function SSL_CTX_use_certificate_chain_file that sets the 12580 "extra_cert"s in addition to the certificate. (This makes sense 12581 only for "PEM" format files, as chains as a whole are not 12582 DER-encoded.) 12583 [Bodo Moeller] 12584 12585 *) Support verify_depth from the SSL API. 12586 x509_vfy.c had what can be considered an off-by-one-error: 12587 Its depth (which was not part of the external interface) 12588 was actually counting the number of certificates in a chain; 12589 now it really counts the depth. 12590 [Bodo Moeller] 12591 12592 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used 12593 instead of X509err, which often resulted in confusing error 12594 messages since the error codes are not globally unique 12595 (e.g. an alleged error in ssl3_accept when a certificate 12596 didn't match the private key). 12597 12598 *) New function SSL_CTX_set_session_id_context that allows to set a default 12599 value (so that you don't need SSL_set_session_id_context for each 12600 connection using the SSL_CTX). 12601 [Bodo Moeller] 12602 12603 *) OAEP decoding bug fix. 12604 [Ulf Möller] 12605 12606 *) Support INSTALL_PREFIX for package builders, as proposed by 12607 David Harris. 12608 [Bodo Moeller] 12609 12610 *) New Configure options "threads" and "no-threads". For systems 12611 where the proper compiler options are known (currently Solaris 12612 and Linux), "threads" is the default. 12613 [Bodo Moeller] 12614 12615 *) New script util/mklink.pl as a faster substitute for util/mklink.sh. 12616 [Bodo Moeller] 12617 12618 *) Install various scripts to $(OPENSSLDIR)/misc, not to 12619 $(INSTALLTOP)/bin -- they shouldn't clutter directories 12620 such as /usr/local/bin. 12621 [Bodo Moeller] 12622 12623 *) "make linux-shared" to build shared libraries. 12624 [Niels Poppe <niels@netbox.org>] 12625 12626 *) New Configure option no-<cipher> (rsa, idea, rc5, ...). 12627 [Ulf Möller] 12628 12629 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for 12630 extension adding in x509 utility. 12631 [Steve Henson] 12632 12633 *) Remove NOPROTO sections and error code comments. 12634 [Ulf Möller] 12635 12636 *) Partial rewrite of the DEF file generator to now parse the ANSI 12637 prototypes. 12638 [Steve Henson] 12639 12640 *) New Configure options --prefix=DIR and --openssldir=DIR. 12641 [Ulf Möller] 12642 12643 *) Complete rewrite of the error code script(s). It is all now handled 12644 by one script at the top level which handles error code gathering, 12645 header rewriting and C source file generation. It should be much better 12646 than the old method: it now uses a modified version of Ulf's parser to 12647 read the ANSI prototypes in all header files (thus the old K&R definitions 12648 aren't needed for error creation any more) and do a better job of 12649 translating function codes into names. The old 'ASN1 error code embedded 12650 in a comment' is no longer necessary and it doesn't use .err files which 12651 have now been deleted. Also the error code call doesn't have to appear all 12652 on one line (which resulted in some large lines...). 12653 [Steve Henson] 12654 12655 *) Change #include filenames from <foo.h> to <openssl/foo.h>. 12656 [Bodo Moeller] 12657 12658 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return 12659 0 (which usually indicates a closed connection), but continue reading. 12660 [Bodo Moeller] 12661 12662 *) Fix some race conditions. 12663 [Bodo Moeller] 12664 12665 *) Add support for CRL distribution points extension. Add Certificate 12666 Policies and CRL distribution points documentation. 12667 [Steve Henson] 12668 12669 *) Move the autogenerated header file parts to crypto/opensslconf.h. 12670 [Ulf Möller] 12671 12672 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of 12673 8 of keying material. Merlin has also confirmed interop with this fix 12674 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0. 12675 [Merlin Hughes <merlin@baltimore.ie>] 12676 12677 *) Fix lots of warnings. 12678 [Richard Levitte <levitte@stacken.kth.se>] 12679 12680 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if 12681 the directory spec didn't end with a LIST_SEPARATOR_CHAR. 12682 [Richard Levitte <levitte@stacken.kth.se>] 12683 12684 *) Fix problems with sizeof(long) == 8. 12685 [Andy Polyakov <appro@fy.chalmers.se>] 12686 12687 *) Change functions to ANSI C. 12688 [Ulf Möller] 12689 12690 *) Fix typos in error codes. 12691 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller] 12692 12693 *) Remove defunct assembler files from Configure. 12694 [Ulf Möller] 12695 12696 *) SPARC v8 assembler BIGNUM implementation. 12697 [Andy Polyakov <appro@fy.chalmers.se>] 12698 12699 *) Support for Certificate Policies extension: both print and set. 12700 Various additions to support the r2i method this uses. 12701 [Steve Henson] 12702 12703 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could 12704 return a const string when you are expecting an allocated buffer. 12705 [Ben Laurie] 12706 12707 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE 12708 types DirectoryString and DisplayText. 12709 [Steve Henson] 12710 12711 *) Add code to allow r2i extensions to access the configuration database, 12712 add an LHASH database driver and add several ctx helper functions. 12713 [Steve Henson] 12714 12715 *) Fix an evil bug in bn_expand2() which caused various BN functions to 12716 fail when they extended the size of a BIGNUM. 12717 [Steve Henson] 12718 12719 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to 12720 support typesafe stack. 12721 [Steve Henson] 12722 12723 *) Fix typo in SSL_[gs]et_options(). 