CHANGES

   1
   2  OpenSSL CHANGES
   3  _______________
   4
   5  Changes between 1.1.0e and 1.1.1 [xx XXX xxxx]
   6
   7   *) Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
   8      with Z.  These are meant to replace LONG and ZLONG and to be size safe.
   9      The use of LONG and ZLONG is discouraged and scheduled for deprecation
  10      in OpenSSL 1.2.0.
  11      [Richard Levitte]
  12
  13   *) Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
  14      'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
  15      [Richard Levitte, Andy Polyakov]
  16
  17   *) Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
  18      does for RSA, etc.
  19      [Richard Levitte]
  20
  21   *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
  22      platform rather than 'mingw'.
  23      [Richard Levitte]
  24
  25   *) The functions X509_STORE_add_cert and X509_STORE_add_crl return
  26      success if they are asked to add an object which already exists
  27      in the store. This change cascades to other functions which load
  28      certificates and CRLs.
  29      [Paul Dale]
  30
  31   *) x86_64 assembly pack: annotate code with DWARF CFI directives to
  32      facilitate stack unwinding even from assembly subroutines.
  33      [Andy Polyakov]
  34
  35   *) Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
  36      Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
  37      [Richard Levitte]
  38
  39   *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
  40      VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
  41      which is the minimum version we support.
  42      [Richard Levitte]
  43
  44   *) Certificate time validation (X509_cmp_time) enforces stricter
  45      compliance with RFC 5280. Fractional seconds and timezone offsets
  46      are no longer allowed.
  47      [Emilia Käsper]
  48
  49   *) Add support for ARIA
  50      [Paul Dale]
  51
  52   *) Add support for SipHash
  53      [Todd Short]
  54
  55   *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0
  56      or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
  57      prevent issues where no progress is being made and the peer continually
  58      sends unrecognised record types, using up resources processing them.
  59      [Matt Caswell]
  60
  61   *) 'openssl passwd' can now produce SHA256 and SHA512 based output,
  62      using the algorithm defined in
  63      https://www.akkadia.org/drepper/SHA-crypt.txt
  64      [Richard Levitte]
  65
  66   *) Heartbeat support has been removed; the ABI is changed for now.
  67      [Richard Levitte, Rich Salz]
  68
  69   *) Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
  70      [Emilia Käsper]
  71
  72   *) The RSA "null" method, which was partially supported to avoid patent
  73      issues, has been replaced to always returns NULL.
  74      [Rich Salz]
  75
  76  Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
  77
  78   *) Encrypt-Then-Mac renegotiation crash
  79
  80      During a renegotiation handshake if the Encrypt-Then-Mac extension is
  81      negotiated where it was not in the original handshake (or vice-versa) then
  82      this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
  83      and servers are affected.
  84
  85      This issue was reported to OpenSSL by Joe Orton (Red Hat).
  86      (CVE-2017-3733)
  87      [Matt Caswell]
  88
  89  Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
  90
  91   *) Truncated packet could crash via OOB read
  92
  93      If one side of an SSL/TLS path is running on a 32-bit host and a specific
  94      cipher is being used, then a truncated packet can cause that host to
  95      perform an out-of-bounds read, usually resulting in a crash.
  96
  97      This issue was reported to OpenSSL by Robert Święcki of Google.
  98      (CVE-2017-3731)
  99      [Andy Polyakov]
 100
 101   *) Bad (EC)DHE parameters cause a client crash
 102
 103      If a malicious server supplies bad parameters for a DHE or ECDHE key
 104      exchange then this can result in the client attempting to dereference a
 105      NULL pointer leading to a client crash. This could be exploited in a Denial
 106      of Service attack.
 107
 108      This issue was reported to OpenSSL by Guido Vranken.
 109      (CVE-2017-3730)
 110      [Matt Caswell]
 111
 112   *) BN_mod_exp may produce incorrect results on x86_64
 113
 114      There is a carry propagating bug in the x86_64 Montgomery squaring
 115      procedure. No EC algorithms are affected. Analysis suggests that attacks
 116      against RSA and DSA as a result of this defect would be very difficult to
 117      perform and are not believed likely. Attacks against DH are considered just
 118      feasible (although very difficult) because most of the work necessary to
 119      deduce information about a private key may be performed offline. The amount
 120      of resources required for such an attack would be very significant and
 121      likely only accessible to a limited number of attackers. An attacker would
 122      additionally need online access to an unpatched system using the target
 123      private key in a scenario with persistent DH parameters and a private
 124      key that is shared between multiple clients. For example this can occur by
 125      default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
 126      similar to CVE-2015-3193 but must be treated as a separate problem.
 127
 128      This issue was reported to OpenSSL by the OSS-Fuzz project.
 129      (CVE-2017-3732)
 130      [Andy Polyakov]
 131
 132  Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
 133
 134   *) ChaCha20/Poly1305 heap-buffer-overflow
 135
 136      TLS connections using *-CHACHA20-POLY1305 ciphersuites are susceptible to
 137      a DoS attack by corrupting larger payloads. This can result in an OpenSSL
 138      crash. This issue is not considered to be exploitable beyond a DoS.
 139
 140      This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
 141      (CVE-2016-7054)
 142      [Richard Levitte]
 143
 144   *) CMS Null dereference
 145
 146      Applications parsing invalid CMS structures can crash with a NULL pointer
 147      dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
 148      type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
 149      structure callback if an attempt is made to free certain invalid encodings.
 150      Only CHOICE structures using a callback which do not handle NULL value are
 151      affected.
 152
 153      This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
 154      (CVE-2016-7053)
 155      [Stephen Henson]
 156
 157   *) Montgomery multiplication may produce incorrect results
 158
 159      There is a carry propagating bug in the Broadwell-specific Montgomery
 160      multiplication procedure that handles input lengths divisible by, but
 161      longer than 256 bits. Analysis suggests that attacks against RSA, DSA
 162      and DH private keys are impossible. This is because the subroutine in
 163      question is not used in operations with the private key itself and an input
 164      of the attacker's direct choice. Otherwise the bug can manifest itself as
 165      transient authentication and key negotiation failures or reproducible
 166      erroneous outcome of public-key operations with specially crafted input.
 167      Among EC algorithms only Brainpool P-512 curves are affected and one
 168      presumably can attack ECDH key negotiation. Impact was not analyzed in
 169      detail, because pre-requisites for attack are considered unlikely. Namely
 170      multiple clients have to choose the curve in question and the server has to
 171      share the private key among them, neither of which is default behaviour.
 172      Even then only clients that chose the curve will be affected.
 173
 174      This issue was publicly reported as transient failures and was not
 175      initially recognized as a security issue. Thanks to Richard Morgan for
 176      providing reproducible case.
 177      (CVE-2016-7055)
 178      [Andy Polyakov]
 179
 180   *) Removed automatic addition of RPATH in shared libraries and executables,
 181      as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
 182      [Richard Levitte]
 183
 184  Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
 185
 186   *) Fix Use After Free for large message sizes
 187
 188      The patch applied to address CVE-2016-6307 resulted in an issue where if a
 189      message larger than approx 16k is received then the underlying buffer to
 190      store the incoming message is reallocated and moved. Unfortunately a
 191      dangling pointer to the old location is left which results in an attempt to
 192      write to the previously freed location. This is likely to result in a
 193      crash, however it could potentially lead to execution of arbitrary code.
 194
 195      This issue only affects OpenSSL 1.1.0a.
 196
 197      This issue was reported to OpenSSL by Robert Święcki.
 198      (CVE-2016-6309)
 199      [Matt Caswell]
 200
 201  Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
 202
 203   *) OCSP Status Request extension unbounded memory growth
 204
 205      A malicious client can send an excessively large OCSP Status Request
 206      extension. If that client continually requests renegotiation, sending a
 207      large OCSP Status Request extension each time, then there will be unbounded
 208      memory growth on the server. This will eventually lead to a Denial Of
 209      Service attack through memory exhaustion. Servers with a default
 210      configuration are vulnerable even if they do not support OCSP. Builds using
 211      the "no-ocsp" build time option are not affected.
 212
 213      This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
 214      (CVE-2016-6304)
 215      [Matt Caswell]
 216
 217   *) SSL_peek() hang on empty record
 218
 219      OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
 220      sends an empty record. This could be exploited by a malicious peer in a
 221      Denial Of Service attack.
 222
 223      This issue was reported to OpenSSL by Alex Gaynor.
 224      (CVE-2016-6305)
 225      [Matt Caswell]
 226
 227   *) Excessive allocation of memory in tls_get_message_header() and
 228      dtls1_preprocess_fragment()
 229
 230      A (D)TLS message includes 3 bytes for its length in the header for the
 231      message. This would allow for messages up to 16Mb in length. Messages of
 232      this length are excessive and OpenSSL includes a check to ensure that a
 233      peer is sending reasonably sized messages in order to avoid too much memory
 234      being consumed to service a connection. A flaw in the logic of version
 235      1.1.0 means that memory for the message is allocated too early, prior to
 236      the excessive message length check. Due to way memory is allocated in
 237      OpenSSL this could mean an attacker could force up to 21Mb to be allocated
 238      to service a connection. This could lead to a Denial of Service through
 239      memory exhaustion. However, the excessive message length check still takes
 240      place, and this would cause the connection to immediately fail. Assuming
 241      that the application calls SSL_free() on the failed connection in a timely
 242      manner then the 21Mb of allocated memory will then be immediately freed
 243      again. Therefore the excessive memory allocation will be transitory in
 244      nature. This then means that there is only a security impact if:
 245
 246      1) The application does not call SSL_free() in a timely manner in the event
 247      that the connection fails
 248      or
 249      2) The application is working in a constrained environment where there is
 250      very little free memory
 251      or
 252      3) The attacker initiates multiple connection attempts such that there are
 253      multiple connections in a state where memory has been allocated for the
 254      connection; SSL_free() has not yet been called; and there is insufficient
 255      memory to service the multiple requests.
 256
 257      Except in the instance of (1) above any Denial Of Service is likely to be
 258      transitory because as soon as the connection fails the memory is
 259      subsequently freed again in the SSL_free() call. However there is an
 260      increased risk during this period of application crashes due to the lack of
 261      memory - which would then mean a more serious Denial of Service.
 262
 263      This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
 264      (CVE-2016-6307 and CVE-2016-6308)
 265      [Matt Caswell]
 266
 267   *) solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
 268      had to be removed. Primary reason is that vendor assembler can't
 269      assemble our modules with -KPIC flag. As result it, assembly
 270      support, was not even available as option. But its lack means
 271      lack of side-channel resistant code, which is incompatible with
 272      security by todays standards. Fortunately gcc is readily available
 273      prepackaged option, which we firmly point at...
 274      [Andy Polyakov]
 275
 276  Changes between 1.0.2h and 1.1.0  [25 Aug 2016]
 277
 278   *) Windows command-line tool supports UTF-8 opt-in option for arguments
 279      and console input. Setting OPENSSL_WIN32_UTF8 environment variable
 280      (to any value) allows Windows user to access PKCS#12 file generated
 281      with Windows CryptoAPI and protected with non-ASCII password, as well
 282      as files generated under UTF-8 locale on Linux also protected with
 283      non-ASCII password.
 284      [Andy Polyakov]
 285
 286   *) To mitigate the SWEET32 attack (CVE-2016-2183), 3DES cipher suites
 287      have been disabled by default and removed from DEFAULT, just like RC4.
 288      See the RC4 item below to re-enable both.
 289      [Rich Salz]
 290
 291   *) The method for finding the storage location for the Windows RAND seed file
 292      has changed. First we check %RANDFILE%. If that is not set then we check
 293      the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
 294      all else fails we fall back to C:\.
 295      [Matt Caswell]
 296
 297   *) The EVP_EncryptUpdate() function has had its return type changed from void
 298      to int. A return of 0 indicates and error while a return of 1 indicates
 299      success.
 300      [Matt Caswell]
 301
 302   *) The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
 303      DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
 304      off the constant time implementation for RSA, DSA and DH have been made
 305      no-ops and deprecated.
 306      [Matt Caswell]
 307
 308   *) Windows RAND implementation was simplified to only get entropy by
 309      calling CryptGenRandom(). Various other RAND-related tickets
 310      were also closed.
 311      [Joseph Wylie Yandle, Rich Salz]
 312
 313   *) The stack and lhash API's were renamed to start with OPENSSL_SK_
 314      and OPENSSL_LH_, respectively.  The old names are available
 315      with API compatibility.  They new names are now completely documented.
 316      [Rich Salz]
 317
 318   *) Unify TYPE_up_ref(obj) methods signature.
 319      SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
 320      X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
 321      int (instead of void) like all others TYPE_up_ref() methods.
 322      So now these methods also check the return value of CRYPTO_atomic_add(),
 323      and the validity of object reference counter.
 324      [fdasilvayy@gmail.com]
 325
 326   *) With Windows Visual Studio builds, the .pdb files are installed
 327      alongside the installed libraries and executables.  For a static
 328      library installation, ossl_static.pdb is the associate compiler
 329      generated .pdb file to be used when linking programs.
 330      [Richard Levitte]
 331
 332   *) Remove openssl.spec.  Packaging files belong with the packagers.
 333      [Richard Levitte]
 334
 335   *) Automatic Darwin/OSX configuration has had a refresh, it will now
 336      recognise x86_64 architectures automatically.  You can still decide
 337      to build for a different bitness with the environment variable
 338      KERNEL_BITS (can be 32 or 64), for example:
 339
 340          KERNEL_BITS=32 ./config
 341
 342      [Richard Levitte]
 343
 344   *) Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
 345      256 bit AES and HMAC with SHA256.
 346      [Steve Henson]
 347
 348   *) Remove support for MIPS o32 ABI on IRIX (and IRIX only).
 349      [Andy Polyakov]
 350
 351   *) Triple-DES ciphers have been moved from HIGH to MEDIUM.
 352      [Rich Salz]
 353
 354   *) To enable users to have their own config files and build file templates,
 355      Configure looks in the directory indicated by the environment variable
 356      OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
 357      directory.  On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
 358      name and is used as is.
 359      [Richard Levitte]
 360
 361   *) The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
 362      X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD.  The unused type
 363      X509_CERT_FILE_CTX was removed.
 364      [Rich Salz]
 365
 366   *) "shared" builds are now the default. To create only static libraries use
 367      the "no-shared" Configure option.
 368      [Matt Caswell]
 369
 370   *) Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
 371      All of these option have not worked for some while and are fundamental
 372      algorithms.
 373      [Matt Caswell]
 374
 375   *) Make various cleanup routines no-ops and mark them as deprecated. Most
 376      global cleanup functions are no longer required because they are handled
 377      via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
 378      Explicitly de-initing can cause problems (e.g. where a library that uses
 379      OpenSSL de-inits, but an application is still using it). The affected
 380      functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
 381      EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
 382      RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
 383      COMP_zlib_cleanup().
 384      [Matt Caswell]
 385
 386   *) --strict-warnings no longer enables runtime debugging options
 387      such as REF_DEBUG. Instead, debug options are automatically
 388      enabled with '--debug' builds.
 389      [Andy Polyakov, Emilia Käsper]
 390
 391   *) Made DH and DH_METHOD opaque. The structures for managing DH objects
 392      have been moved out of the public header files. New functions for managing
 393      these have been added.
 394      [Matt Caswell]
 395
 396   *) Made RSA and RSA_METHOD opaque. The structures for managing RSA
 397      objects have been moved out of the public header files. New
 398      functions for managing these have been added.
 399      [Richard Levitte]
 400
 401   *) Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
 402      have been moved out of the public header files. New functions for managing
 403      these have been added.
 404      [Matt Caswell]
 405
 406   *) Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
 407      moved out of the public header files. New functions for managing these
 408      have been added.
 409      [Matt Caswell]
 410
 411   *) Removed no-rijndael as a config option. Rijndael is an old name for AES.
 412      [Matt Caswell]
 413
 414   *) Removed the mk1mf build scripts.
 415      [Richard Levitte]
 416
 417   *) Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
 418      it is always safe to #include a header now.
 419      [Rich Salz]
 420
 421   *) Removed the aged BC-32 config and all its supporting scripts
 422      [Richard Levitte]
 423
 424   *) Removed support for Ultrix, Netware, and OS/2.
 425      [Rich Salz]
 426
 427   *) Add support for HKDF.
 428      [Alessandro Ghedini]
 429
 430   *) Add support for blake2b and blake2s
 431      [Bill Cox]
 432
 433   *) Added support for "pipelining". Ciphers that have the
 434      EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
 435      encryptions/decryptions simultaneously. There are currently no built-in
 436      ciphers with this property but the expectation is that engines will be able
 437      to offer it to significantly improve throughput. Support has been extended
 438      into libssl so that multiple records for a single connection can be
 439      processed in one go (for >=TLS 1.1).
 440      [Matt Caswell]
 441
 442   *) Added the AFALG engine. This is an async capable engine which is able to
 443      offload work to the Linux kernel. In this initial version it only supports
 444      AES128-CBC. The kernel must be version 4.1.0 or greater.
 445      [Catriona Lucey]
 446
 447   *) OpenSSL now uses a new threading API. It is no longer necessary to
 448      set locking callbacks to use OpenSSL in a multi-threaded environment. There
 449      are two supported threading models: pthreads and windows threads. It is
 450      also possible to configure OpenSSL at compile time for "no-threads". The
 451      old threading API should no longer be used. The functions have been
 452      replaced with "no-op" compatibility macros.
 453      [Alessandro Ghedini, Matt Caswell]
 454
 455   *) Modify behavior of ALPN to invoke callback after SNI/servername
 456      callback, such that updates to the SSL_CTX affect ALPN.
 457      [Todd Short]
 458
 459   *) Add SSL_CIPHER queries for authentication and key-exchange.
 460      [Todd Short]
 461
 462   *) Changes to the DEFAULT cipherlist:
 463        - Prefer (EC)DHE handshakes over plain RSA.
 464        - Prefer AEAD ciphers over legacy ciphers.
 465        - Prefer ECDSA over RSA when both certificates are available.
 466        - Prefer TLSv1.2 ciphers/PRF.
 467        - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
 468          default cipherlist.
 469      [Emilia Käsper]
 470
 471   *) Change the ECC default curve list to be this, in order: x25519,
 472      secp256r1, secp521r1, secp384r1.
 473      [Rich Salz]
 474
 475   *) RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
 476      disabled by default. They can be re-enabled using the
 477      enable-weak-ssl-ciphers option to Configure.
 478      [Matt Caswell]
 479
 480   *) If the server has ALPN configured, but supports no protocols that the
 481      client advertises, send a fatal "no_application_protocol" alert.
 482      This behaviour is SHALL in RFC 7301, though it isn't universally
 483      implemented by other servers.
 484      [Emilia Käsper]
 485
 486   *) Add X25519 support.
 487      Add ASN.1 and EVP_PKEY methods for X25519. This includes support
 488      for public and private key encoding using the format documented in
 489      draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
 490      key generation and key derivation.
 491
 492      TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
 493      X25519(29).
 494      [Steve Henson]
 495
 496   *) Deprecate SRP_VBASE_get_by_user.
 497      SRP_VBASE_get_by_user had inconsistent memory management behaviour.
 498      In order to fix an unavoidable memory leak (CVE-2016-0798),
 499      SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
 500      seed, even if the seed is configured.
 501
 502      Users should use SRP_VBASE_get1_by_user instead. Note that in
 503      SRP_VBASE_get1_by_user, caller must free the returned value. Note
 504      also that even though configuring the SRP seed attempts to hide
 505      invalid usernames by continuing the handshake with fake
 506      credentials, this behaviour is not constant time and no strong
 507      guarantees are made that the handshake is indistinguishable from
 508      that of a valid user.
 509      [Emilia Käsper]
 510
 511   *) Configuration change; it's now possible to build dynamic engines
 512      without having to build shared libraries and vice versa.  This
 513      only applies to the engines in engines/, those in crypto/engine/
 514      will always be built into libcrypto (i.e. "static").
 515
 516      Building dynamic engines is enabled by default; to disable, use
 517      the configuration option "disable-dynamic-engine".
 518
 519      The only requirements for building dynamic engines are the
 520      presence of the DSO module and building with position independent
 521      code, so they will also automatically be disabled if configuring
 522      with "disable-dso" or "disable-pic".
 523
 524      The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
 525      are also taken away from openssl/opensslconf.h, as they are
 526      irrelevant.
 527      [Richard Levitte]
 528
 529   *) Configuration change; if there is a known flag to compile
 530      position independent code, it will always be applied on the
 531      libcrypto and libssl object files, and never on the application
 532      object files.  This means other libraries that use routines from
 533      libcrypto / libssl can be made into shared libraries regardless
 534      of how OpenSSL was configured.
 535
 536      If this isn't desirable, the configuration options "disable-pic"
 537      or "no-pic" can be used to disable the use of PIC.  This will
 538      also disable building shared libraries and dynamic engines.
 539      [Richard Levitte]
 540
 541   *) Removed JPAKE code.  It was experimental and has no wide use.
 542      [Rich Salz]
 543
 544   *) The INSTALL_PREFIX Makefile variable has been renamed to
 545      DESTDIR.  That makes for less confusion on what this variable
 546      is for.  Also, the configuration option --install_prefix is
 547      removed.
 548      [Richard Levitte]
 549
 550   *) Heartbeat for TLS has been removed and is disabled by default
 551      for DTLS; configure with enable-heartbeats.  Code that uses the
 552      old #define's might need to be updated.
 553      [Emilia Käsper, Rich Salz]
 554
 555   *) Rename REF_CHECK to REF_DEBUG.
 556      [Rich Salz]
 557
 558   *) New "unified" build system
 559
 560      The "unified" build system is aimed to be a common system for all
 561      platforms we support.  With it comes new support for VMS.
 562
 563      This system builds supports building in a different directory tree
 564      than the source tree.  It produces one Makefile (for unix family
 565      or lookalikes), or one descrip.mms (for VMS).
 566
 567      The source of information to make the Makefile / descrip.mms is
 568      small files called 'build.info', holding the necessary
 569      information for each directory with source to compile, and a
 570      template in Configurations, like unix-Makefile.tmpl or
 571      descrip.mms.tmpl.
 572
 573      With this change, the library names were also renamed on Windows
 574      and on VMS.  They now have names that are closer to the standard
 575      on Unix, and include the major version number, and in certain
 576      cases, the architecture they are built for.  See "Notes on shared
 577      libraries" in INSTALL.
 578
 579      We rely heavily on the perl module Text::Template.
 580      [Richard Levitte]
 581
 582   *) Added support for auto-initialisation and de-initialisation of the library.
 583      OpenSSL no longer requires explicit init or deinit routines to be called,
 584      except in certain circumstances. See the OPENSSL_init_crypto() and
 585      OPENSSL_init_ssl() man pages for further information.
 586      [Matt Caswell]
 587
 588   *) The arguments to the DTLSv1_listen function have changed. Specifically the
 589      "peer" argument is now expected to be a BIO_ADDR object.
 590
 591   *) Rewrite of BIO networking library. The BIO library lacked consistent
 592      support of IPv6, and adding it required some more extensive
 593      modifications.  This introduces the BIO_ADDR and BIO_ADDRINFO types,
 594      which hold all types of addresses and chains of address information.
 595      It also introduces a new API, with functions like BIO_socket,
 596      BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
 597      The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
 598      have been adapted accordingly.
 599      [Richard Levitte]
 600
 601   *) RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
 602      the leading 0-byte.
 603      [Emilia Käsper]
 604
 605   *) CRIME protection: disable compression by default, even if OpenSSL is
 606      compiled with zlib enabled. Applications can still enable compression
 607      by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
 608      using the SSL_CONF library to configure compression.
 609      [Emilia Käsper]
 610
 611   *) The signature of the session callback configured with
 612      SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
 613      was explicitly marked as 'const unsigned char*' instead of
 614      'unsigned char*'.
 615      [Emilia Käsper]
 616
 617   *) Always DPURIFY. Remove the use of uninitialized memory in the
 618      RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
 619      [Emilia Käsper]
 620
 621   *) Removed many obsolete configuration items, including
 622         DES_PTR, DES_RISC1, DES_RISC2, DES_INT
 623         MD2_CHAR, MD2_INT, MD2_LONG
 624         BF_PTR, BF_PTR2
 625         IDEA_SHORT, IDEA_LONG
 626         RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
 627      [Rich Salz, with advice from Andy Polyakov]
 628
 629   *) Many BN internals have been moved to an internal header file.
 630      [Rich Salz with help from Andy Polyakov]
 631
 632   *) Configuration and writing out the results from it has changed.
 633      Files such as Makefile include/openssl/opensslconf.h and are now
 634      produced through general templates, such as Makefile.in and
 635      crypto/opensslconf.h.in and some help from the perl module
 636      Text::Template.
 637
 638      Also, the center of configuration information is no longer
 639      Makefile.  Instead, Configure produces a perl module in
 640      configdata.pm which holds most of the config data (in the hash
 641      table %config), the target data that comes from the target
 642      configuration in one of the Configurations/*.conf files (in
 643      %target).
 644      [Richard Levitte]
 645
 646   *) To clarify their intended purposes, the Configure options
 647      --prefix and --openssldir change their semantics, and become more
 648      straightforward and less interdependent.
 649
 650      --prefix shall be used exclusively to give the location INSTALLTOP
 651      where programs, scripts, libraries, include files and manuals are
 652      going to be installed.  The default is now /usr/local.
 653
 654      --openssldir shall be used exclusively to give the default
 655      location OPENSSLDIR where certificates, private keys, CRLs are
 656      managed.  This is also where the default openssl.cnf gets
 657      installed.
 658      If the directory given with this option is a relative path, the
 659      values of both the --prefix value and the --openssldir value will
 660      be combined to become OPENSSLDIR.
 661      The default for --openssldir is INSTALLTOP/ssl.
 662
 663      Anyone who uses --openssldir to specify where OpenSSL is to be
 664      installed MUST change to use --prefix instead.
 665      [Richard Levitte]
 666
 667   *) The GOST engine was out of date and therefore it has been removed. An up
 668      to date GOST engine is now being maintained in an external repository.
 669      See: https://wiki.openssl.org/index.php/Binaries. Libssl still retains
 670      support for GOST ciphersuites (these are only activated if a GOST engine
 671      is present).
 672      [Matt Caswell]
 673
 674   *) EGD is no longer supported by default; use enable-egd when
 675      configuring.
 676      [Ben Kaduk and Rich Salz]
 677
 678   *) The distribution now has Makefile.in files, which are used to
 679      create Makefile's when Configure is run.  *Configure must be run
 680      before trying to build now.*
 681      [Rich Salz]
 682
 683   *) The return value for SSL_CIPHER_description() for error conditions
 684      has changed.
 685      [Rich Salz]
 686
 687   *) Support for RFC6698/RFC7671 DANE TLSA peer authentication.
 688
 689      Obtaining and performing DNSSEC validation of TLSA records is
 690      the application's responsibility.  The application provides
 691      the TLSA records of its choice to OpenSSL, and these are then
 692      used to authenticate the peer.
 693
 694      The TLSA records need not even come from DNS.  They can, for
 695      example, be used to implement local end-entity certificate or
 696      trust-anchor "pinning", where the "pin" data takes the form
 697      of TLSA records, which can augment or replace verification
 698      based on the usual WebPKI public certification authorities.
 699      [Viktor Dukhovni]
 700
 701   *) Revert default OPENSSL_NO_DEPRECATED setting.  Instead OpenSSL
 702      continues to support deprecated interfaces in default builds.
 703      However, applications are strongly advised to compile their
 704      source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
 705      the declarations of all interfaces deprecated in 0.9.8, 1.0.0
 706      or the 1.1.0 releases.
 707
 708      In environments in which all applications have been ported to
 709      not use any deprecated interfaces OpenSSL's Configure script
 710      should be used with the --api=1.1.0 option to entirely remove
 711      support for the deprecated features from the library and
 712      unconditionally disable them in the installed headers.
 713      Essentially the same effect can be achieved with the "no-deprecated"
 714      argument to Configure, except that this will always restrict
 715      the build to just the latest API, rather than a fixed API
 716      version.
 717
 718      As applications are ported to future revisions of the API,
 719      they should update their compile-time OPENSSL_API_COMPAT define
 720      accordingly, but in most cases should be able to continue to
 721      compile with later releases.
 722
 723      The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
 724      0x10000000L and 0x00908000L, respectively.  However those
 725      versions did not support the OPENSSL_API_COMPAT feature, and
 726      so applications are not typically tested for explicit support
 727      of just the undeprecated features of either release.
 728      [Viktor Dukhovni]
 729
 730   *) Add support for setting the minimum and maximum supported protocol.
 731      It can bet set via the SSL_set_min_proto_version() and
 732      SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
 733      MaxProtcol.  It's recommended to use the new APIs to disable
 734      protocols instead of disabling individual protocols using
 735      SSL_set_options() or SSL_CONF's Protocol.  This change also
 736      removes support for disabling TLS 1.2 in the OpenSSL TLS
 737      client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
 738      [Kurt Roeckx]
 739
 740   *) Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
 741      [Andy Polyakov]
 742
 743   *) New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
 744      and integrates ECDSA and ECDH functionality into EC. Implementations can
 745      now redirect key generation and no longer need to convert to or from
 746      ECDSA_SIG format.
 747
 748      Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
 749      include the ec.h header file instead.
 750      [Steve Henson]
 751
 752   *) Remove support for all 40 and 56 bit ciphers.  This includes all the export
 753      ciphers who are no longer supported and drops support the ephemeral RSA key
 754      exchange. The LOW ciphers currently doesn't have any ciphers in it.
 755      [Kurt Roeckx]
 756
 757   *) Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
 758      opaque.  For HMAC_CTX, the following constructors and destructors
 759      were added:
 760
 761         HMAC_CTX *HMAC_CTX_new(void);
 762         void HMAC_CTX_free(HMAC_CTX *ctx);
 763
 764      For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
 765      destroy such methods has been added.  See EVP_MD_meth_new(3) and
 766      EVP_CIPHER_meth_new(3) for documentation.
 767
 768      Additional changes:
 769      1) EVP_MD_CTX_cleanup(), EVP_CIPHER_CTX_cleanup() and
 770         HMAC_CTX_cleanup() were removed.  HMAC_CTX_reset() and
 771         EVP_MD_CTX_reset() should be called instead to reinitialise
 772         an already created structure.
 773      2) For consistency with the majority of our object creators and
 774         destructors, EVP_MD_CTX_(create|destroy) were renamed to
 775         EVP_MD_CTX_(new|free).  The old names are retained as macros
 776         for deprecated builds.
 777      [Richard Levitte]
 778
 779   *) Added ASYNC support. Libcrypto now includes the async sub-library to enable
 780      cryptographic operations to be performed asynchronously as long as an
 781      asynchronous capable engine is used. See the ASYNC_start_job() man page for
 782      further details. Libssl has also had this capability integrated with the
 783      introduction of the new mode SSL_MODE_ASYNC and associated error
 784      SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
 785      pages. This work was developed in partnership with Intel Corp.
 786      [Matt Caswell]
 787
 788   *) SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
 789      always enabled now.  If you want to disable the support you should
 790      exclude it using the list of supported ciphers. This also means that the
 791      "-no_ecdhe" option has been removed from s_server.
 792      [Kurt Roeckx]
 793
 794   *) SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
 795      SSL_{CTX_}set1_curves() which can set a list.
 796      [Kurt Roeckx]
 797
 798   *) Remove support for SSL_{CTX_}set_tmp_ecdh_callback().  You should set the
 799      curve you want to support using SSL_{CTX_}set1_curves().
 800      [Kurt Roeckx]
 801
 802   *) State machine rewrite. The state machine code has been significantly
 803      refactored in order to remove much duplication of code and solve issues
 804      with the old code (see ssl/statem/README for further details). This change
 805      does have some associated API changes. Notably the SSL_state() function
 806      has been removed and replaced by SSL_get_state which now returns an
 807      "OSSL_HANDSHAKE_STATE" instead of an int. SSL_set_state() has been removed
 808      altogether. The previous handshake states defined in ssl.h and ssl3.h have
 809      also been removed.
 810      [Matt Caswell]
 811
 812   *) All instances of the string "ssleay" in the public API were replaced
 813      with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
 814      Some error codes related to internal RSA_eay API's were renamed.
 815      [Rich Salz]
 816
 817   *) The demo files in crypto/threads were moved to demo/threads.
 818      [Rich Salz]
 819
 820   *) Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
 821      sureware and ubsec.
 822      [Matt Caswell, Rich Salz]
 823
 824   *) New ASN.1 embed macro.
 825
 826      New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
 827      structure is not allocated: it is part of the parent. That is instead of
 828
 829      FOO *x;
 830
 831      it must be:
 832
 833      FOO x;
 834
 835      This reduces memory fragmentation and make it impossible to accidentally
 836      set a mandatory field to NULL.
 837
 838      This currently only works for some fields specifically a SEQUENCE, CHOICE,
 839      or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
 840      equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
 841      SEQUENCE OF.
 842      [Steve Henson]
 843
 844   *) Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
 845      [Emilia Käsper]
 846
 847   *) Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
 848      in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
 849      an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
 850      DES and RC4 ciphersuites.
 851      [Matt Caswell]
 852
 853   *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
 854      This changes the decoding behaviour for some invalid messages,
 855      though the change is mostly in the more lenient direction, and
 856      legacy behaviour is preserved as much as possible.
 857      [Emilia Käsper]
 858
 859   *) Fix no-stdio build.
 860     [ David Woodhouse <David.Woodhouse@intel.com> and also
 861       Ivan Nestlerode <ivan.nestlerode@sonos.com> ]
 862
 863   *) New testing framework
 864      The testing framework has been largely rewritten and is now using
 865      perl and the perl modules Test::Harness and an extended variant of
 866      Test::More called OpenSSL::Test to do its work.  All test scripts in
 867      test/ have been rewritten into test recipes, and all direct calls to
 868      executables in test/Makefile have become individual recipes using the
 869      simplified testing OpenSSL::Test::Simple.
 870
 871      For documentation on our testing modules, do:
 872
 873         perldoc test/testlib/OpenSSL/Test/Simple.pm
 874         perldoc test/testlib/OpenSSL/Test.pm
 875
 876      [Richard Levitte]
 877
 878   *) Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
 879      are used; the latter aborts on memory leaks (usually checked on exit).
 880      Some undocumented "set malloc, etc., hooks" functions were removed
 881      and others were changed.  All are now documented.
 882      [Rich Salz]
 883
 884   *) In DSA_generate_parameters_ex, if the provided seed is too short,
 885      return an error
 886      [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
 887
 888   *) Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
 889      from RFC4279, RFC4785, RFC5487, RFC5489.
 890
 891      Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
 892      original RSA_PSK patch.
 893      [Steve Henson]
 894
 895   *) Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
 896      era flag was never set throughout the codebase (only read). Also removed
 897      SSL3_FLAGS_POP_BUFFER which was only used if
 898      SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
 899      [Matt Caswell]
 900
 901   *) Changed the default name options in the "ca", "crl", "req" and "x509"
 902      to be "oneline" instead of "compat".
 903      [Richard Levitte]
 904
 905   *) Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
 906      not aware of clients that still exhibit this bug, and the workaround
 907      hasn't been working properly for a while.
 908      [Emilia Käsper]
 909
 910   *) The return type of BIO_number_read() and BIO_number_written() as well as
 911      the corresponding num_read and num_write members in the BIO structure has
 912      changed from unsigned long to uint64_t. On platforms where an unsigned
 913      long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
 914      transferred.
 915      [Matt Caswell]
 916
 917   *) Given the pervasive nature of TLS extensions it is inadvisable to run
 918      OpenSSL without support for them. It also means that maintaining
 919      the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
 920      not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
 921      [Matt Caswell]
 922
 923   *) Removed support for the two export grade static DH ciphersuites
 924      EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
 925      were newly added (along with a number of other static DH ciphersuites) to
 926      1.0.2. However the two export ones have *never* worked since they were
 927      introduced. It seems strange in any case to be adding new export
 928      ciphersuites, and given "logjam" it also does not seem correct to fix them.
 929      [Matt Caswell]
 930
 931   *) Version negotiation has been rewritten. In particular SSLv23_method(),
 932      SSLv23_client_method() and SSLv23_server_method() have been deprecated,
 933      and turned into macros which simply call the new preferred function names
 934      TLS_method(), TLS_client_method() and TLS_server_method(). All new code
 935      should use the new names instead. Also as part of this change the ssl23.h
 936      header file has been removed.
 937      [Matt Caswell]
 938
 939   *) Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
 940      code and the associated standard is no longer considered fit-for-purpose.
 941      [Matt Caswell]
 942
 943   *) RT2547 was closed.  When generating a private key, try to make the
 944      output file readable only by the owner.  This behavior change might
 945      be noticeable when interacting with other software.
 946
 947   *) Documented all exdata functions.  Added CRYPTO_free_ex_index.
 948      Added a test.
 949      [Rich Salz]
 950
 951   *) Added HTTP GET support to the ocsp command.
 952      [Rich Salz]
 953
 954   *) Changed default digest for the dgst and enc commands from MD5 to
 955      sha256
 956      [Rich Salz]
 957
 958   *) RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
 959      [Matt Caswell]
 960
 961   *) Added support for TLS extended master secret from
 962      draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
 963      initial patch which was a great help during development.
 964      [Steve Henson]
 965
 966   *) All libssl internal structures have been removed from the public header
 967      files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
 968      now redundant). Users should not attempt to access internal structures
 969      directly. Instead they should use the provided API functions.
 970      [Matt Caswell]
 971
 972   *) config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
 973      Access to deprecated functions can be re-enabled by running config with
 974      "enable-deprecated". In addition applications wishing to use deprecated
 975      functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
 976      will, by default, disable some transitive includes that previously existed
 977      in the header files (e.g. ec.h will no longer, by default, include bn.h)
 978      [Matt Caswell]
 979
 980   *) Added support for OCB mode. OpenSSL has been granted a patent license
 981      compatible with the OpenSSL license for use of OCB. Details are available
 982      at https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf. Support
 983      for OCB can be removed by calling config with no-ocb.
 984      [Matt Caswell]
 985
 986   *) SSLv2 support has been removed.  It still supports receiving a SSLv2
 987      compatible client hello.
 988      [Kurt Roeckx]
 989
 990   *) Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
 991      done while fixing the error code for the key-too-small case.
 992      [Annie Yousar <a.yousar@informatik.hu-berlin.de>]
 993
 994   *) CA.sh has been removed; use CA.pl instead.
 995      [Rich Salz]
 996
 997   *) Removed old DES API.
 998      [Rich Salz]
 999
1000   *) Remove various unsupported platforms:
1001         Sony NEWS4
1002         BEOS and BEOS_R5
1003         NeXT
1004         SUNOS
1005         MPE/iX
1006         Sinix/ReliantUNIX RM400
1007         DGUX
1008         NCR
1009         Tandem
1010         Cray
1011         16-bit platforms such as WIN16
1012      [Rich Salz]
1013
1014   *) Clean up OPENSSL_NO_xxx #define's
1015         Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
1016         Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
1017         OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
1018         OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
1019         OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
1020         Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
1021         OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
1022         OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
1023         OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
1024         Remove MS_STATIC; it's a relic from platforms <32 bits.
1025      [Rich Salz]
1026
1027   *) Cleaned up dead code
1028         Remove all but one '#ifdef undef' which is to be looked at.
1029      [Rich Salz]
1030
1031   *) Clean up calling of xxx_free routines.
1032         Just like free(), fix most of the xxx_free routines to accept
1033         NULL.  Remove the non-null checks from callers.  Save much code.
1034      [Rich Salz]
1035
1036   *) Add secure heap for storage of private keys (when possible).
1037      Add BIO_s_secmem(), CBIGNUM, etc.
1038      Contributed by Akamai Technologies under our Corporate CLA.
1039      [Rich Salz]
1040
1041   *) Experimental support for a new, fast, unbiased prime candidate generator,
1042      bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
1043      [Felix Laurie von Massenbach <felix@erbridge.co.uk>]
1044
1045   *) New output format NSS in the sess_id command line tool. This allows
1046      exporting the session id and the master key in NSS keylog format.
1047      [Martin Kaiser <martin@kaiser.cx>]
1048
1049   *) Harmonize version and its documentation. -f flag is used to display
1050      compilation flags.
1051      [mancha <mancha1@zoho.com>]
1052
1053   *) Fix eckey_priv_encode so it immediately returns an error upon a failure
1054      in i2d_ECPrivateKey.  Thanks to Ted Unangst for feedback on this issue.
1055      [mancha <mancha1@zoho.com>]
1056
1057   *) Fix some double frees. These are not thought to be exploitable.
1058      [mancha <mancha1@zoho.com>]
1059
1060   *) A missing bounds check in the handling of the TLS heartbeat extension
1061      can be used to reveal up to 64k of memory to a connected client or
1062      server.
1063
1064      Thanks for Neel Mehta of Google Security for discovering this bug and to
1065      Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
1066      preparing the fix (CVE-2014-0160)
1067      [Adam Langley, Bodo Moeller]
1068
1069   *) Fix for the attack described in the paper "Recovering OpenSSL
1070      ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
1071      by Yuval Yarom and Naomi Benger. Details can be obtained from:
1072      http://eprint.iacr.org/2014/140
1073
1074      Thanks to Yuval Yarom and Naomi Benger for discovering this
1075      flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
1076      [Yuval Yarom and Naomi Benger]
1077
1078   *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
1079      this fixes a limitation in previous versions of OpenSSL.
1080      [Steve Henson]
1081
1082   *) Experimental encrypt-then-mac support.
1083
1084      Experimental support for encrypt then mac from
1085      draft-gutmann-tls-encrypt-then-mac-02.txt
1086
1087      To enable it set the appropriate extension number (0x42 for the test
1088      server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
1089
1090      For non-compliant peers (i.e. just about everything) this should have no
1091      effect.
1092
1093      WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
1094
1095      [Steve Henson]
1096
1097   *) Add EVP support for key wrapping algorithms, to avoid problems with
1098      existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
1099      the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
1100      algorithms and include tests cases.
1101      [Steve Henson]
1102
1103   *) Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
1104      enveloped data.
1105      [Steve Henson]
1106
1107   *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
1108      MGF1 digest and OAEP label.
1109      [Steve Henson]
1110
1111   *) Make openssl verify return errors.
1112      [Chris Palmer <palmer@google.com> and Ben Laurie]
1113
1114   *) New function ASN1_TIME_diff to calculate the difference between two
1115      ASN1_TIME structures or one structure and the current time.
1116      [Steve Henson]
1117
1118   *) Update fips_test_suite to support multiple command line options. New
1119      test to induce all self test errors in sequence and check expected
1120      failures.
1121      [Steve Henson]
1122
1123   *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
1124      sign or verify all in one operation.
1125      [Steve Henson]
1126
1127   *) Add fips_algvs: a multicall fips utility incorporating all the algorithm
1128      test programs and fips_test_suite. Includes functionality to parse
1129      the minimal script output of fipsalgest.pl directly.
1130      [Steve Henson]
1131
1132   *) Add authorisation parameter to FIPS_module_mode_set().
1133      [Steve Henson]
1134
1135   *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
1136      [Steve Henson]
1137
1138   *) Use separate DRBG fields for internal and external flags. New function
1139      FIPS_drbg_health_check() to perform on demand health checking. Add
1140      generation tests to fips_test_suite with reduced health check interval to
1141      demonstrate periodic health checking. Add "nodh" option to
1142      fips_test_suite to skip very slow DH test.
1143      [Steve Henson]
1144
1145   *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
1146      based on NID.
1147      [Steve Henson]
1148
1149   *) More extensive health check for DRBG checking many more failure modes.
1150      New function FIPS_selftest_drbg_all() to handle every possible DRBG
1151      combination: call this in fips_test_suite.