12724 [Nils Frostberg <nils@medcom.se>] 12725 12726 *) Delete various functions and files that belonged to the (now obsolete) 12727 old X509V3 handling code. 12728 [Steve Henson] 12729 12730 *) New Configure option "rsaref". 12731 [Ulf Möller] 12732 12733 *) Don't auto-generate pem.h. 12734 [Bodo Moeller] 12735 12736 *) Introduce type-safe ASN.1 SETs. 12737 [Ben Laurie] 12738 12739 *) Convert various additional casted stacks to type-safe STACK_OF() variants. 12740 [Ben Laurie, Ralf S. Engelschall, Steve Henson] 12741 12742 *) Introduce type-safe STACKs. This will almost certainly break lots of code 12743 that links with OpenSSL (well at least cause lots of warnings), but fear 12744 not: the conversion is trivial, and it eliminates loads of evil casts. A 12745 few STACKed things have been converted already. Feel free to convert more. 12746 In the fullness of time, I'll do away with the STACK type altogether. 12747 [Ben Laurie] 12748 12749 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate 12750 specified in <certfile> by updating the entry in the index.txt file. 12751 This way one no longer has to edit the index.txt file manually for 12752 revoking a certificate. The -revoke option does the gory details now. 12753 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall] 12754 12755 *) Fix `openssl crl -noout -text' combination where `-noout' killed the 12756 `-text' option at all and this way the `-noout -text' combination was 12757 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'. 12758 [Ralf S. Engelschall] 12759 12760 *) Make sure a corresponding plain text error message exists for the 12761 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a 12762 verify callback function determined that a certificate was revoked. 12763 [Ralf S. Engelschall] 12764 12765 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for 12766 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test 12767 all available ciphers including rc5, which was forgotten until now. 12768 In order to let the testing shell script know which algorithms 12769 are available, a new (up to now undocumented) command 12770 "openssl list-cipher-commands" is used. 12771 [Bodo Moeller] 12772 12773 *) Bugfix: s_client occasionally would sleep in select() when 12774 it should have checked SSL_pending() first. 12775 [Bodo Moeller] 12776 12777 *) New functions DSA_do_sign and DSA_do_verify to provide access to 12778 the raw DSA values prior to ASN.1 encoding. 12779 [Ulf Möller] 12780 12781 *) Tweaks to Configure 12782 [Niels Poppe <niels@netbox.org>] 12783 12784 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support, 12785 yet... 12786 [Steve Henson] 12787 12788 *) New variables $(RANLIB) and $(PERL) in the Makefiles. 12789 [Ulf Möller] 12790 12791 *) New config option to avoid instructions that are illegal on the 80386. 12792 The default code is faster, but requires at least a 486. 12793 [Ulf Möller] 12794 12795 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and 12796 SSL2_SERVER_VERSION (not used at all) macros, which are now the 12797 same as SSL2_VERSION anyway. 12798 [Bodo Moeller] 12799 12800 *) New "-showcerts" option for s_client. 12801 [Bodo Moeller] 12802 12803 *) Still more PKCS#12 integration. Add pkcs12 application to openssl 12804 application. Various cleanups and fixes. 12805 [Steve Henson] 12806 12807 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and 12808 modify error routines to work internally. Add error codes and PBE init 12809 to library startup routines. 12810 [Steve Henson] 12811 12812 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and 12813 packing functions to asn1 and evp. Changed function names and error 12814 codes along the way. 12815 [Steve Henson] 12816 12817 *) PKCS12 integration: and so it begins... First of several patches to 12818 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12 12819 objects to objects.h 12820 [Steve Henson] 12821 12822 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1 12823 and display support for Thawte strong extranet extension. 12824 [Steve Henson] 12825 12826 *) Add LinuxPPC support. 12827 [Jeff Dubrule <igor@pobox.org>] 12828 12829 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to 12830 bn_div_words in alpha.s. 12831 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie] 12832 12833 *) Make sure the RSA OAEP test is skipped under -DRSAref because 12834 OAEP isn't supported when OpenSSL is built with RSAref. 12835 [Ulf Moeller <ulf@fitug.de>] 12836 12837 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 12838 so they no longer are missing under -DNOPROTO. 12839 [Soren S. Jorvang <soren@t.dk>] 12840 12841 12842 Changes between 0.9.1c and 0.9.2b [22 Mar 1999] 12843 12844 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still 12845 doesn't work when the session is reused. Coming soon! 12846 [Ben Laurie] 12847 12848 *) Fix a security hole, that allows sessions to be reused in the wrong 12849 context thus bypassing client cert protection! All software that uses 12850 client certs and session caches in multiple contexts NEEDS PATCHING to 12851 allow session reuse! A fuller solution is in the works. 12852 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)] 12853 12854 *) Some more source tree cleanups (removed obsolete files 12855 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed 12856 permission on "config" script to be executable) and a fix for the INSTALL 12857 document. 12858 [Ulf Moeller <ulf@fitug.de>] 12859 12860 *) Remove some legacy and erroneous uses of malloc, free instead of 12861 Malloc, Free. 12862 [Lennart Bang <lob@netstream.se>, with minor changes by Steve] 12863 12864 *) Make rsa_oaep_test return non-zero on error. 