1152      [Steve Henson]
1153
1154   *) Add support for canonical generation of DSA parameter 'g'. See
1155      FIPS 186-3 A.2.3.
1156
1157   *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
1158      POST to handle HMAC cases.
1159      [Steve Henson]
1160
1161   *) Add functions FIPS_module_version() and FIPS_module_version_text()
1162      to return numerical and string versions of the FIPS module number.
1163      [Steve Henson]
1164
1165   *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
1166      FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
1167      outside the validated module in the FIPS capable OpenSSL.
1168      [Steve Henson]
1169
1170   *) Minor change to DRBG entropy callback semantics. In some cases
1171      there is no multiple of the block length between min_len and
1172      max_len. Allow the callback to return more than max_len bytes
1173      of entropy but discard any extra: it is the callback's responsibility
1174      to ensure that the extra data discarded does not impact the
1175      requested amount of entropy.
1176      [Steve Henson]
1177
1178   *) Add PRNG security strength checks to RSA, DSA and ECDSA using
1179      information in FIPS186-3, SP800-57 and SP800-131A.
1180      [Steve Henson]
1181
1182   *) CCM support via EVP. Interface is very similar to GCM case except we
1183      must supply all data in one chunk (i.e. no update, final) and the
1184      message length must be supplied if AAD is used. Add algorithm test
1185      support.
1186      [Steve Henson]
1187
1188   *) Initial version of POST overhaul. Add POST callback to allow the status
1189      of POST to be monitored and/or failures induced. Modify fips_test_suite
1190      to use callback. Always run all selftests even if one fails.
1191      [Steve Henson]
1192
1193   *) XTS support including algorithm test driver in the fips_gcmtest program.
1194      Note: this does increase the maximum key length from 32 to 64 bytes but
1195      there should be no binary compatibility issues as existing applications
1196      will never use XTS mode.
1197      [Steve Henson]
1198
1199   *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
1200      to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
1201      performs algorithm blocking for unapproved PRNG types. Also do not
1202      set PRNG type in FIPS_mode_set(): leave this to the application.
1203      Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
1204      the standard OpenSSL PRNG: set additional data to a date time vector.
1205      [Steve Henson]
1206
1207   *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*.
1208      This shouldn't present any incompatibility problems because applications
1209      shouldn't be using these directly and any that are will need to rethink
1210      anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
1211      [Steve Henson]
1212
1213   *) Extensive self tests and health checking required by SP800-90 DRBG.
1214      Remove strength parameter from FIPS_drbg_instantiate and always
1215      instantiate at maximum supported strength.
1216      [Steve Henson]
1217
1218   *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
1219      [Steve Henson]
1220
1221   *) New algorithm test program fips_dhvs to handle DH primitives only testing.
1222      [Steve Henson]
1223
1224   *) New function DH_compute_key_padded() to compute a DH key and pad with
1225      leading zeroes if needed: this complies with SP800-56A et al.
1226      [Steve Henson]
1227
1228   *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
1229      anything, incomplete, subject to change and largely untested at present.
1230      [Steve Henson]
1231
1232   *) Modify fipscanisteronly build option to only build the necessary object
1233      files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
1234      [Steve Henson]
1235
1236   *) Add experimental option FIPSSYMS to give all symbols in
1237      fipscanister.o and FIPS or fips prefix. This will avoid
1238      conflicts with future versions of OpenSSL. Add perl script
1239      util/fipsas.pl to preprocess assembly language source files
1240      and rename any affected symbols.
1241      [Steve Henson]
1242
1243   *) Add selftest checks and algorithm block of non-fips algorithms in
1244      FIPS mode. Remove DES2 from selftests.
1245      [Steve Henson]
1246
1247   *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
1248      return internal method without any ENGINE dependencies. Add new
1249      tiny fips sign and verify functions.
1250      [Steve Henson]
1251
1252   *) New build option no-ec2m to disable characteristic 2 code.
1253      [Steve Henson]
1254
1255   *) New build option "fipscanisteronly". This only builds fipscanister.o
1256      and (currently) associated fips utilities. Uses the file Makefile.fips
1257      instead of Makefile.org as the prototype.
1258      [Steve Henson]
1259
1260   *) Add some FIPS mode restrictions to GCM. Add internal IV generator.
1261      Update fips_gcmtest to use IV generator.
1262      [Steve Henson]
1263
1264   *) Initial, experimental EVP support for AES-GCM. AAD can be input by
1265      setting output buffer to NULL. The *Final function must be
1266      called although it will not retrieve any additional data. The tag
1267      can be set or retrieved with a ctrl. The IV length is by default 12
1268      bytes (96 bits) but can be set to an alternative value. If the IV
1269      length exceeds the maximum IV length (currently 16 bytes) it cannot be
1270      set before the key.
1271      [Steve Henson]
1272
1273   *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
1274      underlying do_cipher function handles all cipher semantics itself
1275      including padding and finalisation. This is useful if (for example)
1276      an ENGINE cipher handles block padding itself. The behaviour of
1277      do_cipher is subtly changed if this flag is set: the return value
1278      is the number of characters written to the output buffer (zero is
1279      no longer an error code) or a negative error code. Also if the
1280      input buffer is NULL and length 0 finalisation should be performed.
1281      [Steve Henson]
1282
1283   *) If a candidate issuer certificate is already part of the constructed
1284      path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
1285      [Steve Henson]
1286
1287   *) Improve forward-security support: add functions
1288
1289        void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
1290        void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
1291
1292      for use by SSL/TLS servers; the callback function will be called whenever a
1293      new session is created, and gets to decide whether the session may be
1294      cached to make it resumable (return 0) or not (return 1).  (As by the
1295      SSL/TLS protocol specifications, the session_id sent by the server will be
1296      empty to indicate that the session is not resumable; also, the server will
1297      not generate RFC 4507 (RFC 5077) session tickets.)
1298
1299      A simple reasonable callback implementation is to return is_forward_secure.
1300      This parameter will be set to 1 or 0 depending on the ciphersuite selected
1301      by the SSL/TLS server library, indicating whether it can provide forward
1302      security.
1303      [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)]
1304
1305   *) New -verify_name option in command line utilities to set verification
1306      parameters by name.
1307      [Steve Henson]
1308
1309   *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
1310      Add CMAC pkey methods.
1311      [Steve Henson]
1312
1313   *) Experimental renegotiation in s_server -www mode. If the client
1314      browses /reneg connection is renegotiated. If /renegcert it is
1315      renegotiated requesting a certificate.
1316      [Steve Henson]
1317
1318   *) Add an "external" session cache for debugging purposes to s_server. This
1319      should help trace issues which normally are only apparent in deployed
1320      multi-process servers.
1321      [Steve Henson]
1322
1323   *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
1324      return value is ignored. NB. The functions RAND_add(), RAND_seed(),
1325      BIO_set_cipher() and some obscure PEM functions were changed so they
1326      can now return an error. The RAND changes required a change to the
1327      RAND_METHOD structure.
1328      [Steve Henson]
1329
1330   *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of
1331      a gcc attribute to warn if the result of a function is ignored. This
1332      is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
1333      whose return value is often ignored.
1334      [Steve Henson]
1335
1336   *) New -noct, -requestct, -requirect and -ctlogfile options for s_client.
1337      These allow SCTs (signed certificate timestamps) to be requested and
1338      validated when establishing a connection.
1339      [Rob Percival <robpercival@google.com>]
1340
1341  Changes between 1.0.2g and 1.0.2h [3 May 2016]
1342
1343   *) Prevent padding oracle in AES-NI CBC MAC check
1344
1345      A MITM attacker can use a padding oracle attack to decrypt traffic
1346      when the connection uses an AES CBC cipher and the server support
1347      AES-NI.
1348
1349      This issue was introduced as part of the fix for Lucky 13 padding
1350      attack (CVE-2013-0169). The padding check was rewritten to be in
1351      constant time by making sure that always the same bytes are read and
1352      compared against either the MAC or padding bytes. But it no longer
1353      checked that there was enough data to have both the MAC and padding
1354      bytes.
1355
1356      This issue was reported by Juraj Somorovsky using TLS-Attacker.
1357      (CVE-2016-2107)
1358      [Kurt Roeckx]
1359
1360   *) Fix EVP_EncodeUpdate overflow
1361
1362      An overflow can occur in the EVP_EncodeUpdate() function which is used for
1363      Base64 encoding of binary data. If an attacker is able to supply very large
1364      amounts of input data then a length check can overflow resulting in a heap
1365      corruption.
1366
1367      Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
1368      the PEM_write_bio* family of functions. These are mainly used within the
1369      OpenSSL command line applications, so any application which processes data
1370      from an untrusted source and outputs it as a PEM file should be considered
1371      vulnerable to this issue. User applications that call these APIs directly
1372      with large amounts of untrusted data may also be vulnerable.
1373
1374      This issue was reported by Guido Vranken.
1375      (CVE-2016-2105)
1376      [Matt Caswell]
1377
1378   *) Fix EVP_EncryptUpdate overflow
1379
1380      An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
1381      is able to supply very large amounts of input data after a previous call to
1382      EVP_EncryptUpdate() with a partial block then a length check can overflow
1383      resulting in a heap corruption. Following an analysis of all OpenSSL
1384      internal usage of the EVP_EncryptUpdate() function all usage is one of two
1385      forms. The first form is where the EVP_EncryptUpdate() call is known to be
1386      the first called function after an EVP_EncryptInit(), and therefore that
1387      specific call must be safe. The second form is where the length passed to
1388      EVP_EncryptUpdate() can be seen from the code to be some small value and
1389      therefore there is no possibility of an overflow. Since all instances are
1390      one of these two forms, it is believed that there can be no overflows in
1391      internal code due to this problem. It should be noted that
1392      EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
1393      Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
1394      of these calls have also been analysed too and it is believed there are no
1395      instances in internal usage where an overflow could occur.
1396
1397      This issue was reported by Guido Vranken.
1398      (CVE-2016-2106)
1399      [Matt Caswell]
1400
1401   *) Prevent ASN.1 BIO excessive memory allocation
1402
1403      When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
1404      a short invalid encoding can cause allocation of large amounts of memory
1405      potentially consuming excessive resources or exhausting memory.
1406
1407      Any application parsing untrusted data through d2i BIO functions is
1408      affected. The memory based functions such as d2i_X509() are *not* affected.
1409      Since the memory based functions are used by the TLS library, TLS
1410      applications are not affected.
1411
1412      This issue was reported by Brian Carpenter.
1413      (CVE-2016-2109)
1414      [Stephen Henson]
1415
1416   *) EBCDIC overread
1417
1418      ASN1 Strings that are over 1024 bytes can cause an overread in applications
1419      using the X509_NAME_oneline() function on EBCDIC systems. This could result
1420      in arbitrary stack data being returned in the buffer.
1421
1422      This issue was reported by Guido Vranken.
1423      (CVE-2016-2176)
1424      [Matt Caswell]
1425
1426   *) Modify behavior of ALPN to invoke callback after SNI/servername
1427      callback, such that updates to the SSL_CTX affect ALPN.
1428      [Todd Short]
1429
1430   *) Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
1431      default.
1432      [Kurt Roeckx]
1433
1434   *) Only remove the SSLv2 methods with the no-ssl2-method option. When the
1435      methods are enabled and ssl2 is disabled the methods return NULL.
1436      [Kurt Roeckx]
1437
1438  Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
1439
1440   * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
1441     Builds that are not configured with "enable-weak-ssl-ciphers" will not
1442     provide any "EXPORT" or "LOW" strength ciphers.
1443     [Viktor Dukhovni]
1444
1445   * Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
1446     is by default disabled at build-time.  Builds that are not configured with
1447     "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
1448     users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
1449     will need to explicitly call either of:
1450
1451         SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
1452     or
1453         SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
1454
1455     as appropriate.  Even if either of those is used, or the application
1456     explicitly uses the version-specific SSLv2_method() or its client and
1457     server variants, SSLv2 ciphers vulnerable to exhaustive search key
1458     recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
1459     ciphers, and SSLv2 56-bit DES are no longer available.
1460     (CVE-2016-0800)
1461     [Viktor Dukhovni]
1462
1463   *) Fix a double-free in DSA code
1464
1465      A double free bug was discovered when OpenSSL parses malformed DSA private
1466      keys and could lead to a DoS attack or memory corruption for applications
1467      that receive DSA private keys from untrusted sources.  This scenario is
1468      considered rare.
1469
1470      This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
1471      libFuzzer.
1472      (CVE-2016-0705)
1473      [Stephen Henson]
1474
1475   *) Disable SRP fake user seed to address a server memory leak.
1476
1477      Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
1478
1479      SRP_VBASE_get_by_user had inconsistent memory management behaviour.
1480      In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
1481      was changed to ignore the "fake user" SRP seed, even if the seed
1482      is configured.
1483
1484      Users should use SRP_VBASE_get1_by_user instead. Note that in
1485      SRP_VBASE_get1_by_user, caller must free the returned value. Note
1486      also that even though configuring the SRP seed attempts to hide
1487      invalid usernames by continuing the handshake with fake
1488      credentials, this behaviour is not constant time and no strong
1489      guarantees are made that the handshake is indistinguishable from
1490      that of a valid user.
1491      (CVE-2016-0798)
1492      [Emilia Käsper]
1493
1494   *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
1495
1496      In the BN_hex2bn function the number of hex digits is calculated using an
1497      int value |i|. Later |bn_expand| is called with a value of |i * 4|. For
1498      large values of |i| this can result in |bn_expand| not allocating any
1499      memory because |i * 4| is negative. This can leave the internal BIGNUM data
1500      field as NULL leading to a subsequent NULL ptr deref. For very large values
1501      of |i|, the calculation |i * 4| could be a positive value smaller than |i|.
1502      In this case memory is allocated to the internal BIGNUM data field, but it
1503      is insufficiently sized leading to heap corruption. A similar issue exists
1504      in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
1505      is ever called by user applications with very large untrusted hex/dec data.
1506      This is anticipated to be a rare occurrence.
1507
1508      All OpenSSL internal usage of these functions use data that is not expected
1509      to be untrusted, e.g. config file data or application command line
1510      arguments. If user developed applications generate config file data based
1511      on untrusted data then it is possible that this could also lead to security
1512      consequences. This is also anticipated to be rare.
1513
1514      This issue was reported to OpenSSL by Guido Vranken.
1515      (CVE-2016-0797)
1516      [Matt Caswell]
1517
1518   *) Fix memory issues in BIO_*printf functions
1519
1520      The internal |fmtstr| function used in processing a "%s" format string in
1521      the BIO_*printf functions could overflow while calculating the length of a
1522      string and cause an OOB read when printing very long strings.
1523
1524      Additionally the internal |doapr_outch| function can attempt to write to an
1525      OOB memory location (at an offset from the NULL pointer) in the event of a
1526      memory allocation failure. In 1.0.2 and below this could be caused where
1527      the size of a buffer to be allocated is greater than INT_MAX. E.g. this
1528      could be in processing a very long "%s" format string. Memory leaks can
1529      also occur.
1530
1531      The first issue may mask the second issue dependent on compiler behaviour.
1532      These problems could enable attacks where large amounts of untrusted data
1533      is passed to the BIO_*printf functions. If applications use these functions
1534      in this way then they could be vulnerable. OpenSSL itself uses these
1535      functions when printing out human-readable dumps of ASN.1 data. Therefore
1536      applications that print this data could be vulnerable if the data is from
1537      untrusted sources. OpenSSL command line applications could also be
1538      vulnerable where they print out ASN.1 data, or if untrusted data is passed
1539      as command line arguments.
1540
1541      Libssl is not considered directly vulnerable. Additionally certificates etc
1542      received via remote connections via libssl are also unlikely to be able to
1543      trigger these issues because of message size limits enforced within libssl.
1544
1545      This issue was reported to OpenSSL Guido Vranken.
1546      (CVE-2016-0799)
1547      [Matt Caswell]
1548
1549   *) Side channel attack on modular exponentiation
1550
1551      A side-channel attack was found which makes use of cache-bank conflicts on
1552      the Intel Sandy-Bridge microarchitecture which could lead to the recovery
1553      of RSA keys.  The ability to exploit this issue is limited as it relies on
1554      an attacker who has control of code in a thread running on the same
1555      hyper-threaded core as the victim thread which is performing decryptions.
1556
1557      This issue was reported to OpenSSL by Yuval Yarom, The University of
1558      Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
1559      Nadia Heninger, University of Pennsylvania with more information at
1560      http://cachebleed.info.
1561      (CVE-2016-0702)
1562      [Andy Polyakov]
1563
1564   *) Change the req app to generate a 2048-bit RSA/DSA key by default,
1565      if no keysize is specified with default_bits. This fixes an
1566      omission in an earlier change that changed all RSA/DSA key generation
1567      apps to use 2048 bits by default.
1568      [Emilia Käsper]
1569
1570  Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
1571   *) DH small subgroups
1572
1573      Historically OpenSSL only ever generated DH parameters based on "safe"
1574      primes. More recently (in version 1.0.2) support was provided for
1575      generating X9.42 style parameter files such as those required for RFC 5114
1576      support. The primes used in such files may not be "safe". Where an
1577      application is using DH configured with parameters based on primes that are
1578      not "safe" then an attacker could use this fact to find a peer's private
1579      DH exponent. This attack requires that the attacker complete multiple
1580      handshakes in which the peer uses the same private DH exponent. For example
1581      this could be used to discover a TLS server's private DH exponent if it's
1582      reusing the private DH exponent or it's using a static DH ciphersuite.
1583
1584      OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
1585      TLS. It is not on by default. If the option is not set then the server
1586      reuses the same private DH exponent for the life of the server process and
1587      would be vulnerable to this attack. It is believed that many popular
1588      applications do set this option and would therefore not be at risk.
1589
1590      The fix for this issue adds an additional check where a "q" parameter is
1591      available (as is the case in X9.42 based parameters). This detects the
1592      only known attack, and is the only possible defense for static DH
1593      ciphersuites. This could have some performance impact.
1594
1595      Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
1596      default and cannot be disabled. This could have some performance impact.
1597
1598      This issue was reported to OpenSSL by Antonio Sanso (Adobe).
1599      (CVE-2016-0701)
1600      [Matt Caswell]
1601
1602   *) SSLv2 doesn't block disabled ciphers
1603
1604      A malicious client can negotiate SSLv2 ciphers that have been disabled on
1605      the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
1606      been disabled, provided that the SSLv2 protocol was not also disabled via
1607      SSL_OP_NO_SSLv2.
1608
1609      This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
1610      and Sebastian Schinzel.
1611      (CVE-2015-3197)
1612      [Viktor Dukhovni]
1613
1614  Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
1615
1616   *) BN_mod_exp may produce incorrect results on x86_64
1617
1618      There is a carry propagating bug in the x86_64 Montgomery squaring
1619      procedure. No EC algorithms are affected. Analysis suggests that attacks
1620      against RSA and DSA as a result of this defect would be very difficult to
1621      perform and are not believed likely. Attacks against DH are considered just
1622      feasible (although very difficult) because most of the work necessary to
1623      deduce information about a private key may be performed offline. The amount
1624      of resources required for such an attack would be very significant and
1625      likely only accessible to a limited number of attackers. An attacker would
1626      additionally need online access to an unpatched system using the target
1627      private key in a scenario with persistent DH parameters and a private
1628      key that is shared between multiple clients. For example this can occur by
1629      default in OpenSSL DHE based SSL/TLS ciphersuites.
1630
1631      This issue was reported to OpenSSL by Hanno Böck.
1632      (CVE-2015-3193)
1633      [Andy Polyakov]
1634
1635   *) Certificate verify crash with missing PSS parameter
1636
1637      The signature verification routines will crash with a NULL pointer
1638      dereference if presented with an ASN.1 signature using the RSA PSS
1639      algorithm and absent mask generation function parameter. Since these
1640      routines are used to verify certificate signature algorithms this can be
1641      used to crash any certificate verification operation and exploited in a
1642      DoS attack. Any application which performs certificate verification is
1643      vulnerable including OpenSSL clients and servers which enable client
1644      authentication.
1645
1646      This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
1647      (CVE-2015-3194)
1648      [Stephen Henson]
1649
1650   *) X509_ATTRIBUTE memory leak
1651
1652      When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
1653      memory. This structure is used by the PKCS#7 and CMS routines so any
1654      application which reads PKCS#7 or CMS data from untrusted sources is
1655      affected. SSL/TLS is not affected.
1656
1657      This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
1658      libFuzzer.
1659      (CVE-2015-3195)
1660      [Stephen Henson]
1661
1662   *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
1663      This changes the decoding behaviour for some invalid messages,
1664      though the change is mostly in the more lenient direction, and
1665      legacy behaviour is preserved as much as possible.
1666      [Emilia Käsper]
1667
1668   *) In DSA_generate_parameters_ex, if the provided seed is too short,
1669      return an error
1670      [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
1671
1672  Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
1673
1674   *) Alternate chains certificate forgery
1675
1676      During certificate verification, OpenSSL will attempt to find an
1677      alternative certificate chain if the first attempt to build such a chain
1678      fails. An error in the implementation of this logic can mean that an
1679      attacker could cause certain checks on untrusted certificates to be
1680      bypassed, such as the CA flag, enabling them to use a valid leaf
1681      certificate to act as a CA and "issue" an invalid certificate.
1682
1683      This issue was reported to OpenSSL by Adam Langley/David Benjamin
1684      (Google/BoringSSL).
1685      [Matt Caswell]
1686
1687  Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
1688
1689   *) Fix HMAC ABI incompatibility. The previous version introduced an ABI
1690      incompatibility in the handling of HMAC. The previous ABI has now been
1691      restored.
1692      [Matt Caswell]
1693
1694  Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
1695
1696   *) Malformed ECParameters causes infinite loop
1697
1698      When processing an ECParameters structure OpenSSL enters an infinite loop
1699      if the curve specified is over a specially malformed binary polynomial
1700      field.
1701
1702      This can be used to perform denial of service against any
1703      system which processes public keys, certificate requests or
1704      certificates.  This includes TLS clients and TLS servers with
1705      client authentication enabled.
1706
1707      This issue was reported to OpenSSL by Joseph Barr-Pixton.
1708      (CVE-2015-1788)
1709      [Andy Polyakov]
1710
1711   *) Exploitable out-of-bounds read in X509_cmp_time
1712
1713      X509_cmp_time does not properly check the length of the ASN1_TIME
1714      string and can read a few bytes out of bounds. In addition,
1715      X509_cmp_time accepts an arbitrary number of fractional seconds in the
1716      time string.
1717
1718      An attacker can use this to craft malformed certificates and CRLs of
1719      various sizes and potentially cause a segmentation fault, resulting in
1720      a DoS on applications that verify certificates or CRLs. TLS clients
1721      that verify CRLs are affected. TLS clients and servers with client
1722      authentication enabled may be affected if they use custom verification
1723      callbacks.
1724
1725      This issue was reported to OpenSSL by Robert Swiecki (Google), and
1726      independently by Hanno Böck.
1727      (CVE-2015-1789)
1728      [Emilia Käsper]
1729
1730   *) PKCS7 crash with missing EnvelopedContent
1731
1732      The PKCS#7 parsing code does not handle missing inner EncryptedContent
1733      correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
1734      with missing content and trigger a NULL pointer dereference on parsing.
1735
1736      Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
1737      structures from untrusted sources are affected. OpenSSL clients and
1738      servers are not affected.
1739
1740      This issue was reported to OpenSSL by Michal Zalewski (Google).
1741      (CVE-2015-1790)
1742      [Emilia Käsper]
1743
1744   *) CMS verify infinite loop with unknown hash function
1745
1746      When verifying a signedData message the CMS code can enter an infinite loop
1747      if presented with an unknown hash function OID. This can be used to perform
1748      denial of service against any system which verifies signedData messages using
1749      the CMS code.
1750      This issue was reported to OpenSSL by Johannes Bauer.
1751      (CVE-2015-1792)
1752      [Stephen Henson]
1753
1754   *) Race condition handling NewSessionTicket
1755
1756      If a NewSessionTicket is received by a multi-threaded client when attempting to
1757      reuse a previous ticket then a race condition can occur potentially leading to
1758      a double free of the ticket data.
1759      (CVE-2015-1791)
1760      [Matt Caswell]
1761
1762   *) Only support 256-bit or stronger elliptic curves with the
1763      'ecdh_auto' setting (server) or by default (client). Of supported
1764      curves, prefer P-256 (both).
1765      [Emilia Kasper]
1766
1767  Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
1768
1769   *) ClientHello sigalgs DoS fix
1770
1771      If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
1772      invalid signature algorithms extension a NULL pointer dereference will
1773      occur. This can be exploited in a DoS attack against the server.
1774
1775      This issue was was reported to OpenSSL by David Ramos of Stanford
1776      University.
1777      (CVE-2015-0291)
1778      [Stephen Henson and Matt Caswell]
1779
1780   *) Multiblock corrupted pointer fix
1781
1782      OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
1783      feature only applies on 64 bit x86 architecture platforms that support AES
1784      NI instructions. A defect in the implementation of "multiblock" can cause
1785      OpenSSL's internal write buffer to become incorrectly set to NULL when
1786      using non-blocking IO. Typically, when the user application is using a
1787      socket BIO for writing, this will only result in a failed connection.
1788      However if some other BIO is used then it is likely that a segmentation
1789      fault will be triggered, thus enabling a potential DoS attack.
1790
1791      This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
1792      (CVE-2015-0290)
1793      [Matt Caswell]
1794
1795   *) Segmentation fault in DTLSv1_listen fix
1796
1797      The DTLSv1_listen function is intended to be stateless and processes the
1798      initial ClientHello from many peers. It is common for user code to loop
1799      over the call to DTLSv1_listen until a valid ClientHello is received with
1800      an associated cookie. A defect in the implementation of DTLSv1_listen means
1801      that state is preserved in the SSL object from one invocation to the next
1802      that can lead to a segmentation fault. Errors processing the initial
1803      ClientHello can trigger this scenario. An example of such an error could be
1804      that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
1805      server.
1806
1807      This issue was reported to OpenSSL by Per Allansson.
1808      (CVE-2015-0207)
1809      [Matt Caswell]
1810
1811   *) Segmentation fault in ASN1_TYPE_cmp fix
1812
1813      The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
1814      made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
1815      certificate signature algorithm consistency this can be used to crash any
1816      certificate verification operation and exploited in a DoS attack. Any
1817      application which performs certificate verification is vulnerable including
1818      OpenSSL clients and servers which enable client authentication.
1819      (CVE-2015-0286)
1820      [Stephen Henson]
1821
1822   *) Segmentation fault for invalid PSS parameters fix
1823
1824      The signature verification routines will crash with a NULL pointer
1825      dereference if presented with an ASN.1 signature using the RSA PSS
1826      algorithm and invalid parameters. Since these routines are used to verify
1827      certificate signature algorithms this can be used to crash any
1828      certificate verification operation and exploited in a DoS attack. Any
1829      application which performs certificate verification is vulnerable including
1830      OpenSSL clients and servers which enable client authentication.
1831
1832      This issue was was reported to OpenSSL by Brian Carpenter.
1833      (CVE-2015-0208)
1834      [Stephen Henson]
1835
1836   *) ASN.1 structure reuse memory corruption fix
1837
1838      Reusing a structure in ASN.1 parsing may allow an attacker to cause
1839      memory corruption via an invalid write. Such reuse is and has been
1840      strongly discouraged and is believed to be rare.
1841
1842      Applications that parse structures containing CHOICE or ANY DEFINED BY
1843      components may be affected. Certificate parsing (d2i_X509 and related
1844      functions) are however not affected. OpenSSL clients and servers are
1845      not affected.
1846      (CVE-2015-0287)
1847      [Stephen Henson]
1848
1849   *) PKCS7 NULL pointer dereferences fix
1850
1851      The PKCS#7 parsing code does not handle missing outer ContentInfo
1852      correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
1853      missing content and trigger a NULL pointer dereference on parsing.
1854
1855      Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
1856      otherwise parse PKCS#7 structures from untrusted sources are
1857      affected. OpenSSL clients and servers are not affected.
1858
1859      This issue was reported to OpenSSL by Michal Zalewski (Google).
1860      (CVE-2015-0289)
1861      [Emilia Käsper]
1862
1863   *) DoS via reachable assert in SSLv2 servers fix
1864
1865      A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
1866      servers that both support SSLv2 and enable export cipher suites by sending
1867      a specially crafted SSLv2 CLIENT-MASTER-KEY message.
1868
1869      This issue was discovered by Sean Burford (Google) and Emilia Käsper
1870      (OpenSSL development team).
1871      (CVE-2015-0293)
1872      [Emilia Käsper]
1873
1874   *) Empty CKE with client auth and DHE fix
1875
1876      If client auth is used then a server can seg fault in the event of a DHE
1877      ciphersuite being selected and a zero length ClientKeyExchange message
1878      being sent by the client. This could be exploited in a DoS attack.
1879      (CVE-2015-1787)
1880      [Matt Caswell]
1881
1882   *) Handshake with unseeded PRNG fix
1883
1884      Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
1885      with an unseeded PRNG. The conditions are:
1886      - The client is on a platform where the PRNG has not been seeded
1887      automatically, and the user has not seeded manually
1888      - A protocol specific client method version has been used (i.e. not
1889      SSL_client_methodv23)
1890      - A ciphersuite is used that does not require additional random data from
1891      the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
1892
1893      If the handshake succeeds then the client random that has been used will
1894      have been generated from a PRNG with insufficient entropy and therefore the
1895      output may be predictable.
1896
1897      For example using the following command with an unseeded openssl will
1898      succeed on an unpatched platform:
1899
1900      openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
1901      (CVE-2015-0285)
1902      [Matt Caswell]
1903
1904   *) Use After Free following d2i_ECPrivatekey error fix
1905
1906      A malformed EC private key file consumed via the d2i_ECPrivateKey function
1907      could cause a use after free condition. This, in turn, could cause a double
1908      free in several private key parsing functions (such as d2i_PrivateKey
1909      or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
1910      for applications that receive EC private keys from untrusted
1911      sources. This scenario is considered rare.
1912
1913      This issue was discovered by the BoringSSL project and fixed in their
1914      commit 517073cd4b.
1915      (CVE-2015-0209)
1916      [Matt Caswell]
1917
1918   *) X509_to_X509_REQ NULL pointer deref fix
1919
1920      The function X509_to_X509_REQ will crash with a NULL pointer dereference if
1921      the certificate key is invalid. This function is rarely used in practice.
1922
1923      This issue was discovered by Brian Carpenter.
1924      (CVE-2015-0288)
1925      [Stephen Henson]
1926
1927   *) Removed the export ciphers from the DEFAULT ciphers
1928      [Kurt Roeckx]
1929
1930  Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
1931
1932   *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
1933      ARMv5 through ARMv8, as opposite to "locking" it to single one.
1934      So far those who have to target multiple platforms would compromise
1935      and argue that binary targeting say ARMv5 would still execute on
1936      ARMv8. "Universal" build resolves this compromise by providing
1937      near-optimal performance even on newer platforms.
1938      [Andy Polyakov]
1939
1940   *) Accelerated NIST P-256 elliptic curve implementation for x86_64
1941      (other platforms pending).
1942      [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov]
1943
1944   *) Add support for the SignedCertificateTimestampList certificate and
1945      OCSP response extensions from RFC6962.
1946      [Rob Stradling]
1947
1948   *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
1949      for corner cases. (Certain input points at infinity could lead to
1950      bogus results, with non-infinity inputs mapped to infinity too.)
1951      [Bodo Moeller]
1952
1953   *) Initial support for PowerISA 2.0.7, first implemented in POWER8.
1954      This covers AES, SHA256/512 and GHASH. "Initial" means that most
1955      common cases are optimized and there still is room for further
1956      improvements. Vector Permutation AES for Altivec is also added.
1957      [Andy Polyakov]
1958
1959   *) Add support for little-endian ppc64 Linux target.
1960      [Marcelo Cerri (IBM)]
1961
1962   *) Initial support for AMRv8 ISA crypto extensions. This covers AES,
1963      SHA1, SHA256 and GHASH. "Initial" means that most common cases
1964      are optimized and there still is room for further improvements.
1965      Both 32- and 64-bit modes are supported.
1966      [Andy Polyakov, Ard Biesheuvel (Linaro)]
1967
1968   *) Improved ARMv7 NEON support.
1969      [Andy Polyakov]
1970
1971   *) Support for SPARC Architecture 2011 crypto extensions, first
1972      implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
1973      SHA256/512, MD5, GHASH and modular exponentiation.
1974      [Andy Polyakov, David Miller]
1975
1976   *) Accelerated modular exponentiation for Intel processors, a.k.a.
1977      RSAZ.
1978      [Shay Gueron & Vlad Krasnov (Intel Corp)]
1979
1980   *) Support for new and upcoming Intel processors, including AVX2,
1981      BMI and SHA ISA extensions. This includes additional "stitched"
1982      implementations, AESNI-SHA256 and GCM, and multi-buffer support
1983      for TLS encrypt.
1984
1985      This work was sponsored by Intel Corp.
1986      [Andy Polyakov]
1987
1988   *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
1989      supports both DTLS 1.2 and 1.0 and should use whatever version the peer
1990      supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
1991      [Steve Henson]
1992
1993   *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
1994      this fixes a limitation in previous versions of OpenSSL.
1995      [Steve Henson]
1996
1997   *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
1998      MGF1 digest and OAEP label.
1999      [Steve Henson]
2000
2001   *) Add EVP support for key wrapping algorithms, to avoid problems with
2002      existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
2003      the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
2004      algorithms and include tests cases.
2005      [Steve Henson]
2006
2007   *) Add functions to allocate and set the fields of an ECDSA_METHOD
2008      structure.
2009      [Douglas E. Engert, Steve Henson]
2010
2011   *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
2012      difference in days and seconds between two tm or ASN1_TIME structures.
2013      [Steve Henson]
2014
2015   *) Add -rev test option to s_server to just reverse order of characters
2016      received by client and send back to server. Also prints an abbreviated
2017      summary of the connection parameters.
2018      [Steve Henson]
2019
2020   *) New option -brief for s_client and s_server to print out a brief summary
2021      of connection parameters.
2022      [Steve Henson]
2023
2024   *) Add callbacks for arbitrary TLS extensions.
2025      [Trevor Perrin <trevp@trevp.net> and Ben Laurie]
2026
2027   *) New option -crl_download in several openssl utilities to download CRLs
2028      from CRLDP extension in certificates.
2029      [Steve Henson]
2030
2031   *) New options -CRL and -CRLform for s_client and s_server for CRLs.
2032      [Steve Henson]
2033
2034   *) New function X509_CRL_diff to generate a delta CRL from the difference
2035      of two full CRLs. Add support to "crl" utility.
2036      [Steve Henson]
2037
2038   *) New functions to set lookup_crls function and to retrieve
2039      X509_STORE from X509_STORE_CTX.
2040      [Steve Henson]
2041
2042   *) Print out deprecated issuer and subject unique ID fields in
2043      certificates.
2044      [Steve Henson]
2045
2046   *) Extend OCSP I/O functions so they can be used for simple general purpose
2047      HTTP as well as OCSP. New wrapper function which can be used to download
2048      CRLs using the OCSP API.
2049      [Steve Henson]
2050
2051   *) Delegate command line handling in s_client/s_server to SSL_CONF APIs.
2052      [Steve Henson]
2053
2054   *) SSL_CONF* functions. These provide a common framework for application
2055      configuration using configuration files or command lines.
2056      [Steve Henson]
2057
2058   *) SSL/TLS tracing code. This parses out SSL/TLS records using the
2059      message callback and prints the results. Needs compile time option
2060      "enable-ssl-trace". New options to s_client and s_server to enable
2061      tracing.
2062      [Steve Henson]
2063
2064   *) New ctrl and macro to retrieve supported points extensions.
2065      Print out extension in s_server and s_client.
2066      [Steve Henson]
2067
2068   *) New functions to retrieve certificate signature and signature
2069      OID NID.
2070      [Steve Henson]
2071
2072   *) Add functions to retrieve and manipulate the raw cipherlist sent by a
2073      client to OpenSSL.
2074      [Steve Henson]
2075
2076   *) New Suite B modes for TLS code. These use and enforce the requirements
2077      of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
2078      only use Suite B curves. The Suite B modes can be set by using the
2079      strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
2080      [Steve Henson]
2081
2082   *) New chain verification flags for Suite B levels of security. Check
2083      algorithms are acceptable when flags are set in X509_verify_cert.
2084      [Steve Henson]
2085
2086   *) Make tls1_check_chain return a set of flags indicating checks passed
2087      by a certificate chain. Add additional tests to handle client
2088      certificates: checks for matching certificate type and issuer name
2089      comparison.
2090      [Steve Henson]
2091
2092   *) If an attempt is made to use a signature algorithm not in the peer
2093      preference list abort the handshake. If client has no suitable
2094      signature algorithms in response to a certificate request do not
2095      use the certificate.
2096      [Steve Henson]
2097
2098   *) If server EC tmp key is not in client preference list abort handshake.
2099      [Steve Henson]
2100
2101   *) Add support for certificate stores in CERT structure. This makes it
2102      possible to have different stores per SSL structure or one store in
2103      the parent SSL_CTX. Include distinct stores for certificate chain
2104      verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
2105      to build and store a certificate chain in CERT structure: returning
2106      an error if the chain cannot be built: this will allow applications
2107      to test if a chain is correctly configured.
2108
2109      Note: if the CERT based stores are not set then the parent SSL_CTX
2110      store is used to retain compatibility with existing behaviour.
2111
2112      [Steve Henson]
2113
2114   *) New function ssl_set_client_disabled to set a ciphersuite disabled
2115      mask based on the current session, check mask when sending client
2116      hello and checking the requested ciphersuite.
2117      [Steve Henson]
2118
2119   *) New ctrls to retrieve and set certificate types in a certificate
2120      request message. Print out received values in s_client. If certificate
2121      types is not set with custom values set sensible values based on
2122      supported signature algorithms.
2123      [Steve Henson]
2124
2125   *) Support for distinct client and server supported signature algorithms.
2126      [Steve Henson]
2127
2128   *) Add certificate callback. If set this is called whenever a certificate
2129      is required by client or server. An application can decide which
2130      certificate chain to present based on arbitrary criteria: for example
2131      supported signature algorithms. Add very simple example to s_server.
2132      This fixes many of the problems and restrictions of the existing client
2133      certificate callback: for example you can now clear an existing
2134      certificate and specify the whole chain.
2135      [Steve Henson]
2136
2137   *) Add new "valid_flags" field to CERT_PKEY structure which determines what
2138      the certificate can be used for (if anything). Set valid_flags field
2139      in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
2140      to have similar checks in it.
2141
2142      Add new "cert_flags" field to CERT structure and include a "strict mode".
2143      This enforces some TLS certificate requirements (such as only permitting
2144      certificate signature algorithms contained in the supported algorithms
2145      extension) which some implementations ignore: this option should be used
2146      with caution as it could cause interoperability issues.
2147      [Steve Henson]
2148
2149   *) Update and tidy signature algorithm extension processing. Work out
2150      shared signature algorithms based on preferences and peer algorithms
2151      and print them out in s_client and s_server. Abort handshake if no
2152      shared signature algorithms.
2153      [Steve Henson]
2154
2155   *) Add new functions to allow customised supported signature algorithms
2156      for SSL and SSL_CTX structures. Add options to s_client and s_server
2157      to support them.
2158      [Steve Henson]
2159
2160   *) New function SSL_certs_clear() to delete all references to certificates
2161      from an SSL structure. Before this once a certificate had been added
2162      it couldn't be removed.
2163      [Steve Henson]
2164
2165   *) Integrate hostname, email address and IP address checking with certificate
2166      verification. New verify options supporting checking in openssl utility.
2167      [Steve Henson]
2168
2169   *) Fixes and wildcard matching support to hostname and email checking
2170      functions. Add manual page.
2171      [Florian Weimer (Red Hat Product Security Team)]
2172
2173   *) New functions to check a hostname email or IP address against a
2174      certificate. Add options x509 utility to print results of checks against
2175      a certificate.
2176      [Steve Henson]
2177
2178   *) Fix OCSP checking.
2179      [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie]
2180
2181   *) Initial experimental support for explicitly trusted non-root CAs.
2182      OpenSSL still tries to build a complete chain to a root but if an
2183      intermediate CA has a trust setting included that is used. The first
2184      setting is used: whether to trust (e.g., -addtrust option to the x509
2185      utility) or reject.
2186      [Steve Henson]
2187
2188   *) Add -trusted_first option which attempts to find certificates in the
2189      trusted store even if an untrusted chain is also supplied.
2190      [Steve Henson]
2191
2192   *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
2193      platform support for Linux and Android.
2194      [Andy Polyakov]
2195
2196   *) Support for linux-x32, ILP32 environment in x86_64 framework.
2197      [Andy Polyakov]
2198
2199   *) Experimental multi-implementation support for FIPS capable OpenSSL.
2200      When in FIPS mode the approved implementations are used as normal,
2201      when not in FIPS mode the internal unapproved versions are used instead.
2202      This means that the FIPS capable OpenSSL isn't forced to use the
2203      (often lower performance) FIPS implementations outside FIPS mode.
2204      [Steve Henson]
2205
2206   *) Transparently support X9.42 DH parameters when calling
2207      PEM_read_bio_DHparameters. This means existing applications can handle
2208      the new parameter format automatically.
2209      [Steve Henson]
2210
2211   *) Initial experimental support for X9.42 DH parameter format: mainly
2212      to support use of 'q' parameter for RFC5114 parameters.
2213      [Steve Henson]
2214
2215   *) Add DH parameters from RFC5114 including test data to dhtest.
2216      [Steve Henson]
2217
2218   *) Support for automatic EC temporary key parameter selection. If enabled
2219      the most preferred EC parameters are automatically used instead of
2220      hardcoded fixed parameters. Now a server just has to call:
2221      SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
2222      support ECDH and use the most appropriate parameters.
2223      [Steve Henson]
2224
2225   *) Enhance and tidy EC curve and point format TLS extension code. Use
2226      static structures instead of allocation if default values are used.
2227      New ctrls to set curves we wish to support and to retrieve shared curves.
2228      Print out shared curves in s_server. New options to s_server and s_client
2229      to set list of supported curves.
2230      [Steve Henson]
2231
2232   *) New ctrls to retrieve supported signature algorithms and
2233      supported curve values as an array of NIDs. Extend openssl utility
2234      to print out received values.
2235      [Steve Henson]
2236
2237   *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
2238      between NIDs and the more common NIST names such as "P-256". Enhance
2239      ecparam utility and ECC method to recognise the NIST names for curves.
2240      [Steve Henson]
2241
2242   *) Enhance SSL/TLS certificate chain handling to support different
2243      chains for each certificate instead of one chain in the parent SSL_CTX.
2244      [Steve Henson]
2245
2246   *) Support for fixed DH ciphersuite client authentication: where both
2247      server and client use DH certificates with common parameters.
2248      [Steve Henson]
2249
2250   *) Support for fixed DH ciphersuites: those requiring DH server
2251      certificates.
2252      [Steve Henson]
2253
2254   *) New function i2d_re_X509_tbs for re-encoding the TBS portion of
2255      the certificate.
2256      Note: Related 1.0.2-beta specific macros X509_get_cert_info,
2257      X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
2258      X509_CINF_get_signature were reverted post internal team review.
2259
2260  Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
2261
2262   *) Build fixes for the Windows and OpenVMS platforms
2263      [Matt Caswell and Richard Levitte]
2264
2265  Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
2266
2267   *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
2268      message can cause a segmentation fault in OpenSSL due to a NULL pointer
2269      dereference. This could lead to a Denial Of Service attack. Thanks to
2270      Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
2271      (CVE-2014-3571)
2272      [Steve Henson]
2273
2274   *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
2275      dtls1_buffer_record function under certain conditions. In particular this
2276      could occur if an attacker sent repeated DTLS records with the same
2277      sequence number but for the next epoch. The memory leak could be exploited
2278      by an attacker in a Denial of Service attack through memory exhaustion.
2279      Thanks to Chris Mueller for reporting this issue.
2280      (CVE-2015-0206)
2281      [Matt Caswell]
2282
2283   *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
2284      built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
2285      method would be set to NULL which could later result in a NULL pointer
2286      dereference. Thanks to Frank Schmirler for reporting this issue.
2287      (CVE-2014-3569)
2288      [Kurt Roeckx]
2289
2290   *) Abort handshake if server key exchange message is omitted for ephemeral
2291      ECDH ciphersuites.
2292
2293      Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
2294      reporting this issue.
2295      (CVE-2014-3572)
2296      [Steve Henson]
2297
2298   *) Remove non-export ephemeral RSA code on client and server. This code
2299      violated the TLS standard by allowing the use of temporary RSA keys in
2300      non-export ciphersuites and could be used by a server to effectively
2301      downgrade the RSA key length used to a value smaller than the server
2302      certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
2303      INRIA or reporting this issue.
2304      (CVE-2015-0204)
2305      [Steve Henson]
2306
2307   *) Fixed issue where DH client certificates are accepted without verification.
2308      An OpenSSL server will accept a DH certificate for client authentication
2309      without the certificate verify message. This effectively allows a client to
2310      authenticate without the use of a private key. This only affects servers
2311      which trust a client certificate authority which issues certificates
2312      containing DH keys: these are extremely rare and hardly ever encountered.
2313      Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
2314      this issue.
2315      (CVE-2015-0205)
2316      [Steve Henson]
2317
2318   *) Ensure that the session ID context of an SSL is updated when its
2319      SSL_CTX is updated via SSL_set_SSL_CTX.
2320
2321      The session ID context is typically set from the parent SSL_CTX,
2322      and can vary with the CTX.
2323      [Adam Langley]
2324
2325   *) Fix various certificate fingerprint issues.
2326
2327      By using non-DER or invalid encodings outside the signed portion of a
2328      certificate the fingerprint can be changed without breaking the signature.
2329      Although no details of the signed portion of the certificate can be changed
2330      this can cause problems with some applications: e.g. those using the
2331      certificate fingerprint for blacklists.
2332
2333      1. Reject signatures with non zero unused bits.
2334
2335      If the BIT STRING containing the signature has non zero unused bits reject
2336      the signature. All current signature algorithms require zero unused bits.
2337
2338      2. Check certificate algorithm consistency.
2339
2340      Check the AlgorithmIdentifier inside TBS matches the one in the
2341      certificate signature. NB: this will result in signature failure
2342      errors for some broken certificates.
2343
2344      Thanks to Konrad Kraszewski from Google for reporting this issue.
2345
2346      3. Check DSA/ECDSA signatures use DER.
2347
2348      Re-encode DSA/ECDSA signatures and compare with the original received
2349      signature. Return an error if there is a mismatch.
2350
2351      This will reject various cases including garbage after signature
2352      (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
2353      program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
2354      (negative or with leading zeroes).
2355
2356      Further analysis was conducted and fixes were developed by Stephen Henson
2357      of the OpenSSL core team.
2358
2359      (CVE-2014-8275)
2360      [Steve Henson]
2361
2362    *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
2363       results on some platforms, including x86_64. This bug occurs at random
2364       with a very low probability, and is not known to be exploitable in any
2365       way, though its exact impact is difficult to determine. Thanks to Pieter
2366       Wuille (Blockstream) who reported this issue and also suggested an initial
2367       fix. Further analysis was conducted by the OpenSSL development team and
2368       Adam Langley of Google. The final fix was developed by Andy Polyakov of
2369       the OpenSSL core team.
2370       (CVE-2014-3570)
2371       [Andy Polyakov]
2372
2373    *) Do not resume sessions on the server if the negotiated protocol
2374       version does not match the session's version. Resuming with a different
2375       version, while not strictly forbidden by the RFC, is of questionable
2376       sanity and breaks all known clients.
2377       [David Benjamin, Emilia Käsper]
2378
2379    *) Tighten handling of the ChangeCipherSpec (CCS) message: reject
2380       early CCS messages during renegotiation. (Note that because
2381       renegotiation is encrypted, this early CCS was not exploitable.)
2382       [Emilia Käsper]
2383
2384    *) Tighten client-side session ticket handling during renegotiation:
2385       ensure that the client only accepts a session ticket if the server sends
2386       the extension anew in the ServerHello. Previously, a TLS client would
2387       reuse the old extension state and thus accept a session ticket if one was
2388       announced in the initial ServerHello.
2389
2390       Similarly, ensure that the client requires a session ticket if one
2391       was advertised in the ServerHello. Previously, a TLS client would
2392       ignore a missing NewSessionTicket message.
2393       [Emilia Käsper]
2394
2395  Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
2396
2397   *) SRTP Memory Leak.
2398
2399      A flaw in the DTLS SRTP extension parsing code allows an attacker, who
2400      sends a carefully crafted handshake message, to cause OpenSSL to fail
2401      to free up to 64k of memory causing a memory leak. This could be
2402      exploited in a Denial Of Service attack. This issue affects OpenSSL
2403      1.0.1 server implementations for both SSL/TLS and DTLS regardless of
2404      whether SRTP is used or configured. Implementations of OpenSSL that
2405      have been compiled with OPENSSL_NO_SRTP defined are not affected.
2406
2407      The fix was developed by the OpenSSL team.
2408      (CVE-2014-3513)
2409      [OpenSSL team]
2410
2411   *) Session Ticket Memory Leak.
2412
2413      When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
2414      integrity of that ticket is first verified. In the event of a session
2415      ticket integrity check failing, OpenSSL will fail to free memory
2416      causing a memory leak. By sending a large number of invalid session
2417      tickets an attacker could exploit this issue in a Denial Of Service
2418      attack.
2419      (CVE-2014-3567)
2420      [Steve Henson]
2421
2422   *) Build option no-ssl3 is incomplete.
2423
2424      When OpenSSL is configured with "no-ssl3" as a build option, servers
2425      could accept and complete a SSL 3.0 handshake, and clients could be
2426      configured to send them.
2427      (CVE-2014-3568)
2428      [Akamai and the OpenSSL team]
2429
2430   *) Add support for TLS_FALLBACK_SCSV.
2431      Client applications doing fallback retries should call
2432      SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
2433      (CVE-2014-3566)
2434      [Adam Langley, Bodo Moeller]
2435
2436   *) Add additional DigestInfo checks.
2437
2438      Re-encode DigestInto in DER and check against the original when
2439      verifying RSA signature: this will reject any improperly encoded
2440      DigestInfo structures.
2441
2442      Note: this is a precautionary measure and no attacks are currently known.
2443
2444      [Steve Henson]
2445
2446  Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
2447
2448   *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
2449      SRP code can be overrun an internal buffer. Add sanity check that
2450      g, A, B < N to SRP code.
2451
2452      Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
2453      Group for discovering this issue.
2454      (CVE-2014-3512)
2455      [Steve Henson]
2456
2457   *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
2458      TLS 1.0 instead of higher protocol versions when the ClientHello message
2459      is badly fragmented. This allows a man-in-the-middle attacker to force a
2460      downgrade to TLS 1.0 even if both the server and the client support a
2461      higher protocol version, by modifying the client's TLS records.
2462
2463      Thanks to David Benjamin and Adam Langley (Google) for discovering and
2464      researching this issue.
2465      (CVE-2014-3511)
2466      [David Benjamin]
2467
2468   *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
2469      to a denial of service attack. A malicious server can crash the client
2470      with a null pointer dereference (read) by specifying an anonymous (EC)DH
2471      ciphersuite and sending carefully crafted handshake messages.
2472
2473      Thanks to Felix Gröbert (Google) for discovering and researching this
2474      issue.
2475      (CVE-2014-3510)
2476      [Emilia Käsper]
2477
2478   *) By sending carefully crafted DTLS packets an attacker could cause openssl
2479      to leak memory. This can be exploited through a Denial of Service attack.
2480      Thanks to Adam Langley for discovering and researching this issue.
2481      (CVE-2014-3507)
2482      [Adam Langley]
2483
2484   *) An attacker can force openssl to consume large amounts of memory whilst
2485      processing DTLS handshake messages. This can be exploited through a
2486      Denial of Service attack.
2487      Thanks to Adam Langley for discovering and researching this issue.
2488      (CVE-2014-3506)
2489      [Adam Langley]
2490
2491   *) An attacker can force an error condition which causes openssl to crash
2492      whilst processing DTLS packets due to memory being freed twice. This
2493      can be exploited through a Denial of Service attack.
2494      Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
2495      this issue.
2496      (CVE-2014-3505)
2497      [Adam Langley]
2498
2499   *) If a multithreaded client connects to a malicious server using a resumed
2500      session and the server sends an ec point format extension it could write
2501      up to 255 bytes to freed memory.
2502
2503      Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
2504      issue.
2505      (CVE-2014-3509)
2506      [Gabor Tyukasz]
2507
2508   *) A malicious server can crash an OpenSSL client with a null pointer
2509      dereference (read) by specifying an SRP ciphersuite even though it was not
2510      properly negotiated with the client. This can be exploited through a
2511      Denial of Service attack.
2512
2513      Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
2514      discovering and researching this issue.
2515      (CVE-2014-5139)
2516      [Steve Henson]
2517
2518   *) A flaw in OBJ_obj2txt may cause pretty printing functions such as
2519      X509_name_oneline, X509_name_print_ex et al. to leak some information
2520      from the stack. Applications may be affected if they echo pretty printing
2521      output to the attacker.
2522
2523      Thanks to Ivan Fratric (Google) for discovering this issue.
2524      (CVE-2014-3508)
2525      [Emilia Käsper, and Steve Henson]
2526
2527   *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
2528      for corner cases. (Certain input points at infinity could lead to
2529      bogus results, with non-infinity inputs mapped to infinity too.)
2530      [Bodo Moeller]
2531
2532  Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
2533
2534   *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
2535      handshake can force the use of weak keying material in OpenSSL
2536      SSL/TLS clients and servers.
2537
2538      Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
2539      researching this issue. (CVE-2014-0224)
2540      [KIKUCHI Masashi, Steve Henson]
2541
2542   *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
2543      OpenSSL DTLS client the code can be made to recurse eventually crashing
2544      in a DoS attack.
2545
2546      Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
2547      (CVE-2014-0221)
2548      [Imre Rad, Steve Henson]
2549
2550   *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
2551      be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
2552      client or server. This is potentially exploitable to run arbitrary
2553      code on a vulnerable client or server.
2554
2555      Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195)
2556      [Jüri Aedla, Steve Henson]
2557
2558   *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
2559      are subject to a denial of service attack.
2560
2561      Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
2562      this issue. (CVE-2014-3470)
2563      [Felix Gröbert, Ivan Fratric, Steve Henson]
2564
2565   *) Harmonize version and its documentation. -f flag is used to display
2566      compilation flags.
2567      [mancha <mancha1@zoho.com>]
2568
2569   *) Fix eckey_priv_encode so it immediately returns an error upon a failure
2570      in i2d_ECPrivateKey.
2571      [mancha <mancha1@zoho.com>]
2572
2573   *) Fix some double frees. These are not thought to be exploitable.
2574      [mancha <mancha1@zoho.com>]
2575
2576  Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
2577
2578   *) A missing bounds check in the handling of the TLS heartbeat extension
2579      can be used to reveal up to 64k of memory to a connected client or
2580      server.
2581
2582      Thanks for Neel Mehta of Google Security for discovering this bug and to
2583      Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
2584      preparing the fix (CVE-2014-0160)
2585      [Adam Langley, Bodo Moeller]
2586
2587   *) Fix for the attack described in the paper "Recovering OpenSSL
2588      ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
2589      by Yuval Yarom and Naomi Benger. Details can be obtained from:
2590      http://eprint.iacr.org/2014/140
2591
2592      Thanks to Yuval Yarom and Naomi Benger for discovering this
2593      flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
2594      [Yuval Yarom and Naomi Benger]
2595
2596   *) TLS pad extension: draft-agl-tls-padding-03
2597
2598      Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
2599      TLS client Hello record length value would otherwise be > 255 and
2600      less that 512 pad with a dummy extension containing zeroes so it
2601      is at least 512 bytes long.
2602
2603      [Adam Langley, Steve Henson]
2604
2605  Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
2606
2607   *) Fix for TLS record tampering bug. A carefully crafted invalid
2608      handshake could crash OpenSSL with a NULL pointer exception.
2609      Thanks to Anton Johansson for reporting this issues.
2610      (CVE-2013-4353)
2611
2612   *) Keep original DTLS digest and encryption contexts in retransmission
2613      structures so we can use the previous session parameters if they need
2614      to be resent. (CVE-2013-6450)
2615      [Steve Henson]
2616
2617   *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
2618      avoids preferring ECDHE-ECDSA ciphers when the client appears to be
2619      Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
2620      several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
2621      is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
2622      10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
2623      [Rob Stradling, Adam Langley]
2624
2625  Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
2626
2627   *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI
2628      supporting platforms or when small records were transferred.
2629      [Andy Polyakov, Steve Henson]
2630
2631  Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
2632
2633   *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
2634
2635      This addresses the flaw in CBC record processing discovered by
2636      Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
2637      at: http://www.isg.rhul.ac.uk/tls/
2638
2639      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
2640      Security Group at Royal Holloway, University of London
2641      (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
2642      Emilia Käsper for the initial patch.
2643      (CVE-2013-0169)
2644      [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
2645
2646   *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
2647      ciphersuites which can be exploited in a denial of service attack.
2648      Thanks go to and to Adam Langley <agl@chromium.org> for discovering
2649      and detecting this bug and to Wolfgang Ettlinger
2650      <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
2651      (CVE-2012-2686)
2652      [Adam Langley]
2653
2654   *) Return an error when checking OCSP signatures when key is NULL.
2655      This fixes a DoS attack. (CVE-2013-0166)
2656      [Steve Henson]
2657
2658   *) Make openssl verify return errors.
2659      [Chris Palmer <palmer@google.com> and Ben Laurie]
2660
2661   *) Call OCSP Stapling callback after ciphersuite has been chosen, so
2662      the right response is stapled. Also change SSL_get_certificate()
2663      so it returns the certificate actually sent.
2664      See http://rt.openssl.org/Ticket/Display.html?id=2836.
2665      [Rob Stradling <rob.stradling@comodo.com>]
2666
2667   *) Fix possible deadlock when decoding public keys.
2668      [Steve Henson]
2669
2670   *) Don't use TLS 1.0 record version number in initial client hello
2671      if renegotiating.
2672      [Steve Henson]
2673
2674  Changes between 1.0.1b and 1.0.1c [10 May 2012]
2675
2676   *) Sanity check record length before skipping explicit IV in TLS
2677      1.2, 1.1 and DTLS to fix DoS attack.
2678
2679      Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
2680      fuzzing as a service testing platform.
2681      (CVE-2012-2333)
2682      [Steve Henson]
2683
2684   *) Initialise tkeylen properly when encrypting CMS messages.
2685      Thanks to Solar Designer of Openwall for reporting this issue.
2686      [Steve Henson]
2687
2688   *) In FIPS mode don't try to use composite ciphers as they are not
2689      approved.
2690      [Steve Henson]
2691
2692  Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
2693
2694   *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
2695      1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
2696      mean any application compiled against OpenSSL 1.0.0 headers setting
2697      SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng
2698      TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
2699      0x10000000L Any application which was previously compiled against
2700      OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
2701      will need to be recompiled as a result. Letting be results in
2702      inability to disable specifically TLS 1.1 and in client context,
2703      in unlike event, limit maximum offered version to TLS 1.0 [see below].
2704      [Steve Henson]
2705
2706   *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not
2707      disable just protocol X, but all protocols above X *if* there are
2708      protocols *below* X still enabled. In more practical terms it means
2709      that if application wants to disable TLS1.0 in favor of TLS1.1 and
2710      above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass
2711      SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to
2712      client side.
2713      [Andy Polyakov]
2714
2715  Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
2716
2717   *) Check for potentially exploitable overflows in asn1_d2i_read_bio
2718      BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
2719      in CRYPTO_realloc_clean.
2720
2721      Thanks to Tavis Ormandy, Google Security Team, for discovering this
2722      issue and to Adam Langley <agl@chromium.org> for fixing it.
2723      (CVE-2012-2110)
2724      [Adam Langley (Google), Tavis Ormandy, Google Security Team]
2725
2726   *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
2727      [Adam Langley]
2728
2729   *) Workarounds for some broken servers that "hang" if a client hello
2730      record length exceeds 255 bytes.
2731
2732      1. Do not use record version number > TLS 1.0 in initial client
2733         hello: some (but not all) hanging servers will now work.
2734      2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
2735         the number of ciphers sent in the client hello. This should be
2736         set to an even number, such as 50, for example by passing:
2737         -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
2738         Most broken servers should now work.
2739      3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
2740         TLS 1.2 client support entirely.
2741      [Steve Henson]
2742
2743   *) Fix SEGV in Vector Permutation AES module observed in OpenSSH.
2744      [Andy Polyakov]
2745
2746  Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
2747
2748   *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET
2749      STRING form instead of a DigestInfo.
2750      [Steve Henson]
2751
2752   *) The format used for MDC2 RSA signatures is inconsistent between EVP
2753      and the RSA_sign/RSA_verify functions. This was made more apparent when
2754      OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
2755      those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
2756      the correct format in RSA_verify so both forms transparently work.
2757      [Steve Henson]
2758
2759   *) Some servers which support TLS 1.0 can choke if we initially indicate
2760      support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
2761      encrypted premaster secret. As a workaround use the maximum permitted
2762      client version in client hello, this should keep such servers happy
2763      and still work with previous versions of OpenSSL.
2764      [Steve Henson]
2765
2766   *) Add support for TLS/DTLS heartbeats.
2767      [Robin Seggelmann <seggelmann@fh-muenster.de>]
2768
2769   *) Add support for SCTP.
2770      [Robin Seggelmann <seggelmann@fh-muenster.de>]
2771
2772   *) Improved PRNG seeding for VOS.
2773      [Paul Green <Paul.Green@stratus.com>]
2774
2775   *) Extensive assembler packs updates, most notably:
2776
2777         - x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
2778         - x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
2779         - x86_64:       bit-sliced AES implementation;
2780         - ARM:          NEON support, contemporary platforms optimizations;
2781         - s390x:        z196 support;
2782         - *:            GHASH and GF(2^m) multiplication implementations;
2783
2784      [Andy Polyakov]
2785
2786   *) Make TLS-SRP code conformant with RFC 5054 API cleanup
2787      (removal of unnecessary code)
2788      [Peter Sylvester <peter.sylvester@edelweb.fr>]
2789
2790   *) Add TLS key material exporter from RFC 5705.
2791      [Eric Rescorla]
2792
2793   *) Add DTLS-SRTP negotiation from RFC 5764.
2794      [Eric Rescorla]
2795
2796   *) Add Next Protocol Negotiation,
2797      http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
2798      disabled with a no-npn flag to config or Configure. Code donated
2799      by Google.
2800      [Adam Langley <agl@google.com> and Ben Laurie]
2801
2802   *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
2803      NIST-P256, NIST-P521, with constant-time single point multiplication on
2804      typical inputs. Compiler support for the nonstandard type __uint128_t is
2805      required to use this (present in gcc 4.4 and later, for 64-bit builds).
2806      Code made available under Apache License version 2.0.
2807
2808      Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
2809      line to include this in your build of OpenSSL, and run "make depend" (or
2810      "make update"). This enables the following EC_METHODs:
2811
2812          EC_GFp_nistp224_method()
2813          EC_GFp_nistp256_method()
2814          EC_GFp_nistp521_method()
2815
2816      EC_GROUP_new_by_curve_name() will automatically use these (while
2817      EC_GROUP_new_curve_GFp() currently prefers the more flexible
2818      implementations).
2819      [Emilia Käsper, Adam Langley, Bodo Moeller (Google)]
2820
2821   *) Use type ossl_ssize_t instad of ssize_t which isn't available on
2822      all platforms. Move ssize_t definition from e_os.h to the public
2823      header file e_os2.h as it now appears in public header file cms.h
2824      [Steve Henson]
2825
2826   *) New -sigopt option to the ca, req and x509 utilities. Additional
2827      signature parameters can be passed using this option and in
2828      particular PSS.
2829      [Steve Henson]
2830
2831   *) Add RSA PSS signing function. This will generate and set the
2832      appropriate AlgorithmIdentifiers for PSS based on those in the
2833      corresponding EVP_MD_CTX structure. No application support yet.
2834      [Steve Henson]
2835
2836   *) Support for companion algorithm specific ASN1 signing routines.
2837      New function ASN1_item_sign_ctx() signs a pre-initialised
2838      EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
2839      the appropriate parameters.
2840      [Steve Henson]
2841
2842   *) Add new algorithm specific ASN1 verification initialisation function
2843      to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
2844      handling will be the same no matter what EVP_PKEY_METHOD is used.
2845      Add a PSS handler to support verification of PSS signatures: checked
2846      against a number of sample certificates.
2847      [Steve Henson]
2848
2849   *) Add signature printing for PSS. Add PSS OIDs.
2850      [Steve Henson, Martin Kaiser <lists@kaiser.cx>]
2851
2852   *) Add algorithm specific signature printing. An individual ASN1 method
2853      can now print out signatures instead of the standard hex dump.
2854
2855      More complex signatures (e.g. PSS) can print out more meaningful
2856      information. Include DSA version that prints out the signature
2857      parameters r, s.
2858      [Steve Henson]
2859
2860   *) Password based recipient info support for CMS library: implementing
2861      RFC3211.
2862      [Steve Henson]
2863
2864   *) Split password based encryption into PBES2 and PBKDF2 functions. This
2865      neatly separates the code into cipher and PBE sections and is required
2866      for some algorithms that split PBES2 into separate pieces (such as
2867      password based CMS).
2868      [Steve Henson]
2869
2870   *) Session-handling fixes:
2871      - Fix handling of connections that are resuming with a session ID,
2872        but also support Session Tickets.
2873      - Fix a bug that suppressed issuing of a new ticket if the client
2874        presented a ticket with an expired session.
2875      - Try to set the ticket lifetime hint to something reasonable.
2876      - Make tickets shorter by excluding irrelevant information.
2877      - On the client side, don't ignore renewed tickets.
2878      [Adam Langley, Bodo Moeller (Google)]
2879
2880   *) Fix PSK session representation.
2881      [Bodo Moeller]
2882
2883   *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
2884
2885      This work was sponsored by Intel.
2886      [Andy Polyakov]
2887
2888   *) Add GCM support to TLS library. Some custom code is needed to split
2889      the IV between the fixed (from PRF) and explicit (from TLS record)
2890      portions. This adds all GCM ciphersuites supported by RFC5288 and
2891      RFC5289. Generalise some AES* cipherstrings to include GCM and
2892      add a special AESGCM string for GCM only.
2893      [Steve Henson]
2894
2895   *) Expand range of ctrls for AES GCM. Permit setting invocation
2896      field on decrypt and retrieval of invocation field only on encrypt.
2897      [Steve Henson]
2898
2899   *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
2900      As required by RFC5289 these ciphersuites cannot be used if for
2901      versions of TLS earlier than 1.2.
2902      [Steve Henson]
2903
2904   *) For FIPS capable OpenSSL interpret a NULL default public key method
2905      as unset and return the appropriate default but do *not* set the default.
2906      This means we can return the appropriate method in applications that
2907      switch between FIPS and non-FIPS modes.
2908      [Steve Henson]
2909
2910   *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
2911      ENGINE is used then we cannot handle that in the FIPS module so we
2912      keep original code iff non-FIPS operations are allowed.
2913      [Steve Henson]
2914
2915   *) Add -attime option to openssl utilities.
2916      [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson]
2917
2918   *) Redirect DSA and DH operations to FIPS module in FIPS mode.
2919      [Steve Henson]
2920
2921   *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
2922      FIPS EC methods unconditionally for now.
2923      [Steve Henson]
2924
2925   *) New build option no-ec2m to disable characteristic 2 code.
2926      [Steve Henson]
2927
2928   *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
2929      all cases can be covered as some introduce binary incompatibilities.
2930      [Steve Henson]
2931
2932   *) Redirect RSA operations to FIPS module including keygen,
2933      encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
2934      [Steve Henson]
2935
2936   *) Add similar low level API blocking to ciphers.
2937      [Steve Henson]
2938
2939   *) Low level digest APIs are not approved in FIPS mode: any attempt
2940      to use these will cause a fatal error. Applications that *really* want
2941      to use them can use the private_* version instead.
2942      [Steve Henson]
2943
2944   *) Redirect cipher operations to FIPS module for FIPS builds.
2945      [Steve Henson]
2946
2947   *) Redirect digest operations to FIPS module for FIPS builds.
2948      [Steve Henson]
2949
2950   *) Update build system to add "fips" flag which will link in fipscanister.o
2951      for static and shared library builds embedding a signature if needed.
2952      [Steve Henson]
2953
2954   *) Output TLS supported curves in preference order instead of numerical
2955      order. This is currently hardcoded for the highest order curves first.
2956      This should be configurable so applications can judge speed vs strength.
2957      [Steve Henson]
2958
2959   *) Add TLS v1.2 server support for client authentication.
2960      [Steve Henson]
2961
2962   *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
2963      and enable MD5.
2964      [Steve Henson]
2965
2966   *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
2967      FIPS modules versions.
2968      [Steve Henson]
2969
2970   *) Add TLS v1.2 client side support for client authentication. Keep cache
2971      of handshake records longer as we don't know the hash algorithm to use
2972      until after the certificate request message is received.
2973      [Steve Henson]
2974
2975   *) Initial TLS v1.2 client support. Add a default signature algorithms
2976      extension including all the algorithms we support. Parse new signature
2977      format in client key exchange. Relax some ECC signing restrictions for
2978      TLS v1.2 as indicated in RFC5246.
2979      [Steve Henson]
2980
2981   *) Add server support for TLS v1.2 signature algorithms extension. Switch
2982      to new signature format when needed using client digest preference.
2983      All server ciphersuites should now work correctly in TLS v1.2. No client
2984      support yet and no support for client certificates.
2985      [Steve Henson]
2986
2987   *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
2988      to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
2989      ciphersuites. At present only RSA key exchange ciphersuites work with
2990      TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
2991      SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
2992      and version checking.
2993      [Steve Henson]
2994
2995   *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
2996      with this defined it will not be affected by any changes to ssl internal
2997      structures. Add several utility functions to allow openssl application
2998      to work with OPENSSL_NO_SSL_INTERN defined.
2999      [Steve Henson]
3000
3001   *) Add SRP support.
3002      [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie]
3003
3004   *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
3005      [Steve Henson]
3006
3007   *) Permit abbreviated handshakes when renegotiating using the function
3008      SSL_renegotiate_abbreviated().
3009      [Robin Seggelmann <seggelmann@fh-muenster.de>]
3010
3011   *) Add call to ENGINE_register_all_complete() to
3012      ENGINE_load_builtin_engines(), so some implementations get used
3013      automatically instead of needing explicit application support.
3014      [Steve Henson]
3015
3016   *) Add support for TLS key exporter as described in RFC5705.
3017      [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson]
3018
3019   *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
3020      a few changes are required:
3021
3022        Add SSL_OP_NO_TLSv1_1 flag.
3023        Add TLSv1_1 methods.
3024        Update version checking logic to handle version 1.1.
3025        Add explicit IV handling (ported from DTLS code).
3026        Add command line options to s_client/s_server.
3027      [Steve Henson]
3028
3029  Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
3030
3031   *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
3032      in CMS and PKCS7 code. When RSA decryption fails use a random key for
3033      content decryption and always return the same error. Note: this attack
3034      needs on average 2^20 messages so it only affects automated senders. The
3035      old behaviour can be re-enabled in the CMS code by setting the
3036      CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
3037      an MMA defence is not necessary.
3038      Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
3039      this issue. (CVE-2012-0884)
3040      [Steve Henson]
3041
3042   *) Fix CVE-2011-4619: make sure we really are receiving a
3043      client hello before rejecting multiple SGC restarts. Thanks to
3044      Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
3045      [Steve Henson]
3046
3047  Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
3048
3049   *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
3050      Thanks to Antonio Martin, Enterprise Secure Access Research and
3051      Development, Cisco Systems, Inc. for discovering this bug and
3052      preparing a fix. (CVE-2012-0050)
3053      [Antonio Martin]
3054
3055  Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
3056
3057   *) Nadhem Alfardan and Kenny Paterson have discovered an extension
3058      of the Vaudenay padding oracle attack on CBC mode encryption
3059      which enables an efficient plaintext recovery attack against
3060      the OpenSSL implementation of DTLS. Their attack exploits timing
3061      differences arising during decryption processing. A research
3062      paper describing this attack can be found at:
3063                   http://www.isg.rhul.ac.uk/~kp/dtls.pdf
3064      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
3065      Security Group at Royal Holloway, University of London
3066      (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
3067      <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
3068      for preparing the fix. (CVE-2011-4108)
3069      [Robin Seggelmann, Michael Tuexen]
3070
3071   *) Clear bytes used for block padding of SSL 3.0 records.
3072      (CVE-2011-4576)
3073      [Adam Langley (Google)]
3074
3075   *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
3076      Kadianakis <desnacked@gmail.com> for discovering this issue and
3077      Adam Langley for preparing the fix. (CVE-2011-4619)
3078      [Adam Langley (Google)]
3079
3080   *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027)
3081      [Andrey Kulikov <amdeich@gmail.com>]
3082
3083   *) Prevent malformed RFC3779 data triggering an assertion failure.
3084      Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
3085      and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577)
3086      [Rob Austein <sra@hactrn.net>]
3087
3088   *) Improved PRNG seeding for VOS.
3089      [Paul Green <Paul.Green@stratus.com>]
3090
3091   *) Fix ssl_ciph.c set-up race.
3092      [Adam Langley (Google)]
3093
3094   *) Fix spurious failures in ecdsatest.c.
3095      [Emilia Käsper (Google)]
3096
3097   *) Fix the BIO_f_buffer() implementation (which was mixing different
3098      interpretations of the '..._len' fields).
3099      [Adam Langley (Google)]
3100
3101   *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
3102      BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
3103      threads won't reuse the same blinding coefficients.
3104
3105      This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
3106      lock to call BN_BLINDING_invert_ex, and avoids one use of
3107      BN_BLINDING_update for each BN_BLINDING structure (previously,
3108      the last update always remained unused).
3109      [Emilia Käsper (Google)]
3110
3111   *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
3112      [Bob Buckholz (Google)]
3113
3114  Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
3115
3116   *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
3117      by initialising X509_STORE_CTX properly. (CVE-2011-3207)
3118      [Kaspar Brand <ossl@velox.ch>]
3119
3120   *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
3121      for multi-threaded use of ECDH. (CVE-2011-3210)
3122      [Adam Langley (Google)]
3123
3124   *) Fix x509_name_ex_d2i memory leak on bad inputs.
3125      [Bodo Moeller]
3126
3127   *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
3128      signature public key algorithm by using OID xref utilities instead.
3129      Before this you could only use some ECC ciphersuites with SHA1 only.
3130      [Steve Henson]
3131
3132   *) Add protection against ECDSA timing attacks as mentioned in the paper
3133      by Billy Bob Brumley and Nicola Tuveri, see:
3134
3135         http://eprint.iacr.org/2011/232.pdf
3136
3137      [Billy Bob Brumley and Nicola Tuveri]
3138
3139  Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
3140
3141   *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
3142      [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
3143
3144   *) Fix bug in string printing code: if *any* escaping is enabled we must
3145      escape the escape character (backslash) or the resulting string is
3146      ambiguous.
3147      [Steve Henson]
3148
3149  Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
3150
3151   *) Disable code workaround for ancient and obsolete Netscape browsers
3152      and servers: an attacker can use it in a ciphersuite downgrade attack.
3153      Thanks to Martin Rex for discovering this bug. CVE-2010-4180
3154      [Steve Henson]
3155
3156   *) Fixed J-PAKE implementation error, originally discovered by
3157      Sebastien Martini, further info and confirmation from Stefan
3158      Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
3159      [Ben Laurie]
3160
3161  Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
3162
3163   *) Fix extension code to avoid race conditions which can result in a buffer
3164      overrun vulnerability: resumed sessions must not be modified as they can
3165      be shared by multiple threads. CVE-2010-3864
3166      [Steve Henson]
3167
3168   *) Fix WIN32 build system to correctly link an ENGINE directory into
3169      a DLL.
3170      [Steve Henson]
3171
3172  Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
3173
3174   *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover
3175      (CVE-2010-1633)
3176      [Steve Henson, Peter-Michael Hager <hager@dortmund.net>]
3177
3178  Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
3179
3180   *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
3181      context. The operation can be customised via the ctrl mechanism in
3182      case ENGINEs want to include additional functionality.
3183      [Steve Henson]
3184
3185   *) Tolerate yet another broken PKCS#8 key format: private key value negative.
3186      [Steve Henson]
3187
3188   *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
3189      output hashes compatible with older versions of OpenSSL.
3190      [Willy Weisz <weisz@vcpc.univie.ac.at>]
3191
3192   *) Fix compression algorithm handling: if resuming a session use the
3193      compression algorithm of the resumed session instead of determining
3194      it from client hello again. Don't allow server to change algorithm.
3195      [Steve Henson]
3196
3197   *) Add load_crls() function to apps tidying load_certs() too. Add option
3198      to verify utility to allow additional CRLs to be included.
3199      [Steve Henson]
3200
3201   *) Update OCSP request code to permit adding custom headers to the request:
3202      some responders need this.
3203      [Steve Henson]
3204
3205   *) The function EVP_PKEY_sign() returns <=0 on error: check return code
3206      correctly.
3207      [Julia Lawall <julia@diku.dk>]
3208
3209   *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
3210      needlessly dereferenced structures, used obsolete functions and
3211      didn't handle all updated verify codes correctly.
3212      [Steve Henson]
3213
3214   *) Disable MD2 in the default configuration.
3215      [Steve Henson]
3216
3217   *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
3218      indicate the initial BIO being pushed or popped. This makes it possible
3219      to determine whether the BIO is the one explicitly called or as a result
3220      of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
3221      it handles reference counts correctly and doesn't zero out the I/O bio
3222      when it is not being explicitly popped. WARNING: applications which
3223      included workarounds for the old buggy behaviour will need to be modified
3224      or they could free up already freed BIOs.
3225      [Steve Henson]
3226
3227   *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
3228      renaming to all platforms (within the 0.9.8 branch, this was
3229      done conditionally on Netware platforms to avoid a name clash).
3230      [Guenter <lists@gknw.net>]
3231
3232   *) Add ECDHE and PSK support to DTLS.
3233      [Michael Tuexen <tuexen@fh-muenster.de>]
3234
3235   *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
3236      be used on C++.
3237      [Steve Henson]
3238
3239   *) Add "missing" function EVP_MD_flags() (without this the only way to
3240      retrieve a digest flags is by accessing the structure directly. Update
3241      EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
3242      or cipher is registered as in the "from" argument. Print out all
3243      registered digests in the dgst usage message instead of manually
3244      attempting to work them out.
3245      [Steve Henson]
3246
3247   *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
3248      this allows the use of compression and extensions. Change default cipher
3249      string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
3250      by default unless an application cipher string requests it.
3251      [Steve Henson]
3252
3253   *) Alter match criteria in PKCS12_parse(). It used to try to use local
3254      key ids to find matching certificates and keys but some PKCS#12 files
3255      don't follow the (somewhat unwritten) rules and this strategy fails.
3256      Now just gather all certificates together and the first private key
3257      then look for the first certificate that matches the key.
3258      [Steve Henson]
3259
3260   *) Support use of registered digest and cipher names for dgst and cipher
3261      commands instead of having to add each one as a special case. So now
3262      you can do:
3263
3264         openssl sha256 foo
3265
3266      as well as:
3267
3268         openssl dgst -sha256 foo
3269
3270      and this works for ENGINE based algorithms too.
3271
3272      [Steve Henson]
3273
3274   *) Update Gost ENGINE to support parameter files.
3275      [Victor B. Wagner <vitus@cryptocom.ru>]
3276
3277   *) Support GeneralizedTime in ca utility.
3278      [Oliver Martin <oliver@volatilevoid.net>, Steve Henson]
3279
3280   *) Enhance the hash format used for certificate directory links. The new
3281      form uses the canonical encoding (meaning equivalent names will work
3282      even if they aren't identical) and uses SHA1 instead of MD5. This form
3283      is incompatible with the older format and as a result c_rehash should
3284      be used to rebuild symbolic links.
3285      [Steve Henson]
3286
3287   *) Make PKCS#8 the default write format for private keys, replacing the
3288      traditional format. This form is standardised, more secure and doesn't
3289      include an implicit MD5 dependency.
3290      [Steve Henson]
3291
3292   *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
3293      committed to OpenSSL should pass this lot as a minimum.
3294      [Steve Henson]
3295
3296   *) Add session ticket override functionality for use by EAP-FAST.
3297      [Jouni Malinen <j@w1.fi>]
3298
3299   *) Modify HMAC functions to return a value. Since these can be implemented
3300      in an ENGINE errors can occur.
3301      [Steve Henson]
3302
3303   *) Type-checked OBJ_bsearch_ex.
3304      [Ben Laurie]
3305
3306   *) Type-checked OBJ_bsearch. Also some constification necessitated
3307      by type-checking.  Still to come: TXT_DB, bsearch(?),
3308      OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
3309      CONF_VALUE.
3310      [Ben Laurie]
3311
3312   *) New function OPENSSL_gmtime_adj() to add a specific number of days and
3313      seconds to a tm structure directly, instead of going through OS
3314      specific date routines. This avoids any issues with OS routines such
3315      as the year 2038 bug. New *_adj() functions for ASN1 time structures
3316      and X509_time_adj_ex() to cover the extended range. The existing
3317      X509_time_adj() is still usable and will no longer have any date issues.
3318      [Steve Henson]
3319
3320   *) Delta CRL support. New use deltas option which will attempt to locate
3321      and search any appropriate delta CRLs available.
3322
3323      This work was sponsored by Google.
3324      [Steve Henson]
3325
3326   *) Support for CRLs partitioned by reason code. Reorganise CRL processing
3327      code and add additional score elements. Validate alternate CRL paths
3328      as part of the CRL checking and indicate a new error "CRL path validation
3329      error" in this case. Applications wanting additional details can use
3330      the verify callback and check the new "parent" field. If this is not
3331      NULL CRL path validation is taking place. Existing applications won't
3332      see this because it requires extended CRL support which is off by
3333      default.
3334
3335      This work was sponsored by Google.
3336      [Steve Henson]
3337
3338   *) Support for freshest CRL extension.
3339
3340      This work was sponsored by Google.
3341      [Steve Henson]
3342
3343   *) Initial indirect CRL support. Currently only supported in the CRLs
3344      passed directly and not via lookup. Process certificate issuer
3345      CRL entry extension and lookup CRL entries by bother issuer name
3346      and serial number. Check and process CRL issuer entry in IDP extension.
3347
3348      This work was sponsored by Google.
3349      [Steve Henson]
3350
3351   *) Add support for distinct certificate and CRL paths. The CRL issuer
3352      certificate is validated separately in this case. Only enabled if
3353      an extended CRL support flag is set: this flag will enable additional
3354      CRL functionality in future.
3355
3356      This work was sponsored by Google.
3357      [Steve Henson]
3358
3359   *) Add support for policy mappings extension.
3360
3361      This work was sponsored by Google.
3362      [Steve Henson]
3363
3364   *) Fixes to pathlength constraint, self issued certificate handling,
3365      policy processing to align with RFC3280 and PKITS tests.
3366
3367      This work was sponsored by Google.
3368      [Steve Henson]
3369
3370   *) Support for name constraints certificate extension. DN, email, DNS
3371      and URI types are currently supported.
3372
3373      This work was sponsored by Google.
3374      [Steve Henson]
3375
3376   *) To cater for systems that provide a pointer-based thread ID rather
3377      than numeric, deprecate the current numeric thread ID mechanism and
3378      replace it with a structure and associated callback type. This
3379      mechanism allows a numeric "hash" to be extracted from a thread ID in
3380      either case, and on platforms where pointers are larger than 'long',
3381      mixing is done to help ensure the numeric 'hash' is usable even if it
3382      can't be guaranteed unique. The default mechanism is to use "&errno"
3383      as a pointer-based thread ID to distinguish between threads.
3384
3385      Applications that want to provide their own thread IDs should now use
3386      CRYPTO_THREADID_set_callback() to register a callback that will call
3387      either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
3388
3389      Note that ERR_remove_state() is now deprecated, because it is tied
3390      to the assumption that thread IDs are numeric.  ERR_remove_state(0)
3391      to free the current thread's error state should be replaced by
3392      ERR_remove_thread_state(NULL).
3393
3394      (This new approach replaces the functions CRYPTO_set_idptr_callback(),
3395      CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
3396      OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
3397      application was previously providing a numeric thread callback that
3398      was inappropriate for distinguishing threads, then uniqueness might
3399      have been obtained with &errno that happened immediately in the
3400      intermediate development versions of OpenSSL; this is no longer the
3401      case, the numeric thread callback will now override the automatic use
3402      of &errno.)
3403      [Geoff Thorpe, with help from Bodo Moeller]
3404
3405   *) Initial support for different CRL issuing certificates. This covers a
3406      simple case where the self issued certificates in the chain exist and
3407      the real CRL issuer is higher in the existing chain.
3408
3409      This work was sponsored by Google.
3410      [Steve Henson]
3411
3412   *) Removed effectively defunct crypto/store from the build.
3413      [Ben Laurie]
3414
3415   *) Revamp of STACK to provide stronger type-checking. Still to come:
3416      TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
3417      ASN1_STRING, CONF_VALUE.
3418      [Ben Laurie]
3419
3420   *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
3421      RAM on SSL connections.  This option can save about 34k per idle SSL.
3422      [Nick Mathewson]
3423
3424   *) Revamp of LHASH to provide stronger type-checking. Still to come:
3425      STACK, TXT_DB, bsearch, qsort.
3426      [Ben Laurie]
3427
3428   *) Initial support for Cryptographic Message Syntax (aka CMS) based
3429      on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
3430      support for data, signedData, compressedData, digestedData and
3431      encryptedData, envelopedData types included. Scripts to check against
3432      RFC4134 examples draft and interop and consistency checks of many
3433      content types and variants.
3434      [Steve Henson]
3435
3436   *) Add options to enc utility to support use of zlib compression BIO.
3437      [Steve Henson]
3438
3439   *) Extend mk1mf to support importing of options and assembly language
3440      files from Configure script, currently only included in VC-WIN32.
3441      The assembly language rules can now optionally generate the source
3442      files from the associated perl scripts.
3443      [Steve Henson]
3444
3445   *) Implement remaining functionality needed to support GOST ciphersuites.
3446      Interop testing has been performed using CryptoPro implementations.
3447      [Victor B. Wagner <vitus@cryptocom.ru>]
3448
3449   *) s390x assembler pack.
3450      [Andy Polyakov]
3451
3452   *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
3453      "family."
3454      [Andy Polyakov]
3455
3456   *) Implement Opaque PRF Input TLS extension as specified in
3457      draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
3458      official specification yet and no extension type assignment by
3459      IANA exists, this extension (for now) will have to be explicitly
3460      enabled when building OpenSSL by providing the extension number
3461      to use.  For example, specify an option
3462
3463          -DTLSEXT_TYPE_opaque_prf_input=0x9527
3464
3465      to the "config" or "Configure" script to enable the extension,
3466      assuming extension number 0x9527 (which is a completely arbitrary
3467      and unofficial assignment based on the MD5 hash of the Internet
3468      Draft).  Note that by doing so, you potentially lose
3469      interoperability with other TLS implementations since these might
3470      be using the same extension number for other purposes.