12865 [Ulf Moeller <ulf@fitug.de>] 12866 12867 *) Add support for native Solaris shared libraries. Configure 12868 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice 12869 if someone would make that last step automatic. 12870 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>] 12871 12872 *) ctx_size was not built with the right compiler during "make links". Fixed. 12873 [Ben Laurie] 12874 12875 *) Change the meaning of 'ALL' in the cipher list. It now means "everything 12876 except NULL ciphers". This means the default cipher list will no longer 12877 enable NULL ciphers. They need to be specifically enabled e.g. with 12878 the string "DEFAULT:eNULL". 12879 [Steve Henson] 12880 12881 *) Fix to RSA private encryption routines: if p < q then it would 12882 occasionally produce an invalid result. This will only happen with 12883 externally generated keys because OpenSSL (and SSLeay) ensure p > q. 12884 [Steve Henson] 12885 12886 *) Be less restrictive and allow also `perl util/perlpath.pl 12887 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin', 12888 because this way one can also use an interpreter named `perl5' (which is 12889 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still 12890 installed as `perl'). 12891 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 12892 12893 *) Let util/clean-depend.pl work also with older Perl 5.00x versions. 12894 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 12895 12896 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add 12897 advapi32.lib to Win32 build and change the pem test comparison 12898 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the 12899 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h 12900 and crypto/des/ede_cbcm_enc.c. 12901 [Steve Henson] 12902 12903 *) DES quad checksum was broken on big-endian architectures. Fixed. 12904 [Ben Laurie] 12905 12906 *) Comment out two functions in bio.h that aren't implemented. Fix up the 12907 Win32 test batch file so it (might) work again. The Win32 test batch file 12908 is horrible: I feel ill.... 12909 [Steve Henson] 12910 12911 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected 12912 in e_os.h. Audit of header files to check ANSI and non ANSI 12913 sections: 10 functions were absent from non ANSI section and not exported 12914 from Windows DLLs. Fixed up libeay.num for new functions. 12915 [Steve Henson] 12916 12917 *) Make `openssl version' output lines consistent. 12918 [Ralf S. Engelschall] 12919 12920 *) Fix Win32 symbol export lists for BIO functions: Added 12921 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data 12922 to ms/libeay{16,32}.def. 12923 [Ralf S. Engelschall] 12924 12925 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled 12926 fine under Unix and passes some trivial tests I've now added. But the 12927 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was 12928 added to make sure no one expects that this stuff really works in the 12929 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources 12930 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and 12931 openssl_bio.xs. 12932 [Ralf S. Engelschall] 12933 12934 *) Fix the generation of two part addresses in perl. 12935 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie] 12936 12937 *) Add config entry for Linux on MIPS. 12938 [John Tobey <jtobey@channel1.com>] 12939 12940 *) Make links whenever Configure is run, unless we are on Windoze. 12941 [Ben Laurie] 12942 12943 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf. 12944 Currently only issuerAltName and AuthorityKeyIdentifier make any sense 12945 in CRLs. 12946 [Steve Henson] 12947 12948 *) Add a useful kludge to allow package maintainers to specify compiler and 12949 other platforms details on the command line without having to patch the 12950 Configure script every time: One now can use ``perl Configure 12951 <id>:<details>'', i.e. platform ids are allowed to have details appended 12952 to them (separated by colons). This is treated as there would be a static 12953 pre-configured entry in Configure's %table under key <id> with value 12954 <details> and ``perl Configure <id>'' is called. So, when you want to 12955 perform a quick test-compile under FreeBSD 3.1 with pgcc and without 12956 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"'' 12957 now, which overrides the FreeBSD-elf entry on-the-fly. 12958 [Ralf S. Engelschall] 12959 12960 *) Disable new TLS1 ciphersuites by default: they aren't official yet. 12961 [Ben Laurie] 12962 12963 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified 12964 on the `perl Configure ...' command line. This way one can compile 12965 OpenSSL libraries with Position Independent Code (PIC) which is needed 12966 for linking it into DSOs. 12967 [Ralf S. Engelschall] 12968 12969 *) Remarkably, export ciphers were totally broken and no-one had noticed! 12970 Fixed. 12971 [Ben Laurie] 12972 12973 *) Cleaned up the LICENSE document: The official contact for any license 12974 questions now is the OpenSSL core team under openssl-core@openssl.org. 12975 And add a paragraph about the dual-license situation to make sure people 12976 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply 12977 to the OpenSSL toolkit. 12978 [Ralf S. Engelschall] 12979 12980 *) General source tree makefile cleanups: Made `making xxx in yyy...' 12981 display consistent in the source tree and replaced `/bin/rm' by `rm'. 12982 Additionally cleaned up the `make links' target: Remove unnecessary 12983 semicolons, subsequent redundant removes, inline point.sh into mklink.sh 12984 to speed processing and no longer clutter the display with confusing 12985 stuff. Instead only the actually done links are displayed. 12986 [Ralf S. Engelschall] 12987 12988 *) Permit null encryption ciphersuites, used for authentication only. It used 12989 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this. 12990 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null 12991 encryption. 12992 [Ben Laurie] 12993 12994 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder 12995 signed attributes when verifying signatures (this would break them), 12996 the detached data encoding was wrong and public keys obtained using 12997 X509_get_pubkey() weren't freed. 12998 [Steve Henson] 12999 13000 *) Add text documentation for the BUFFER functions. Also added a work around 13001 to a Win95 console bug. This was triggered by the password read stuff: the 13002 last character typed gets carried over to the next fread(). If you were 13003 generating a new cert request using 'req' for example then the last 13004 character of the passphrase would be CR which would then enter the first 13005 field as blank. 13006 [Steve Henson] 13007 13008 *) Added the new `Includes OpenSSL Cryptography Software' button as 13009 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay 13010 button and can be used by applications based on OpenSSL to show the 13011 relationship to the OpenSSL project. 13012 [Ralf S. Engelschall] 13013 13014 *) Remove confusing variables in function signatures in files 13015 ssl/ssl_lib.c and ssl/ssl.h. 13016 [Lennart Bong <lob@kulthea.stacken.kth.se>] 13017 13018 *) Don't install bss_file.c under PREFIX/include/ 13019 [Lennart Bong <lob@kulthea.stacken.kth.se>] 13020 13021 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle 13022 functions that return function pointers and has support for NT specific 13023 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various 13024 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from 13025 unsigned to signed types: this was killing the Win32 compile. 13026 [Steve Henson] 13027 13028 *) Add new certificate file to stack functions, 13029 SSL_add_dir_cert_subjects_to_stack() and 13030 SSL_add_file_cert_subjects_to_stack(). These largely supplant 13031 SSL_load_client_CA_file(), and can be used to add multiple certs easily 13032 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()). 13033 This means that Apache-SSL and similar packages don't have to mess around 13034 to add as many CAs as they want to the preferred list. 13035 [Ben Laurie] 13036 13037 *) Experiment with doxygen documentation. Currently only partially applied to 13038 ssl/ssl_lib.c. 13039 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with 13040 openssl.doxy as the configuration file. 13041 [Ben Laurie] 13042 13043 *) Get rid of remaining C++-style comments which strict C compilers hate. 13044 [Ralf S. Engelschall, pointed out by Carlos Amengual] 13045 13046 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not 13047 compiled in by default: it has problems with large keys. 13048 [Steve Henson] 13049 13050 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and 13051 DH private keys and/or callback functions which directly correspond to 13052 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This 13053 is needed for applications which have to configure certificates on a 13054 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis 13055 (e.g. s_server). 13056 For the RSA certificate situation is makes no difference, but 13057 for the DSA certificate situation this fixes the "no shared cipher" 13058 problem where the OpenSSL cipher selection procedure failed because the 13059 temporary keys were not overtaken from the context and the API provided 13060 no way to reconfigure them. 13061 The new functions now let applications reconfigure the stuff and they 13062 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh, 13063 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new 13064 non-public-API function ssl_cert_instantiate() is used as a helper 13065 function and also to reduce code redundancy inside ssl_rsa.c. 13066 [Ralf S. Engelschall] 13067 13068 *) Move s_server -dcert and -dkey options out of the undocumented feature 13069 area because they are useful for the DSA situation and should be 13070 recognized by the users. 13071 [Ralf S. Engelschall] 13072 13073 *) Fix the cipher decision scheme for export ciphers: the export bits are 13074 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within 13075 SSL_EXP_MASK. So, the original variable has to be used instead of the 13076 already masked variable. 13077 [Richard Levitte <levitte@stacken.kth.se>] 13078 13079 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c 13080 [Richard Levitte <levitte@stacken.kth.se>] 13081 13082 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal() 13083 from `int' to `unsigned int' because it's a length and initialized by 13084 EVP_DigestFinal() which expects an `unsigned int *'. 13085 [Richard Levitte <levitte@stacken.kth.se>] 13086 13087 *) Don't hard-code path to Perl interpreter on shebang line of Configure 13088 script. Instead use the usual Shell->Perl transition trick. 13089 [Ralf S. Engelschall] 13090 13091 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates 13092 (in addition to RSA certificates) to match the behaviour of `openssl dsa 13093 -noout -modulus' as it's already the case for `openssl rsa -noout 13094 -modulus'. For RSA the -modulus is the real "modulus" while for DSA 13095 currently the public key is printed (a decision which was already done by 13096 `openssl dsa -modulus' in the past) which serves a similar purpose. 