3471
3472      SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
3473      opaque PRF input value to use in the handshake.  This will create
3474      an interal copy of the length-'len' string at 'src', and will
3475      return non-zero for success.
3476
3477      To get more control and flexibility, provide a callback function
3478      by using
3479
3480           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
3481           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
3482
3483      where
3484
3485           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
3486           void *arg;
3487
3488      Callback function 'cb' will be called in handshakes, and is
3489      expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
3490      Argument 'arg' is for application purposes (the value as given to
3491      SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
3492      be provided to the callback function).  The callback function
3493      has to return non-zero to report success: usually 1 to use opaque
3494      PRF input just if possible, or 2 to enforce use of the opaque PRF
3495      input.  In the latter case, the library will abort the handshake
3496      if opaque PRF input is not successfully negotiated.
3497
3498      Arguments 'peerinput' and 'len' given to the callback function
3499      will always be NULL and 0 in the case of a client.  A server will
3500      see the client's opaque PRF input through these variables if
3501      available (NULL and 0 otherwise).  Note that if the server
3502      provides an opaque PRF input, the length must be the same as the
3503      length of the client's opaque PRF input.
3504
3505      Note that the callback function will only be called when creating
3506      a new session (session resumption can resume whatever was
3507      previously negotiated), and will not be called in SSL 2.0
3508      handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
3509      SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
3510      for applications that need to enforce opaque PRF input.
3511
3512      [Bodo Moeller]
3513
3514   *) Update ssl code to support digests other than SHA1+MD5 for handshake
3515      MAC.
3516
3517      [Victor B. Wagner <vitus@cryptocom.ru>]
3518
3519   *) Add RFC4507 support to OpenSSL. This includes the corrections in
3520      RFC4507bis. The encrypted ticket format is an encrypted encoded
3521      SSL_SESSION structure, that way new session features are automatically
3522      supported.
3523
3524      If a client application caches session in an SSL_SESSION structure
3525      support is transparent because tickets are now stored in the encoded
3526      SSL_SESSION.
3527
3528      The SSL_CTX structure automatically generates keys for ticket
3529      protection in servers so again support should be possible
3530      with no application modification.
3531
3532      If a client or server wishes to disable RFC4507 support then the option
3533      SSL_OP_NO_TICKET can be set.
3534
3535      Add a TLS extension debugging callback to allow the contents of any client
3536      or server extensions to be examined.
3537
3538      This work was sponsored by Google.
3539      [Steve Henson]
3540
3541   *) Final changes to avoid use of pointer pointer casts in OpenSSL.
3542      OpenSSL should now compile cleanly on gcc 4.2
3543      [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson]
3544
3545   *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
3546      support including streaming MAC support: this is required for GOST
3547      ciphersuite support.
3548      [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson]
3549
3550   *) Add option -stream to use PKCS#7 streaming in smime utility. New
3551      function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
3552      to output in BER and PEM format.
3553      [Steve Henson]
3554
3555   *) Experimental support for use of HMAC via EVP_PKEY interface. This
3556      allows HMAC to be handled via the EVP_DigestSign*() interface. The
3557      EVP_PKEY "key" in this case is the HMAC key, potentially allowing
3558      ENGINE support for HMAC keys which are unextractable. New -mac and
3559      -macopt options to dgst utility.
3560      [Steve Henson]
3561
3562   *) New option -sigopt to dgst utility. Update dgst to use
3563      EVP_Digest{Sign,Verify}*. These two changes make it possible to use
3564      alternative signing parameters such as X9.31 or PSS in the dgst
3565      utility.
3566      [Steve Henson]
3567
3568   *) Change ssl_cipher_apply_rule(), the internal function that does
3569      the work each time a ciphersuite string requests enabling
3570      ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
3571      removing ("!foo+bar") a class of ciphersuites: Now it maintains
3572      the order of disabled ciphersuites such that those ciphersuites
3573      that most recently went from enabled to disabled not only stay
3574      in order with respect to each other, but also have higher priority
3575      than other disabled ciphersuites the next time ciphersuites are
3576      enabled again.
3577
3578      This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
3579      the same ciphersuites as with "HIGH" alone, but in a specific
3580      order where the PSK ciphersuites come first (since they are the
3581      most recently disabled ciphersuites when "HIGH" is parsed).
3582
3583      Also, change ssl_create_cipher_list() (using this new
3584      funcionality) such that between otherwise identical
3585      cihpersuites, ephemeral ECDH is preferred over ephemeral DH in
3586      the default order.
3587      [Bodo Moeller]
3588
3589   *) Change ssl_create_cipher_list() so that it automatically
3590      arranges the ciphersuites in reasonable order before starting
3591      to process the rule string.  Thus, the definition for "DEFAULT"
3592      (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
3593      remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH".
3594      This makes it much easier to arrive at a reasonable default order
3595      in applications for which anonymous ciphers are OK (meaning
3596      that you can't actually use DEFAULT).
3597      [Bodo Moeller; suggested by Victor Duchovni]
3598
3599   *) Split the SSL/TLS algorithm mask (as used for ciphersuite string
3600      processing) into multiple integers instead of setting
3601      "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
3602      "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
3603      (These masks as well as the individual bit definitions are hidden
3604      away into the non-exported interface ssl/ssl_locl.h, so this
3605      change to the definition of the SSL_CIPHER structure shouldn't
3606      affect applications.)  This give us more bits for each of these
3607      categories, so there is no longer a need to coagulate AES128 and
3608      AES256 into a single algorithm bit, and to coagulate Camellia128
3609      and Camellia256 into a single algorithm bit, which has led to all
3610      kinds of kludges.
3611
3612      Thus, among other things, the kludge introduced in 0.9.7m and
3613      0.9.8e for masking out AES256 independently of AES128 or masking
3614      out Camellia256 independently of AES256 is not needed here in 0.9.9.
3615
3616      With the change, we also introduce new ciphersuite aliases that
3617      so far were missing: "AES128", "AES256", "CAMELLIA128", and
3618      "CAMELLIA256".
3619      [Bodo Moeller]
3620
3621   *) Add support for dsa-with-SHA224 and dsa-with-SHA256.
3622      Use the leftmost N bytes of the signature input if the input is
3623      larger than the prime q (with N being the size in bytes of q).
3624      [Nils Larsch]
3625
3626   *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
3627      it yet and it is largely untested.
3628      [Steve Henson]
3629
3630   *) Add support for the ecdsa-with-SHA224/256/384/512 signature types.
3631      [Nils Larsch]
3632
3633   *) Initial incomplete changes to avoid need for function casts in OpenSSL
3634      some compilers (gcc 4.2 and later) reject their use. Safestack is
3635      reimplemented.  Update ASN1 to avoid use of legacy functions.
3636      [Steve Henson]
3637
3638   *) Win32/64 targets are linked with Winsock2.
3639      [Andy Polyakov]
3640
3641   *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
3642      to external functions. This can be used to increase CRL handling
3643      efficiency especially when CRLs are very large by (for example) storing
3644      the CRL revoked certificates in a database.
3645      [Steve Henson]
3646
3647   *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so
3648      new CRLs added to a directory can be used. New command line option
3649      -verify_return_error to s_client and s_server. This causes real errors
3650      to be returned by the verify callback instead of carrying on no matter
3651      what. This reflects the way a "real world" verify callback would behave.
3652      [Steve Henson]
3653
3654   *) GOST engine, supporting several GOST algorithms and public key formats.
3655      Kindly donated by Cryptocom.
3656      [Cryptocom]
3657
3658   *) Partial support for Issuing Distribution Point CRL extension. CRLs
3659      partitioned by DP are handled but no indirect CRL or reason partitioning
3660      (yet). Complete overhaul of CRL handling: now the most suitable CRL is
3661      selected via a scoring technique which handles IDP and AKID in CRLs.
3662      [Steve Henson]
3663
3664   *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
3665      will ultimately be used for all verify operations: this will remove the
3666      X509_STORE dependency on certificate verification and allow alternative
3667      lookup methods.  X509_STORE based implementations of these two callbacks.
3668      [Steve Henson]
3669
3670   *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
3671      Modify get_crl() to find a valid (unexpired) CRL if possible.
3672      [Steve Henson]
3673
3674   *) New function X509_CRL_match() to check if two CRLs are identical. Normally
3675      this would be called X509_CRL_cmp() but that name is already used by
3676      a function that just compares CRL issuer names. Cache several CRL
3677      extensions in X509_CRL structure and cache CRLDP in X509.
3678      [Steve Henson]
3679
3680   *) Store a "canonical" representation of X509_NAME structure (ASN1 Name)
3681      this maps equivalent X509_NAME structures into a consistent structure.
3682      Name comparison can then be performed rapidly using memcmp().
3683      [Steve Henson]
3684
3685   *) Non-blocking OCSP request processing. Add -timeout option to ocsp
3686      utility.
3687      [Steve Henson]
3688
3689   *) Allow digests to supply their own micalg string for S/MIME type using
3690      the ctrl EVP_MD_CTRL_MICALG.
3691      [Steve Henson]
3692
3693   *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the
3694      EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
3695      ctrl. It can then customise the structure before and/or after signing
3696      if necessary.
3697      [Steve Henson]
3698
3699   *) New function OBJ_add_sigid() to allow application defined signature OIDs
3700      to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
3701      to free up any added signature OIDs.
3702      [Steve Henson]
3703
3704   *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
3705      EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
3706      digest and cipher tables. New options added to openssl utility:
3707      list-message-digest-algorithms and list-cipher-algorithms.
3708      [Steve Henson]
3709
3710   *) Change the array representation of binary polynomials: the list
3711      of degrees of non-zero coefficients is now terminated with -1.
3712      Previously it was terminated with 0, which was also part of the
3713      value; thus, the array representation was not applicable to
3714      polynomials where t^0 has coefficient zero.  This change makes
3715      the array representation useful in a more general context.
3716      [Douglas Stebila]
3717
3718   *) Various modifications and fixes to SSL/TLS cipher string
3719      handling.  For ECC, the code now distinguishes between fixed ECDH
3720      with RSA certificates on the one hand and with ECDSA certificates
3721      on the other hand, since these are separate ciphersuites.  The
3722      unused code for Fortezza ciphersuites has been removed.
3723
3724      For consistency with EDH, ephemeral ECDH is now called "EECDH"
3725      (not "ECDHE").  For consistency with the code for DH
3726      certificates, use of ECDH certificates is now considered ECDH
3727      authentication, not RSA or ECDSA authentication (the latter is
3728      merely the CA's signing algorithm and not actively used in the
3729      protocol).
3730
3731      The temporary ciphersuite alias "ECCdraft" is no longer
3732      available, and ECC ciphersuites are no longer excluded from "ALL"
3733      and "DEFAULT".  The following aliases now exist for RFC 4492
3734      ciphersuites, most of these by analogy with the DH case:
3735
3736          kECDHr   - ECDH cert, signed with RSA
3737          kECDHe   - ECDH cert, signed with ECDSA
3738          kECDH    - ECDH cert (signed with either RSA or ECDSA)
3739          kEECDH   - ephemeral ECDH
3740          ECDH     - ECDH cert or ephemeral ECDH
3741
3742          aECDH    - ECDH cert
3743          aECDSA   - ECDSA cert
3744          ECDSA    - ECDSA cert
3745
3746          AECDH    - anonymous ECDH
3747          EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
3748
3749      [Bodo Moeller]
3750
3751   *) Add additional S/MIME capabilities for AES and GOST ciphers if supported.
3752      Use correct micalg parameters depending on digest(s) in signed message.
3753      [Steve Henson]
3754
3755   *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
3756      an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
3757      [Steve Henson]
3758
3759   *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
3760      an engine to register a method. Add ENGINE lookups for methods and
3761      functional reference processing.
3762      [Steve Henson]
3763
3764   *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of
3765      EVP_{Sign,Verify}* which allow an application to customise the signature
3766      process.
3767      [Steve Henson]
3768
3769   *) New -resign option to smime utility. This adds one or more signers
3770      to an existing PKCS#7 signedData structure. Also -md option to use an
3771      alternative message digest algorithm for signing.
3772      [Steve Henson]
3773
3774   *) Tidy up PKCS#7 routines and add new functions to make it easier to
3775      create PKCS7 structures containing multiple signers. Update smime
3776      application to support multiple signers.
3777      [Steve Henson]
3778
3779   *) New -macalg option to pkcs12 utility to allow setting of an alternative
3780      digest MAC.
3781      [Steve Henson]
3782
3783   *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
3784      Reorganize PBE internals to lookup from a static table using NIDs,
3785      add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
3786      EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
3787      PRF which will be automatically used with PBES2.
3788      [Steve Henson]
3789
3790   *) Replace the algorithm specific calls to generate keys in "req" with the
3791      new API.
3792      [Steve Henson]
3793
3794   *) Update PKCS#7 enveloped data routines to use new API. This is now
3795      supported by any public key method supporting the encrypt operation. A
3796      ctrl is added to allow the public key algorithm to examine or modify
3797      the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
3798      a no op.
3799      [Steve Henson]
3800
3801   *) Add a ctrl to asn1 method to allow a public key algorithm to express
3802      a default digest type to use. In most cases this will be SHA1 but some
3803      algorithms (such as GOST) need to specify an alternative digest. The
3804      return value indicates how strong the preference is 1 means optional and
3805      2 is mandatory (that is it is the only supported type). Modify
3806      ASN1_item_sign() to accept a NULL digest argument to indicate it should
3807      use the default md. Update openssl utilities to use the default digest
3808      type for signing if it is not explicitly indicated.
3809      [Steve Henson]
3810
3811   *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
3812      EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
3813      signing method from the key type. This effectively removes the link
3814      between digests and public key types.
3815      [Steve Henson]
3816
3817   *) Add an OID cross reference table and utility functions. Its purpose is to
3818      translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
3819      rsaEncryption. This will allow some of the algorithm specific hackery
3820      needed to use the correct OID to be removed.
3821      [Steve Henson]
3822
3823   *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
3824      structures for PKCS7_sign(). They are now set up by the relevant public
3825      key ASN1 method.
3826      [Steve Henson]
3827
3828   *) Add provisional EC pkey method with support for ECDSA and ECDH.
3829      [Steve Henson]
3830
3831   *) Add support for key derivation (agreement) in the API, DH method and
3832      pkeyutl.
3833      [Steve Henson]
3834
3835   *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
3836      public and private key formats. As a side effect these add additional
3837      command line functionality not previously available: DSA signatures can be
3838      generated and verified using pkeyutl and DH key support and generation in
3839      pkey, genpkey.
3840      [Steve Henson]
3841
3842   *) BeOS support.
3843      [Oliver Tappe <zooey@hirschkaefer.de>]
3844
3845   *) New make target "install_html_docs" installs HTML renditions of the
3846      manual pages.
3847      [Oliver Tappe <zooey@hirschkaefer.de>]
3848
3849   *) New utility "genpkey" this is analogous to "genrsa" etc except it can
3850      generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
3851      support key and parameter generation and add initial key generation
3852      functionality for RSA.
3853      [Steve Henson]
3854
3855   *) Add functions for main EVP_PKEY_method operations. The undocumented
3856      functions EVP_PKEY_{encrypt,decrypt} have been renamed to
3857      EVP_PKEY_{encrypt,decrypt}_old.
3858      [Steve Henson]
3859
3860   *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public
3861      key API, doesn't do much yet.
3862      [Steve Henson]
3863
3864   *) New function EVP_PKEY_asn1_get0_info() to retrieve information about
3865      public key algorithms. New option to openssl utility:
3866      "list-public-key-algorithms" to print out info.
3867      [Steve Henson]
3868
3869   *) Implement the Supported Elliptic Curves Extension for
3870      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
3871      [Douglas Stebila]
3872
3873   *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
3874      EVP_CIPHER structures to avoid later problems in EVP_cleanup().
3875      [Steve Henson]
3876
3877   *) New utilities pkey and pkeyparam. These are similar to algorithm specific
3878      utilities such as rsa, dsa, dsaparam etc except they process any key
3879      type.
3880      [Steve Henson]
3881
3882   *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
3883      functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
3884      EVP_PKEY_print_param() to print public key data from an EVP_PKEY
3885      structure.
3886      [Steve Henson]
3887
3888   *) Initial support for pluggable public key ASN1.
3889      De-spaghettify the public key ASN1 handling. Move public and private
3890      key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
3891      algorithm specific handling to a single module within the relevant
3892      algorithm directory. Add functions to allow (near) opaque processing
3893      of public and private key structures.
3894      [Steve Henson]
3895
3896   *) Implement the Supported Point Formats Extension for
3897      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
3898      [Douglas Stebila]
3899
3900   *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
3901      for the psk identity [hint] and the psk callback functions to the
3902      SSL_SESSION, SSL and SSL_CTX structure.
3903
3904      New ciphersuites:
3905          PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
3906          PSK-AES256-CBC-SHA
3907
3908      New functions:
3909          SSL_CTX_use_psk_identity_hint
3910          SSL_get_psk_identity_hint
3911          SSL_get_psk_identity
3912          SSL_use_psk_identity_hint
3913
3914      [Mika Kousa and Pasi Eronen of Nokia Corporation]
3915
3916   *) Add RFC 3161 compliant time stamp request creation, response generation
3917      and response verification functionality.
3918      [Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project]
3919
3920   *) Add initial support for TLS extensions, specifically for the server_name
3921      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
3922      have new members for a host name.  The SSL data structure has an
3923      additional member SSL_CTX *initial_ctx so that new sessions can be
3924      stored in that context to allow for session resumption, even after the
3925      SSL has been switched to a new SSL_CTX in reaction to a client's
3926      server_name extension.
3927
3928      New functions (subject to change):
3929
3930          SSL_get_servername()
3931          SSL_get_servername_type()
3932          SSL_set_SSL_CTX()
3933
3934      New CTRL codes and macros (subject to change):
3935
3936          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
3937                                  - SSL_CTX_set_tlsext_servername_callback()
3938          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
3939                                       - SSL_CTX_set_tlsext_servername_arg()
3940          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
3941
3942      openssl s_client has a new '-servername ...' option.
3943
3944      openssl s_server has new options '-servername_host ...', '-cert2 ...',
3945      '-key2 ...', '-servername_fatal' (subject to change).  This allows
3946      testing the HostName extension for a specific single host name ('-cert'
3947      and '-key' remain fallbacks for handshakes without HostName
3948      negotiation).  If the unrecognized_name alert has to be sent, this by
3949      default is a warning; it becomes fatal with the '-servername_fatal'
3950      option.
3951
3952      [Peter Sylvester,  Remy Allais, Christophe Renou]
3953
3954   *) Whirlpool hash implementation is added.
3955      [Andy Polyakov]
3956
3957   *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
3958      bn(64,32). Because of instruction set limitations it doesn't have
3959      any negative impact on performance. This was done mostly in order
3960      to make it possible to share assembler modules, such as bn_mul_mont
3961      implementations, between 32- and 64-bit builds without hassle.
3962      [Andy Polyakov]
3963
3964   *) Move code previously exiled into file crypto/ec/ec2_smpt.c
3965      to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
3966      macro.
3967      [Bodo Moeller]
3968
3969   *) New candidate for BIGNUM assembler implementation, bn_mul_mont,
3970      dedicated Montgomery multiplication procedure, is introduced.
3971      BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
3972      "64-bit" performance on certain 32-bit targets.
3973      [Andy Polyakov]
3974
3975   *) New option SSL_OP_NO_COMP to disable use of compression selectively
3976      in SSL structures. New SSL ctrl to set maximum send fragment size.
3977      Save memory by seeting the I/O buffer sizes dynamically instead of
3978      using the maximum available value.
3979      [Steve Henson]
3980
3981   *) New option -V for 'openssl ciphers'. This prints the ciphersuite code
3982      in addition to the text details.
3983      [Bodo Moeller]
3984
3985   *) Very, very preliminary EXPERIMENTAL support for printing of general
3986      ASN1 structures. This currently produces rather ugly output and doesn't
3987      handle several customised structures at all.
3988      [Steve Henson]
3989
3990   *) Integrated support for PVK file format and some related formats such
3991      as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
3992      these in the 'rsa' and 'dsa' utilities.
3993      [Steve Henson]
3994
3995   *) Support for PKCS#1 RSAPublicKey format on rsa utility command line.
3996      [Steve Henson]
3997
3998   *) Remove the ancient ASN1_METHOD code. This was only ever used in one
3999      place for the (very old) "NETSCAPE" format certificates which are now
4000      handled using new ASN1 code equivalents.
4001      [Steve Henson]
4002
4003   *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
4004      pointer and make the SSL_METHOD parameter in SSL_CTX_new,
4005      SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
4006      [Nils Larsch]
4007
4008   *) Modify CRL distribution points extension code to print out previously
4009      unsupported fields. Enhance extension setting code to allow setting of
4010      all fields.
4011      [Steve Henson]
4012
4013   *) Add print and set support for Issuing Distribution Point CRL extension.
4014      [Steve Henson]
4015
4016   *) Change 'Configure' script to enable Camellia by default.
4017      [NTT]
4018
4019  Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
4020
4021   *) When rejecting SSL/TLS records due to an incorrect version number, never
4022      update s->server with a new major version number.  As of
4023      - OpenSSL 0.9.8m if 'short' is a 16-bit type,
4024      - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
4025      the previous behavior could result in a read attempt at NULL when
4026      receiving specific incorrect SSL/TLS records once record payload
4027      protection is active.  (CVE-2010-0740)
4028      [Bodo Moeller, Adam Langley <agl@chromium.org>]
4029
4030   *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
4031      could be crashed if the relevant tables were not present (e.g. chrooted).
4032      [Tomas Hoger <thoger@redhat.com>]
4033
4034  Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
4035
4036   *) Always check bn_wexpend() return values for failure.  (CVE-2009-3245)
4037      [Martin Olsson, Neel Mehta]
4038
4039   *) Fix X509_STORE locking: Every 'objs' access requires a lock (to
4040      accommodate for stack sorting, always a write lock!).
4041      [Bodo Moeller]
4042
4043   *) On some versions of WIN32 Heap32Next is very slow. This can cause
4044      excessive delays in the RAND_poll(): over a minute. As a workaround
4045      include a time check in the inner Heap32Next loop too.
4046      [Steve Henson]
4047
4048   *) The code that handled flushing of data in SSL/TLS originally used the
4049      BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
4050      the problem outlined in PR#1949. The fix suggested there however can
4051      trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
4052      of Apache). So instead simplify the code to flush unconditionally.
4053      This should be fine since flushing with no data to flush is a no op.
4054      [Steve Henson]
4055
4056   *) Handle TLS versions 2.0 and later properly and correctly use the
4057      highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
4058      off ancient servers have a habit of sticking around for a while...
4059      [Steve Henson]
4060
4061   *) Modify compression code so it frees up structures without using the
4062      ex_data callbacks. This works around a problem where some applications
4063      call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
4064      restarting) then use compression (e.g. SSL with compression) later.
4065      This results in significant per-connection memory leaks and
4066      has caused some security issues including CVE-2008-1678 and
4067      CVE-2009-4355.
4068      [Steve Henson]
4069
4070   *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
4071      change when encrypting or decrypting.
4072      [Bodo Moeller]
4073
4074   *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
4075      connect and renegotiate with servers which do not support RI.
4076      Until RI is more widely deployed this option is enabled by default.
4077      [Steve Henson]
4078
4079   *) Add "missing" ssl ctrls to clear options and mode.
4080      [Steve Henson]
4081
4082   *) If client attempts to renegotiate and doesn't support RI respond with
4083      a no_renegotiation alert as required by RFC5746.  Some renegotiating
4084      TLS clients will continue a connection gracefully when they receive
4085      the alert. Unfortunately OpenSSL mishandled this alert and would hang
4086      waiting for a server hello which it will never receive. Now we treat a
4087      received no_renegotiation alert as a fatal error. This is because
4088      applications requesting a renegotiation might well expect it to succeed
4089      and would have no code in place to handle the server denying it so the
4090      only safe thing to do is to terminate the connection.
4091      [Steve Henson]
4092
4093   *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
4094      peer supports secure renegotiation and 0 otherwise. Print out peer
4095      renegotiation support in s_client/s_server.
4096      [Steve Henson]
4097
4098   *) Replace the highly broken and deprecated SPKAC certification method with
4099      the updated NID creation version. This should correctly handle UTF8.
4100      [Steve Henson]
4101
4102   *) Implement RFC5746. Re-enable renegotiation but require the extension
4103      as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
4104      turns out to be a bad idea. It has been replaced by
4105      SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
4106      SSL_CTX_set_options(). This is really not recommended unless you
4107      know what you are doing.
4108      [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson]
4109
4110   *) Fixes to stateless session resumption handling. Use initial_ctx when
4111      issuing and attempting to decrypt tickets in case it has changed during
4112      servername handling. Use a non-zero length session ID when attempting
4113      stateless session resumption: this makes it possible to determine if
4114      a resumption has occurred immediately after receiving server hello
4115      (several places in OpenSSL subtly assume this) instead of later in
4116      the handshake.
4117      [Steve Henson]
4118
4119   *) The functions ENGINE_ctrl(), OPENSSL_isservice(),
4120      CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
4121      fixes for a few places where the return code is not checked
4122      correctly.
4123      [Julia Lawall <julia@diku.dk>]
4124
4125   *) Add --strict-warnings option to Configure script to include devteam
4126      warnings in other configurations.
4127      [Steve Henson]
4128
4129   *) Add support for --libdir option and LIBDIR variable in makefiles. This
4130      makes it possible to install openssl libraries in locations which
4131      have names other than "lib", for example "/usr/lib64" which some
4132      systems need.
4133      [Steve Henson, based on patch from Jeremy Utley]
4134
4135   *) Don't allow the use of leading 0x80 in OIDs. This is a violation of
4136      X690 8.9.12 and can produce some misleading textual output of OIDs.
4137      [Steve Henson, reported by Dan Kaminsky]
4138
4139   *) Delete MD2 from algorithm tables. This follows the recommendation in
4140      several standards that it is not used in new applications due to
4141      several cryptographic weaknesses. For binary compatibility reasons
4142      the MD2 API is still compiled in by default.
4143      [Steve Henson]
4144
4145   *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
4146      and restored.
4147      [Steve Henson]
4148
4149   *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
4150      OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
4151      clash.
4152      [Guenter <lists@gknw.net>]
4153
4154   *) Fix the server certificate chain building code to use X509_verify_cert(),
4155      it used to have an ad-hoc builder which was unable to cope with anything
4156      other than a simple chain.
4157      [David Woodhouse <dwmw2@infradead.org>, Steve Henson]
4158
4159   *) Don't check self signed certificate signatures in X509_verify_cert()
4160      by default (a flag can override this): it just wastes time without
4161      adding any security. As a useful side effect self signed root CAs
4162      with non-FIPS digests are now usable in FIPS mode.
4163      [Steve Henson]
4164
4165   *) In dtls1_process_out_of_seq_message() the check if the current message
4166      is already buffered was missing. For every new message was memory
4167      allocated, allowing an attacker to perform an denial of service attack
4168      with sending out of seq handshake messages until there is no memory
4169      left. Additionally every future messege was buffered, even if the
4170      sequence number made no sense and would be part of another handshake.
4171      So only messages with sequence numbers less than 10 in advance will be
4172      buffered.  (CVE-2009-1378)
4173      [Robin Seggelmann, discovered by Daniel Mentz]
4174
4175   *) Records are buffered if they arrive with a future epoch to be
4176      processed after finishing the corresponding handshake. There is
4177      currently no limitation to this buffer allowing an attacker to perform
4178      a DOS attack with sending records with future epochs until there is no
4179      memory left. This patch adds the pqueue_size() function to determine
4180      the size of a buffer and limits the record buffer to 100 entries.
4181      (CVE-2009-1377)
4182      [Robin Seggelmann, discovered by Daniel Mentz]
4183
4184   *) Keep a copy of frag->msg_header.frag_len so it can be used after the
4185      parent structure is freed.  (CVE-2009-1379)
4186      [Daniel Mentz]
4187
4188   *) Handle non-blocking I/O properly in SSL_shutdown() call.
4189      [Darryl Miles <darryl-mailinglists@netbauds.net>]
4190
4191   *) Add 2.5.4.* OIDs
4192      [Ilya O. <vrghost@gmail.com>]
4193
4194  Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
4195
4196   *) Disable renegotiation completely - this fixes a severe security
4197      problem (CVE-2009-3555) at the cost of breaking all
4198      renegotiation. Renegotiation can be re-enabled by setting
4199      SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
4200      run-time. This is really not recommended unless you know what
4201      you're doing.
4202      [Ben Laurie]
4203
4204  Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
4205
4206   *) Don't set val to NULL when freeing up structures, it is freed up by
4207      underlying code. If sizeof(void *) > sizeof(long) this can result in
4208      zeroing past the valid field. (CVE-2009-0789)
4209      [Paolo Ganci <Paolo.Ganci@AdNovum.CH>]
4210
4211   *) Fix bug where return value of CMS_SignerInfo_verify_content() was not
4212      checked correctly. This would allow some invalid signed attributes to
4213      appear to verify correctly. (CVE-2009-0591)
4214      [Ivan Nestlerode <inestlerode@us.ibm.com>]
4215
4216   *) Reject UniversalString and BMPString types with invalid lengths. This
4217      prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
4218      a legal length. (CVE-2009-0590)
4219      [Steve Henson]
4220
4221   *) Set S/MIME signing as the default purpose rather than setting it
4222      unconditionally. This allows applications to override it at the store
4223      level.
4224      [Steve Henson]
4225
4226   *) Permit restricted recursion of ASN1 strings. This is needed in practice
4227      to handle some structures.
4228      [Steve Henson]
4229
4230   *) Improve efficiency of mem_gets: don't search whole buffer each time
4231      for a '\n'
4232      [Jeremy Shapiro <jnshapir@us.ibm.com>]
4233
4234   *) New -hex option for openssl rand.
4235      [Matthieu Herrb]
4236
4237   *) Print out UTF8String and NumericString when parsing ASN1.
4238      [Steve Henson]
4239
4240   *) Support NumericString type for name components.
4241      [Steve Henson]
4242
4243   *) Allow CC in the environment to override the automatically chosen
4244      compiler. Note that nothing is done to ensure flags work with the
4245      chosen compiler.
4246      [Ben Laurie]
4247
4248  Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
4249
4250   *) Properly check EVP_VerifyFinal() and similar return values
4251      (CVE-2008-5077).
4252      [Ben Laurie, Bodo Moeller, Google Security Team]
4253
4254   *) Enable TLS extensions by default.
4255      [Ben Laurie]
4256
4257   *) Allow the CHIL engine to be loaded, whether the application is
4258      multithreaded or not. (This does not release the developer from the
4259      obligation to set up the dynamic locking callbacks.)
4260      [Sander Temme <sander@temme.net>]
4261
4262   *) Use correct exit code if there is an error in dgst command.
4263      [Steve Henson; problem pointed out by Roland Dirlewanger]
4264
4265   *) Tweak Configure so that you need to say "experimental-jpake" to enable
4266      JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
4267      [Bodo Moeller]
4268
4269   *) Add experimental JPAKE support, including demo authentication in
4270      s_client and s_server.
4271      [Ben Laurie]
4272
4273   *) Set the comparison function in v3_addr_canonize().
4274      [Rob Austein <sra@hactrn.net>]
4275
4276   *) Add support for XMPP STARTTLS in s_client.
4277      [Philip Paeps <philip@freebsd.org>]
4278
4279   *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
4280      to ensure that even with this option, only ciphersuites in the
4281      server's preference list will be accepted.  (Note that the option
4282      applies only when resuming a session, so the earlier behavior was
4283      just about the algorithm choice for symmetric cryptography.)
4284      [Bodo Moeller]
4285
4286  Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
4287
4288   *) Fix NULL pointer dereference if a DTLS server received
4289      ChangeCipherSpec as first record (CVE-2009-1386).
4290      [PR #1679]
4291
4292   *) Fix a state transition in s3_srvr.c and d1_srvr.c
4293      (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
4294      [Nagendra Modadugu]
4295
4296   *) The fix in 0.9.8c that supposedly got rid of unsafe
4297      double-checked locking was incomplete for RSA blinding,
4298      addressing just one layer of what turns out to have been
4299      doubly unsafe triple-checked locking.
4300
4301      So now fix this for real by retiring the MONT_HELPER macro
4302      in crypto/rsa/rsa_eay.c.
4303
4304      [Bodo Moeller; problem pointed out by Marius Schilder]
4305
4306   *) Various precautionary measures:
4307
4308      - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
4309
4310      - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
4311        (NB: This would require knowledge of the secret session ticket key
4312        to exploit, in which case you'd be SOL either way.)
4313
4314      - Change bn_nist.c so that it will properly handle input BIGNUMs
4315        outside the expected range.
4316
4317      - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
4318        builds.
4319
4320      [Neel Mehta, Bodo Moeller]
4321
4322   *) Allow engines to be "soft loaded" - i.e. optionally don't die if
4323      the load fails. Useful for distros.
4324      [Ben Laurie and the FreeBSD team]
4325
4326   *) Add support for Local Machine Keyset attribute in PKCS#12 files.
4327      [Steve Henson]
4328
4329   *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
4330      [Huang Ying]
4331
4332   *) Expand ENGINE to support engine supplied SSL client certificate functions.
4333
4334      This work was sponsored by Logica.
4335      [Steve Henson]
4336
4337   *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
4338      keystores. Support for SSL/TLS client authentication too.
4339      Not compiled unless enable-capieng specified to Configure.
4340
4341      This work was sponsored by Logica.
4342      [Steve Henson]
4343
4344   *) Fix bug in X509_ATTRIBUTE creation: don't set attribute using
4345      ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
4346      attribute creation routines such as certificate requests and PKCS#12
4347      files.
4348      [Steve Henson]
4349
4350  Changes between 0.9.8g and 0.9.8h  [28 May 2008]
4351
4352   *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
4353      handshake which could lead to a cilent crash as found using the
4354      Codenomicon TLS test suite (CVE-2008-1672)
4355      [Steve Henson, Mark Cox]
4356
4357   *) Fix double free in TLS server name extensions which could lead to
4358      a remote crash found by Codenomicon TLS test suite (CVE-2008-0891)
4359      [Joe Orton]
4360
4361   *) Clear error queue in SSL_CTX_use_certificate_chain_file()
4362
4363      Clear the error queue to ensure that error entries left from
4364      older function calls do not interfere with the correct operation.
4365      [Lutz Jaenicke, Erik de Castro Lopo]
4366
4367   *) Remove root CA certificates of commercial CAs:
4368
4369      The OpenSSL project does not recommend any specific CA and does not
4370      have any policy with respect to including or excluding any CA.
4371      Therefore it does not make any sense to ship an arbitrary selection
4372      of root CA certificates with the OpenSSL software.
4373      [Lutz Jaenicke]
4374
4375   *) RSA OAEP patches to fix two separate invalid memory reads.
4376      The first one involves inputs when 'lzero' is greater than
4377      'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
4378      before the beginning of from). The second one involves inputs where
4379      the 'db' section contains nothing but zeroes (there is a one-byte
4380      invalid read after the end of 'db').
4381      [Ivan Nestlerode <inestlerode@us.ibm.com>]
4382
4383   *) Partial backport from 0.9.9-dev:
4384
4385      Introduce bn_mul_mont (dedicated Montgomery multiplication
4386      procedure) as a candidate for BIGNUM assembler implementation.
4387      While 0.9.9-dev uses assembler for various architectures, only
4388      x86_64 is available by default here in the 0.9.8 branch, and
4389      32-bit x86 is available through a compile-time setting.
4390
4391      To try the 32-bit x86 assembler implementation, use Configure
4392      option "enable-montasm" (which exists only for this backport).
4393
4394      As "enable-montasm" for 32-bit x86 disclaims code stability
4395      anyway, in this constellation we activate additional code
4396      backported from 0.9.9-dev for further performance improvements,
4397      namely BN_from_montgomery_word.  (To enable this otherwise,
4398      e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
4399
4400      [Andy Polyakov (backport partially by Bodo Moeller)]
4401
4402   *) Add TLS session ticket callback. This allows an application to set
4403      TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
4404      values. This is useful for key rollover for example where several key
4405      sets may exist with different names.
4406      [Steve Henson]
4407
4408   *) Reverse ENGINE-internal logic for caching default ENGINE handles.
4409      This was broken until now in 0.9.8 releases, such that the only way
4410      a registered ENGINE could be used (assuming it initialises
4411      successfully on the host) was to explicitly set it as the default
4412      for the relevant algorithms. This is in contradiction with 0.9.7
4413      behaviour and the documentation. With this fix, when an ENGINE is
4414      registered into a given algorithm's table of implementations, the
4415      'uptodate' flag is reset so that auto-discovery will be used next
4416      time a new context for that algorithm attempts to select an
4417      implementation.
4418      [Ian Lister (tweaked by Geoff Thorpe)]
4419
4420   *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
4421      implementation in the following ways:
4422
4423      Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
4424      hard coded.
4425
4426      Lack of BER streaming support means one pass streaming processing is
4427      only supported if data is detached: setting the streaming flag is
4428      ignored for embedded content.
4429
4430      CMS support is disabled by default and must be explicitly enabled
4431      with the enable-cms configuration option.
4432      [Steve Henson]
4433
4434   *) Update the GMP engine glue to do direct copies between BIGNUM and
4435      mpz_t when openssl and GMP use the same limb size. Otherwise the
4436      existing "conversion via a text string export" trick is still used.
4437      [Paul Sheer <paulsheer@gmail.com>]
4438
4439   *) Zlib compression BIO. This is a filter BIO which compressed and
4440      uncompresses any data passed through it.
4441      [Steve Henson]
4442
4443   *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
4444      RFC3394 compatible AES key wrapping.
4445      [Steve Henson]
4446
4447   *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
4448      sets string data without copying. X509_ALGOR_set0() and
4449      X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
4450      data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
4451      from an X509_ATTRIBUTE structure optionally checking it occurs only
4452      once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
4453      data.
4454      [Steve Henson]
4455
4456   *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
4457      to get the expected BN_FLG_CONSTTIME behavior.
4458      [Bodo Moeller (Google)]
4459
4460   *) Netware support:
4461
4462      - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
4463      - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
4464      - added some more tests to do_tests.pl
4465      - fixed RunningProcess usage so that it works with newer LIBC NDKs too
4466      - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
4467      - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
4468        netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
4469      - various changes to netware.pl to enable gcc-cross builds on Win32
4470        platform
4471      - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
4472      - various changes to fix missing prototype warnings
4473      - fixed x86nasm.pl to create correct asm files for NASM COFF output
4474      - added AES, WHIRLPOOL and CPUID assembler code to build files
4475      - added missing AES assembler make rules to mk1mf.pl
4476      - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
4477      [Guenter Knauf <eflash@gmx.net>]
4478
4479   *) Implement certificate status request TLS extension defined in RFC3546.
4480      A client can set the appropriate parameters and receive the encoded
4481      OCSP response via a callback. A server can query the supplied parameters
4482      and set the encoded OCSP response in the callback. Add simplified examples
4483      to s_client and s_server.
4484      [Steve Henson]
4485
4486  Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
4487
4488   *) Fix various bugs:
4489      + Binary incompatibility of ssl_ctx_st structure
4490      + DTLS interoperation with non-compliant servers
4491      + Don't call get_session_cb() without proposed session
4492      + Fix ia64 assembler code
4493      [Andy Polyakov, Steve Henson]
4494
4495  Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
4496
4497   *) DTLS Handshake overhaul. There were longstanding issues with
4498      OpenSSL DTLS implementation, which were making it impossible for
4499      RFC 4347 compliant client to communicate with OpenSSL server.
4500      Unfortunately just fixing these incompatibilities would "cut off"
4501      pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
4502      server keeps tolerating non RFC compliant syntax. The opposite is
4503      not true, 0.9.8f client can not communicate with earlier server.
4504      This update even addresses CVE-2007-4995.
4505      [Andy Polyakov]
4506
4507   *) Changes to avoid need for function casts in OpenSSL: some compilers
4508      (gcc 4.2 and later) reject their use.
4509      [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
4510       Steve Henson]
4511
4512   *) Add RFC4507 support to OpenSSL. This includes the corrections in
4513      RFC4507bis. The encrypted ticket format is an encrypted encoded
4514      SSL_SESSION structure, that way new session features are automatically
4515      supported.
4516
4517      If a client application caches session in an SSL_SESSION structure
4518      support is transparent because tickets are now stored in the encoded
4519      SSL_SESSION.
4520
4521      The SSL_CTX structure automatically generates keys for ticket
4522      protection in servers so again support should be possible
4523      with no application modification.
4524
4525      If a client or server wishes to disable RFC4507 support then the option
4526      SSL_OP_NO_TICKET can be set.
4527
4528      Add a TLS extension debugging callback to allow the contents of any client
4529      or server extensions to be examined.
4530
4531      This work was sponsored by Google.
4532      [Steve Henson]
4533
4534   *) Add initial support for TLS extensions, specifically for the server_name
4535      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
4536      have new members for a host name.  The SSL data structure has an
4537      additional member SSL_CTX *initial_ctx so that new sessions can be
4538      stored in that context to allow for session resumption, even after the
4539      SSL has been switched to a new SSL_CTX in reaction to a client's
4540      server_name extension.
4541
4542      New functions (subject to change):
4543
4544          SSL_get_servername()
4545          SSL_get_servername_type()
4546          SSL_set_SSL_CTX()
4547
4548      New CTRL codes and macros (subject to change):
4549
4550          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
4551                                  - SSL_CTX_set_tlsext_servername_callback()
4552          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
4553                                       - SSL_CTX_set_tlsext_servername_arg()
4554          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
4555
4556      openssl s_client has a new '-servername ...' option.
4557
4558      openssl s_server has new options '-servername_host ...', '-cert2 ...',
4559      '-key2 ...', '-servername_fatal' (subject to change).  This allows
4560      testing the HostName extension for a specific single host name ('-cert'
4561      and '-key' remain fallbacks for handshakes without HostName
4562      negotiation).  If the unrecognized_name alert has to be sent, this by
4563      default is a warning; it becomes fatal with the '-servername_fatal'
4564      option.
4565
4566      [Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson]
4567
4568   *) Add AES and SSE2 assembly language support to VC++ build.
4569      [Steve Henson]
4570
4571   *) Mitigate attack on final subtraction in Montgomery reduction.
4572      [Andy Polyakov]
4573
4574   *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
4575      (which previously caused an internal error).
4576      [Bodo Moeller]
4577
4578   *) Squeeze another 10% out of IGE mode when in != out.
4579      [Ben Laurie]
4580
4581   *) AES IGE mode speedup.
4582      [Dean Gaudet (Google)]
4583
4584   *) Add the Korean symmetric 128-bit cipher SEED (see
4585      http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
4586      add SEED ciphersuites from RFC 4162:
4587
4588         TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
4589         TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
4590         TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
4591         TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
4592
4593      To minimize changes between patchlevels in the OpenSSL 0.9.8
4594      series, SEED remains excluded from compilation unless OpenSSL
4595      is configured with 'enable-seed'.
4596      [KISA, Bodo Moeller]
4597
4598   *) Mitigate branch prediction attacks, which can be practical if a
4599      single processor is shared, allowing a spy process to extract
4600      information.  For detailed background information, see
4601      http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
4602      J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
4603      and Necessary Software Countermeasures").  The core of the change
4604      are new versions BN_div_no_branch() and
4605      BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
4606      respectively, which are slower, but avoid the security-relevant
4607      conditional branches.  These are automatically called by BN_div()
4608      and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
4609      of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
4610      remove a conditional branch.
4611
4612      BN_FLG_CONSTTIME is the new name for the previous
4613      BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
4614      modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
4615      in the exponent causes BN_mod_exp_mont() to use the alternative
4616      implementation in BN_mod_exp_mont_consttime().)  The old name
4617      remains as a deprecated alias.
4618
4619      Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
4620      RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
4621      constant-time implementations for more than just exponentiation.
4622      Here too the old name is kept as a deprecated alias.
4623
4624      BN_BLINDING_new() will now use BN_dup() for the modulus so that
4625      the BN_BLINDING structure gets an independent copy of the
4626      modulus.  This means that the previous "BIGNUM *m" argument to
4627      BN_BLINDING_new() and to BN_BLINDING_create_param() now
4628      essentially becomes "const BIGNUM *m", although we can't actually
4629      change this in the header file before 0.9.9.  It allows
4630      RSA_setup_blinding() to use BN_with_flags() on the modulus to
4631      enable BN_FLG_CONSTTIME.
4632
4633      [Matthew D Wood (Intel Corp)]
4634
4635   *) In the SSL/TLS server implementation, be strict about session ID
4636      context matching (which matters if an application uses a single
4637      external cache for different purposes).  Previously,
4638      out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
4639      set.  This did ensure strict client verification, but meant that,
4640      with applications using a single external cache for quite
4641      different requirements, clients could circumvent ciphersuite
4642      restrictions for a given session ID context by starting a session
4643      in a different context.
4644      [Bodo Moeller]
4645
4646   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
4647      a ciphersuite string such as "DEFAULT:RSA" cannot enable
4648      authentication-only ciphersuites.
4649      [Bodo Moeller]
4650
4651   *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
4652      not complete and could lead to a possible single byte overflow
4653      (CVE-2007-5135) [Ben Laurie]
4654
4655  Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
4656
4657   *) Since AES128 and AES256 (and similarly Camellia128 and
4658      Camellia256) share a single mask bit in the logic of
4659      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
4660      kludge to work properly if AES128 is available and AES256 isn't
4661      (or if Camellia128 is available and Camellia256 isn't).
4662      [Victor Duchovni]
4663
4664   *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
4665      (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
4666      When a point or a seed is encoded in a BIT STRING, we need to
4667      prevent the removal of trailing zero bits to get the proper DER
4668      encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
4669      of a NamedBitList, for which trailing 0 bits need to be removed.)
4670      [Bodo Moeller]
4671
4672   *) Have SSL/TLS server implementation tolerate "mismatched" record
4673      protocol version while receiving ClientHello even if the
4674      ClientHello is fragmented.  (The server can't insist on the
4675      particular protocol version it has chosen before the ServerHello
4676      message has informed the client about his choice.)
4677      [Bodo Moeller]
4678
4679   *) Add RFC 3779 support.
4680      [Rob Austein for ARIN, Ben Laurie]
4681
4682   *) Load error codes if they are not already present instead of using a
4683      static variable. This allows them to be cleanly unloaded and reloaded.
4684      Improve header file function name parsing.
4685      [Steve Henson]
4686
4687   *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
4688      or CAPABILITY handshake as required by RFCs.
4689      [Goetz Babin-Ebell]
4690
4691  Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
4692
4693   *) Introduce limits to prevent malicious keys being able to
4694      cause a denial of service.  (CVE-2006-2940)
4695      [Steve Henson, Bodo Moeller]
4696
4697   *) Fix ASN.1 parsing of certain invalid structures that can result
4698      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
4699
4700   *) Fix buffer overflow in SSL_get_shared_ciphers() function.
4701      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
4702
4703   *) Fix SSL client code which could crash if connecting to a
4704      malicious SSLv2 server.  (CVE-2006-4343)
4705      [Tavis Ormandy and Will Drewry, Google Security Team]
4706
4707   *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
4708      match only those.  Before that, "AES256-SHA" would be interpreted
4709      as a pattern and match "AES128-SHA" too (since AES128-SHA got
4710      the same strength classification in 0.9.7h) as we currently only
4711      have a single AES bit in the ciphersuite description bitmap.
4712      That change, however, also applied to ciphersuite strings such as
4713      "RC4-MD5" that intentionally matched multiple ciphersuites --
4714      namely, SSL 2.0 ciphersuites in addition to the more common ones
4715      from SSL 3.0/TLS 1.0.
4716
4717      So we change the selection algorithm again: Naming an explicit
4718      ciphersuite selects this one ciphersuite, and any other similar
4719      ciphersuite (same bitmap) from *other* protocol versions.
4720      Thus, "RC4-MD5" again will properly select both the SSL 2.0
4721      ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
4722
4723      Since SSL 2.0 does not have any ciphersuites for which the
4724      128/256 bit distinction would be relevant, this works for now.
4725      The proper fix will be to use different bits for AES128 and
4726      AES256, which would have avoided the problems from the beginning;
4727      however, bits are scarce, so we can only do this in a new release
4728      (not just a patchlevel) when we can change the SSL_CIPHER
4729      definition to split the single 'unsigned long mask' bitmap into
4730      multiple values to extend the available space.
4731
4732      [Bodo Moeller]
4733
4734  Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
4735
4736   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
4737      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
4738
4739   *) Add AES IGE and biIGE modes.
4740      [Ben Laurie]
4741
4742   *) Change the Unix randomness entropy gathering to use poll() when
4743      possible instead of select(), since the latter has some
4744      undesirable limitations.
4745      [Darryl Miles via Richard Levitte and Bodo Moeller]
4746
4747   *) Disable "ECCdraft" ciphersuites more thoroughly.  Now special
4748      treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
4749      cannot be implicitly activated as part of, e.g., the "AES" alias.
4750      However, please upgrade to OpenSSL 0.9.9[-dev] for
4751      non-experimental use of the ECC ciphersuites to get TLS extension
4752      support, which is required for curve and point format negotiation
4753      to avoid potential handshake problems.
4754      [Bodo Moeller]
4755
4756   *) Disable rogue ciphersuites:
4757
4758       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
4759       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
4760       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
4761
4762      The latter two were purportedly from
4763      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
4764      appear there.
4765
4766      Also deactivate the remaining ciphersuites from
4767      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
4768      unofficial, and the ID has long expired.
4769      [Bodo Moeller]
4770
4771   *) Fix RSA blinding Heisenbug (problems sometimes occurred on
4772      dual-core machines) and other potential thread-safety issues.
4773      [Bodo Moeller]
4774
4775   *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
4776      versions), which is now available for royalty-free use
4777      (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
4778      Also, add Camellia TLS ciphersuites from RFC 4132.
4779
4780      To minimize changes between patchlevels in the OpenSSL 0.9.8
4781      series, Camellia remains excluded from compilation unless OpenSSL
4782      is configured with 'enable-camellia'.
4783      [NTT]
4784
4785   *) Disable the padding bug check when compression is in use. The padding
4786      bug check assumes the first packet is of even length, this is not
4787      necessarily true if compresssion is enabled and can result in false
4788      positives causing handshake failure. The actual bug test is ancient
4789      code so it is hoped that implementations will either have fixed it by
4790      now or any which still have the bug do not support compression.
4791      [Steve Henson]
4792
4793  Changes between 0.9.8a and 0.9.8b  [04 May 2006]
4794
4795   *) When applying a cipher rule check to see if string match is an explicit
4796      cipher suite and only match that one cipher suite if it is.
4797      [Steve Henson]
4798
4799   *) Link in manifests for VC++ if needed.
4800      [Austin Ziegler <halostatue@gmail.com>]
4801
4802   *) Update support for ECC-based TLS ciphersuites according to
4803      draft-ietf-tls-ecc-12.txt with proposed changes (but without
4804      TLS extensions, which are supported starting with the 0.9.9
4805      branch, not in the OpenSSL 0.9.8 branch).
4806      [Douglas Stebila]
4807
4808   *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
4809      opaque EVP_CIPHER_CTX handling.
4810      [Steve Henson]
4811
4812   *) Fixes and enhancements to zlib compression code. We now only use
4813      "zlib1.dll" and use the default __cdecl calling convention on Win32
4814      to conform with the standards mentioned here:
4815            http://www.zlib.net/DLL_FAQ.txt
4816      Static zlib linking now works on Windows and the new --with-zlib-include
4817      --with-zlib-lib options to Configure can be used to supply the location
4818      of the headers and library. Gracefully handle case where zlib library
4819      can't be loaded.
4820      [Steve Henson]
4821
4822   *) Several fixes and enhancements to the OID generation code. The old code
4823      sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
4824      handle numbers larger than ULONG_MAX, truncated printing and had a
4825      non standard OBJ_obj2txt() behaviour.
4826      [Steve Henson]
4827
4828   *) Add support for building of engines under engine/ as shared libraries
4829      under VC++ build system.
4830      [Steve Henson]
4831
4832   *) Corrected the numerous bugs in the Win32 path splitter in DSO.
4833      Hopefully, we will not see any false combination of paths any more.
4834      [Richard Levitte]
4835
4836  Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
4837
4838   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
4839      (part of SSL_OP_ALL).  This option used to disable the
4840      countermeasure against man-in-the-middle protocol-version
4841      rollback in the SSL 2.0 server implementation, which is a bad
4842      idea.  (CVE-2005-2969)
4843
4844      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
4845      for Information Security, National Institute of Advanced Industrial
4846      Science and Technology [AIST], Japan)]
4847
4848   *) Add two function to clear and return the verify parameter flags.
4849      [Steve Henson]
4850
4851   *) Keep cipherlists sorted in the source instead of sorting them at
4852      runtime, thus removing the need for a lock.
4853      [Nils Larsch]
4854
4855   *) Avoid some small subgroup attacks in Diffie-Hellman.
4856      [Nick Mathewson and Ben Laurie]
4857
4858   *) Add functions for well-known primes.
4859      [Nick Mathewson]
4860
4861   *) Extended Windows CE support.
4862      [Satoshi Nakamura and Andy Polyakov]
4863
4864   *) Initialize SSL_METHOD structures at compile time instead of during
4865      runtime, thus removing the need for a lock.
4866      [Steve Henson]
4867
4868   *) Make PKCS7_decrypt() work even if no certificate is supplied by
4869      attempting to decrypt each encrypted key in turn. Add support to
4870      smime utility.
4871      [Steve Henson]
4872
4873  Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
4874
4875   [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
4876   OpenSSL 0.9.8.]
4877
4878   *) Add libcrypto.pc and libssl.pc for those who feel they need them.
4879      [Richard Levitte]
4880
4881   *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
4882      key into the same file any more.
4883      [Richard Levitte]
4884
4885   *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
4886      [Andy Polyakov]
4887
4888   *) Add -utf8 command line and config file option to 'ca'.
4889      [Stefan <stf@udoma.org]
4890
4891   *) Removed the macro des_crypt(), as it seems to conflict with some
4892      libraries.  Use DES_crypt().
4893      [Richard Levitte]
4894
4895   *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
4896      involves renaming the source and generated shared-libs for
4897      both. The engines will accept the corrected or legacy ids
4898      ('ncipher' and '4758_cca' respectively) when binding. NB,
4899      this only applies when building 'shared'.
4900      [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe]
4901
4902   *) Add attribute functions to EVP_PKEY structure. Modify
4903      PKCS12_create() to recognize a CSP name attribute and
4904      use it. Make -CSP option work again in pkcs12 utility.
4905      [Steve Henson]
4906
4907   *) Add new functionality to the bn blinding code:
4908      - automatic re-creation of the BN_BLINDING parameters after
4909        a fixed number of uses (currently 32)
4910      - add new function for parameter creation
4911      - introduce flags to control the update behaviour of the
4912        BN_BLINDING parameters
4913      - hide BN_BLINDING structure
4914      Add a second BN_BLINDING slot to the RSA structure to improve
4915      performance when a single RSA object is shared among several
4916      threads.
4917      [Nils Larsch]
4918
4919   *) Add support for DTLS.
4920      [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie]
4921
4922   *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
4923      to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
4924      [Walter Goulet]
4925
4926   *) Remove buggy and incomplete DH cert support from
4927      ssl/ssl_rsa.c and ssl/s3_both.c
4928      [Nils Larsch]
4929
4930   *) Use SHA-1 instead of MD5 as the default digest algorithm for
4931      the apps/openssl applications.
4932      [Nils Larsch]
4933
4934   *) Compile clean with "-Wall -Wmissing-prototypes
4935      -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
4936      DEBUG_SAFESTACK must also be set.
4937      [Ben Laurie]
4938
4939   *) Change ./Configure so that certain algorithms can be disabled by default.
4940      The new counterpiece to "no-xxx" is "enable-xxx".
4941
4942      The patented RC5 and MDC2 algorithms will now be disabled unless
4943      "enable-rc5" and "enable-mdc2", respectively, are specified.
4944
4945      (IDEA remains enabled despite being patented.  This is because IDEA
4946      is frequently required for interoperability, and there is no license
4947      fee for non-commercial use.  As before, "no-idea" can be used to
4948      avoid this algorithm.)
4949
4950      [Bodo Moeller]
4951
4952   *) Add processing of proxy certificates (see RFC 3820).  This work was
4953      sponsored by KTH (The Royal Institute of Technology in Stockholm) and
4954      EGEE (Enabling Grids for E-science in Europe).
4955      [Richard Levitte]
4956
4957   *) RC4 performance overhaul on modern architectures/implementations, such
4958      as Intel P4, IA-64 and AMD64.
4959      [Andy Polyakov]
4960
4961   *) New utility extract-section.pl. This can be used specify an alternative
4962      section number in a pod file instead of having to treat each file as
4963      a separate case in Makefile. This can be done by adding two lines to the
4964      pod file:
4965
4966      =for comment openssl_section:XXX
4967
4968      The blank line is mandatory.
4969
4970      [Steve Henson]
4971
4972   *) New arguments -certform, -keyform and -pass for s_client and s_server
4973      to allow alternative format key and certificate files and passphrase
4974      sources.
4975      [Steve Henson]
4976
4977   *) New structure X509_VERIFY_PARAM which combines current verify parameters,
4978      update associated structures and add various utility functions.
4979
4980      Add new policy related verify parameters, include policy checking in
4981      standard verify code. Enhance 'smime' application with extra parameters
4982      to support policy checking and print out.
4983      [Steve Henson]
4984
4985   *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
4986      Nehemiah processors. These extensions support AES encryption in hardware
4987      as well as RNG (though RNG support is currently disabled).
4988      [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
4989
4990   *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
4991      [Geoff Thorpe]
4992
4993   *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
4994      [Andy Polyakov and a number of other people]
4995
4996   *) Improved PowerPC platform support. Most notably BIGNUM assembler
4997      implementation contributed by IBM.
4998      [Suresh Chari, Peter Waltenberg, Andy Polyakov]
4999
5000   *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
5001      exponent rather than 'unsigned long'. There is a corresponding change to
5002      the new 'rsa_keygen' element of the RSA_METHOD structure.
5003      [Jelte Jansen, Geoff Thorpe]
5004
5005   *) Functionality for creating the initial serial number file is now
5006      moved from CA.pl to the 'ca' utility with a new option -create_serial.
5007
5008      (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
5009      number file to 1, which is bound to cause problems.  To avoid
5010      the problems while respecting compatibility between different 0.9.7
5011      patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
5012      CA.pl for serial number initialization.  With the new release 0.9.8,
5013      we can fix the problem directly in the 'ca' utility.)
5014      [Steve Henson]
5015
5016   *) Reduced header interdepencies by declaring more opaque objects in
5017      ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
5018      give fewer recursive includes, which could break lazy source code - so
5019      this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
5020      developers should define this symbol when building and using openssl to
5021      ensure they track the recommended behaviour, interfaces, [etc], but
5022      backwards-compatible behaviour prevails when this isn't defined.
5023      [Geoff Thorpe]
5024
5025   *) New function X509_POLICY_NODE_print() which prints out policy nodes.
5026      [Steve Henson]
5027
5028   *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
5029      This will generate a random key of the appropriate length based on the
5030      cipher context. The EVP_CIPHER can provide its own random key generation
5031      routine to support keys of a specific form. This is used in the des and
5032      3des routines to generate a key of the correct parity. Update S/MIME
5033      code to use new functions and hence generate correct parity DES keys.
5034      Add EVP_CHECK_DES_KEY #define to return an error if the key is not
5035      valid (weak or incorrect parity).
5036      [Steve Henson]
5037
5038   *) Add a local set of CRLs that can be used by X509_verify_cert() as well
5039      as looking them up. This is useful when the verified structure may contain
5040      CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
5041      present unless the new PKCS7_NO_CRL flag is asserted.
5042      [Steve Henson]
5043
5044   *) Extend ASN1 oid configuration module. It now additionally accepts the
5045      syntax:
5046
5047      shortName = some long name, 1.2.3.4
5048      [Steve Henson]
5049
5050   *) Reimplemented the BN_CTX implementation. There is now no more static
5051      limitation on the number of variables it can handle nor the depth of the
5052      "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
5053      information can now expand as required, and rather than having a single
5054      static array of bignums, BN_CTX now uses a linked-list of such arrays
5055      allowing it to expand on demand whilst maintaining the usefulness of
5056      BN_CTX's "bundling".
5057      [Geoff Thorpe]
5058
5059   *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
5060      to allow all RSA operations to function using a single BN_CTX.
5061      [Geoff Thorpe]
5062
5063   *) Preliminary support for certificate policy evaluation and checking. This
5064      is initially intended to pass the tests outlined in "Conformance Testing
5065      of Relying Party Client Certificate Path Processing Logic" v1.07.
5066      [Steve Henson]
5067
5068   *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
5069      remained unused and not that useful. A variety of other little bignum
5070      tweaks and fixes have also been made continuing on from the audit (see
5071      below).
5072      [Geoff Thorpe]
5073
5074   *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
5075      associated ASN1, EVP and SSL functions and old ASN1 macros.
5076      [Richard Levitte]
5077
5078   *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
5079      and this should never fail. So the return value from the use of
5080      BN_set_word() (which can fail due to needless expansion) is now deprecated;
5081      if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
5082      [Geoff Thorpe]
5083
5084   *) BN_CTX_get() should return zero-valued bignums, providing the same
5085      initialised value as BN_new().
5086      [Geoff Thorpe, suggested by Ulf Möller]
5087
5088   *) Support for inhibitAnyPolicy certificate extension.
5089      [Steve Henson]
5090
5091   *) An audit of the BIGNUM code is underway, for which debugging code is
5092      enabled when BN_DEBUG is defined. This makes stricter enforcements on what
5093      is considered valid when processing BIGNUMs, and causes execution to
5094      assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
5095      further steps are taken to deliberately pollute unused data in BIGNUM
5096      structures to try and expose faulty code further on. For now, openssl will
5097      (in its default mode of operation) continue to tolerate the inconsistent
5098      forms that it has tolerated in the past, but authors and packagers should
5099      consider trying openssl and their own applications when compiled with
5100      these debugging symbols defined. It will help highlight potential bugs in
5101      their own code, and will improve the test coverage for OpenSSL itself. At
5102      some point, these tighter rules will become openssl's default to improve
5103      maintainability, though the assert()s and other overheads will remain only
5104      in debugging configurations. See bn.h for more details.
5105      [Geoff Thorpe, Nils Larsch, Ulf Möller]
5106
5107   *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
5108      that can only be obtained through BN_CTX_new() (which implicitly
5109      initialises it). The presence of this function only made it possible
5110      to overwrite an existing structure (and cause memory leaks).
5111      [Geoff Thorpe]
5112
5113   *) Because of the callback-based approach for implementing LHASH as a
5114      template type, lh_insert() adds opaque objects to hash-tables and
5115      lh_doall() or lh_doall_arg() are typically used with a destructor callback
5116      to clean up those corresponding objects before destroying the hash table
5117      (and losing the object pointers). So some over-zealous constifications in
5118      LHASH have been relaxed so that lh_insert() does not take (nor store) the
5119      objects as "const" and the lh_doall[_arg] callback wrappers are not
5120      prototyped to have "const" restrictions on the object pointers they are
5121      given (and so aren't required to cast them away any more).
5122      [Geoff Thorpe]
5123
5124   *) The tmdiff.h API was so ugly and minimal that our own timing utility
5125      (speed) prefers to use its own implementation. The two implementations
5126      haven't been consolidated as yet (volunteers?) but the tmdiff API has had
5127      its object type properly exposed (MS_TM) instead of casting to/from "char
5128      *". This may still change yet if someone realises MS_TM and "ms_time_***"
5129      aren't necessarily the greatest nomenclatures - but this is what was used
5130      internally to the implementation so I've used that for now.
5131      [Geoff Thorpe]
5132
5133   *) Ensure that deprecated functions do not get compiled when
5134      OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
5135      the self-tests were still using deprecated key-generation functions so
5136      these have been updated also.
5137      [Geoff Thorpe]
5138
5139   *) Reorganise PKCS#7 code to separate the digest location functionality
5140      into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest().
5141      New function PKCS7_set_digest() to set the digest type for PKCS#7
5142      digestedData type. Add additional code to correctly generate the
5143      digestedData type and add support for this type in PKCS7 initialization
5144      functions.
5145      [Steve Henson]
5146
5147   *) New function PKCS7_set0_type_other() this initializes a PKCS7
5148      structure of type "other".
5149      [Steve Henson]
5150
5151   *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
5152      sure the loop does correctly stop and breaking ("division by zero")
5153      modulus operations are not performed. The (pre-generated) prime
5154      table crypto/bn/bn_prime.h was already correct, but it could not be
5155      re-generated on some platforms because of the "division by zero"
5156      situation in the script.
5157      [Ralf S. Engelschall]
5158
5159   *) Update support for ECC-based TLS ciphersuites according to
5160      draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
5161      SHA-1 now is only used for "small" curves (where the
5162      representation of a field element takes up to 24 bytes); for
5163      larger curves, the field element resulting from ECDH is directly
5164      used as premaster secret.
5165      [Douglas Stebila (Sun Microsystems Laboratories)]
5166
5167   *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
5168      curve secp160r1 to the tests.
5169      [Douglas Stebila (Sun Microsystems Laboratories)]
5170
5171   *) Add the possibility to load symbols globally with DSO.
5172      [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
5173
5174   *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
5175      control of the error stack.
5176      [Richard Levitte]
5177
5178   *) Add support for STORE in ENGINE.
5179      [Richard Levitte]
5180
5181   *) Add the STORE type.  The intention is to provide a common interface
5182      to certificate and key stores, be they simple file-based stores, or
5183      HSM-type store, or LDAP stores, or...
5184      NOTE: The code is currently UNTESTED and isn't really used anywhere.
5185      [Richard Levitte]
5186
5187   *) Add a generic structure called OPENSSL_ITEM.  This can be used to
5188      pass a list of arguments to any function as well as provide a way
5189      for a function to pass data back to the caller.
5190      [Richard Levitte]
5191
5192   *) Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
5193      works like BUF_strdup() but can be used to duplicate a portion of
5194      a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
5195      a memory area.
5196      [Richard Levitte]
5197
5198   *) Add the function sk_find_ex() which works like sk_find(), but will
5199      return an index to an element even if an exact match couldn't be
5200      found.  The index is guaranteed to point at the element where the
5201      searched-for key would be inserted to preserve sorting order.
5202      [Richard Levitte]
5203
5204   *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
5205      takes an extra flags argument for optional functionality.  Currently,
5206      the following flags are defined:
5207
5208         OBJ_BSEARCH_VALUE_ON_NOMATCH
5209         This one gets OBJ_bsearch_ex() to return a pointer to the first
5210         element where the comparing function returns a negative or zero
5211         number.
5212
5213         OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
5214         This one gets OBJ_bsearch_ex() to return a pointer to the first
5215         element where the comparing function returns zero.  This is useful
5216         if there are more than one element where the comparing function
5217         returns zero.
5218      [Richard Levitte]
5219
5220   *) Make it possible to create self-signed certificates with 'openssl ca'
5221      in such a way that the self-signed certificate becomes part of the
5222      CA database and uses the same mechanisms for serial number generation
5223      as all other certificate signing.  The new flag '-selfsign' enables
5224      this functionality.  Adapt CA.sh and CA.pl.in.
5225      [Richard Levitte]
5226
5227   *) Add functionality to check the public key of a certificate request
5228      against a given private.  This is useful to check that a certificate
5229      request can be signed by that key (self-signing).
5230      [Richard Levitte]
5231
5232   *) Make it possible to have multiple active certificates with the same
5233      subject in the CA index file.  This is done only if the keyword
5234      'unique_subject' is set to 'no' in the main CA section (default
5235      if 'CA_default') of the configuration file.  The value is saved
5236      with the database itself in a separate index attribute file,
5237      named like the index file with '.attr' appended to the name.
5238      [Richard Levitte]
5239
5240   *) Generate muti valued AVAs using '+' notation in config files for
5241      req and dirName.
5242      [Steve Henson]
5243
5244   *) Support for nameConstraints certificate extension.
5245      [Steve Henson]
5246
5247   *) Support for policyConstraints certificate extension.
5248      [Steve Henson]
5249
5250   *) Support for policyMappings certificate extension.
5251      [Steve Henson]
5252
5253   *) Make sure the default DSA_METHOD implementation only uses its
5254      dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
5255      and change its own handlers to be NULL so as to remove unnecessary
5256      indirection. This lets alternative implementations fallback to the
5257      default implementation more easily.
5258      [Geoff Thorpe]
5259
5260   *) Support for directoryName in GeneralName related extensions
5261      in config files.
5262      [Steve Henson]
5263
5264   *) Make it possible to link applications using Makefile.shared.
5265      Make that possible even when linking against static libraries!
5266      [Richard Levitte]
5267
5268   *) Support for single pass processing for S/MIME signing. This now
5269      means that S/MIME signing can be done from a pipe, in addition
5270      cleartext signing (multipart/signed type) is effectively streaming
5271      and the signed data does not need to be all held in memory.
5272
5273      This is done with a new flag PKCS7_STREAM. When this flag is set
5274      PKCS7_sign() only initializes the PKCS7 structure and the actual signing
5275      is done after the data is output (and digests calculated) in
5276      SMIME_write_PKCS7().
5277      [Steve Henson]
5278
5279   *) Add full support for -rpath/-R, both in shared libraries and
5280      applications, at least on the platforms where it's known how
5281      to do it.
5282      [Richard Levitte]
5283
5284   *) In crypto/ec/ec_mult.c, implement fast point multiplication with
5285      precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
5286      will now compute a table of multiples of the generator that
5287      makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
5288      faster (notably in the case of a single point multiplication,
5289      scalar * generator).
5290      [Nils Larsch, Bodo Moeller]
5291
5292   *) IPv6 support for certificate extensions. The various extensions
5293      which use the IP:a.b.c.d can now take IPv6 addresses using the
5294      formats of RFC1884 2.2 . IPv6 addresses are now also displayed
5295      correctly.
5296      [Steve Henson]
5297
5298   *) Added an ENGINE that implements RSA by performing private key
5299      exponentiations with the GMP library. The conversions to and from
5300      GMP's mpz_t format aren't optimised nor are any montgomery forms
5301      cached, and on x86 it appears OpenSSL's own performance has caught up.
5302      However there are likely to be other architectures where GMP could
5303      provide a boost. This ENGINE is not built in by default, but it can be
5304      specified at Configure time and should be accompanied by the necessary
5305      linker additions, eg;
5306          ./config -DOPENSSL_USE_GMP -lgmp
5307      [Geoff Thorpe]
5308
5309   *) "openssl engine" will not display ENGINE/DSO load failure errors when
5310      testing availability of engines with "-t" - the old behaviour is
5311      produced by increasing the feature's verbosity with "-tt".
5312      [Geoff Thorpe]
5313
5314   *) ECDSA routines: under certain error conditions uninitialized BN objects
5315      could be freed. Solution: make sure initialization is performed early
5316      enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
5317      via PR#459)
5318      [Lutz Jaenicke]
5319
5320   *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
5321      and DH_METHOD (eg. by ENGINE implementations) to override the normal
5322      software implementations. For DSA and DH, parameter generation can
5323      also be overridden by providing the appropriate method callbacks.
5324      [Geoff Thorpe]
5325
5326   *) Change the "progress" mechanism used in key-generation and
5327      primality testing to functions that take a new BN_GENCB pointer in
5328      place of callback/argument pairs. The new API functions have "_ex"
5329      postfixes and the older functions are reimplemented as wrappers for
5330      the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
5331      declarations of the old functions to help (graceful) attempts to
5332      migrate to the new functions. Also, the new key-generation API
5333      functions operate on a caller-supplied key-structure and return
5334      success/failure rather than returning a key or NULL - this is to
5335      help make "keygen" another member function of RSA_METHOD etc.
5336
5337      Example for using the new callback interface:
5338
5339           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
5340           void *my_arg = ...;
5341           BN_GENCB my_cb;
5342
5343           BN_GENCB_set(&my_cb, my_callback, my_arg);
5344
5345           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
5346           /* For the meaning of a, b in calls to my_callback(), see the
5347            * documentation of the function that calls the callback.
5348            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
5349            * my_callback should return 1 if it wants BN_is_prime_ex()
5350            * to continue, or 0 to stop.
5351            */
5352
5353      [Geoff Thorpe]
5354
5355   *) Change the ZLIB compression method to be stateful, and make it
5356      available to TLS with the number defined in
5357      draft-ietf-tls-compression-04.txt.
5358      [Richard Levitte]
5359
5360   *) Add the ASN.1 structures and functions for CertificatePair, which
5361      is defined as follows (according to X.509_4thEditionDraftV6.pdf):
5362
5363      CertificatePair ::= SEQUENCE {
5364         forward         [0]     Certificate OPTIONAL,
5365         reverse         [1]     Certificate OPTIONAL,
5366         -- at least one of the pair shall be present -- }
5367
5368      Also implement the PEM functions to read and write certificate
5369      pairs, and defined the PEM tag as "CERTIFICATE PAIR".
5370
5371      This needed to be defined, mostly for the sake of the LDAP
5372      attribute crossCertificatePair, but may prove useful elsewhere as
5373      well.
5374      [Richard Levitte]
5375
5376   *) Make it possible to inhibit symlinking of shared libraries in
5377      Makefile.shared, for Cygwin's sake.
5378      [Richard Levitte]
5379
5380   *) Extend the BIGNUM API by creating a function
5381           void BN_set_negative(BIGNUM *a, int neg);
5382      and a macro that behave like
5383           int  BN_is_negative(const BIGNUM *a);
5384
5385      to avoid the need to access 'a->neg' directly in applications.
5386      [Nils Larsch]
5387
5388   *) Implement fast modular reduction for pseudo-Mersenne primes
5389      used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
5390      EC_GROUP_new_curve_GFp() will now automatically use this
5391      if applicable.
5392      [Nils Larsch <nla@trustcenter.de>]
5393
5394   *) Add new lock type (CRYPTO_LOCK_BN).
5395      [Bodo Moeller]
5396
5397   *) Change the ENGINE framework to automatically load engines
5398      dynamically from specific directories unless they could be
5399      found to already be built in or loaded.  Move all the
5400      current engines except for the cryptodev one to a new
5401      directory engines/.
5402      The engines in engines/ are built as shared libraries if
5403      the "shared" options was given to ./Configure or ./config.
5404      Otherwise, they are inserted in libcrypto.a.
5405      /usr/local/ssl/engines is the default directory for dynamic
5406      engines, but that can be overridden at configure time through
5407      the usual use of --prefix and/or --openssldir, and at run
5408      time with the environment variable OPENSSL_ENGINES.
5409      [Geoff Thorpe and Richard Levitte]
5410
5411   *) Add Makefile.shared, a helper makefile to build shared
5412      libraries.  Adapt Makefile.org.
5413      [Richard Levitte]
5414
5415   *) Add version info to Win32 DLLs.
5416      [Peter 'Luna' Runestig" <peter@runestig.com>]
5417
5418   *) Add new 'medium level' PKCS#12 API. Certificates and keys
5419      can be added using this API to created arbitrary PKCS#12
5420      files while avoiding the low level API.
5421
5422      New options to PKCS12_create(), key or cert can be NULL and
5423      will then be omitted from the output file. The encryption
5424      algorithm NIDs can be set to -1 for no encryption, the mac
5425      iteration count can be set to 0 to omit the mac.
5426
5427      Enhance pkcs12 utility by making the -nokeys and -nocerts
5428      options work when creating a PKCS#12 file. New option -nomac
5429      to omit the mac, NONE can be set for an encryption algorithm.
5430      New code is modified to use the enhanced PKCS12_create()
5431      instead of the low level API.
5432      [Steve Henson]
5433
5434   *) Extend ASN1 encoder to support indefinite length constructed
5435      encoding. This can output sequences tags and octet strings in
5436      this form. Modify pk7_asn1.c to support indefinite length
5437      encoding. This is experimental and needs additional code to
5438      be useful, such as an ASN1 bio and some enhanced streaming
5439      PKCS#7 code.
5440
5441      Extend template encode functionality so that tagging is passed
5442      down to the template encoder.
5443      [Steve Henson]
5444
5445   *) Let 'openssl req' fail if an argument to '-newkey' is not
5446      recognized instead of using RSA as a default.
5447      [Bodo Moeller]
5448
5449   *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
5450      As these are not official, they are not included in "ALL";
5451      the "ECCdraft" ciphersuite group alias can be used to select them.
5452      [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
5453
5454   *) Add ECDH engine support.
5455      [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
5456
5457   *) Add ECDH in new directory crypto/ecdh/.
5458      [Douglas Stebila (Sun Microsystems Laboratories)]
5459
5460   *) Let BN_rand_range() abort with an error after 100 iterations
5461      without success (which indicates a broken PRNG).
5462      [Bodo Moeller]
5463
5464   *) Change BN_mod_sqrt() so that it verifies that the input value
5465      is really the square of the return value.  (Previously,
5466      BN_mod_sqrt would show GIGO behaviour.)
5467      [Bodo Moeller]
5468
5469   *) Add named elliptic curves over binary fields from X9.62, SECG,
5470      and WAP/WTLS; add OIDs that were still missing.
5471
5472      [Sheueling Chang Shantz and Douglas Stebila
5473      (Sun Microsystems Laboratories)]
5474
5475   *) Extend the EC library for elliptic curves over binary fields
5476      (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
5477      New EC_METHOD:
5478
5479           EC_GF2m_simple_method
5480
5481      New API functions:
5482
5483           EC_GROUP_new_curve_GF2m
5484           EC_GROUP_set_curve_GF2m
5485           EC_GROUP_get_curve_GF2m
5486           EC_POINT_set_affine_coordinates_GF2m
5487           EC_POINT_get_affine_coordinates_GF2m
5488           EC_POINT_set_compressed_coordinates_GF2m
5489
5490      Point compression for binary fields is disabled by default for
5491      patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
5492      enable it).
5493
5494      As binary polynomials are represented as BIGNUMs, various members
5495      of the EC_GROUP and EC_POINT data structures can be shared
5496      between the implementations for prime fields and binary fields;
5497      the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
5498      are essentially identical to their ..._GFp counterparts.
5499      (For simplicity, the '..._GFp' prefix has been dropped from
5500      various internal method names.)
5501
5502      An internal 'field_div' method (similar to 'field_mul' and
5503      'field_sqr') has been added; this is used only for binary fields.
5504
5505      [Sheueling Chang Shantz and Douglas Stebila
5506      (Sun Microsystems Laboratories)]
5507
5508   *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
5509      through methods ('mul', 'precompute_mult').
5510
5511      The generic implementations (now internally called 'ec_wNAF_mul'
5512      and 'ec_wNAF_precomputed_mult') remain the default if these
5513      methods are undefined.
5514
5515      [Sheueling Chang Shantz and Douglas Stebila
5516      (Sun Microsystems Laboratories)]
5517
5518   *) New function EC_GROUP_get_degree, which is defined through
5519      EC_METHOD.  For curves over prime fields, this returns the bit
5520      length of the modulus.
5521
5522      [Sheueling Chang Shantz and Douglas Stebila
5523      (Sun Microsystems Laboratories)]
5524
5525   *) New functions EC_GROUP_dup, EC_POINT_dup.
5526      (These simply call ..._new  and ..._copy).
5527
5528      [Sheueling Chang Shantz and Douglas Stebila
5529      (Sun Microsystems Laboratories)]
5530
5531   *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
5532      Polynomials are represented as BIGNUMs (where the sign bit is not
5533      used) in the following functions [macros]:
5534
5535           BN_GF2m_add
5536           BN_GF2m_sub             [= BN_GF2m_add]
5537           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
5538           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
5539           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
5540           BN_GF2m_mod_inv
5541           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
5542           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
5543           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
5544           BN_GF2m_cmp             [= BN_ucmp]
5545
5546      (Note that only the 'mod' functions are actually for fields GF(2^m).
5547      BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
5548
5549      For some functions, an the irreducible polynomial defining a
5550      field can be given as an 'unsigned int[]' with strictly
5551      decreasing elements giving the indices of those bits that are set;
5552      i.e., p[] represents the polynomial
5553           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
5554      where
5555           p[0] > p[1] > ... > p[k] = 0.
5556      This applies to the following functions:
5557
5558           BN_GF2m_mod_arr
5559           BN_GF2m_mod_mul_arr
5560           BN_GF2m_mod_sqr_arr
5561           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
5562           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
5563           BN_GF2m_mod_exp_arr
5564           BN_GF2m_mod_sqrt_arr
5565           BN_GF2m_mod_solve_quad_arr
5566           BN_GF2m_poly2arr
5567           BN_GF2m_arr2poly
5568
5569      Conversion can be performed by the following functions:
5570
5571           BN_GF2m_poly2arr
5572           BN_GF2m_arr2poly
5573
5574      bntest.c has additional tests for binary polynomial arithmetic.
5575
5576      Two implementations for BN_GF2m_mod_div() are available.
5577      The default algorithm simply uses BN_GF2m_mod_inv() and
5578      BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
5579      if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
5580      copyright notice in crypto/bn/bn_gf2m.c before enabling it).
5581
5582      [Sheueling Chang Shantz and Douglas Stebila
5583      (Sun Microsystems Laboratories)]
5584
5585   *) Add new error code 'ERR_R_DISABLED' that can be used when some
5586      functionality is disabled at compile-time.
5587      [Douglas Stebila <douglas.stebila@sun.com>]
5588
5589   *) Change default behaviour of 'openssl asn1parse' so that more
5590      information is visible when viewing, e.g., a certificate:
5591
5592      Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
5593      mode the content of non-printable OCTET STRINGs is output in a
5594      style similar to INTEGERs, but with '[HEX DUMP]' prepended to
5595      avoid the appearance of a printable string.
5596      [Nils Larsch <nla@trustcenter.de>]
5597
5598   *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
5599      functions
5600           EC_GROUP_set_asn1_flag()
5601           EC_GROUP_get_asn1_flag()
5602           EC_GROUP_set_point_conversion_form()
5603           EC_GROUP_get_point_conversion_form()
5604      These control ASN1 encoding details:
5605      - Curves (i.e., groups) are encoded explicitly unless asn1_flag
5606        has been set to OPENSSL_EC_NAMED_CURVE.
5607      - Points are encoded in uncompressed form by default; options for
5608        asn1_for are as for point2oct, namely
5609           POINT_CONVERSION_COMPRESSED
5610           POINT_CONVERSION_UNCOMPRESSED
5611           POINT_CONVERSION_HYBRID
5612
5613      Also add 'seed' and 'seed_len' members to EC_GROUP with access
5614      functions
5615           EC_GROUP_set_seed()
5616           EC_GROUP_get0_seed()
5617           EC_GROUP_get_seed_len()
5618      This is used only for ASN1 purposes (so far).
5619      [Nils Larsch <nla@trustcenter.de>]
5620
5621   *) Add 'field_type' member to EC_METHOD, which holds the NID
5622      of the appropriate field type OID.  The new function
5623      EC_METHOD_get_field_type() returns this value.
5624      [Nils Larsch <nla@trustcenter.de>]
5625
5626   *) Add functions
5627           EC_POINT_point2bn()
5628           EC_POINT_bn2point()
5629           EC_POINT_point2hex()
5630           EC_POINT_hex2point()
5631      providing useful interfaces to EC_POINT_point2oct() and
5632      EC_POINT_oct2point().
5633      [Nils Larsch <nla@trustcenter.de>]
5634
5635   *) Change internals of the EC library so that the functions
5636           EC_GROUP_set_generator()
5637           EC_GROUP_get_generator()
5638           EC_GROUP_get_order()
5639           EC_GROUP_get_cofactor()
5640      are implemented directly in crypto/ec/ec_lib.c and not dispatched
5641      to methods, which would lead to unnecessary code duplication when
5642      adding different types of curves.
5643      [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
5644
5645   *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
5646      arithmetic, and such that modified wNAFs are generated
5647      (which avoid length expansion in many cases).
5648      [Bodo Moeller]
5649
5650   *) Add a function EC_GROUP_check_discriminant() (defined via
5651      EC_METHOD) that verifies that the curve discriminant is non-zero.
5652
5653      Add a function EC_GROUP_check() that makes some sanity tests
5654      on a EC_GROUP, its generator and order.  This includes
5655      EC_GROUP_check_discriminant().
5656      [Nils Larsch <nla@trustcenter.de>]
5657
5658   *) Add ECDSA in new directory crypto/ecdsa/.
5659
5660      Add applications 'openssl ecparam' and 'openssl ecdsa'
5661      (these are based on 'openssl dsaparam' and 'openssl dsa').
5662
5663      ECDSA support is also included in various other files across the
5664      library.  Most notably,
5665      - 'openssl req' now has a '-newkey ecdsa:file' option;
5666      - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
5667      - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
5668        d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
5669        them suitable for ECDSA where domain parameters must be
5670        extracted before the specific public key;
5671      - ECDSA engine support has been added.
5672      [Nils Larsch <nla@trustcenter.de>]
5673
5674   *) Include some named elliptic curves, and add OIDs from X9.62,
5675      SECG, and WAP/WTLS.  Each curve can be obtained from the new
5676      function
5677           EC_GROUP_new_by_curve_name(),
5678      and the list of available named curves can be obtained with
5679           EC_get_builtin_curves().
5680      Also add a 'curve_name' member to EC_GROUP objects, which can be
5681      accessed via
5682          EC_GROUP_set_curve_name()
5683          EC_GROUP_get_curve_name()
5684      [Nils Larsch <larsch@trustcenter.de, Bodo Moeller]
5685
5686   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
5687      was actually never needed) and in BN_mul().  The removal in BN_mul()
5688      required a small change in bn_mul_part_recursive() and the addition
5689      of the functions bn_cmp_part_words(), bn_sub_part_words() and
5690      bn_add_part_words(), which do the same thing as bn_cmp_words(),
5691      bn_sub_words() and bn_add_words() except they take arrays with
5692      differing sizes.
5693      [Richard Levitte]
5694
5695  Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
5696
5697   *) Cleanse PEM buffers before freeing them since they may contain
5698      sensitive data.
5699      [Benjamin Bennett <ben@psc.edu>]
5700
5701   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
5702      a ciphersuite string such as "DEFAULT:RSA" cannot enable
5703      authentication-only ciphersuites.
5704      [Bodo Moeller]
5705
5706   *) Since AES128 and AES256 share a single mask bit in the logic of
5707      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
5708      kludge to work properly if AES128 is available and AES256 isn't.
5709      [Victor Duchovni]
5710
5711   *) Expand security boundary to match 1.1.1 module.
5712      [Steve Henson]
5713
5714   *) Remove redundant features: hash file source, editing of test vectors
5715      modify fipsld to use external fips_premain.c signature.
5716      [Steve Henson]
5717
5718   *) New perl script mkfipsscr.pl to create shell scripts or batch files to
5719      run algorithm test programs.
5720      [Steve Henson]
5721
5722   *) Make algorithm test programs more tolerant of whitespace.