13097 Additionally the NO_RSA no longer completely removes the whole -modulus 13098 option; it now only avoids using the RSA stuff. Same applies to NO_DSA 13099 now, too. 13100 [Ralf S. Engelschall] 13101 13102 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested 13103 BIO. See the source (crypto/evp/bio_ok.c) for more info. 13104 [Arne Ansper <arne@ats.cyber.ee>] 13105 13106 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs 13107 to be added. Now both 'req' and 'ca' can use new objects defined in the 13108 config file. 13109 [Steve Henson] 13110 13111 *) Add cool BIO that does syslog (or event log on NT). 13112 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie] 13113 13114 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5, 13115 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and 13116 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher 13117 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt. 13118 [Ben Laurie] 13119 13120 *) Add preliminary config info for new extension code. 13121 [Steve Henson] 13122 13123 *) Make RSA_NO_PADDING really use no padding. 13124 [Ulf Moeller <ulf@fitug.de>] 13125 13126 *) Generate errors when private/public key check is done. 13127 [Ben Laurie] 13128 13129 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support 13130 for some CRL extensions and new objects added. 13131 [Steve Henson] 13132 13133 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private 13134 key usage extension and fuller support for authority key id. 13135 [Steve Henson] 13136 13137 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved 13138 padding method for RSA, which is recommended for new applications in PKCS 13139 #1 v2.0 (RFC 2437, October 1998). 13140 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical 13141 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure 13142 against Bleichbacher's attack on RSA. 13143 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by 13144 Ben Laurie] 13145 13146 *) Updates to the new SSL compression code 13147 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13148 13149 *) Fix so that the version number in the master secret, when passed 13150 via RSA, checks that if TLS was proposed, but we roll back to SSLv3 13151 (because the server will not accept higher), that the version number 13152 is 0x03,0x01, not 0x03,0x00 13153 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13154 13155 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory 13156 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes 13157 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c 13158 [Steve Henson] 13159 13160 *) Support for RAW extensions where an arbitrary extension can be 13161 created by including its DER encoding. See apps/openssl.cnf for 13162 an example. 13163 [Steve Henson] 13164 13165 *) Make sure latest Perl versions don't interpret some generated C array 13166 code as Perl array code in the crypto/err/err_genc.pl script. 13167 [Lars Weber <3weber@informatik.uni-hamburg.de>] 13168 13169 *) Modify ms/do_ms.bat to not generate assembly language makefiles since 13170 not many people have the assembler. Various Win32 compilation fixes and 13171 update to the INSTALL.W32 file with (hopefully) more accurate Win32 13172 build instructions. 13173 [Steve Henson] 13174 13175 *) Modify configure script 'Configure' to automatically create crypto/date.h 13176 file under Win32 and also build pem.h from pem.org. New script 13177 util/mkfiles.pl to create the MINFO file on environments that can't do a 13178 'make files': perl util/mkfiles.pl >MINFO should work. 13179 [Steve Henson] 13180 13181 *) Major rework of DES function declarations, in the pursuit of correctness 13182 and purity. As a result, many evil casts evaporated, and some weirdness, 13183 too. You may find this causes warnings in your code. Zapping your evil 13184 casts will probably fix them. Mostly. 13185 [Ben Laurie] 13186 13187 *) Fix for a typo in asn1.h. Bug fix to object creation script 13188 obj_dat.pl. It considered a zero in an object definition to mean 13189 "end of object": none of the objects in objects.h have any zeros 13190 so it wasn't spotted. 13191 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>] 13192 13193 *) Add support for Triple DES Cipher Block Chaining with Output Feedback 13194 Masking (CBCM). In the absence of test vectors, the best I have been able 13195 to do is check that the decrypt undoes the encrypt, so far. Send me test 13196 vectors if you have them. 13197 [Ben Laurie] 13198 13199 *) Correct calculation of key length for export ciphers (too much space was 13200 allocated for null ciphers). This has not been tested! 13201 [Ben Laurie] 13202 13203 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage 13204 message is now correct (it understands "crypto" and "ssl" on its 13205 command line). There is also now an "update" option. This will update 13206 the util/ssleay.num and util/libeay.num files with any new functions. 13207 If you do a: 13208 perl util/mkdef.pl crypto ssl update 13209 it will update them. 13210 [Steve Henson] 13211 13212 *) Overhauled the Perl interface (perl/*): 13213 - ported BN stuff to OpenSSL's different BN library 13214 - made the perl/ source tree CVS-aware 13215 - renamed the package from SSLeay to OpenSSL (the files still contain 13216 their history because I've copied them in the repository) 13217 - removed obsolete files (the test scripts will be replaced 13218 by better Test::Harness variants in the future) 13219 [Ralf S. Engelschall] 13220 13221 *) First cut for a very conservative source tree cleanup: 13222 1. merge various obsolete readme texts into doc/ssleay.txt 13223 where we collect the old documents and readme texts. 