5723      [Steve Henson]
5724
5725   *) Have SSL/TLS server implementation tolerate "mismatched" record
5726      protocol version while receiving ClientHello even if the
5727      ClientHello is fragmented.  (The server can't insist on the
5728      particular protocol version it has chosen before the ServerHello
5729      message has informed the client about his choice.)
5730      [Bodo Moeller]
5731
5732   *) Load error codes if they are not already present instead of using a
5733      static variable. This allows them to be cleanly unloaded and reloaded.
5734      [Steve Henson]
5735
5736  Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
5737
5738   *) Introduce limits to prevent malicious keys being able to
5739      cause a denial of service.  (CVE-2006-2940)
5740      [Steve Henson, Bodo Moeller]
5741
5742   *) Fix ASN.1 parsing of certain invalid structures that can result
5743      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
5744
5745   *) Fix buffer overflow in SSL_get_shared_ciphers() function.
5746      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
5747
5748   *) Fix SSL client code which could crash if connecting to a
5749      malicious SSLv2 server.  (CVE-2006-4343)
5750      [Tavis Ormandy and Will Drewry, Google Security Team]
5751
5752   *) Change ciphersuite string processing so that an explicit
5753      ciphersuite selects this one ciphersuite (so that "AES256-SHA"
5754      will no longer include "AES128-SHA"), and any other similar
5755      ciphersuite (same bitmap) from *other* protocol versions (so that
5756      "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
5757      SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
5758      changes from 0.9.8b and 0.9.8d.
5759      [Bodo Moeller]
5760
5761  Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
5762
5763   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
5764      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
5765
5766   *) Change the Unix randomness entropy gathering to use poll() when
5767      possible instead of select(), since the latter has some
5768      undesirable limitations.
5769      [Darryl Miles via Richard Levitte and Bodo Moeller]
5770
5771   *) Disable rogue ciphersuites:
5772
5773       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
5774       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
5775       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
5776
5777      The latter two were purportedly from
5778      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
5779      appear there.
5780
5781      Also deactive the remaining ciphersuites from
5782      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
5783      unofficial, and the ID has long expired.
5784      [Bodo Moeller]
5785
5786   *) Fix RSA blinding Heisenbug (problems sometimes occurred on
5787      dual-core machines) and other potential thread-safety issues.
5788      [Bodo Moeller]
5789
5790  Changes between 0.9.7i and 0.9.7j  [04 May 2006]
5791
5792   *) Adapt fipsld and the build system to link against the validated FIPS
5793      module in FIPS mode.
5794      [Steve Henson]
5795
5796   *) Fixes for VC++ 2005 build under Windows.
5797      [Steve Henson]
5798
5799   *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
5800      from a Windows bash shell such as MSYS. It is autodetected from the
5801      "config" script when run from a VC++ environment. Modify standard VC++
5802      build to use fipscanister.o from the GNU make build.
5803      [Steve Henson]
5804
5805  Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
5806
5807   *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
5808      The value now differs depending on if you build for FIPS or not.
5809      BEWARE!  A program linked with a shared FIPSed libcrypto can't be
5810      safely run with a non-FIPSed libcrypto, as it may crash because of
5811      the difference induced by this change.
5812      [Andy Polyakov]
5813
5814  Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
5815
5816   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
5817      (part of SSL_OP_ALL).  This option used to disable the
5818      countermeasure against man-in-the-middle protocol-version
5819      rollback in the SSL 2.0 server implementation, which is a bad
5820      idea.  (CVE-2005-2969)
5821
5822      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
5823      for Information Security, National Institute of Advanced Industrial
5824      Science and Technology [AIST], Japan)]
5825
5826   *) Minimal support for X9.31 signatures and PSS padding modes. This is
5827      mainly for FIPS compliance and not fully integrated at this stage.
5828      [Steve Henson]
5829
5830   *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
5831      the exponentiation using a fixed-length exponent.  (Otherwise,
5832      the information leaked through timing could expose the secret key
5833      after many signatures; cf. Bleichenbacher's attack on DSA with
5834      biased k.)
5835      [Bodo Moeller]
5836
5837   *) Make a new fixed-window mod_exp implementation the default for
5838      RSA, DSA, and DH private-key operations so that the sequence of
5839      squares and multiplies and the memory access pattern are
5840      independent of the particular secret key.  This will mitigate
5841      cache-timing and potential related attacks.
5842
5843      BN_mod_exp_mont_consttime() is the new exponentiation implementation,
5844      and this is automatically used by BN_mod_exp_mont() if the new flag
5845      BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
5846      will use this BN flag for private exponents unless the flag
5847      RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
5848      DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
5849
5850      [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
5851
5852   *) Change the client implementation for SSLv23_method() and
5853      SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
5854      Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
5855      (Previously, the SSL 2.0 backwards compatible Client Hello
5856      message format would be used even with SSL_OP_NO_SSLv2.)
5857      [Bodo Moeller]
5858
5859   *) Add support for smime-type MIME parameter in S/MIME messages which some
5860      clients need.
5861      [Steve Henson]
5862
5863   *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
5864      a threadsafe manner. Modify rsa code to use new function and add calls
5865      to dsa and dh code (which had race conditions before).
5866      [Steve Henson]
5867
5868   *) Include the fixed error library code in the C error file definitions
5869      instead of fixing them up at runtime. This keeps the error code
5870      structures constant.
5871      [Steve Henson]
5872
5873  Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
5874
5875   [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
5876   OpenSSL 0.9.8.]
5877
5878   *) Fixes for newer kerberos headers. NB: the casts are needed because
5879      the 'length' field is signed on one version and unsigned on another
5880      with no (?) obvious way to tell the difference, without these VC++
5881      complains. Also the "definition" of FAR (blank) is no longer included
5882      nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
5883      some needed definitions.
5884      [Steve Henson]
5885
5886   *) Undo Cygwin change.
5887      [Ulf Möller]
5888
5889   *) Added support for proxy certificates according to RFC 3820.
5890      Because they may be a security thread to unaware applications,
5891      they must be explicitly allowed in run-time.  See
5892      docs/HOWTO/proxy_certificates.txt for further information.
5893      [Richard Levitte]
5894
5895  Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
5896
5897   *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
5898      server and client random values. Previously
5899      (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
5900      less random data when sizeof(time_t) > 4 (some 64 bit platforms).
5901
5902      This change has negligible security impact because:
5903
5904      1. Server and client random values still have 24 bytes of pseudo random
5905         data.
5906
5907      2. Server and client random values are sent in the clear in the initial
5908         handshake.
5909
5910      3. The master secret is derived using the premaster secret (48 bytes in
5911         size for static RSA ciphersuites) as well as client server and random
5912         values.
5913
5914      The OpenSSL team would like to thank the UK NISCC for bringing this issue
5915      to our attention.
5916
5917      [Stephen Henson, reported by UK NISCC]
5918
5919   *) Use Windows randomness collection on Cygwin.
5920      [Ulf Möller]
5921
5922   *) Fix hang in EGD/PRNGD query when communication socket is closed
5923      prematurely by EGD/PRNGD.
5924      [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
5925
5926   *) Prompt for pass phrases when appropriate for PKCS12 input format.
5927      [Steve Henson]
5928
5929   *) Back-port of selected performance improvements from development
5930      branch, as well as improved support for PowerPC platforms.
5931      [Andy Polyakov]
5932
5933   *) Add lots of checks for memory allocation failure, error codes to indicate
5934      failure and freeing up memory if a failure occurs.
5935      [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
5936
5937   *) Add new -passin argument to dgst.
5938      [Steve Henson]
5939
5940   *) Perform some character comparisons of different types in X509_NAME_cmp:
5941      this is needed for some certificates that re-encode DNs into UTF8Strings
5942      (in violation of RFC3280) and can't or won't issue name rollover
5943      certificates.
5944      [Steve Henson]
5945
5946   *) Make an explicit check during certificate validation to see that
5947      the CA setting in each certificate on the chain is correct.  As a
5948      side effect always do the following basic checks on extensions,
5949      not just when there's an associated purpose to the check:
5950
5951       - if there is an unhandled critical extension (unless the user
5952         has chosen to ignore this fault)
5953       - if the path length has been exceeded (if one is set at all)
5954       - that certain extensions fit the associated purpose (if one has
5955         been given)
5956      [Richard Levitte]
5957
5958  Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
5959
5960   *) Avoid a race condition when CRLs are checked in a multi threaded
5961      environment. This would happen due to the reordering of the revoked
5962      entries during signature checking and serial number lookup. Now the
5963      encoding is cached and the serial number sort performed under a lock.
5964      Add new STACK function sk_is_sorted().
5965      [Steve Henson]
5966
5967   *) Add Delta CRL to the extension code.
5968      [Steve Henson]
5969
5970   *) Various fixes to s3_pkt.c so alerts are sent properly.
5971      [David Holmes <d.holmes@f5.com>]
5972
5973   *) Reduce the chances of duplicate issuer name and serial numbers (in
5974      violation of RFC3280) using the OpenSSL certificate creation utilities.
5975      This is done by creating a random 64 bit value for the initial serial
5976      number when a serial number file is created or when a self signed
5977      certificate is created using 'openssl req -x509'. The initial serial
5978      number file is created using 'openssl x509 -next_serial' in CA.pl
5979      rather than being initialized to 1.
5980      [Steve Henson]
5981
5982  Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
5983
5984   *) Fix null-pointer assignment in do_change_cipher_spec() revealed
5985      by using the Codenomicon TLS Test Tool (CVE-2004-0079)
5986      [Joe Orton, Steve Henson]
5987
5988   *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
5989      (CVE-2004-0112)
5990      [Joe Orton, Steve Henson]
5991
5992   *) Make it possible to have multiple active certificates with the same
5993      subject in the CA index file.  This is done only if the keyword
5994      'unique_subject' is set to 'no' in the main CA section (default
5995      if 'CA_default') of the configuration file.  The value is saved
5996      with the database itself in a separate index attribute file,
5997      named like the index file with '.attr' appended to the name.
5998      [Richard Levitte]
5999
6000   *) X509 verify fixes. Disable broken certificate workarounds when
6001      X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
6002      keyUsage extension present. Don't accept CRLs with unhandled critical
6003      extensions: since verify currently doesn't process CRL extensions this
6004      rejects a CRL with *any* critical extensions. Add new verify error codes
6005      for these cases.
6006      [Steve Henson]
6007
6008   *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
6009      A clarification of RFC2560 will require the use of OCTET STRINGs and
6010      some implementations cannot handle the current raw format. Since OpenSSL
6011      copies and compares OCSP nonces as opaque blobs without any attempt at
6012      parsing them this should not create any compatibility issues.
6013      [Steve Henson]
6014
6015   *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
6016      calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
6017      this HMAC (and other) operations are several times slower than OpenSSL
6018      < 0.9.7.
6019      [Steve Henson]
6020
6021   *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
6022      [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
6023
6024   *) Use the correct content when signing type "other".
6025      [Steve Henson]
6026
6027  Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
6028
6029   *) Fix various bugs revealed by running the NISCC test suite:
6030
6031      Stop out of bounds reads in the ASN1 code when presented with
6032      invalid tags (CVE-2003-0543 and CVE-2003-0544).
6033
6034      Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
6035
6036      If verify callback ignores invalid public key errors don't try to check
6037      certificate signature with the NULL public key.
6038
6039      [Steve Henson]
6040
6041   *) New -ignore_err option in ocsp application to stop the server
6042      exiting on the first error in a request.
6043      [Steve Henson]
6044
6045   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
6046      if the server requested one: as stated in TLS 1.0 and SSL 3.0
6047      specifications.
6048      [Steve Henson]
6049
6050   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
6051      extra data after the compression methods not only for TLS 1.0
6052      but also for SSL 3.0 (as required by the specification).
6053      [Bodo Moeller; problem pointed out by Matthias Loepfe]
6054
6055   *) Change X509_certificate_type() to mark the key as exported/exportable
6056      when it's 512 *bits* long, not 512 bytes.
6057      [Richard Levitte]
6058
6059   *) Change AES_cbc_encrypt() so it outputs exact multiple of
6060      blocks during encryption.
6061      [Richard Levitte]
6062
6063   *) Various fixes to base64 BIO and non blocking I/O. On write
6064      flushes were not handled properly if the BIO retried. On read
6065      data was not being buffered properly and had various logic bugs.
6066      This also affects blocking I/O when the data being decoded is a
6067      certain size.
6068      [Steve Henson]
6069
6070   *) Various S/MIME bugfixes and compatibility changes:
6071      output correct application/pkcs7 MIME type if
6072      PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
6073      Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
6074      of files as .eml work). Correctly handle very long lines in MIME
6075      parser.
6076      [Steve Henson]
6077
6078  Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
6079
6080   *) Countermeasure against the Klima-Pokorny-Rosa extension of
6081      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
6082      a protocol version number mismatch like a decryption error
6083      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
6084      [Bodo Moeller]
6085
6086   *) Turn on RSA blinding by default in the default implementation
6087      to avoid a timing attack. Applications that don't want it can call
6088      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
6089      They would be ill-advised to do so in most cases.
6090      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
6091
6092   *) Change RSA blinding code so that it works when the PRNG is not
6093      seeded (in this case, the secret RSA exponent is abused as
6094      an unpredictable seed -- if it is not unpredictable, there
6095      is no point in blinding anyway).  Make RSA blinding thread-safe
6096      by remembering the creator's thread ID in rsa->blinding and
6097      having all other threads use local one-time blinding factors
6098      (this requires more computation than sharing rsa->blinding, but
6099      avoids excessive locking; and if an RSA object is not shared
6100      between threads, blinding will still be very fast).
6101      [Bodo Moeller]
6102
6103   *) Fixed a typo bug that would cause ENGINE_set_default() to set an
6104      ENGINE as defaults for all supported algorithms irrespective of
6105      the 'flags' parameter. 'flags' is now honoured, so applications
6106      should make sure they are passing it correctly.
6107      [Geoff Thorpe]
6108
6109   *) Target "mingw" now allows native Windows code to be generated in
6110      the Cygwin environment as well as with the MinGW compiler.
6111      [Ulf Moeller]
6112
6113  Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
6114
6115   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
6116      via timing by performing a MAC computation even if incorrect
6117      block cipher padding has been found.  This is a countermeasure
6118      against active attacks where the attacker has to distinguish
6119      between bad padding and a MAC verification error. (CVE-2003-0078)
6120
6121      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
6122      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
6123      Martin Vuagnoux (EPFL, Ilion)]
6124
6125   *) Make the no-err option work as intended.  The intention with no-err
6126      is not to have the whole error stack handling routines removed from
6127      libcrypto, it's only intended to remove all the function name and
6128      reason texts, thereby removing some of the footprint that may not
6129      be interesting if those errors aren't displayed anyway.
6130
6131      NOTE: it's still possible for any application or module to have it's
6132      own set of error texts inserted.  The routines are there, just not
6133      used by default when no-err is given.
6134      [Richard Levitte]
6135
6136   *) Add support for FreeBSD on IA64.
6137      [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
6138
6139   *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
6140      Kerberos function mit_des_cbc_cksum().  Before this change,
6141      the value returned by DES_cbc_cksum() was like the one from
6142      mit_des_cbc_cksum(), except the bytes were swapped.
6143      [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
6144
6145   *) Allow an application to disable the automatic SSL chain building.
6146      Before this a rather primitive chain build was always performed in
6147      ssl3_output_cert_chain(): an application had no way to send the
6148      correct chain if the automatic operation produced an incorrect result.
6149
6150      Now the chain builder is disabled if either:
6151
6152      1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
6153
6154      2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
6155
6156      The reasoning behind this is that an application would not want the
6157      auto chain building to take place if extra chain certificates are
6158      present and it might also want a means of sending no additional
6159      certificates (for example the chain has two certificates and the
6160      root is omitted).
6161      [Steve Henson]
6162
6163   *) Add the possibility to build without the ENGINE framework.
6164      [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
6165
6166   *) Under Win32 gmtime() can return NULL: check return value in
6167      OPENSSL_gmtime(). Add error code for case where gmtime() fails.
6168      [Steve Henson]
6169
6170   *) DSA routines: under certain error conditions uninitialized BN objects
6171      could be freed. Solution: make sure initialization is performed early
6172      enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
6173      Nils Larsch <nla@trustcenter.de> via PR#459)
6174      [Lutz Jaenicke]
6175
6176   *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
6177      checked on reconnect on the client side, therefore session resumption
6178      could still fail with a "ssl session id is different" error. This
6179      behaviour is masked when SSL_OP_ALL is used due to
6180      SSL_OP_MICROSOFT_SESS_ID_BUG being set.
6181      Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
6182      followup to PR #377.
6183      [Lutz Jaenicke]
6184
6185   *) IA-32 assembler support enhancements: unified ELF targets, support
6186      for SCO/Caldera platforms, fix for Cygwin shared build.
6187      [Andy Polyakov]
6188
6189   *) Add support for FreeBSD on sparc64.  As a consequence, support for
6190      FreeBSD on non-x86 processors is separate from x86 processors on
6191      the config script, much like the NetBSD support.
6192      [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
6193
6194  Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
6195
6196   [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
6197   OpenSSL 0.9.7.]
6198
6199   *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
6200      code (06) was taken as the first octet of the session ID and the last
6201      octet was ignored consequently. As a result SSLv2 client side session
6202      caching could not have worked due to the session ID mismatch between
6203      client and server.
6204      Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
6205      PR #377.
6206      [Lutz Jaenicke]
6207
6208   *) Change the declaration of needed Kerberos libraries to use EX_LIBS
6209      instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
6210      removed entirely.
6211      [Richard Levitte]
6212
6213   *) The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
6214      seems that in spite of existing for more than a year, many application
6215      author have done nothing to provide the necessary callbacks, which
6216      means that this particular engine will not work properly anywhere.
6217      This is a very unfortunate situation which forces us, in the name
6218      of usability, to give the hw_ncipher.c a static lock, which is part
6219      of libcrypto.
6220      NOTE: This is for the 0.9.7 series ONLY.  This hack will never
6221      appear in 0.9.8 or later.  We EXPECT application authors to have
6222      dealt properly with this when 0.9.8 is released (unless we actually
6223      make such changes in the libcrypto locking code that changes will
6224      have to be made anyway).
6225      [Richard Levitte]
6226
6227   *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
6228      octets have been read, EOF or an error occurs. Without this change
6229      some truncated ASN1 structures will not produce an error.
6230      [Steve Henson]
6231
6232   *) Disable Heimdal support, since it hasn't been fully implemented.
6233      Still give the possibility to force the use of Heimdal, but with
6234      warnings and a request that patches get sent to openssl-dev.
6235      [Richard Levitte]
6236
6237   *) Add the VC-CE target, introduce the WINCE sysname, and add
6238      INSTALL.WCE and appropriate conditionals to make it build.
6239      [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
6240
6241   *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
6242      cygssl-x.y.z.dll, where x, y and z are the major, minor and
6243      edit numbers of the version.
6244      [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
6245
6246   *) Introduce safe string copy and catenation functions
6247      (BUF_strlcpy() and BUF_strlcat()).
6248      [Ben Laurie (CHATS) and Richard Levitte]
6249
6250   *) Avoid using fixed-size buffers for one-line DNs.
6251      [Ben Laurie (CHATS)]
6252
6253   *) Add BUF_MEM_grow_clean() to avoid information leakage when
6254      resizing buffers containing secrets, and use where appropriate.
6255      [Ben Laurie (CHATS)]
6256
6257   *) Avoid using fixed size buffers for configuration file location.
6258      [Ben Laurie (CHATS)]
6259
6260   *) Avoid filename truncation for various CA files.
6261      [Ben Laurie (CHATS)]
6262
6263   *) Use sizeof in preference to magic numbers.
6264      [Ben Laurie (CHATS)]
6265
6266   *) Avoid filename truncation in cert requests.
6267      [Ben Laurie (CHATS)]
6268
6269   *) Add assertions to check for (supposedly impossible) buffer
6270      overflows.
6271      [Ben Laurie (CHATS)]
6272
6273   *) Don't cache truncated DNS entries in the local cache (this could
6274      potentially lead to a spoofing attack).
6275      [Ben Laurie (CHATS)]
6276
6277   *) Fix various buffers to be large enough for hex/decimal
6278      representations in a platform independent manner.
6279      [Ben Laurie (CHATS)]
6280
6281   *) Add CRYPTO_realloc_clean() to avoid information leakage when
6282      resizing buffers containing secrets, and use where appropriate.
6283      [Ben Laurie (CHATS)]
6284
6285   *) Add BIO_indent() to avoid much slightly worrying code to do
6286      indents.
6287      [Ben Laurie (CHATS)]
6288
6289   *) Convert sprintf()/BIO_puts() to BIO_printf().
6290      [Ben Laurie (CHATS)]
6291
6292   *) buffer_gets() could terminate with the buffer only half
6293      full. Fixed.
6294      [Ben Laurie (CHATS)]
6295
6296   *) Add assertions to prevent user-supplied crypto functions from
6297      overflowing internal buffers by having large block sizes, etc.
6298      [Ben Laurie (CHATS)]
6299
6300   *) New OPENSSL_assert() macro (similar to assert(), but enabled
6301      unconditionally).
6302      [Ben Laurie (CHATS)]
6303
6304   *) Eliminate unused copy of key in RC4.
6305      [Ben Laurie (CHATS)]
6306
6307   *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
6308      [Ben Laurie (CHATS)]
6309
6310   *) Fix off-by-one error in EGD path.
6311      [Ben Laurie (CHATS)]
6312
6313   *) If RANDFILE path is too long, ignore instead of truncating.
6314      [Ben Laurie (CHATS)]
6315
6316   *) Eliminate unused and incorrectly sized X.509 structure
6317      CBCParameter.
6318      [Ben Laurie (CHATS)]
6319
6320   *) Eliminate unused and dangerous function knumber().
6321      [Ben Laurie (CHATS)]
6322
6323   *) Eliminate unused and dangerous structure, KSSL_ERR.
6324      [Ben Laurie (CHATS)]
6325
6326   *) Protect against overlong session ID context length in an encoded
6327      session object. Since these are local, this does not appear to be
6328      exploitable.
6329      [Ben Laurie (CHATS)]
6330
6331   *) Change from security patch (see 0.9.6e below) that did not affect
6332      the 0.9.6 release series:
6333
6334      Remote buffer overflow in SSL3 protocol - an attacker could
6335      supply an oversized master key in Kerberos-enabled versions.
6336      (CVE-2002-0657)
6337      [Ben Laurie (CHATS)]
6338
6339   *) Change the SSL kerb5 codes to match RFC 2712.
6340      [Richard Levitte]
6341
6342   *) Make -nameopt work fully for req and add -reqopt switch.
6343      [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
6344
6345   *) The "block size" for block ciphers in CFB and OFB mode should be 1.
6346      [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
6347
6348   *) Make sure tests can be performed even if the corresponding algorithms
6349      have been removed entirely.  This was also the last step to make
6350      OpenSSL compilable with DJGPP under all reasonable conditions.
6351      [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
6352
6353   *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
6354      to allow version independent disabling of normally unselected ciphers,
6355      which may be activated as a side-effect of selecting a single cipher.
6356
6357      (E.g., cipher list string "RSA" enables ciphersuites that are left
6358      out of "ALL" because they do not provide symmetric encryption.
6359      "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
6360      [Lutz Jaenicke, Bodo Moeller]
6361
6362   *) Add appropriate support for separate platform-dependent build
6363      directories.  The recommended way to make a platform-dependent
6364      build directory is the following (tested on Linux), maybe with
6365      some local tweaks:
6366
6367         # Place yourself outside of the OpenSSL source tree.  In
6368         # this example, the environment variable OPENSSL_SOURCE
6369         # is assumed to contain the absolute OpenSSL source directory.
6370         mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
6371         cd objtree/"`uname -s`-`uname -r`-`uname -m`"
6372         (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
6373                 mkdir -p `dirname $F`
6374                 ln -s $OPENSSL_SOURCE/$F $F
6375         done
6376
6377      To be absolutely sure not to disturb the source tree, a "make clean"
6378      is a good thing.  If it isn't successful, don't worry about it,
6379      it probably means the source directory is very clean.
6380      [Richard Levitte]
6381
6382   *) Make sure any ENGINE control commands make local copies of string
6383      pointers passed to them whenever necessary. Otherwise it is possible
6384      the caller may have overwritten (or deallocated) the original string
6385      data when a later ENGINE operation tries to use the stored values.
6386      [Götz Babin-Ebell <babinebell@trustcenter.de>]
6387
6388   *) Improve diagnostics in file reading and command-line digests.
6389      [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
6390
6391   *) Add AES modes CFB and OFB to the object database.  Correct an
6392      error in AES-CFB decryption.
6393      [Richard Levitte]
6394
6395   *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
6396      allows existing EVP_CIPHER_CTX structures to be reused after
6397      calling EVP_*Final(). This behaviour is used by encryption
6398      BIOs and some applications. This has the side effect that
6399      applications must explicitly clean up cipher contexts with
6400      EVP_CIPHER_CTX_cleanup() or they will leak memory.
6401      [Steve Henson]
6402
6403   *) Check the values of dna and dnb in bn_mul_recursive before calling
6404      bn_mul_comba (a non zero value means the a or b arrays do not contain
6405      n2 elements) and fallback to bn_mul_normal if either is not zero.
6406      [Steve Henson]
6407
6408   *) Fix escaping of non-ASCII characters when using the -subj option
6409      of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
6410      [Lutz Jaenicke]
6411
6412   *) Make object definitions compliant to LDAP (RFC2256): SN is the short
6413      form for "surname", serialNumber has no short form.
6414      Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
6415      therefore remove "mail" short name for "internet 7".
6416      The OID for unique identifiers in X509 certificates is
6417      x500UniqueIdentifier, not uniqueIdentifier.
6418      Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
6419      [Lutz Jaenicke]
6420
6421   *) Add an "init" command to the ENGINE config module and auto initialize
6422      ENGINEs. Without any "init" command the ENGINE will be initialized
6423      after all ctrl commands have been executed on it. If init=1 the
6424      ENGINE is initailized at that point (ctrls before that point are run
6425      on the uninitialized ENGINE and after on the initialized one). If
6426      init=0 then the ENGINE will not be iniatialized at all.
6427      [Steve Henson]
6428
6429   *) Fix the 'app_verify_callback' interface so that the user-defined
6430      argument is actually passed to the callback: In the
6431      SSL_CTX_set_cert_verify_callback() prototype, the callback
6432      declaration has been changed from
6433           int (*cb)()
6434      into
6435           int (*cb)(X509_STORE_CTX *,void *);
6436      in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
6437           i=s->ctx->app_verify_callback(&ctx)
6438      has been changed into
6439           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
6440
6441      To update applications using SSL_CTX_set_cert_verify_callback(),
6442      a dummy argument can be added to their callback functions.
6443      [D. K. Smetters <smetters@parc.xerox.com>]
6444
6445   *) Added the '4758cca' ENGINE to support IBM 4758 cards.
6446      [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
6447
6448   *) Add and OPENSSL_LOAD_CONF define which will cause
6449      OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
6450      This allows older applications to transparently support certain
6451      OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
6452      Two new functions OPENSSL_add_all_algorithms_noconf() which will never
6453      load the config file and OPENSSL_add_all_algorithms_conf() which will
6454      always load it have also been added.
6455      [Steve Henson]
6456
6457   *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
6458      Adjust NIDs and EVP layer.
6459      [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
6460
6461   *) Config modules support in openssl utility.
6462
6463      Most commands now load modules from the config file,
6464      though in a few (such as version) this isn't done
6465      because it couldn't be used for anything.
6466
6467      In the case of ca and req the config file used is
6468      the same as the utility itself: that is the -config
6469      command line option can be used to specify an
6470      alternative file.
6471      [Steve Henson]
6472
6473   *) Move default behaviour from OPENSSL_config(). If appname is NULL
6474      use "openssl_conf" if filename is NULL use default openssl config file.
6475      [Steve Henson]
6476
6477   *) Add an argument to OPENSSL_config() to allow the use of an alternative
6478      config section name. Add a new flag to tolerate a missing config file
6479      and move code to CONF_modules_load_file().
6480      [Steve Henson]
6481
6482   *) Support for crypto accelerator cards from Accelerated Encryption
6483      Processing, www.aep.ie.  (Use engine 'aep')
6484      The support was copied from 0.9.6c [engine] and adapted/corrected
6485      to work with the new engine framework.
6486      [AEP Inc. and Richard Levitte]
6487
6488   *) Support for SureWare crypto accelerator cards from Baltimore
6489      Technologies.  (Use engine 'sureware')
6490      The support was copied from 0.9.6c [engine] and adapted
6491      to work with the new engine framework.
6492      [Richard Levitte]
6493
6494   *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
6495      make the newer ENGINE framework commands for the CHIL engine work.
6496      [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
6497
6498   *) Make it possible to produce shared libraries on ReliantUNIX.
6499      [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
6500
6501   *) Add the configuration target debug-linux-ppro.
6502      Make 'openssl rsa' use the general key loading routines
6503      implemented in apps.c, and make those routines able to
6504      handle the key format FORMAT_NETSCAPE and the variant
6505      FORMAT_IISSGC.
6506      [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
6507
6508  *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
6509      [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
6510
6511   *) Add -keyform to rsautl, and document -engine.
6512      [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
6513
6514   *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
6515      BIO_R_NO_SUCH_FILE error code rather than the generic
6516      ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
6517      [Ben Laurie]
6518
6519   *) Add new functions
6520           ERR_peek_last_error
6521           ERR_peek_last_error_line
6522           ERR_peek_last_error_line_data.
6523      These are similar to
6524           ERR_peek_error
6525           ERR_peek_error_line
6526           ERR_peek_error_line_data,
6527      but report on the latest error recorded rather than the first one
6528      still in the error queue.
6529      [Ben Laurie, Bodo Moeller]
6530
6531   *) default_algorithms option in ENGINE config module. This allows things
6532      like:
6533      default_algorithms = ALL
6534      default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
6535      [Steve Henson]
6536
6537   *) Preliminary ENGINE config module.
6538      [Steve Henson]
6539
6540   *) New experimental application configuration code.
6541      [Steve Henson]
6542
6543   *) Change the AES code to follow the same name structure as all other
6544      symmetric ciphers, and behave the same way.  Move everything to
6545      the directory crypto/aes, thereby obsoleting crypto/rijndael.
6546      [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
6547
6548   *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
6549      [Ben Laurie and Theo de Raadt]
6550
6551   *) Add option to output public keys in req command.
6552      [Massimiliano Pala madwolf@openca.org]
6553
6554   *) Use wNAFs in EC_POINTs_mul() for improved efficiency
6555      (up to about 10% better than before for P-192 and P-224).
6556      [Bodo Moeller]
6557
6558   *) New functions/macros
6559
6560           SSL_CTX_set_msg_callback(ctx, cb)
6561           SSL_CTX_set_msg_callback_arg(ctx, arg)
6562           SSL_set_msg_callback(ssl, cb)
6563           SSL_set_msg_callback_arg(ssl, arg)
6564
6565      to request calling a callback function
6566
6567           void cb(int write_p, int version, int content_type,
6568                   const void *buf, size_t len, SSL *ssl, void *arg)
6569
6570      whenever a protocol message has been completely received
6571      (write_p == 0) or sent (write_p == 1).  Here 'version' is the
6572      protocol version  according to which the SSL library interprets
6573      the current protocol message (SSL2_VERSION, SSL3_VERSION, or
6574      TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
6575      the content type as defined in the SSL 3.0/TLS 1.0 protocol
6576      specification (change_cipher_spec(20), alert(21), handshake(22)).
6577      'buf' and 'len' point to the actual message, 'ssl' to the
6578      SSL object, and 'arg' is the application-defined value set by
6579      SSL[_CTX]_set_msg_callback_arg().
6580
6581      'openssl s_client' and 'openssl s_server' have new '-msg' options
6582      to enable a callback that displays all protocol messages.
6583      [Bodo Moeller]
6584
6585   *) Change the shared library support so shared libraries are built as
6586      soon as the corresponding static library is finished, and thereby get
6587      openssl and the test programs linked against the shared library.
6588      This still only happens when the keyword "shard" has been given to
6589      the configuration scripts.
6590
6591      NOTE: shared library support is still an experimental thing, and
6592      backward binary compatibility is still not guaranteed.
6593      ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
6594
6595   *) Add support for Subject Information Access extension.
6596      [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
6597
6598   *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
6599      additional bytes when new memory had to be allocated, not just
6600      when reusing an existing buffer.
6601      [Bodo Moeller]
6602
6603   *) New command line and configuration option 'utf8' for the req command.
6604      This allows field values to be specified as UTF8 strings.
6605      [Steve Henson]
6606
6607   *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
6608      runs for the former and machine-readable output for the latter.
6609      [Ben Laurie]
6610
6611   *) Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
6612      of the e-mail address in the DN (i.e., it will go into a certificate
6613      extension only).  The new configuration file option 'email_in_dn = no'
6614      has the same effect.
6615      [Massimiliano Pala madwolf@openca.org]
6616
6617   *) Change all functions with names starting with des_ to be starting
6618      with DES_ instead.  Add wrappers that are compatible with libdes,
6619      but are named _ossl_old_des_*.  Finally, add macros that map the
6620      des_* symbols to the corresponding _ossl_old_des_* if libdes
6621      compatibility is desired.  If OpenSSL 0.9.6c compatibility is
6622      desired, the des_* symbols will be mapped to DES_*, with one
6623      exception.
6624
6625      Since we provide two compatibility mappings, the user needs to
6626      define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
6627      compatibility is desired.  The default (i.e., when that macro
6628      isn't defined) is OpenSSL 0.9.6c compatibility.
6629
6630      There are also macros that enable and disable the support of old
6631      des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
6632      and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
6633      are defined, the default will apply: to support the old des routines.
6634
6635      In either case, one must include openssl/des.h to get the correct
6636      definitions.  Do not try to just include openssl/des_old.h, that
6637      won't work.
6638
6639      NOTE: This is a major break of an old API into a new one.  Software
6640      authors are encouraged to switch to the DES_ style functions.  Some
6641      time in the future, des_old.h and the libdes compatibility functions
6642      will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
6643      default), and then completely removed.
6644      [Richard Levitte]
6645
6646   *) Test for certificates which contain unsupported critical extensions.
6647      If such a certificate is found during a verify operation it is
6648      rejected by default: this behaviour can be overridden by either
6649      handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
6650      by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
6651      X509_supported_extension() has also been added which returns 1 if a
6652      particular extension is supported.
6653      [Steve Henson]
6654
6655   *) Modify the behaviour of EVP cipher functions in similar way to digests
6656      to retain compatibility with existing code.
6657      [Steve Henson]
6658
6659   *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
6660      compatibility with existing code. In particular the 'ctx' parameter does
6661      not have to be to be initialized before the call to EVP_DigestInit() and
6662      it is tidied up after a call to EVP_DigestFinal(). New function
6663      EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
6664      EVP_MD_CTX_copy() changed to not require the destination to be
6665      initialized valid and new function EVP_MD_CTX_copy_ex() added which
6666      requires the destination to be valid.
6667
6668      Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
6669      EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
6670      [Steve Henson]
6671
6672   *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
6673      so that complete 'Handshake' protocol structures are kept in memory
6674      instead of overwriting 'msg_type' and 'length' with 'body' data.
6675      [Bodo Moeller]
6676
6677   *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
6678      [Massimo Santin via Richard Levitte]
6679
6680   *) Major restructuring to the underlying ENGINE code. This includes
6681      reduction of linker bloat, separation of pure "ENGINE" manipulation
6682      (initialisation, etc) from functionality dealing with implementations
6683      of specific crypto iterfaces. This change also introduces integrated
6684      support for symmetric ciphers and digest implementations - so ENGINEs
6685      can now accelerate these by providing EVP_CIPHER and EVP_MD
6686      implementations of their own. This is detailed in crypto/engine/README
6687      as it couldn't be adequately described here. However, there are a few
6688      API changes worth noting - some RSA, DSA, DH, and RAND functions that
6689      were changed in the original introduction of ENGINE code have now
6690      reverted back - the hooking from this code to ENGINE is now a good
6691      deal more passive and at run-time, operations deal directly with
6692      RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
6693      dereferencing through an ENGINE pointer any more. Also, the ENGINE
6694      functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
6695      they were not being used by the framework as there is no concept of a
6696      BIGNUM_METHOD and they could not be generalised to the new
6697      'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
6698      ENGINE_cpy() has been removed as it cannot be consistently defined in
6699      the new code.
6700      [Geoff Thorpe]
6701
6702   *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
6703      [Steve Henson]
6704
6705   *) Change mkdef.pl to sort symbols that get the same entry number,
6706      and make sure the automatically generated functions ERR_load_*
6707      become part of libeay.num as well.
6708      [Richard Levitte]
6709
6710   *) New function SSL_renegotiate_pending().  This returns true once
6711      renegotiation has been requested (either SSL_renegotiate() call
6712      or HelloRequest/ClientHello received from the peer) and becomes
6713      false once a handshake has been completed.
6714      (For servers, SSL_renegotiate() followed by SSL_do_handshake()
6715      sends a HelloRequest, but does not ensure that a handshake takes
6716      place.  SSL_renegotiate_pending() is useful for checking if the
6717      client has followed the request.)
6718      [Bodo Moeller]
6719
6720   *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
6721      By default, clients may request session resumption even during
6722      renegotiation (if session ID contexts permit); with this option,
6723      session resumption is possible only in the first handshake.
6724
6725      SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
6726      more bits available for options that should not be part of
6727      SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
6728      [Bodo Moeller]
6729
6730   *) Add some demos for certificate and certificate request creation.
6731      [Steve Henson]
6732
6733   *) Make maximum certificate chain size accepted from the peer application
6734      settable (SSL*_get/set_max_cert_list()), as proposed by
6735      "Douglas E. Engert" <deengert@anl.gov>.
6736      [Lutz Jaenicke]
6737
6738   *) Add support for shared libraries for Unixware-7
6739      (Boyd Lynn Gerber <gerberb@zenez.com>).
6740      [Lutz Jaenicke]
6741
6742   *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
6743      be done prior to destruction. Use this to unload error strings from
6744      ENGINEs that load their own error strings. NB: This adds two new API
6745      functions to "get" and "set" this destroy handler in an ENGINE.
6746      [Geoff Thorpe]
6747
6748   *) Alter all existing ENGINE implementations (except "openssl" and
6749      "openbsd") to dynamically instantiate their own error strings. This
6750      makes them more flexible to be built both as statically-linked ENGINEs
6751      and self-contained shared-libraries loadable via the "dynamic" ENGINE.
6752      Also, add stub code to each that makes building them as self-contained
6753      shared-libraries easier (see README.ENGINE).
6754      [Geoff Thorpe]
6755
6756   *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
6757      implementations into applications that are completely implemented in
6758      self-contained shared-libraries. The "dynamic" ENGINE exposes control
6759      commands that can be used to configure what shared-library to load and
6760      to control aspects of the way it is handled. Also, made an update to
6761      the README.ENGINE file that brings its information up-to-date and
6762      provides some information and instructions on the "dynamic" ENGINE
6763      (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
6764      [Geoff Thorpe]
6765
6766   *) Make it possible to unload ranges of ERR strings with a new
6767      "ERR_unload_strings" function.
6768      [Geoff Thorpe]
6769
6770   *) Add a copy() function to EVP_MD.
6771      [Ben Laurie]
6772
6773   *) Make EVP_MD routines take a context pointer instead of just the
6774      md_data void pointer.
6775      [Ben Laurie]
6776
6777   *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
6778      that the digest can only process a single chunk of data
6779      (typically because it is provided by a piece of
6780      hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
6781      is only going to provide a single chunk of data, and hence the
6782      framework needn't accumulate the data for oneshot drivers.
6783      [Ben Laurie]
6784
6785   *) As with "ERR", make it possible to replace the underlying "ex_data"
6786      functions. This change also alters the storage and management of global
6787      ex_data state - it's now all inside ex_data.c and all "class" code (eg.
6788      RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
6789      index counters. The API functions that use this state have been changed
6790      to take a "class_index" rather than pointers to the class's local STACK
6791      and counter, and there is now an API function to dynamically create new
6792      classes. This centralisation allows us to (a) plug a lot of the
6793      thread-safety problems that existed, and (b) makes it possible to clean
6794      up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
6795      such data would previously have always leaked in application code and
6796      workarounds were in place to make the memory debugging turn a blind eye
6797      to it. Application code that doesn't use this new function will still
6798      leak as before, but their memory debugging output will announce it now
6799      rather than letting it slide.
6800
6801      Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
6802      induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
6803      has a return value to indicate success or failure.
6804      [Geoff Thorpe]
6805
6806   *) Make it possible to replace the underlying "ERR" functions such that the
6807      global state (2 LHASH tables and 2 locks) is only used by the "default"
6808      implementation. This change also adds two functions to "get" and "set"
6809      the implementation prior to it being automatically set the first time
6810      any other ERR function takes place. Ie. an application can call "get",
6811      pass the return value to a module it has just loaded, and that module
6812      can call its own "set" function using that value. This means the
6813      module's "ERR" operations will use (and modify) the error state in the
6814      application and not in its own statically linked copy of OpenSSL code.
6815      [Geoff Thorpe]
6816
6817   *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
6818      reference counts. This performs normal REF_PRINT/REF_CHECK macros on
6819      the operation, and provides a more encapsulated way for external code
6820      (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
6821      to use these functions rather than manually incrementing the counts.
6822
6823      Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
6824      [Geoff Thorpe]
6825
6826   *) Add EVP test program.
6827      [Ben Laurie]
6828
6829   *) Add symmetric cipher support to ENGINE. Expect the API to change!
6830      [Ben Laurie]
6831
6832   *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
6833      X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
6834      X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
6835      These allow a CRL to be built without having to access X509_CRL fields
6836      directly. Modify 'ca' application to use new functions.
6837      [Steve Henson]
6838
6839   *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
6840      bug workarounds. Rollback attack detection is a security feature.
6841      The problem will only arise on OpenSSL servers when TLSv1 is not
6842      available (sslv3_server_method() or SSL_OP_NO_TLSv1).
6843      Software authors not wanting to support TLSv1 will have special reasons
6844      for their choice and can explicitly enable this option.
6845      [Bodo Moeller, Lutz Jaenicke]
6846
6847   *) Rationalise EVP so it can be extended: don't include a union of
6848      cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
6849      (similar to those existing for EVP_CIPHER_CTX).
6850      Usage example:
6851
6852          EVP_MD_CTX md;
6853
6854          EVP_MD_CTX_init(&md);             /* new function call */
6855          EVP_DigestInit(&md, EVP_sha1());
6856          EVP_DigestUpdate(&md, in, len);
6857          EVP_DigestFinal(&md, out, NULL);
6858          EVP_MD_CTX_cleanup(&md);          /* new function call */
6859
6860      [Ben Laurie]
6861
6862   *) Make DES key schedule conform to the usual scheme, as well as
6863      correcting its structure. This means that calls to DES functions
6864      now have to pass a pointer to a des_key_schedule instead of a
6865      plain des_key_schedule (which was actually always a pointer
6866      anyway): E.g.,
6867
6868          des_key_schedule ks;
6869
6870          des_set_key_checked(..., &ks);
6871          des_ncbc_encrypt(..., &ks, ...);
6872
6873      (Note that a later change renames 'des_...' into 'DES_...'.)
6874      [Ben Laurie]
6875
6876   *) Initial reduction of linker bloat: the use of some functions, such as
6877      PEM causes large amounts of unused functions to be linked in due to
6878      poor organisation. For example pem_all.c contains every PEM function
6879      which has a knock on effect of linking in large amounts of (unused)
6880      ASN1 code. Grouping together similar functions and splitting unrelated
6881      functions prevents this.
6882      [Steve Henson]
6883
6884   *) Cleanup of EVP macros.
6885      [Ben Laurie]
6886
6887   *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
6888      correct _ecb suffix.
6889      [Ben Laurie]
6890
6891   *) Add initial OCSP responder support to ocsp application. The
6892      revocation information is handled using the text based index
6893      use by the ca application. The responder can either handle
6894      requests generated internally, supplied in files (for example
6895      via a CGI script) or using an internal minimal server.
6896      [Steve Henson]
6897
6898   *) Add configuration choices to get zlib compression for TLS.
6899      [Richard Levitte]
6900
6901   *) Changes to Kerberos SSL for RFC 2712 compliance:
6902      1.  Implemented real KerberosWrapper, instead of just using
6903          KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
6904      2.  Implemented optional authenticator field of KerberosWrapper.
6905
6906      Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
6907      and authenticator structs; see crypto/krb5/.
6908
6909      Generalized Kerberos calls to support multiple Kerberos libraries.
6910      [Vern Staats <staatsvr@asc.hpc.mil>,
6911       Jeffrey Altman <jaltman@columbia.edu>
6912       via Richard Levitte]
6913
6914   *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
6915      already does with RSA. testdsa.h now has 'priv_key/pub_key'
6916      values for each of the key sizes rather than having just
6917      parameters (and 'speed' generating keys each time).
6918      [Geoff Thorpe]
6919
6920   *) Speed up EVP routines.
6921      Before:
6922 encrypt
6923 type              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
6924 des-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
6925 des-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
6926 des-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
6927 decrypt
6928 des-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
6929 des-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
6930 des-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
6931      After:
6932 encrypt
6933 des-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
6934 decrypt
6935 des-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
6936      [Ben Laurie]
6937
6938   *) Added the OS2-EMX target.
6939      ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
6940
6941   *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
6942      to support NCONF routines in extension code. New function CONF_set_nconf()
6943      to allow functions which take an NCONF to also handle the old LHASH
6944      structure: this means that the old CONF compatible routines can be
6945      retained (in particular wrt extensions) without having to duplicate the
6946      code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
6947      [Steve Henson]
6948
6949   *) Enhance the general user interface with mechanisms for inner control
6950      and with possibilities to have yes/no kind of prompts.
6951      [Richard Levitte]
6952
6953   *) Change all calls to low level digest routines in the library and
6954      applications to use EVP. Add missing calls to HMAC_cleanup() and
6955      don't assume HMAC_CTX can be copied using memcpy().
6956      [Verdon Walker <VWalker@novell.com>, Steve Henson]
6957
6958   *) Add the possibility to control engines through control names but with
6959      arbitrary arguments instead of just a string.
6960      Change the key loaders to take a UI_METHOD instead of a callback
6961      function pointer.  NOTE: this breaks binary compatibility with earlier
6962      versions of OpenSSL [engine].
6963      Adapt the nCipher code for these new conditions and add a card insertion
6964      callback.
6965      [Richard Levitte]
6966
6967   *) Enhance the general user interface with mechanisms to better support
6968      dialog box interfaces, application-defined prompts, the possibility
6969      to use defaults (for example default passwords from somewhere else)
6970      and interrupts/cancellations.
6971      [Richard Levitte]
6972
6973   *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
6974      attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
6975      [Steve Henson]
6976
6977   *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
6978      tidy up some unnecessarily weird code in 'sk_new()').
6979      [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
6980
6981   *) Change the key loading routines for ENGINEs to use the same kind
6982      callback (pem_password_cb) as all other routines that need this
6983      kind of callback.
6984      [Richard Levitte]
6985
6986   *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
6987      256 bit (=32 byte) keys. Of course seeding with more entropy bytes
6988      than this minimum value is recommended.
6989      [Lutz Jaenicke]
6990
6991   *) New random seeder for OpenVMS, using the system process statistics
6992      that are easily reachable.
6993      [Richard Levitte]
6994
6995   *) Windows apparently can't transparently handle global
6996      variables defined in DLLs. Initialisations such as:
6997
6998         const ASN1_ITEM *it = &ASN1_INTEGER_it;
6999
7000      won't compile. This is used by the any applications that need to
7001      declare their own ASN1 modules. This was fixed by adding the option
7002      EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
7003      needed for static libraries under Win32.
7004      [Steve Henson]
7005
7006   *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
7007      setting of purpose and trust fields. New X509_STORE trust and
7008      purpose functions and tidy up setting in other SSL functions.
7009      [Steve Henson]
7010
7011   *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
7012      structure. These are inherited by X509_STORE_CTX when it is
7013      initialised. This allows various defaults to be set in the
7014      X509_STORE structure (such as flags for CRL checking and custom
7015      purpose or trust settings) for functions which only use X509_STORE_CTX
7016      internally such as S/MIME.
7017
7018      Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
7019      trust settings if they are not set in X509_STORE. This allows X509_STORE
7020      purposes and trust (in S/MIME for example) to override any set by default.
7021
7022      Add command line options for CRL checking to smime, s_client and s_server
7023      applications.
7024      [Steve Henson]
7025
7026   *) Initial CRL based revocation checking. If the CRL checking flag(s)
7027      are set then the CRL is looked up in the X509_STORE structure and
7028      its validity and signature checked, then if the certificate is found
7029      in the CRL the verify fails with a revoked error.
7030
7031      Various new CRL related callbacks added to X509_STORE_CTX structure.
7032
7033      Command line options added to 'verify' application to support this.
7034
7035      This needs some additional work, such as being able to handle multiple
7036      CRLs with different times, extension based lookup (rather than just
7037      by subject name) and ultimately more complete V2 CRL extension
7038      handling.
7039      [Steve Henson]
7040
7041   *) Add a general user interface API (crypto/ui/).  This is designed
7042      to replace things like des_read_password and friends (backward
7043      compatibility functions using this new API are provided).
7044      The purpose is to remove prompting functions from the DES code
7045      section as well as provide for prompting through dialog boxes in
7046      a window system and the like.
7047      [Richard Levitte]
7048
7049   *) Add "ex_data" support to ENGINE so implementations can add state at a
7050      per-structure level rather than having to store it globally.
7051      [Geoff]
7052
7053   *) Make it possible for ENGINE structures to be copied when retrieved by
7054      ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
7055      This causes the "original" ENGINE structure to act like a template,
7056      analogous to the RSA vs. RSA_METHOD type of separation. Because of this
7057      operational state can be localised to each ENGINE structure, despite the
7058      fact they all share the same "methods". New ENGINE structures returned in
7059      this case have no functional references and the return value is the single
7060      structural reference. This matches the single structural reference returned
7061      by ENGINE_by_id() normally, when it is incremented on the pre-existing
7062      ENGINE structure.
7063      [Geoff]
7064
7065   *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
7066      needs to match any other type at all we need to manually clear the
7067      tag cache.
7068      [Steve Henson]
7069
7070   *) Changes to the "openssl engine" utility to include;
7071      - verbosity levels ('-v', '-vv', and '-vvv') that provide information
7072        about an ENGINE's available control commands.
7073      - executing control commands from command line arguments using the
7074        '-pre' and '-post' switches. '-post' is only used if '-t' is
7075        specified and the ENGINE is successfully initialised. The syntax for
7076        the individual commands are colon-separated, for example;
7077          openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
7078      [Geoff]
7079
7080   *) New dynamic control command support for ENGINEs. ENGINEs can now
7081      declare their own commands (numbers), names (strings), descriptions,
7082      and input types for run-time discovery by calling applications. A
7083      subset of these commands are implicitly classed as "executable"
7084      depending on their input type, and only these can be invoked through
7085      the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
7086      can be based on user input, config files, etc). The distinction is
7087      that "executable" commands cannot return anything other than a boolean
7088      result and can only support numeric or string input, whereas some
7089      discoverable commands may only be for direct use through
7090      ENGINE_ctrl(), eg. supporting the exchange of binary data, function
7091      pointers, or other custom uses. The "executable" commands are to
7092      support parameterisations of ENGINE behaviour that can be
7093      unambiguously defined by ENGINEs and used consistently across any
7094      OpenSSL-based application. Commands have been added to all the
7095      existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
7096      control over shared-library paths without source code alterations.
7097      [Geoff]
7098
7099   *) Changed all ENGINE implementations to dynamically allocate their
7100      ENGINEs rather than declaring them statically. Apart from this being
7101      necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
7102      this also allows the implementations to compile without using the
7103      internal engine_int.h header.
7104      [Geoff]
7105
7106   *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
7107      'const' value. Any code that should be able to modify a RAND_METHOD
7108      should already have non-const pointers to it (ie. they should only
7109      modify their own ones).
7110      [Geoff]
7111
7112   *) Made a variety of little tweaks to the ENGINE code.
7113      - "atalla" and "ubsec" string definitions were moved from header files
7114        to C code. "nuron" string definitions were placed in variables
7115        rather than hard-coded - allowing parameterisation of these values
7116        later on via ctrl() commands.
7117      - Removed unused "#if 0"'d code.
7118      - Fixed engine list iteration code so it uses ENGINE_free() to release
7119        structural references.
7120      - Constified the RAND_METHOD element of ENGINE structures.
7121      - Constified various get/set functions as appropriate and added
7122        missing functions (including a catch-all ENGINE_cpy that duplicates
7123        all ENGINE values onto a new ENGINE except reference counts/state).
7124      - Removed NULL parameter checks in get/set functions. Setting a method
7125        or function to NULL is a way of cancelling out a previously set
7126        value.  Passing a NULL ENGINE parameter is just plain stupid anyway
7127        and doesn't justify the extra error symbols and code.
7128      - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
7129        flags from engine_int.h to engine.h.
7130      - Changed prototypes for ENGINE handler functions (init(), finish(),
7131        ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
7132      [Geoff]
7133
7134   *) Implement binary inversion algorithm for BN_mod_inverse in addition
7135      to the algorithm using long division.  The binary algorithm can be
7136      used only if the modulus is odd.  On 32-bit systems, it is faster
7137      only for relatively small moduli (roughly 20-30% for 128-bit moduli,
7138      roughly 5-15% for 256-bit moduli), so we use it only for moduli
7139      up to 450 bits.  In 64-bit environments, the binary algorithm
7140      appears to be advantageous for much longer moduli; here we use it
7141      for moduli up to 2048 bits.
7142      [Bodo Moeller]
7143
7144   *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
7145      could not support the combine flag in choice fields.
7146      [Steve Henson]
7147
7148   *) Add a 'copy_extensions' option to the 'ca' utility. This copies
7149      extensions from a certificate request to the certificate.
7150      [Steve Henson]
7151
7152   *) Allow multiple 'certopt' and 'nameopt' options to be separated
7153      by commas. Add 'namopt' and 'certopt' options to the 'ca' config
7154      file: this allows the display of the certificate about to be
7155      signed to be customised, to allow certain fields to be included
7156      or excluded and extension details. The old system didn't display
7157      multicharacter strings properly, omitted fields not in the policy
7158      and couldn't display additional details such as extensions.
7159      [Steve Henson]
7160
7161   *) Function EC_POINTs_mul for multiple scalar multiplication
7162      of an arbitrary number of elliptic curve points
7163           \sum scalars[i]*points[i],
7164      optionally including the generator defined for the EC_GROUP:
7165           scalar*generator +  \sum scalars[i]*points[i].
7166
7167      EC_POINT_mul is a simple wrapper function for the typical case
7168      that the point list has just one item (besides the optional
7169      generator).
7170      [Bodo Moeller]
7171
7172   *) First EC_METHODs for curves over GF(p):
7173
7174      EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
7175      operations and provides various method functions that can also
7176      operate with faster implementations of modular arithmetic.
7177
7178      EC_GFp_mont_method() reuses most functions that are part of
7179      EC_GFp_simple_method, but uses Montgomery arithmetic.
7180
7181      [Bodo Moeller; point addition and point doubling
7182      implementation directly derived from source code provided by
7183      Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
7184
7185   *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
7186      crypto/ec/ec_lib.c):
7187
7188      Curves are EC_GROUP objects (with an optional group generator)
7189      based on EC_METHODs that are built into the library.
7190
7191      Points are EC_POINT objects based on EC_GROUP objects.
7192
7193      Most of the framework would be able to handle curves over arbitrary
7194      finite fields, but as there are no obvious types for fields other
7195      than GF(p), some functions are limited to that for now.
7196      [Bodo Moeller]
7197
7198   *) Add the -HTTP option to s_server.  It is similar to -WWW, but requires
7199      that the file contains a complete HTTP response.
7200      [Richard Levitte]
7201
7202   *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
7203      change the def and num file printf format specifier from "%-40sXXX"
7204      to "%-39s XXX". The latter will always guarantee a space after the
7205      field while the former will cause them to run together if the field
7206      is 40 of more characters long.
7207      [Steve Henson]
7208
7209   *) Constify the cipher and digest 'method' functions and structures
7210      and modify related functions to take constant EVP_MD and EVP_CIPHER
7211      pointers.
7212      [Steve Henson]
7213
7214   *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
7215      in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
7216      [Bodo Moeller]
7217
7218   *) Modify EVP_Digest*() routines so they now return values. Although the
7219      internal software routines can never fail additional hardware versions
7220      might.
7221      [Steve Henson]
7222
7223   *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
7224
7225      Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
7226      (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
7227
7228      ASN1 error codes
7229           ERR_R_NESTED_ASN1_ERROR
7230           ...
7231           ERR_R_MISSING_ASN1_EOS
7232      were 4 .. 9, conflicting with
7233           ERR_LIB_RSA (= ERR_R_RSA_LIB)
7234           ...
7235           ERR_LIB_PEM (= ERR_R_PEM_LIB).
7236      They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
7237
7238      Add new error code 'ERR_R_INTERNAL_ERROR'.
7239      [Bodo Moeller]
7240
7241   *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
7242      suffices.
7243      [Bodo Moeller]
7244
7245   *) New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
7246      sets the subject name for a new request or supersedes the
7247      subject name in a given request. Formats that can be parsed are
7248           'CN=Some Name, OU=myOU, C=IT'
7249      and
7250           'CN=Some Name/OU=myOU/C=IT'.
7251
7252      Add options '-batch' and '-verbose' to 'openssl req'.
7253      [Massimiliano Pala <madwolf@hackmasters.net>]
7254
7255   *) Introduce the possibility to access global variables through
7256      functions on platform were that's the best way to handle exporting
7257      global variables in shared libraries.  To enable this functionality,
7258      one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
7259      "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
7260      is normally done by Configure or something similar).
7261
7262      To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
7263      in the source file (foo.c) like this:
7264
7265         OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
7266         OPENSSL_IMPLEMENT_GLOBAL(double,bar);
7267
7268      To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
7269      and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
7270
7271         OPENSSL_DECLARE_GLOBAL(int,foo);
7272         #define foo OPENSSL_GLOBAL_REF(foo)
7273         OPENSSL_DECLARE_GLOBAL(double,bar);
7274         #define bar OPENSSL_GLOBAL_REF(bar)
7275
7276      The #defines are very important, and therefore so is including the
7277      header file everywhere where the defined globals are used.
7278
7279      The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
7280      of ASN.1 items, but that structure is a bit different.
7281
7282      The largest change is in util/mkdef.pl which has been enhanced with
7283      better and easier to understand logic to choose which symbols should
7284      go into the Windows .def files as well as a number of fixes and code
7285      cleanup (among others, algorithm keywords are now sorted
7286      lexicographically to avoid constant rewrites).
7287      [Richard Levitte]
7288
7289   *) In BN_div() keep a copy of the sign of 'num' before writing the
7290      result to 'rm' because if rm==num the value will be overwritten
7291      and produce the wrong result if 'num' is negative: this caused
7292      problems with BN_mod() and BN_nnmod().
7293      [Steve Henson]
7294
7295   *) Function OCSP_request_verify(). This checks the signature on an
7296      OCSP request and verifies the signer certificate. The signer
7297      certificate is just checked for a generic purpose and OCSP request
7298      trust settings.
7299      [Steve Henson]
7300
7301   *) Add OCSP_check_validity() function to check the validity of OCSP
7302      responses. OCSP responses are prepared in real time and may only
7303      be a few seconds old. Simply checking that the current time lies
7304      between thisUpdate and nextUpdate max reject otherwise valid responses
7305      caused by either OCSP responder or client clock inaccuracy. Instead
7306      we allow thisUpdate and nextUpdate to fall within a certain period of
7307      the current time. The age of the response can also optionally be
7308      checked. Two new options -validity_period and -status_age added to
7309      ocsp utility.
7310      [Steve Henson]
7311
7312   *) If signature or public key algorithm is unrecognized print out its
7313      OID rather that just UNKNOWN.
7314      [Steve Henson]
7315
7316   *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
7317      OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
7318      ID to be generated from the issuer certificate alone which can then be
7319      passed to OCSP_id_issuer_cmp().
7320      [Steve Henson]
7321
7322   *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
7323      ASN1 modules to export functions returning ASN1_ITEM pointers
7324      instead of the ASN1_ITEM structures themselves. This adds several
7325      new macros which allow the underlying ASN1 function/structure to
7326      be accessed transparently. As a result code should not use ASN1_ITEM
7327      references directly (such as &X509_it) but instead use the relevant
7328      macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
7329      use of the new ASN1 code on platforms where exporting structures
7330      is problematical (for example in shared libraries) but exporting
7331      functions returning pointers to structures is not.
7332      [Steve Henson]
7333
7334   *) Add support for overriding the generation of SSL/TLS session IDs.
7335      These callbacks can be registered either in an SSL_CTX or per SSL.
7336      The purpose of this is to allow applications to control, if they wish,
7337      the arbitrary values chosen for use as session IDs, particularly as it
7338      can be useful for session caching in multiple-server environments. A
7339      command-line switch for testing this (and any client code that wishes
7340      to use such a feature) has been added to "s_server".
7341      [Geoff Thorpe, Lutz Jaenicke]
7342
7343   *) Modify mkdef.pl to recognise and parse preprocessor conditionals
7344      of the form '#if defined(...) || defined(...) || ...' and
7345      '#if !defined(...) && !defined(...) && ...'.  This also avoids
7346      the growing number of special cases it was previously handling.
7347      [Richard Levitte]
7348
7349   *) Make all configuration macros available for application by making
7350      sure they are available in opensslconf.h, by giving them names starting
7351      with "OPENSSL_" to avoid conflicts with other packages and by making
7352      sure e_os2.h will cover all platform-specific cases together with
7353      opensslconf.h.
7354      Additionally, it is now possible to define configuration/platform-
7355      specific names (called "system identities").  In the C code, these
7356      are prefixed with "OPENSSL_SYSNAME_".  e_os2.h will create another
7357      macro with the name beginning with "OPENSSL_SYS_", which is determined
7358      from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
7359      what is available.
7360      [Richard Levitte]
7361
7362   *) New option -set_serial to 'req' and 'x509' this allows the serial
7363      number to use to be specified on the command line. Previously self
7364      signed certificates were hard coded with serial number 0 and the
7365      CA options of 'x509' had to use a serial number in a file which was
7366      auto incremented.
7367      [Steve Henson]
7368
7369   *) New options to 'ca' utility to support V2 CRL entry extensions.
7370      Currently CRL reason, invalidity date and hold instruction are
7371      supported. Add new CRL extensions to V3 code and some new objects.
7372      [Steve Henson]
7373
7374   *) New function EVP_CIPHER_CTX_set_padding() this is used to
7375      disable standard block padding (aka PKCS#5 padding) in the EVP
7376      API, which was previously mandatory. This means that the data is
7377      not padded in any way and so the total length much be a multiple
7378      of the block size, otherwise an error occurs.
7379      [Steve Henson]
7380
7381   *) Initial (incomplete) OCSP SSL support.
7382      [Steve Henson]
7383
7384   *) New function OCSP_parse_url(). This splits up a URL into its host,
7385      port and path components: primarily to parse OCSP URLs. New -url
7386      option to ocsp utility.
7387      [Steve Henson]
7388
7389   *) New nonce behavior. The return value of OCSP_check_nonce() now
7390      reflects the various checks performed. Applications can decide
7391      whether to tolerate certain situations such as an absent nonce
7392      in a response when one was present in a request: the ocsp application
7393      just prints out a warning. New function OCSP_add1_basic_nonce()
7394      this is to allow responders to include a nonce in a response even if
7395      the request is nonce-less.
7396      [Steve Henson]
7397
7398   *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
7399      skipped when using openssl x509 multiple times on a single input file,
7400      e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
7401      [Bodo Moeller]
7402
7403   *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
7404      set string type: to handle setting ASN1_TIME structures. Fix ca
7405      utility to correctly initialize revocation date of CRLs.
7406      [Steve Henson]
7407
7408   *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
7409      the clients preferred ciphersuites and rather use its own preferences.
7410      Should help to work around M$ SGC (Server Gated Cryptography) bug in
7411      Internet Explorer by ensuring unchanged hash method during stepup.
7412      (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
7413      [Lutz Jaenicke]
7414
7415   *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
7416      to aes and add a new 'exist' option to print out symbols that don't
7417      appear to exist.
7418      [Steve Henson]
7419
7420   *) Additional options to ocsp utility to allow flags to be set and
7421      additional certificates supplied.
7422      [Steve Henson]
7423
7424   *) Add the option -VAfile to 'openssl ocsp', so the user can give the
7425      OCSP client a number of certificate to only verify the response
7426      signature against.
7427      [Richard Levitte]
7428
7429   *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
7430      handle the new API. Currently only ECB, CBC modes supported. Add new
7431      AES OIDs.
7432
7433      Add TLS AES ciphersuites as described in RFC3268, "Advanced
7434      Encryption Standard (AES) Ciphersuites for Transport Layer
7435      Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
7436      not enabled by default and were not part of the "ALL" ciphersuite
7437      alias because they were not yet official; they could be
7438      explicitly requested by specifying the "AESdraft" ciphersuite
7439      group alias.  In the final release of OpenSSL 0.9.7, the group
7440      alias is called "AES" and is part of "ALL".)
7441      [Ben Laurie, Steve  Henson, Bodo Moeller]
7442
7443   *) New function OCSP_copy_nonce() to copy nonce value (if present) from
7444      request to response.
7445      [Steve Henson]
7446
7447   *) Functions for OCSP responders. OCSP_request_onereq_count(),
7448      OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
7449      extract information from a certificate request. OCSP_response_create()
7450      creates a response and optionally adds a basic response structure.
7451      OCSP_basic_add1_status() adds a complete single response to a basic
7452      response and returns the OCSP_SINGLERESP structure just added (to allow
7453      extensions to be included for example). OCSP_basic_add1_cert() adds a
7454      certificate to a basic response and OCSP_basic_sign() signs a basic
7455      response with various flags. New helper functions ASN1_TIME_check()
7456      (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
7457      (converts ASN1_TIME to GeneralizedTime).
7458      [Steve Henson]
7459
7460   *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
7461      in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
7462      structure from a certificate. X509_pubkey_digest() digests the public_key
7463      contents: this is used in various key identifiers.
7464      [Steve Henson]
7465
7466   *) Make sk_sort() tolerate a NULL argument.
7467      [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
7468
7469   *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
7470      passed by the function are trusted implicitly. If any of them signed the
7471      response then it is assumed to be valid and is not verified.
7472      [Steve Henson]
7473
7474   *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
7475      to data. This was previously part of the PKCS7 ASN1 code. This
7476      was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
7477      [Steve Henson, reported by Kenneth R. Robinette
7478                                 <support@securenetterm.com>]
7479
7480   *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
7481      routines: without these tracing memory leaks is very painful.
7482      Fix leaks in PKCS12 and PKCS7 routines.
7483      [Steve Henson]
7484
7485   *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
7486      Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
7487      effectively meant GeneralizedTime would never be used. Now it
7488      is initialised to -1 but X509_time_adj() now has to check the value
7489      and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
7490      V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
7491      [Steve Henson, reported by Kenneth R. Robinette
7492                                 <support@securenetterm.com>]
7493
7494   *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
7495      result in a zero length in the ASN1_INTEGER structure which was
7496      not consistent with the structure when d2i_ASN1_INTEGER() was used
7497      and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
7498      to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
7499      where it did not print out a minus for negative ASN1_INTEGER.
7500      [Steve Henson]
7501
7502   *) Add summary printout to ocsp utility. The various functions which
7503      convert status values to strings have been renamed to:
7504      OCSP_response_status_str(), OCSP_cert_status_str() and
7505      OCSP_crl_reason_str() and are no longer static. New options
7506      to verify nonce values and to disable verification. OCSP response
7507      printout format cleaned up.
7508      [Steve Henson]
7509
7510   *) Add additional OCSP certificate checks. These are those specified
7511      in RFC2560. This consists of two separate checks: the CA of the
7512      certificate being checked must either be the OCSP signer certificate
7513      or the issuer of the OCSP signer certificate. In the latter case the
7514      OCSP signer certificate must contain the OCSP signing extended key
7515      usage. This check is performed by attempting to match the OCSP
7516      signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
7517      in the OCSP_CERTID structures of the response.
7518      [Steve Henson]
7519
7520   *) Initial OCSP certificate verification added to OCSP_basic_verify()
7521      and related routines. This uses the standard OpenSSL certificate
7522      verify routines to perform initial checks (just CA validity) and
7523      to obtain the certificate chain. Then additional checks will be
7524      performed on the chain. Currently the root CA is checked to see
7525      if it is explicitly trusted for OCSP signing. This is used to set
7526      a root CA as a global signing root: that is any certificate that
7527      chains to that CA is an acceptable OCSP signing certificate.
7528      [Steve Henson]
7529
7530   *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
7531      extensions from a separate configuration file.
7532      As when reading extensions from the main configuration file,
7533      the '-extensions ...' option may be used for specifying the
7534      section to use.
7535      [Massimiliano Pala <madwolf@comune.modena.it>]
7536
7537   *) New OCSP utility. Allows OCSP requests to be generated or
7538      read. The request can be sent to a responder and the output
7539      parsed, outputed or printed in text form. Not complete yet:
7540      still needs to check the OCSP response validity.
7541      [Steve Henson]
7542
7543   *) New subcommands for 'openssl ca':
7544      'openssl ca -status <serial>' prints the status of the cert with
7545      the given serial number (according to the index file).
7546      'openssl ca -updatedb' updates the expiry status of certificates
7547      in the index file.
7548      [Massimiliano Pala <madwolf@comune.modena.it>]
7549
7550   *) New '-newreq-nodes' command option to CA.pl.  This is like
7551      '-newreq', but calls 'openssl req' with the '-nodes' option
7552      so that the resulting key is not encrypted.
7553      [Damien Miller <djm@mindrot.org>]
7554
7555   *) New configuration for the GNU Hurd.
7556      [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
7557
7558   *) Initial code to implement OCSP basic response verify. This
7559      is currently incomplete. Currently just finds the signer's
7560      certificate and verifies the signature on the response.
7561      [Steve Henson]
7562
7563   *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
7564      value of OPENSSLDIR.  This is available via the new '-d' option
7565      to 'openssl version', and is also included in 'openssl version -a'.
7566      [Bodo Moeller]
7567
7568   *) Allowing defining memory allocation callbacks that will be given
7569      file name and line number information in additional arguments
7570      (a const char* and an int).  The basic functionality remains, as
7571      well as the original possibility to just replace malloc(),
7572      realloc() and free() by functions that do not know about these
7573      additional arguments.  To register and find out the current
7574      settings for extended allocation functions, the following
7575      functions are provided:
7576
7577         CRYPTO_set_mem_ex_functions
7578         CRYPTO_set_locked_mem_ex_functions
7579         CRYPTO_get_mem_ex_functions
7580         CRYPTO_get_locked_mem_ex_functions
7581
7582      These work the same way as CRYPTO_set_mem_functions and friends.
7583      CRYPTO_get_[locked_]mem_functions now writes 0 where such an
7584      extended allocation function is enabled.
7585      Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
7586      a conventional allocation function is enabled.
7587      [Richard Levitte, Bodo Moeller]
7588
7589   *) Finish off removing the remaining LHASH function pointer casts.
7590      There should no longer be any prototype-casting required when using
7591      the LHASH abstraction, and any casts that remain are "bugs". See
7592      the callback types and macros at the head of lhash.h for details
7593      (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
7594      [Geoff Thorpe]
7595
7596   *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
7597      If /dev/[u]random devices are not available or do not return enough
7598      entropy, EGD style sockets (served by EGD or PRNGD) will automatically
7599      be queried.
7600      The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
7601      /etc/entropy will be queried once each in this sequence, querying stops
7602      when enough entropy was collected without querying more sockets.
7603      [Lutz Jaenicke]
7604
7605   *) Change the Unix RAND_poll() variant to be able to poll several
7606      random devices, as specified by DEVRANDOM, until a sufficient amount
7607      of data has been collected.   We spend at most 10 ms on each file
7608      (select timeout) and read in non-blocking mode.  DEVRANDOM now
7609      defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
7610      (previously it was just the string "/dev/urandom"), so on typical
7611      platforms the 10 ms delay will never occur.
7612      Also separate out the Unix variant to its own file, rand_unix.c.
7613      For VMS, there's a currently-empty rand_vms.c.
7614      [Richard Levitte]
7615
7616   *) Move OCSP client related routines to ocsp_cl.c. These
7617      provide utility functions which an application needing
7618      to issue a request to an OCSP responder and analyse the
7619      response will typically need: as opposed to those which an
7620      OCSP responder itself would need which will be added later.
7621
7622      OCSP_request_sign() signs an OCSP request with an API similar
7623      to PKCS7_sign(). OCSP_response_status() returns status of OCSP
7624      response. OCSP_response_get1_basic() extracts basic response
7625      from response. OCSP_resp_find_status(): finds and extracts status
7626      information from an OCSP_CERTID structure (which will be created
7627      when the request structure is built). These are built from lower
7628      level functions which work on OCSP_SINGLERESP structures but
7629      won't normally be used unless the application wishes to examine
7630      extensions in the OCSP response for example.
7631
7632      Replace nonce routines with a pair of functions.
7633      OCSP_request_add1_nonce() adds a nonce value and optionally
7634      generates a random value. OCSP_check_nonce() checks the
7635      validity of the nonce in an OCSP response.
7636      [Steve Henson]
7637
7638   *) Change function OCSP_request_add() to OCSP_request_add0_id().
7639      This doesn't copy the supplied OCSP_CERTID and avoids the
7640      need to free up the newly created id. Change return type
7641      to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
7642      This can then be used to add extensions to the request.
7643      Deleted OCSP_request_new(), since most of its functionality
7644      is now in OCSP_REQUEST_new() (and the case insensitive name
7645      clash) apart from the ability to set the request name which
7646      will be added elsewhere.
7647      [Steve Henson]
7648
7649   *) Update OCSP API. Remove obsolete extensions argument from
7650      various functions. Extensions are now handled using the new
7651      OCSP extension code. New simple OCSP HTTP function which
7652      can be used to send requests and parse the response.
7653      [Steve Henson]
7654
7655   *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
7656      ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
7657      uses the special reorder version of SET OF to sort the attributes
7658      and reorder them to match the encoded order. This resolves a long
7659      standing problem: a verify on a PKCS7 structure just after signing
7660      it used to fail because the attribute order did not match the
7661      encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
7662      it uses the received order. This is necessary to tolerate some broken
7663      software that does not order SET OF. This is handled by encoding
7664      as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
7665      to produce the required SET OF.
7666      [Steve Henson]
7667
7668   *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
7669      OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
7670      files to get correct declarations of the ASN.1 item variables.
7671      [Richard Levitte]
7672
7673   *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
7674      PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
7675      asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
7676      NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
7677      New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
7678      ASN1_ITEM and no wrapper functions.
7679      [Steve Henson]
7680
7681   *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
7682      replace the old function pointer based I/O routines. Change most of
7683      the *_d2i_bio() and *_d2i_fp() functions to use these.
7684      [Steve Henson]
7685
7686   *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
7687      lines, recognice more "algorithms" that can be deselected, and make
7688      it complain about algorithm deselection that isn't recognised.
7689      [Richard Levitte]
7690
7691   *) New ASN1 functions to handle dup, sign, verify, digest, pack and
7692      unpack operations in terms of ASN1_ITEM. Modify existing wrappers
7693      to use new functions. Add NO_ASN1_OLD which can be set to remove
7694      some old style ASN1 functions: this can be used to determine if old
7695      code will still work when these eventually go away.
7696      [Steve Henson]
7697
7698   *) New extension functions for OCSP structures, these follow the
7699      same conventions as certificates and CRLs.
7700      [Steve Henson]
7701
7702   *) New function X509V3_add1_i2d(). This automatically encodes and
7703      adds an extension. Its behaviour can be customised with various
7704      flags to append, replace or delete. Various wrappers added for
7705      certificates and CRLs.
7706      [Steve Henson]
7707
7708   *) Fix to avoid calling the underlying ASN1 print routine when
7709      an extension cannot be parsed. Correct a typo in the
7710      OCSP_SERVICELOC extension. Tidy up print OCSP format.
7711      [Steve Henson]
7712
7713   *) Make mkdef.pl parse some of the ASN1 macros and add appropriate
7714      entries for variables.
7715      [Steve Henson]
7716
7717   *) Add functionality to apps/openssl.c for detecting locking
7718      problems: As the program is single-threaded, all we have
7719      to do is register a locking callback using an array for
7720      storing which locks are currently held by the program.
7721      [Bodo Moeller]
7722
7723   *) Use a lock around the call to CRYPTO_get_ex_new_index() in
7724      SSL_get_ex_data_X509_STORE_idx(), which is used in
7725      ssl_verify_cert_chain() and thus can be called at any time
7726      during TLS/SSL handshakes so that thread-safety is essential.
7727      Unfortunately, the ex_data design is not at all suited
7728      for multi-threaded use, so it probably should be abolished.
7729      [Bodo Moeller]
7730
7731   *) Added Broadcom "ubsec" ENGINE to OpenSSL.
7732      [Broadcom, tweaked and integrated by Geoff Thorpe]
7733
7734   *) Move common extension printing code to new function
7735      X509V3_print_extensions(). Reorganise OCSP print routines and
7736      implement some needed OCSP ASN1 functions. Add OCSP extensions.
7737      [Steve Henson]
7738
7739   *) New function X509_signature_print() to remove duplication in some
7740      print routines.
7741      [Steve Henson]
7742
7743   *) Add a special meaning when SET OF and SEQUENCE OF flags are both
7744      set (this was treated exactly the same as SET OF previously). This
7745      is used to reorder the STACK representing the structure to match the
7746      encoding. This will be used to get round a problem where a PKCS7
7747      structure which was signed could not be verified because the STACK
7748      order did not reflect the encoded order.
7749      [Steve Henson]
7750
7751   *) Reimplement the OCSP ASN1 module using the new code.
7752      [Steve Henson]
7753
7754   *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
7755      for its ASN1 operations. The old style function pointers still exist
7756      for now but they will eventually go away.
7757      [Steve Henson]
7758
7759   *) Merge in replacement ASN1 code from the ASN1 branch. This almost
7760      completely replaces the old ASN1 functionality with a table driven
7761      encoder and decoder which interprets an ASN1_ITEM structure describing
7762      the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
7763      largely maintained. Almost all of the old asn1_mac.h macro based ASN1
7764      has also been converted to the new form.
7765      [Steve Henson]
7766
7767   *) Change BN_mod_exp_recp so that negative moduli are tolerated
7768      (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
7769      so that BN_mod_exp_mont and BN_mod_exp_mont_word work
7770      for negative moduli.
7771      [Bodo Moeller]
7772
7773   *) Fix BN_uadd and BN_usub: Always return non-negative results instead
7774      of not touching the result's sign bit.
7775      [Bodo Moeller]
7776
7777   *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
7778      set.
7779      [Bodo Moeller]
7780
7781   *) Changed the LHASH code to use prototypes for callbacks, and created
7782      macros to declare and implement thin (optionally static) functions
7783      that provide type-safety and avoid function pointer casting for the
7784      type-specific callbacks.
7785      [Geoff Thorpe]
7786
7787   *) Added Kerberos Cipher Suites to be used with TLS, as written in
7788      RFC 2712.
7789      [Veers Staats <staatsvr@asc.hpc.mil>,
7790       Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
7791
7792   *) Reformat the FAQ so the different questions and answers can be divided
7793      in sections depending on the subject.
7794      [Richard Levitte]
7795
7796   *) Have the zlib compression code load ZLIB.DLL dynamically under
7797      Windows.
7798      [Richard Levitte]
7799
7800   *) New function BN_mod_sqrt for computing square roots modulo a prime
7801      (using the probabilistic Tonelli-Shanks algorithm unless
7802      p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
7803      be handled deterministically).
7804      [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
7805
7806   *) Make BN_mod_inverse faster by explicitly handling small quotients
7807      in the Euclid loop. (Speed gain about 20% for small moduli [256 or
7808      512 bits], about 30% for larger ones [1024 or 2048 bits].)
7809      [Bodo Moeller]
7810
7811   *) New function BN_kronecker.
7812      [Bodo Moeller]
7813
7814   *) Fix BN_gcd so that it works on negative inputs; the result is
7815      positive unless both parameters are zero.
7816      Previously something reasonably close to an infinite loop was
7817      possible because numbers could be growing instead of shrinking
7818      in the implementation of Euclid's algorithm.
7819      [Bodo Moeller]
7820
7821   *) Fix BN_is_word() and BN_is_one() macros to take into account the
7822      sign of the number in question.
7823
7824      Fix BN_is_word(a,w) to work correctly for w == 0.
7825
7826      The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
7827      because its test if the absolute value of 'a' equals 'w'.
7828      Note that BN_abs_is_word does *not* handle w == 0 reliably;
7829      it exists mostly for use in the implementations of BN_is_zero(),
7830      BN_is_one(), and BN_is_word().
7831      [Bodo Moeller]
7832
7833   *) New function BN_swap.
7834      [Bodo Moeller]
7835
7836   *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
7837      the exponentiation functions are more likely to produce reasonable
7838      results on negative inputs.
7839      [Bodo Moeller]
7840
7841   *) Change BN_mod_mul so that the result is always non-negative.
7842      Previously, it could be negative if one of the factors was negative;
7843      I don't think anyone really wanted that behaviour.
7844      [Bodo Moeller]
7845
7846   *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
7847      (except for exponentiation, which stays in crypto/bn/bn_exp.c,
7848      and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
7849      and add new functions:
7850
7851           BN_nnmod
7852           BN_mod_sqr
7853           BN_mod_add
7854           BN_mod_add_quick
7855           BN_mod_sub
7856           BN_mod_sub_quick
7857           BN_mod_lshift1
7858           BN_mod_lshift1_quick
7859           BN_mod_lshift
7860           BN_mod_lshift_quick
7861
7862      These functions always generate non-negative results.
7863
7864      BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder  r
7865      such that  |m| < r < 0,  BN_nnmod will output  rem + |m|  instead).
7866
7867      BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
7868      BN_mod_XXX(r, a, [b,] m, ctx), but requires that  a  [and  b]
7869      be reduced modulo  m.
7870      [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
7871
7872 #if 0
7873      The following entry accidentally appeared in the CHANGES file
7874      distributed with OpenSSL 0.9.7.  The modifications described in
7875      it do *not* apply to OpenSSL 0.9.7.
7876
7877   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
7878      was actually never needed) and in BN_mul().  The removal in BN_mul()
7879      required a small change in bn_mul_part_recursive() and the addition
7880      of the functions bn_cmp_part_words(), bn_sub_part_words() and
7881      bn_add_part_words(), which do the same thing as bn_cmp_words(),
7882      bn_sub_words() and bn_add_words() except they take arrays with
7883      differing sizes.
7884      [Richard Levitte]
7885 #endif
7886
7887   *) In 'openssl passwd', verify passwords read from the terminal
7888      unless the '-salt' option is used (which usually means that
7889      verification would just waste user's time since the resulting
7890      hash is going to be compared with some given password hash)
7891      or the new '-noverify' option is used.
7892
7893      This is an incompatible change, but it does not affect
7894      non-interactive use of 'openssl passwd' (passwords on the command
7895      line, '-stdin' option, '-in ...' option) and thus should not
7896      cause any problems.
7897      [Bodo Moeller]
7898
7899   *) Remove all references to RSAref, since there's no more need for it.
7900      [Richard Levitte]
7901
7902   *) Make DSO load along a path given through an environment variable
7903      (SHLIB_PATH) with shl_load().
7904      [Richard Levitte]
7905
7906   *) Constify the ENGINE code as a result of BIGNUM constification.
7907      Also constify the RSA code and most things related to it.  In a
7908      few places, most notable in the depth of the ASN.1 code, ugly
7909      casts back to non-const were required (to be solved at a later
7910      time)
7911      [Richard Levitte]
7912
7913   *) Make it so the openssl application has all engines loaded by default.
7914      [Richard Levitte]
7915
7916   *) Constify the BIGNUM routines a little more.
7917      [Richard Levitte]
7918
7919   *) Add the following functions:
7920
7921         ENGINE_load_cswift()
7922         ENGINE_load_chil()
7923         ENGINE_load_atalla()
7924         ENGINE_load_nuron()
7925         ENGINE_load_builtin_engines()
7926
7927      That way, an application can itself choose if external engines that
7928      are built-in in OpenSSL shall ever be used or not.  The benefit is
7929      that applications won't have to be linked with libdl or other dso
7930      libraries unless it's really needed.
7931
7932      Changed 'openssl engine' to load all engines on demand.
7933      Changed the engine header files to avoid the duplication of some
7934      declarations (they differed!).
7935      [Richard Levitte]
7936
7937   *) 'openssl engine' can now list capabilities.
7938      [Richard Levitte]
7939
7940   *) Better error reporting in 'openssl engine'.
7941      [Richard Levitte]
7942
7943   *) Never call load_dh_param(NULL) in s_server.
7944      [Bodo Moeller]
7945
7946   *) Add engine application.  It can currently list engines by name and
7947      identity, and test if they are actually available.
7948      [Richard Levitte]
7949
7950   *) Improve RPM specification file by forcing symbolic linking and making
7951      sure the installed documentation is also owned by root.root.
7952      [Damien Miller <djm@mindrot.org>]
7953
7954   *) Give the OpenSSL applications more possibilities to make use of
7955      keys (public as well as private) handled by engines.
7956      [Richard Levitte]
7957
7958   *) Add OCSP code that comes from CertCo.
7959      [Richard Levitte]
7960
7961   *) Add VMS support for the Rijndael code.
7962      [Richard Levitte]
7963
7964   *) Added untested support for Nuron crypto accelerator.
7965      [Ben Laurie]
7966
7967   *) Add support for external cryptographic devices.  This code was
7968      previously distributed separately as the "engine" branch.
7969      [Geoff Thorpe, Richard Levitte]
7970
7971   *) Rework the filename-translation in the DSO code. It is now possible to
7972      have far greater control over how a "name" is turned into a filename
7973      depending on the operating environment and any oddities about the
7974      different shared library filenames on each system.
7975      [Geoff Thorpe]
7976
7977   *) Support threads on FreeBSD-elf in Configure.
7978      [Richard Levitte]
7979
7980   *) Fix for SHA1 assembly problem with MASM: it produces
7981      warnings about corrupt line number information when assembling
7982      with debugging information. This is caused by the overlapping
7983      of two sections.
7984      [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
7985
7986   *) NCONF changes.
7987      NCONF_get_number() has no error checking at all.  As a replacement,
7988      NCONF_get_number_e() is defined (_e for "error checking") and is
7989      promoted strongly.  The old NCONF_get_number is kept around for
7990      binary backward compatibility.
7991      Make it possible for methods to load from something other than a BIO,
7992      by providing a function pointer that is given a name instead of a BIO.
7993      For example, this could be used to load configuration data from an
7994      LDAP server.
7995      [Richard Levitte]
7996
7997   *) Fix for non blocking accept BIOs. Added new I/O special reason
7998      BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
7999      with non blocking I/O was not possible because no retry code was
8000      implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
8001      this case.