13224 2. remove the first part of files where I'm already sure that we no 13225 longer need them because of three reasons: either they are just temporary 13226 files which were left by Eric or they are preserved original files where 13227 I've verified that the diff is also available in the CVS via "cvs diff 13228 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for 13229 the crypto/md/ stuff). 13230 [Ralf S. Engelschall] 13231 13232 *) More extension code. Incomplete support for subject and issuer alt 13233 name, issuer and authority key id. Change the i2v function parameters 13234 and add an extra 'crl' parameter in the X509V3_CTX structure: guess 13235 what that's for :-) Fix to ASN1 macro which messed up 13236 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED. 13237 [Steve Henson] 13238 13239 *) Preliminary support for ENUMERATED type. This is largely copied from the 13240 INTEGER code. 13241 [Steve Henson] 13242 13243 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy. 13244 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13245 13246 *) Make sure `make rehash' target really finds the `openssl' program. 13247 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>] 13248 13249 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd 13250 like to hear about it if this slows down other processors. 13251 [Ben Laurie] 13252 13253 *) Add CygWin32 platform information to Configure script. 13254 [Alan Batie <batie@aahz.jf.intel.com>] 13255 13256 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm' 13257 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>] 13258 13259 *) New program nseq to manipulate netscape certificate sequences 13260 [Steve Henson] 13261 13262 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a 13263 few typos. 13264 [Steve Henson] 13265 13266 *) Fixes to BN code. Previously the default was to define BN_RECURSION 13267 but the BN code had some problems that would cause failures when 13268 doing certificate verification and some other functions. 13269 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)] 13270 13271 *) Add ASN1 and PEM code to support netscape certificate sequences. 13272 [Steve Henson] 13273 13274 *) Add ASN1 and PEM code to support netscape certificate sequences. 13275 [Steve Henson] 13276 13277 *) Add several PKIX and private extended key usage OIDs. 13278 [Steve Henson] 13279 13280 *) Modify the 'ca' program to handle the new extension code. Modify 13281 openssl.cnf for new extension format, add comments. 13282 [Steve Henson] 13283 13284 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req' 13285 and add a sample to openssl.cnf so req -x509 now adds appropriate 13286 CA extensions. 13287 [Steve Henson] 13288 13289 *) Continued X509 V3 changes. Add to other makefiles, integrate with the 13290 error code, add initial support to X509_print() and x509 application. 13291 [Steve Henson] 13292 13293 *) Takes a deep breath and start adding X509 V3 extension support code. Add 13294 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this 13295 stuff is currently isolated and isn't even compiled yet. 13296 [Steve Henson] 13297 13298 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL 13299 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print. 13300 Removed the versions check from X509 routines when loading extensions: 13301 this allows certain broken certificates that don't set the version 13302 properly to be processed. 13303 [Steve Henson] 13304 13305 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another 13306 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which 13307 can still be regenerated with "make depend". 13308 [Ben Laurie] 13309 13310 *) Spelling mistake in C version of CAST-128. 13311 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>] 13312 13313 *) Changes to the error generation code. The perl script err-code.pl 13314 now reads in the old error codes and retains the old numbers, only 13315 adding new ones if necessary. It also only changes the .err files if new 13316 codes are added. The makefiles have been modified to only insert errors 13317 when needed (to avoid needlessly modifying header files). This is done 13318 by only inserting errors if the .err file is newer than the auto generated 13319 C file. To rebuild all the error codes from scratch (the old behaviour) 13320 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl 13321 or delete all the .err files. 13322 [Steve Henson] 13323 13324 *) CAST-128 was incorrectly implemented for short keys. The C version has 13325 been fixed, but is untested. The assembler versions are also fixed, but 13326 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing 13327 to regenerate it if needed. 13328 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun 13329 Hagino <itojun@kame.net>] 13330 13331 *) File was opened incorrectly in randfile.c. 13332 [Ulf Möller <ulf@fitug.de>] 13333 13334 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print 13335 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or 13336 GeneralizedTime. ASN1_TIME is the proper type used in certificates et 13337 al: it's just almost always a UTCTime. Note this patch adds new error 13338 codes so do a "make errors" if there are problems. 13339 [Steve Henson] 13340 13341 *) Correct Linux 1 recognition in config. 13342 [Ulf Möller <ulf@fitug.de>] 13343 13344 *) Remove pointless MD5 hash when using DSA keys in ca. 13345 [Anonymous <nobody@replay.com>] 13346 13347 *) Generate an error if given an empty string as a cert directory. Also 13348 generate an error if handed NULL (previously returned 0 to indicate an 13349 error, but didn't set one). 13350 [Ben Laurie, reported by Anonymous <nobody@replay.com>] 13351 13352 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last. 13353 [Ben Laurie] 13354 13355 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct 13356 parameters. This was causing a warning which killed off the Win32 compile. 13357 [Steve Henson] 13358 13359 *) Remove C++ style comments from crypto/bn/bn_local.h. 13360 [Neil Costigan <neil.costigan@celocom.com>] 13361 13362 *) The function OBJ_txt2nid was broken. It was supposed to return a nid 13363 based on a text string, looking up short and long names and finally 13364 "dot" format. The "dot" format stuff didn't work. Added new function 13365 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 13366 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the 13367 OID is not part of the table. 13368 [Steve Henson] 13369 13370 *) Add prototypes to X509 lookup/verify methods, fixing a bug in 13371 X509_LOOKUP_by_alias(). 13372 [Ben Laurie] 13373 13374 *) Sort openssl functions by name. 13375 [Ben Laurie] 13376 13377 *) Get the gendsa program working (hopefully) and add it to app list. Remove 13378 encryption from sample DSA keys (in case anyone is interested the password 13379 was "1234"). 13380 [Steve Henson] 13381 13382 *) Make _all_ *_free functions accept a NULL pointer. 13383 [Frans Heymans <fheymans@isaserver.be>] 13384 13385 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use 13386 NULL pointers. 13387 [Anonymous <nobody@replay.com>] 13388 13389 *) s_server should send the CAfile as acceptable CAs, not its own cert. 13390 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 13391 13392 *) Don't blow it for numeric -newkey arguments to apps/req. 13393 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>] 13394 13395 *) Temp key "for export" tests were wrong in s3_srvr.c. 13396 [Anonymous <nobody@replay.com>] 13397 13398 *) Add prototype for temp key callback functions 13399 SSL_CTX_set_tmp_{rsa,dh}_callback(). 13400 [Ben Laurie] 13401 13402 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and 13403 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey(). 13404 [Steve Henson] 13405 13406 *) X509_name_add_entry() freed the wrong thing after an error. 13407 [Arne Ansper <arne@ats.cyber.ee>] 13408 13409 *) rsa_eay.c would attempt to free a NULL context. 13410 [Arne Ansper <arne@ats.cyber.ee>] 13411 13412 *) BIO_s_socket() had a broken should_retry() on Windoze. 13413 [Arne Ansper <arne@ats.cyber.ee>] 13414 13415 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH. 13416 [Arne Ansper <arne@ats.cyber.ee>] 13417 13418 *) Make sure the already existing X509_STORE->depth variable is initialized 13419 in X509_STORE_new(), but document the fact that this variable is still 13420 unused in the certificate verification process. 13421 [Ralf S. Engelschall] 13422 13423 *) Fix the various library and apps files to free up pkeys obtained from 13424 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions. 13425 [Steve Henson] 13426 13427 *) Fix reference counting in X509_PUBKEY_get(). This makes 13428 demos/maurice/example2.c work, amongst others, probably. 13429 [Steve Henson and Ben Laurie] 13430 13431 *) First cut of a cleanup for apps/. First the `ssleay' program is now named 13432 `openssl' and second, the shortcut symlinks for the `openssl <command>' 13433 are no longer created. This way we have a single and consistent command 13434 line interface `openssl <command>', similar to `cvs <command>'. 13435 [Ralf S. Engelschall, Paul Sutton and Ben Laurie] 13436 13437 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey 13438 BIT STRING wrapper always have zero unused bits. 13439 [Steve Henson] 13440 13441 *) Add CA.pl, perl version of CA.sh, add extended key usage OID. 13442 [Steve Henson] 13443 13444 *) Make the top-level INSTALL documentation easier to understand. 13445 [Paul Sutton] 13446 13447 *) Makefiles updated to exit if an error occurs in a sub-directory 13448 make (including if user presses ^C) [Paul Sutton] 13449 13450 *) Make Montgomery context stuff explicit in RSA data structure. 13451 [Ben Laurie] 13452 13453 *) Fix build order of pem and err to allow for generated pem.h. 13454 [Ben Laurie] 13455 13456 *) Fix renumbering bug in X509_NAME_delete_entry(). 13457 [Ben Laurie] 13458 13459 *) Enhanced the err-ins.pl script so it makes the error library number 13460 global and can add a library name. This is needed for external ASN1 and 13461 other error libraries. 13462 [Steve Henson] 13463 13464 *) Fixed sk_insert which never worked properly. 13465 [Steve Henson] 13466 13467 *) Fix ASN1 macros so they can handle indefinite length constructed 13468 EXPLICIT tags. Some non standard certificates use these: they can now 13469 be read in. 13470 [Steve Henson] 13471 13472 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc) 13473 into a single doc/ssleay.txt bundle. This way the information is still 13474 preserved but no longer messes up this directory. Now it's new room for 13475 the new set of documentation files. 13476 [Ralf S. Engelschall] 13477 13478 *) SETs were incorrectly DER encoded. This was a major pain, because they 13479 shared code with SEQUENCEs, which aren't coded the same. This means that 13480 almost everything to do with SETs or SEQUENCEs has either changed name or 13481 number of arguments. 13482 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>] 13483 13484 *) Fix test data to work with the above. 13485 [Ben Laurie] 13486 13487 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but 13488 was already fixed by Eric for 0.9.1 it seems. 13489 [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>] 13490 13491 *) Autodetect FreeBSD3. 13492 [Ben Laurie] 13493 13494 *) Fix various bugs in Configure. This affects the following platforms: 13495 nextstep 13496 ncr-scde 13497 unixware-2.0 13498 unixware-2.0-pentium 13499 sco5-cc. 13500 [Ben Laurie] 13501 13502 *) Eliminate generated files from CVS. Reorder tests to regenerate files 13503 before they are needed. 13504 [Ben Laurie] 13505 13506 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy). 13507 [Ben Laurie] 13508 13509 13510 Changes between 0.9.1b and 0.9.1c [23-Dec-1998] 13511 13512 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 13513 changed SSLeay to OpenSSL in version strings. 13514 [Ralf S. Engelschall] 13515 13516 *) Some fixups to the top-level documents. 13517 [Paul Sutton] 13518 13519 *) Fixed the nasty bug where rsaref.h was not found under compile-time 13520 because the symlink to include/ was missing. 13521 [Ralf S. Engelschall] 13522 13523 *) Incorporated the popular no-RSA/DSA-only patches 13524 which allow to compile a RSA-free SSLeay. 13525 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall] 13526 13527 *) Fixed nasty rehash problem under `make -f Makefile.ssl links' 13528 when "ssleay" is still not found. 13529 [Ralf S. Engelschall] 13530 13531 *) Added more platforms to Configure: Cray T3E, HPUX 11, 13532 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>] 13533 13534 *) Updated the README file. 13535 [Ralf S. Engelschall] 13536 13537 *) Added various .cvsignore files in the CVS repository subdirs 13538 to make a "cvs update" really silent. 13539 [Ralf S. Engelschall] 13540 13541 *) Recompiled the error-definition header files and added 13542 missing symbols to the Win32 linker tables. 13543 [Ralf S. Engelschall] 13544 13545 *) Cleaned up the top-level documents; 13546 o new files: CHANGES and LICENSE 13547 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 13548 o merged COPYRIGHT into LICENSE 13549 o removed obsolete TODO file 13550 o renamed MICROSOFT to INSTALL.W32 13551 [Ralf S. Engelschall] 13552 13553 *) Removed dummy files from the 0.9.1b source tree: 13554 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi 13555 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f 13556 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f 13557 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f 13558 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f 13559 [Ralf S. Engelschall] 13560 13561 *) Added various platform portability fixes. 13562 [Mark J. Cox] 13563 13564 *) The Genesis of the OpenSSL rpject: 13565 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A. 13566 Young and Tim J. Hudson created while they were working for C2Net until 13567 summer 1998. 13568 [The OpenSSL Project] 13569 13570 13571 Changes between 0.9.0b and 0.9.1b [not released] 13572 13573 *) Updated a few CA certificates under certs/ 13574 [Eric A. Young] 13575 13576 *) Changed some BIGNUM api stuff. 13577 [Eric A. Young] 13578 13579 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 13580 DGUX x86, Linux Alpha, etc. 13581 [Eric A. Young] 13582 13583 *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 13584 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is 13585 available). 13586 [Eric A. Young] 13587 13588 *) Add -strparse option to asn1pars program which parses nested 13589 binary structures 13590 [Dr Stephen Henson <shenson@bigfoot.com>] 13591 13592 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs. 13593 [Eric A. Young] 13594 13595 *) DSA fix for "ca" program. 13596 [Eric A. Young] 13597 13598 *) Added "-genkey" option to "dsaparam" program. 13599 [Eric A. Young] 13600 13601 *) Added RIPE MD160 (rmd160) message digest. 13602 [Eric A. Young] 13603 13604 *) Added -a (all) option to "ssleay version" command. 13605 [Eric A. Young] 13606 13607 *) Added PLATFORM define which is the id given to Configure. 13608 [Eric A. Young] 13609 13610 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking. 13611 [Eric A. Young] 13612 13613 *) Extended the ASN.1 parser routines. 13614 [Eric A. Young] 13615 13616 *) Extended BIO routines to support REUSEADDR, seek, tell, etc. 13617 [Eric A. Young] 13618 13619 *) Added a BN_CTX to the BN library. 13620 [Eric A. Young] 13621 13622 *) Fixed the weak key values in DES library 13623 [Eric A. Young] 13624 13625 *) Changed API in EVP library for cipher aliases. 13626 [Eric A. Young] 13627 13628 *) Added support for RC2/64bit cipher. 13629 [Eric A. Young] 13630 13631 *) Converted the lhash library to the crypto/mem.c functions. 13632 [Eric A. Young] 13633 13634 *) Added more recognized ASN.1 object ids. 13635 [Eric A. Young] 13636 13637 *) Added more RSA padding checks for SSL/TLS. 13638 [Eric A. Young] 13639 13640 *) Added BIO proxy/filter functionality. 13641 [Eric A. Young] 13642 13643 *) Added extra_certs to SSL_CTX which can be used 13644 send extra CA certificates to the client in the CA cert chain sending 13645 process. It can be configured with SSL_CTX_add_extra_chain_cert(). 13646 [Eric A. Young] 13647 13648 *) Now Fortezza is denied in the authentication phase because 13649 this is key exchange mechanism is not supported by SSLeay at all. 13650 [Eric A. Young] 13651 13652 *) Additional PKCS1 checks. 13653 [Eric A. Young] 13654 13655 *) Support the string "TLSv1" for all TLS v1 ciphers. 13656 [Eric A. Young] 13657 13658 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the 13659 ex_data index of the SSL context in the X509_STORE_CTX ex_data. 13660 [Eric A. Young] 13661 13662 *) Fixed a few memory leaks. 13663 [Eric A. Young] 13664 13665 *) Fixed various code and comment typos. 13666 [Eric A. Young] 13667 13668 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 13669 bytes sent in the client random. 13670 [Edward Bishop <ebishop@spyglass.com>] 13671