8002      [Steve Henson]
8003
8004   *) Added the beginnings of Rijndael support.
8005      [Ben Laurie]
8006
8007   *) Fix for bug in DirectoryString mask setting. Add support for
8008      X509_NAME_print_ex() in 'req' and X509_print_ex() function
8009      to allow certificate printing to more controllable, additional
8010      'certopt' option to 'x509' to allow new printing options to be
8011      set.
8012      [Steve Henson]
8013
8014   *) Clean old EAY MD5 hack from e_os.h.
8015      [Richard Levitte]
8016
8017  Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
8018
8019   *) Fix null-pointer assignment in do_change_cipher_spec() revealed
8020      by using the Codenomicon TLS Test Tool (CVE-2004-0079)
8021      [Joe Orton, Steve Henson]
8022
8023  Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
8024
8025   *) Fix additional bug revealed by the NISCC test suite:
8026
8027      Stop bug triggering large recursion when presented with
8028      certain ASN.1 tags (CVE-2003-0851)
8029      [Steve Henson]
8030
8031  Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
8032
8033   *) Fix various bugs revealed by running the NISCC test suite:
8034
8035      Stop out of bounds reads in the ASN1 code when presented with
8036      invalid tags (CVE-2003-0543 and CVE-2003-0544).
8037
8038      If verify callback ignores invalid public key errors don't try to check
8039      certificate signature with the NULL public key.
8040
8041      [Steve Henson]
8042
8043   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
8044      if the server requested one: as stated in TLS 1.0 and SSL 3.0
8045      specifications.
8046      [Steve Henson]
8047
8048   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
8049      extra data after the compression methods not only for TLS 1.0
8050      but also for SSL 3.0 (as required by the specification).
8051      [Bodo Moeller; problem pointed out by Matthias Loepfe]
8052
8053   *) Change X509_certificate_type() to mark the key as exported/exportable
8054      when it's 512 *bits* long, not 512 bytes.
8055      [Richard Levitte]
8056
8057  Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
8058
8059   *) Countermeasure against the Klima-Pokorny-Rosa extension of
8060      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
8061      a protocol version number mismatch like a decryption error
8062      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
8063      [Bodo Moeller]
8064
8065   *) Turn on RSA blinding by default in the default implementation
8066      to avoid a timing attack. Applications that don't want it can call
8067      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
8068      They would be ill-advised to do so in most cases.
8069      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
8070
8071   *) Change RSA blinding code so that it works when the PRNG is not
8072      seeded (in this case, the secret RSA exponent is abused as
8073      an unpredictable seed -- if it is not unpredictable, there
8074      is no point in blinding anyway).  Make RSA blinding thread-safe
8075      by remembering the creator's thread ID in rsa->blinding and
8076      having all other threads use local one-time blinding factors
8077      (this requires more computation than sharing rsa->blinding, but
8078      avoids excessive locking; and if an RSA object is not shared
8079      between threads, blinding will still be very fast).
8080      [Bodo Moeller]
8081
8082  Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
8083
8084   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
8085      via timing by performing a MAC computation even if incorrrect
8086      block cipher padding has been found.  This is a countermeasure
8087      against active attacks where the attacker has to distinguish
8088      between bad padding and a MAC verification error. (CVE-2003-0078)
8089
8090      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
8091      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
8092      Martin Vuagnoux (EPFL, Ilion)]
8093
8094  Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
8095
8096   *) New function OPENSSL_cleanse(), which is used to cleanse a section of
8097      memory from it's contents.  This is done with a counter that will
8098      place alternating values in each byte.  This can be used to solve
8099      two issues: 1) the removal of calls to memset() by highly optimizing
8100      compilers, and 2) cleansing with other values than 0, since those can
8101      be read through on certain media, for example a swap space on disk.
8102      [Geoff Thorpe]
8103
8104   *) Bugfix: client side session caching did not work with external caching,
8105      because the session->cipher setting was not restored when reloading
8106      from the external cache. This problem was masked, when
8107      SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
8108      (Found by Steve Haslam <steve@araqnid.ddts.net>.)
8109      [Lutz Jaenicke]
8110
8111   *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
8112      length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
8113      [Zeev Lieber <zeev-l@yahoo.com>]
8114
8115   *) Undo an undocumented change introduced in 0.9.6e which caused
8116      repeated calls to OpenSSL_add_all_ciphers() and
8117      OpenSSL_add_all_digests() to be ignored, even after calling
8118      EVP_cleanup().
8119      [Richard Levitte]
8120
8121   *) Change the default configuration reader to deal with last line not
8122      being properly terminated.
8123      [Richard Levitte]
8124
8125   *) Change X509_NAME_cmp() so it applies the special rules on handling
8126      DN values that are of type PrintableString, as well as RDNs of type
8127      emailAddress where the value has the type ia5String.
8128      [stefank@valicert.com via Richard Levitte]
8129
8130   *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
8131      the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
8132      doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
8133      the bitwise-OR of the two for use by the majority of applications
8134      wanting this behaviour, and update the docs. The documented
8135      behaviour and actual behaviour were inconsistent and had been
8136      changing anyway, so this is more a bug-fix than a behavioural
8137      change.
8138      [Geoff Thorpe, diagnosed by Nadav Har'El]
8139
8140   *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
8141      (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
8142      [Bodo Moeller]
8143
8144   *) Fix initialization code race conditions in
8145         SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
8146         SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
8147         SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
8148         TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
8149         ssl2_get_cipher_by_char(),
8150         ssl3_get_cipher_by_char().
8151      [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
8152
8153   *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
8154      the cached sessions are flushed, as the remove_cb() might use ex_data
8155      contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
8156      (see [openssl.org #212]).
8157      [Geoff Thorpe, Lutz Jaenicke]
8158
8159   *) Fix typo in OBJ_txt2obj which incorrectly passed the content
8160      length, instead of the encoding length to d2i_ASN1_OBJECT.
8161      [Steve Henson]
8162
8163  Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
8164
8165   *) [In 0.9.6g-engine release:]
8166      Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
8167      [Lynn Gazis <lgazis@rainbow.com>]
8168
8169  Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
8170
8171   *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
8172      and get fix the header length calculation.
8173      [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
8174         Alon Kantor <alonk@checkpoint.com> (and others),
8175         Steve Henson]
8176
8177   *) Use proper error handling instead of 'assertions' in buffer
8178      overflow checks added in 0.9.6e.  This prevents DoS (the
8179      assertions could call abort()).
8180      [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
8181
8182  Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
8183
8184   *) Add various sanity checks to asn1_get_length() to reject
8185      the ASN1 length bytes if they exceed sizeof(long), will appear
8186      negative or the content length exceeds the length of the
8187      supplied buffer.
8188      [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
8189
8190   *) Fix cipher selection routines: ciphers without encryption had no flags
8191      for the cipher strength set and where therefore not handled correctly
8192      by the selection routines (PR #130).
8193      [Lutz Jaenicke]
8194
8195   *) Fix EVP_dsa_sha macro.
8196      [Nils Larsch]
8197
8198   *) New option
8199           SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
8200      for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
8201      that was added in OpenSSL 0.9.6d.
8202
8203      As the countermeasure turned out to be incompatible with some
8204      broken SSL implementations, the new option is part of SSL_OP_ALL.
8205      SSL_OP_ALL is usually employed when compatibility with weird SSL
8206      implementations is desired (e.g. '-bugs' option to 's_client' and
8207      's_server'), so the new option is automatically set in many
8208      applications.
8209      [Bodo Moeller]
8210
8211   *) Changes in security patch:
8212
8213      Changes marked "(CHATS)" were sponsored by the Defense Advanced
8214      Research Projects Agency (DARPA) and Air Force Research Laboratory,
8215      Air Force Materiel Command, USAF, under agreement number
8216      F30602-01-2-0537.
8217
8218   *) Add various sanity checks to asn1_get_length() to reject
8219      the ASN1 length bytes if they exceed sizeof(long), will appear
8220      negative or the content length exceeds the length of the
8221      supplied buffer. (CVE-2002-0659)
8222      [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
8223
8224   *) Assertions for various potential buffer overflows, not known to
8225      happen in practice.
8226      [Ben Laurie (CHATS)]
8227
8228   *) Various temporary buffers to hold ASCII versions of integers were
8229      too small for 64 bit platforms. (CVE-2002-0655)
8230      [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
8231
8232   *) Remote buffer overflow in SSL3 protocol - an attacker could
8233      supply an oversized session ID to a client. (CVE-2002-0656)
8234      [Ben Laurie (CHATS)]
8235
8236   *) Remote buffer overflow in SSL2 protocol - an attacker could
8237      supply an oversized client master key. (CVE-2002-0656)
8238      [Ben Laurie (CHATS)]
8239
8240  Changes between 0.9.6c and 0.9.6d  [9 May 2002]
8241
8242   *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
8243      encoded as NULL) with id-dsa-with-sha1.
8244      [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
8245
8246   *) Check various X509_...() return values in apps/req.c.
8247      [Nils Larsch <nla@trustcenter.de>]
8248
8249   *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
8250      an end-of-file condition would erroneously be flagged, when the CRLF
8251      was just at the end of a processed block. The bug was discovered when
8252      processing data through a buffering memory BIO handing the data to a
8253      BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
8254      <ptsekov@syntrex.com> and Nedelcho Stanev.
8255      [Lutz Jaenicke]
8256
8257   *) Implement a countermeasure against a vulnerability recently found
8258      in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
8259      before application data chunks to avoid the use of known IVs
8260      with data potentially chosen by the attacker.
8261      [Bodo Moeller]
8262
8263   *) Fix length checks in ssl3_get_client_hello().
8264      [Bodo Moeller]
8265
8266   *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
8267      to prevent ssl3_read_internal() from incorrectly assuming that
8268      ssl3_read_bytes() found application data while handshake
8269      processing was enabled when in fact s->s3->in_read_app_data was
8270      merely automatically cleared during the initial handshake.
8271      [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
8272
8273   *) Fix object definitions for Private and Enterprise: they were not
8274      recognized in their shortname (=lowercase) representation. Extend
8275      obj_dat.pl to issue an error when using undefined keywords instead
8276      of silently ignoring the problem (Svenning Sorensen
8277      <sss@sss.dnsalias.net>).
8278      [Lutz Jaenicke]
8279
8280   *) Fix DH_generate_parameters() so that it works for 'non-standard'
8281      generators, i.e. generators other than 2 and 5.  (Previously, the
8282      code did not properly initialise the 'add' and 'rem' values to
8283      BN_generate_prime().)
8284
8285      In the new general case, we do not insist that 'generator' is
8286      actually a primitive root: This requirement is rather pointless;
8287      a generator of the order-q subgroup is just as good, if not
8288      better.
8289      [Bodo Moeller]
8290
8291   *) Map new X509 verification errors to alerts. Discovered and submitted by
8292      Tom Wu <tom@arcot.com>.
8293      [Lutz Jaenicke]
8294
8295   *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
8296      returning non-zero before the data has been completely received
8297      when using non-blocking I/O.
8298      [Bodo Moeller; problem pointed out by John Hughes]
8299
8300   *) Some of the ciphers missed the strength entry (SSL_LOW etc).
8301      [Ben Laurie, Lutz Jaenicke]
8302
8303   *) Fix bug in SSL_clear(): bad sessions were not removed (found by
8304      Yoram Zahavi <YoramZ@gilian.com>).
8305      [Lutz Jaenicke]
8306
8307   *) Add information about CygWin 1.3 and on, and preserve proper
8308      configuration for the versions before that.
8309      [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
8310
8311   *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
8312      check whether we deal with a copy of a session and do not delete from
8313      the cache in this case. Problem reported by "Izhar Shoshani Levi"
8314      <izhar@checkpoint.com>.
8315      [Lutz Jaenicke]
8316
8317   *) Do not store session data into the internal session cache, if it
8318      is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
8319      flag is set). Proposed by Aslam <aslam@funk.com>.
8320      [Lutz Jaenicke]
8321
8322   *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
8323      value is 0.
8324      [Richard Levitte]
8325
8326   *) [In 0.9.6d-engine release:]
8327      Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
8328      [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
8329
8330   *) Add the configuration target linux-s390x.
8331      [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
8332
8333   *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
8334      ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
8335      variable as an indication that a ClientHello message has been
8336      received.  As the flag value will be lost between multiple
8337      invocations of ssl3_accept when using non-blocking I/O, the
8338      function may not be aware that a handshake has actually taken
8339      place, thus preventing a new session from being added to the
8340      session cache.
8341
8342      To avoid this problem, we now set s->new_session to 2 instead of
8343      using a local variable.
8344      [Lutz Jaenicke, Bodo Moeller]
8345
8346   *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
8347      if the SSL_R_LENGTH_MISMATCH error is detected.
8348      [Geoff Thorpe, Bodo Moeller]
8349
8350   *) New 'shared_ldflag' column in Configure platform table.
8351      [Richard Levitte]
8352
8353   *) Fix EVP_CIPHER_mode macro.
8354      ["Dan S. Camper" <dan@bti.net>]
8355
8356   *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
8357      type, we must throw them away by setting rr->length to 0.
8358      [D P Chang <dpc@qualys.com>]
8359
8360  Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
8361
8362   *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
8363      <Dominikus.Scherkl@biodata.com>.  (The previous implementation
8364      worked incorrectly for those cases where  range = 10..._2  and
8365      3*range  is two bits longer than  range.)
8366      [Bodo Moeller]
8367
8368   *) Only add signing time to PKCS7 structures if it is not already
8369      present.
8370      [Steve Henson]
8371
8372   *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
8373      OBJ_ld_ce should be OBJ_id_ce.
8374      Also some ip-pda OIDs in crypto/objects/objects.txt were
8375      incorrect (cf. RFC 3039).
8376      [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
8377
8378   *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
8379      returns early because it has nothing to do.
8380      [Andy Schneider <andy.schneider@bjss.co.uk>]
8381
8382   *) [In 0.9.6c-engine release:]
8383      Fix mutex callback return values in crypto/engine/hw_ncipher.c.
8384      [Andy Schneider <andy.schneider@bjss.co.uk>]
8385
8386   *) [In 0.9.6c-engine release:]
8387      Add support for Cryptographic Appliance's keyserver technology.
8388      (Use engine 'keyclient')
8389      [Cryptographic Appliances and Geoff Thorpe]
8390
8391   *) Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
8392      is called via tools/c89.sh because arguments have to be
8393      rearranged (all '-L' options must appear before the first object
8394      modules).
8395      [Richard Shapiro <rshapiro@abinitio.com>]
8396
8397   *) [In 0.9.6c-engine release:]
8398      Add support for Broadcom crypto accelerator cards, backported
8399      from 0.9.7.
8400      [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
8401
8402   *) [In 0.9.6c-engine release:]
8403      Add support for SureWare crypto accelerator cards from
8404      Baltimore Technologies.  (Use engine 'sureware')
8405      [Baltimore Technologies and Mark Cox]
8406
8407   *) [In 0.9.6c-engine release:]
8408      Add support for crypto accelerator cards from Accelerated
8409      Encryption Processing, www.aep.ie.  (Use engine 'aep')
8410      [AEP Inc. and Mark Cox]
8411
8412   *) Add a configuration entry for gcc on UnixWare.
8413      [Gary Benson <gbenson@redhat.com>]
8414
8415   *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
8416      messages are stored in a single piece (fixed-length part and
8417      variable-length part combined) and fix various bugs found on the way.
8418      [Bodo Moeller]
8419
8420   *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
8421      instead.  BIO_gethostbyname() does not know what timeouts are
8422      appropriate, so entries would stay in cache even when they have
8423      become invalid.
8424      [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
8425
8426   *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
8427      faced with a pathologically small ClientHello fragment that does
8428      not contain client_version: Instead of aborting with an error,
8429      simply choose the highest available protocol version (i.e.,
8430      TLS 1.0 unless it is disabled).  In practice, ClientHello
8431      messages are never sent like this, but this change gives us
8432      strictly correct behaviour at least for TLS.
8433      [Bodo Moeller]
8434
8435   *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
8436      never resets s->method to s->ctx->method when called from within
8437      one of the SSL handshake functions.
8438      [Bodo Moeller; problem pointed out by Niko Baric]
8439
8440   *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
8441      (sent using the client's version number) if client_version is
8442      smaller than the protocol version in use.  Also change
8443      ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
8444      the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
8445      the client will at least see that alert.
8446      [Bodo Moeller]
8447
8448   *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
8449      correctly.
8450      [Bodo Moeller]
8451
8452   *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
8453      client receives HelloRequest while in a handshake.
8454      [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
8455
8456   *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
8457      should end in 'break', not 'goto end' which circumvents various
8458      cleanups done in state SSL_ST_OK.   But session related stuff
8459      must be disabled for SSL_ST_OK in the case that we just sent a
8460      HelloRequest.
8461
8462      Also avoid some overhead by not calling ssl_init_wbio_buffer()
8463      before just sending a HelloRequest.
8464      [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
8465
8466   *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
8467      reveal whether illegal block cipher padding was found or a MAC
8468      verification error occurred.  (Neither SSLerr() codes nor alerts
8469      are directly visible to potential attackers, but the information
8470      may leak via logfiles.)
8471
8472      Similar changes are not required for the SSL 2.0 implementation
8473      because the number of padding bytes is sent in clear for SSL 2.0,
8474      and the extra bytes are just ignored.  However ssl/s2_pkt.c
8475      failed to verify that the purported number of padding bytes is in
8476      the legal range.
8477      [Bodo Moeller]
8478
8479   *) Add OpenUNIX-8 support including shared libraries
8480      (Boyd Lynn Gerber <gerberb@zenez.com>).
8481      [Lutz Jaenicke]
8482
8483   *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
8484      'wristwatch attack' using huge encoding parameters (cf.
8485      James H. Manger's CRYPTO 2001 paper).  Note that the
8486      RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
8487      encoding parameters and hence was not vulnerable.
8488      [Bodo Moeller]
8489
8490   *) BN_sqr() bug fix.
8491      [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
8492
8493   *) Rabin-Miller test analyses assume uniformly distributed witnesses,
8494      so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
8495      followed by modular reduction.
8496      [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
8497
8498   *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
8499      equivalent based on BN_pseudo_rand() instead of BN_rand().
8500      [Bodo Moeller]
8501
8502   *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
8503      This function was broken, as the check for a new client hello message
8504      to handle SGC did not allow these large messages.
8505      (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
8506      [Lutz Jaenicke]
8507
8508   *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
8509      [Lutz Jaenicke]
8510
8511   *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
8512      for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
8513      [Lutz Jaenicke]
8514
8515   *) Rework the configuration and shared library support for Tru64 Unix.
8516      The configuration part makes use of modern compiler features and
8517      still retains old compiler behavior for those that run older versions
8518      of the OS.  The shared library support part includes a variant that
8519      uses the RPATH feature, and is available through the special
8520      configuration target "alpha-cc-rpath", which will never be selected
8521      automatically.
8522      [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
8523
8524   *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
8525      with the same message size as in ssl3_get_certificate_request().
8526      Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
8527      messages might inadvertently be reject as too long.
8528      [Petr Lampa <lampa@fee.vutbr.cz>]
8529
8530   *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
8531      [Andy Polyakov]
8532
8533   *) Modified SSL library such that the verify_callback that has been set
8534      specificly for an SSL object with SSL_set_verify() is actually being
8535      used. Before the change, a verify_callback set with this function was
8536      ignored and the verify_callback() set in the SSL_CTX at the time of
8537      the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
8538      to allow the necessary settings.
8539      [Lutz Jaenicke]
8540
8541   *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
8542      explicitly to NULL, as at least on Solaris 8 this seems not always to be
8543      done automatically (in contradiction to the requirements of the C
8544      standard). This made problems when used from OpenSSH.
8545      [Lutz Jaenicke]
8546
8547   *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
8548      dh->length and always used
8549
8550           BN_rand_range(priv_key, dh->p).
8551
8552      BN_rand_range() is not necessary for Diffie-Hellman, and this
8553      specific range makes Diffie-Hellman unnecessarily inefficient if
8554      dh->length (recommended exponent length) is much smaller than the
8555      length of dh->p.  We could use BN_rand_range() if the order of
8556      the subgroup was stored in the DH structure, but we only have
8557      dh->length.
8558
8559      So switch back to
8560
8561           BN_rand(priv_key, l, ...)
8562
8563      where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
8564      otherwise.
8565      [Bodo Moeller]
8566
8567   *) In
8568
8569           RSA_eay_public_encrypt
8570           RSA_eay_private_decrypt
8571           RSA_eay_private_encrypt (signing)
8572           RSA_eay_public_decrypt (signature verification)
8573
8574      (default implementations for RSA_public_encrypt,
8575      RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
8576      always reject numbers >= n.
8577      [Bodo Moeller]
8578
8579   *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
8580      to synchronize access to 'locking_thread'.  This is necessary on
8581      systems where access to 'locking_thread' (an 'unsigned long'
8582      variable) is not atomic.
8583      [Bodo Moeller]
8584
8585   *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
8586      *before* setting the 'crypto_lock_rand' flag.  The previous code had
8587      a race condition if 0 is a valid thread ID.
8588      [Travis Vitek <vitek@roguewave.com>]
8589
8590   *) Add support for shared libraries under Irix.
8591      [Albert Chin-A-Young <china@thewrittenword.com>]
8592
8593   *) Add configuration option to build on Linux on both big-endian and
8594      little-endian MIPS.
8595      [Ralf Baechle <ralf@uni-koblenz.de>]
8596
8597   *) Add the possibility to create shared libraries on HP-UX.
8598      [Richard Levitte]
8599
8600  Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
8601
8602   *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
8603      to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
8604      Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
8605      PRNG state recovery was possible based on the output of
8606      one PRNG request appropriately sized to gain knowledge on
8607      'md' followed by enough consecutive 1-byte PRNG requests
8608      to traverse all of 'state'.
8609
8610      1. When updating 'md_local' (the current thread's copy of 'md')
8611         during PRNG output generation, hash all of the previous
8612         'md_local' value, not just the half used for PRNG output.
8613
8614      2. Make the number of bytes from 'state' included into the hash
8615         independent from the number of PRNG bytes requested.
8616
8617      The first measure alone would be sufficient to avoid
8618      Markku-Juhani's attack.  (Actually it had never occurred
8619      to me that the half of 'md_local' used for chaining was the
8620      half from which PRNG output bytes were taken -- I had always
8621      assumed that the secret half would be used.)  The second
8622      measure makes sure that additional data from 'state' is never
8623      mixed into 'md_local' in small portions; this heuristically
8624      further strengthens the PRNG.
8625      [Bodo Moeller]
8626
8627   *) Fix crypto/bn/asm/mips3.s.
8628      [Andy Polyakov]
8629
8630   *) When only the key is given to "enc", the IV is undefined. Print out
8631      an error message in this case.
8632      [Lutz Jaenicke]
8633
8634   *) Handle special case when X509_NAME is empty in X509 printing routines.
8635      [Steve Henson]
8636
8637   *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
8638      positive and less than q.
8639      [Bodo Moeller]
8640
8641   *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
8642      used: it isn't thread safe and the add_lock_callback should handle
8643      that itself.
8644      [Paul Rose <Paul.Rose@bridge.com>]
8645
8646   *) Verify that incoming data obeys the block size in
8647      ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
8648      [Bodo Moeller]
8649
8650   *) Fix OAEP check.
8651      [Ulf Möller, Bodo Möller]
8652
8653   *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
8654      RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
8655      when fixing the server behaviour for backwards-compatible 'client
8656      hello' messages.  (Note that the attack is impractical against
8657      SSL 3.0 and TLS 1.0 anyway because length and version checking
8658      means that the probability of guessing a valid ciphertext is
8659      around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
8660      paper.)
8661
8662      Before 0.9.5, the countermeasure (hide the error by generating a
8663      random 'decryption result') did not work properly because
8664      ERR_clear_error() was missing, meaning that SSL_get_error() would
8665      detect the supposedly ignored error.
8666
8667      Both problems are now fixed.
8668      [Bodo Moeller]
8669
8670   *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
8671      (previously it was 1024).
8672      [Bodo Moeller]
8673
8674   *) Fix for compatibility mode trust settings: ignore trust settings
8675      unless some valid trust or reject settings are present.
8676      [Steve Henson]
8677
8678   *) Fix for blowfish EVP: its a variable length cipher.
8679      [Steve Henson]
8680
8681   *) Fix various bugs related to DSA S/MIME verification. Handle missing
8682      parameters in DSA public key structures and return an error in the
8683      DSA routines if parameters are absent.
8684      [Steve Henson]
8685
8686   *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
8687      in the current directory if neither $RANDFILE nor $HOME was set.
8688      RAND_file_name() in 0.9.6a returned NULL in this case.  This has
8689      caused some confusion to Windows users who haven't defined $HOME.
8690      Thus RAND_file_name() is changed again: e_os.h can define a
8691      DEFAULT_HOME, which will be used if $HOME is not set.
8692      For Windows, we use "C:"; on other platforms, we still require
8693      environment variables.
8694
8695   *) Move 'if (!initialized) RAND_poll()' into regions protected by
8696      CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
8697      having multiple threads call RAND_poll() concurrently.
8698      [Bodo Moeller]
8699
8700   *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
8701      combination of a flag and a thread ID variable.
8702      Otherwise while one thread is in ssleay_rand_bytes (which sets the
8703      flag), *other* threads can enter ssleay_add_bytes without obeying
8704      the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
8705      that they do not hold after the first thread unsets add_do_not_lock).
8706      [Bodo Moeller]
8707
8708   *) Change bctest again: '-x' expressions are not available in all
8709      versions of 'test'.
8710      [Bodo Moeller]
8711
8712  Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
8713
8714   *) Fix a couple of memory leaks in PKCS7_dataDecode()
8715      [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
8716
8717   *) Change Configure and Makefiles to provide EXE_EXT, which will contain
8718      the default extension for executables, if any.  Also, make the perl
8719      scripts that use symlink() to test if it really exists and use "cp"
8720      if it doesn't.  All this made OpenSSL compilable and installable in
8721      CygWin.
8722      [Richard Levitte]
8723
8724   *) Fix for asn1_GetSequence() for indefinite length constructed data.
8725      If SEQUENCE is length is indefinite just set c->slen to the total
8726      amount of data available.
8727      [Steve Henson, reported by shige@FreeBSD.org]
8728      [This change does not apply to 0.9.7.]
8729
8730   *) Change bctest to avoid here-documents inside command substitution
8731      (workaround for FreeBSD /bin/sh bug).
8732      For compatibility with Ultrix, avoid shell functions (introduced
8733      in the bctest version that searches along $PATH).
8734      [Bodo Moeller]
8735
8736   *) Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
8737      with des_encrypt() defined on some operating systems, like Solaris
8738      and UnixWare.
8739      [Richard Levitte]
8740
8741   *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
8742      On the Importance of Eliminating Errors in Cryptographic
8743      Computations, J. Cryptology 14 (2001) 2, 101-119,
8744      http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
8745      [Ulf Moeller]
8746
8747   *) MIPS assembler BIGNUM division bug fix.
8748      [Andy Polyakov]
8749
8750   *) Disabled incorrect Alpha assembler code.
8751      [Richard Levitte]
8752
8753   *) Fix PKCS#7 decode routines so they correctly update the length
8754      after reading an EOC for the EXPLICIT tag.
8755      [Steve Henson]
8756      [This change does not apply to 0.9.7.]
8757
8758   *) Fix bug in PKCS#12 key generation routines. This was triggered
8759      if a 3DES key was generated with a 0 initial byte. Include
8760      PKCS12_BROKEN_KEYGEN compilation option to retain the old
8761      (but broken) behaviour.
8762      [Steve Henson]
8763
8764   *) Enhance bctest to search for a working bc along $PATH and print
8765      it when found.
8766      [Tim Rice <tim@multitalents.net> via Richard Levitte]
8767
8768   *) Fix memory leaks in err.c: free err_data string if necessary;
8769      don't write to the wrong index in ERR_set_error_data.
8770      [Bodo Moeller]
8771
8772   *) Implement ssl23_peek (analogous to ssl23_read), which previously
8773      did not exist.
8774      [Bodo Moeller]
8775
8776   *) Replace rdtsc with _emit statements for VC++ version 5.
8777      [Jeremy Cooper <jeremy@baymoo.org>]
8778
8779   *) Make it possible to reuse SSLv2 sessions.
8780      [Richard Levitte]
8781
8782   *) In copy_email() check for >= 0 as a return value for
8783      X509_NAME_get_index_by_NID() since 0 is a valid index.
8784      [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
8785
8786   *) Avoid coredump with unsupported or invalid public keys by checking if
8787      X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
8788      PKCS7_verify() fails with non detached data.
8789      [Steve Henson]
8790
8791   *) Don't use getenv in library functions when run as setuid/setgid.
8792      New function OPENSSL_issetugid().
8793      [Ulf Moeller]
8794
8795   *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
8796      due to incorrect handling of multi-threading:
8797
8798      1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
8799
8800      2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
8801
8802      3. Count how many times MemCheck_off() has been called so that
8803         nested use can be treated correctly.  This also avoids
8804         inband-signalling in the previous code (which relied on the
8805         assumption that thread ID 0 is impossible).
8806      [Bodo Moeller]
8807
8808   *) Add "-rand" option also to s_client and s_server.
8809      [Lutz Jaenicke]
8810
8811   *) Fix CPU detection on Irix 6.x.
8812      [Kurt Hockenbury <khockenb@stevens-tech.edu> and
8813       "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
8814
8815   *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
8816      was empty.
8817      [Steve Henson]
8818      [This change does not apply to 0.9.7.]
8819
8820   *) Use the cached encoding of an X509_NAME structure rather than
8821      copying it. This is apparently the reason for the libsafe "errors"
8822      but the code is actually correct.
8823      [Steve Henson]
8824
8825   *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
8826      Bleichenbacher's DSA attack.
8827      Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
8828      to be set and top=0 forces the highest bit to be set; top=-1 is new
8829      and leaves the highest bit random.
8830      [Ulf Moeller, Bodo Moeller]
8831
8832   *) In the NCONF_...-based implementations for CONF_... queries
8833      (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
8834      a temporary CONF structure with the data component set to NULL
8835      (which gives segmentation faults in lh_retrieve).
8836      Instead, use NULL for the CONF pointer in CONF_get_string and
8837      CONF_get_number (which may use environment variables) and directly
8838      return NULL from CONF_get_section.
8839      [Bodo Moeller]
8840
8841   *) Fix potential buffer overrun for EBCDIC.
8842      [Ulf Moeller]
8843
8844   *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
8845      keyUsage if basicConstraints absent for a CA.
8846      [Steve Henson]
8847
8848   *) Make SMIME_write_PKCS7() write mail header values with a format that
8849      is more generally accepted (no spaces before the semicolon), since
8850      some programs can't parse those values properly otherwise.  Also make
8851      sure BIO's that break lines after each write do not create invalid
8852      headers.
8853      [Richard Levitte]
8854
8855   *) Make the CRL encoding routines work with empty SEQUENCE OF. The
8856      macros previously used would not encode an empty SEQUENCE OF
8857      and break the signature.
8858      [Steve Henson]
8859      [This change does not apply to 0.9.7.]
8860
8861   *) Zero the premaster secret after deriving the master secret in
8862      DH ciphersuites.
8863      [Steve Henson]
8864
8865   *) Add some EVP_add_digest_alias registrations (as found in
8866      OpenSSL_add_all_digests()) to SSL_library_init()
8867      aka OpenSSL_add_ssl_algorithms().  This provides improved
8868      compatibility with peers using X.509 certificates
8869      with unconventional AlgorithmIdentifier OIDs.
8870      [Bodo Moeller]
8871
8872   *) Fix for Irix with NO_ASM.
8873      ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
8874
8875   *) ./config script fixes.
8876      [Ulf Moeller, Richard Levitte]
8877
8878   *) Fix 'openssl passwd -1'.
8879      [Bodo Moeller]
8880
8881   *) Change PKCS12_key_gen_asc() so it can cope with non null
8882      terminated strings whose length is passed in the passlen
8883      parameter, for example from PEM callbacks. This was done
8884      by adding an extra length parameter to asc2uni().
8885      [Steve Henson, reported by <oddissey@samsung.co.kr>]
8886
8887   *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
8888      call failed, free the DSA structure.
8889      [Bodo Moeller]
8890
8891   *) Fix to uni2asc() to cope with zero length Unicode strings.
8892      These are present in some PKCS#12 files.
8893      [Steve Henson]
8894
8895   *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
8896      Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
8897      when writing a 32767 byte record.
8898      [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
8899
8900   *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
8901      obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
8902
8903      (RSA objects have a reference count access to which is protected
8904      by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
8905      so they are meant to be shared between threads.)
8906      [Bodo Moeller, Geoff Thorpe; original patch submitted by
8907      "Reddie, Steven" <Steven.Reddie@ca.com>]
8908
8909   *) Fix a deadlock in CRYPTO_mem_leaks().
8910      [Bodo Moeller]
8911
8912   *) Use better test patterns in bntest.
8913      [Ulf Möller]
8914
8915   *) rand_win.c fix for Borland C.
8916      [Ulf Möller]
8917
8918   *) BN_rshift bugfix for n == 0.
8919      [Bodo Moeller]
8920
8921   *) Add a 'bctest' script that checks for some known 'bc' bugs
8922      so that 'make test' does not abort just because 'bc' is broken.
8923      [Bodo Moeller]
8924
8925   *) Store verify_result within SSL_SESSION also for client side to
8926      avoid potential security hole. (Re-used sessions on the client side
8927      always resulted in verify_result==X509_V_OK, not using the original
8928      result of the server certificate verification.)
8929      [Lutz Jaenicke]
8930
8931   *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
8932      SSL3_RT_APPLICATION_DATA, return 0.
8933      Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
8934      [Bodo Moeller]
8935
8936   *) Fix SSL_peek:
8937      Both ssl2_peek and ssl3_peek, which were totally broken in earlier
8938      releases, have been re-implemented by renaming the previous
8939      implementations of ssl2_read and ssl3_read to ssl2_read_internal
8940      and ssl3_read_internal, respectively, and adding 'peek' parameters
8941      to them.  The new ssl[23]_{read,peek} functions are calls to
8942      ssl[23]_read_internal with the 'peek' flag set appropriately.
8943      A 'peek' parameter has also been added to ssl3_read_bytes, which
8944      does the actual work for ssl3_read_internal.
8945      [Bodo Moeller]
8946
8947   *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
8948      the method-specific "init()" handler. Also clean up ex_data after
8949      calling the method-specific "finish()" handler. Previously, this was
8950      happening the other way round.
8951      [Geoff Thorpe]
8952
8953   *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
8954      The previous value, 12, was not always sufficient for BN_mod_exp().
8955      [Bodo Moeller]
8956
8957   *) Make sure that shared libraries get the internal name engine with
8958      the full version number and not just 0.  This should mark the
8959      shared libraries as not backward compatible.  Of course, this should
8960      be changed again when we can guarantee backward binary compatibility.
8961      [Richard Levitte]
8962
8963   *) Fix typo in get_cert_by_subject() in by_dir.c
8964      [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
8965
8966   *) Rework the system to generate shared libraries:
8967
8968      - Make note of the expected extension for the shared libraries and
8969        if there is a need for symbolic links from for example libcrypto.so.0
8970        to libcrypto.so.0.9.7.  There is extended info in Configure for
8971        that.
8972
8973      - Make as few rebuilds of the shared libraries as possible.
8974
8975      - Still avoid linking the OpenSSL programs with the shared libraries.
8976
8977      - When installing, install the shared libraries separately from the
8978        static ones.
8979      [Richard Levitte]
8980
8981   *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
8982
8983      Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
8984      and not in SSL_clear because the latter is also used by the
8985      accept/connect functions; previously, the settings made by
8986      SSL_set_read_ahead would be lost during the handshake.
8987      [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
8988
8989   *) Correct util/mkdef.pl to be selective about disabled algorithms.
8990      Previously, it would create entries for disabled algorithms no
8991      matter what.
8992      [Richard Levitte]
8993
8994   *) Added several new manual pages for SSL_* function.
8995      [Lutz Jaenicke]
8996
8997  Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
8998
8999   *) In ssl23_get_client_hello, generate an error message when faced
9000      with an initial SSL 3.0/TLS record that is too small to contain the
9001      first two bytes of the ClientHello message, i.e. client_version.
9002      (Note that this is a pathologic case that probably has never happened
9003      in real life.)  The previous approach was to use the version number
9004      from the record header as a substitute; but our protocol choice
9005      should not depend on that one because it is not authenticated
9006      by the Finished messages.
9007      [Bodo Moeller]
9008
9009   *) More robust randomness gathering functions for Windows.
9010      [Jeffrey Altman <jaltman@columbia.edu>]
9011
9012   *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
9013      not set then we don't setup the error code for issuer check errors
9014      to avoid possibly overwriting other errors which the callback does
9015      handle. If an application does set the flag then we assume it knows
9016      what it is doing and can handle the new informational codes
9017      appropriately.
9018      [Steve Henson]
9019
9020   *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
9021      a general "ANY" type, as such it should be able to decode anything
9022      including tagged types. However it didn't check the class so it would
9023      wrongly interpret tagged types in the same way as their universal
9024      counterpart and unknown types were just rejected. Changed so that the
9025      tagged and unknown types are handled in the same way as a SEQUENCE:
9026      that is the encoding is stored intact. There is also a new type
9027      "V_ASN1_OTHER" which is used when the class is not universal, in this
9028      case we have no idea what the actual type is so we just lump them all
9029      together.
9030      [Steve Henson]
9031
9032   *) On VMS, stdout may very well lead to a file that is written to
9033      in a record-oriented fashion.  That means that every write() will
9034      write a separate record, which will be read separately by the
9035      programs trying to read from it.  This can be very confusing.
9036
9037      The solution is to put a BIO filter in the way that will buffer
9038      text until a linefeed is reached, and then write everything a
9039      line at a time, so every record written will be an actual line,
9040      not chunks of lines and not (usually doesn't happen, but I've
9041      seen it once) several lines in one record.  BIO_f_linebuffer() is
9042      the answer.
9043
9044      Currently, it's a VMS-only method, because that's where it has
9045      been tested well enough.
9046      [Richard Levitte]
9047
9048   *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
9049      it can return incorrect results.
9050      (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
9051      but it was in 0.9.6-beta[12].)
9052      [Bodo Moeller]
9053
9054   *) Disable the check for content being present when verifying detached
9055      signatures in pk7_smime.c. Some versions of Netscape (wrongly)
9056      include zero length content when signing messages.
9057      [Steve Henson]
9058
9059   *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
9060      BIO_ctrl (for BIO pairs).
9061      [Bodo Möller]
9062
9063   *) Add DSO method for VMS.
9064      [Richard Levitte]
9065
9066   *) Bug fix: Montgomery multiplication could produce results with the
9067      wrong sign.
9068      [Ulf Möller]
9069
9070   *) Add RPM specification openssl.spec and modify it to build three
9071      packages.  The default package contains applications, application
9072      documentation and run-time libraries.  The devel package contains
9073      include files, static libraries and function documentation.  The
9074      doc package contains the contents of the doc directory.  The original
9075      openssl.spec was provided by Damien Miller <djm@mindrot.org>.
9076      [Richard Levitte]
9077
9078   *) Add a large number of documentation files for many SSL routines.
9079      [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
9080
9081   *) Add a configuration entry for Sony News 4.
9082      [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
9083
9084   *) Don't set the two most significant bits to one when generating a
9085      random number < q in the DSA library.
9086      [Ulf Möller]
9087
9088   *) New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
9089      behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
9090      the underlying transport is blocking) if a handshake took place.
9091      (The default behaviour is needed by applications such as s_client
9092      and s_server that use select() to determine when to use SSL_read;
9093      but for applications that know in advance when to expect data, it
9094      just makes things more complicated.)
9095      [Bodo Moeller]
9096
9097   *) Add RAND_egd_bytes(), which gives control over the number of bytes read
9098      from EGD.
9099      [Ben Laurie]
9100
9101   *) Add a few more EBCDIC conditionals that make `req' and `x509'
9102      work better on such systems.
9103      [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
9104
9105   *) Add two demo programs for PKCS12_parse() and PKCS12_create().
9106      Update PKCS12_parse() so it copies the friendlyName and the
9107      keyid to the certificates aux info.
9108      [Steve Henson]
9109
9110   *) Fix bug in PKCS7_verify() which caused an infinite loop
9111      if there was more than one signature.
9112      [Sven Uszpelkat <su@celocom.de>]
9113
9114   *) Major change in util/mkdef.pl to include extra information
9115      about each symbol, as well as presenting variables as well
9116      as functions.  This change means that there's n more need
9117      to rebuild the .num files when some algorithms are excluded.
9118      [Richard Levitte]
9119
9120   *) Allow the verify time to be set by an application,
9121      rather than always using the current time.
9122      [Steve Henson]
9123
9124   *) Phase 2 verify code reorganisation. The certificate
9125      verify code now looks up an issuer certificate by a
9126      number of criteria: subject name, authority key id
9127      and key usage. It also verifies self signed certificates
9128      by the same criteria. The main comparison function is
9129      X509_check_issued() which performs these checks.
9130
9131      Lot of changes were necessary in order to support this
9132      without completely rewriting the lookup code.
9133
9134      Authority and subject key identifier are now cached.
9135
9136      The LHASH 'certs' is X509_STORE has now been replaced
9137      by a STACK_OF(X509_OBJECT). This is mainly because an
9138      LHASH can't store or retrieve multiple objects with
9139      the same hash value.
9140
9141      As a result various functions (which were all internal
9142      use only) have changed to handle the new X509_STORE
9143      structure. This will break anything that messed round
9144      with X509_STORE internally.
9145
9146      The functions X509_STORE_add_cert() now checks for an
9147      exact match, rather than just subject name.
9148
9149      The X509_STORE API doesn't directly support the retrieval
9150      of multiple certificates matching a given criteria, however
9151      this can be worked round by performing a lookup first
9152      (which will fill the cache with candidate certificates)
9153      and then examining the cache for matches. This is probably
9154      the best we can do without throwing out X509_LOOKUP
9155      entirely (maybe later...).
9156
9157      The X509_VERIFY_CTX structure has been enhanced considerably.
9158
9159      All certificate lookup operations now go via a get_issuer()
9160      callback. Although this currently uses an X509_STORE it
9161      can be replaced by custom lookups. This is a simple way
9162      to bypass the X509_STORE hackery necessary to make this
9163      work and makes it possible to use more efficient techniques
9164      in future. A very simple version which uses a simple
9165      STACK for its trusted certificate store is also provided
9166      using X509_STORE_CTX_trusted_stack().
9167
9168      The verify_cb() and verify() callbacks now have equivalents
9169      in the X509_STORE_CTX structure.
9170
9171      X509_STORE_CTX also has a 'flags' field which can be used
9172      to customise the verify behaviour.
9173      [Steve Henson]
9174
9175   *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
9176      excludes S/MIME capabilities.
9177      [Steve Henson]
9178
9179   *) When a certificate request is read in keep a copy of the
9180      original encoding of the signed data and use it when outputting
9181      again. Signatures then use the original encoding rather than
9182      a decoded, encoded version which may cause problems if the
9183      request is improperly encoded.
9184      [Steve Henson]
9185
9186   *) For consistency with other BIO_puts implementations, call
9187      buffer_write(b, ...) directly in buffer_puts instead of calling
9188      BIO_write(b, ...).
9189
9190      In BIO_puts, increment b->num_write as in BIO_write.
9191      [Peter.Sylvester@EdelWeb.fr]
9192
9193   *) Fix BN_mul_word for the case where the word is 0. (We have to use
9194      BN_zero, we may not return a BIGNUM with an array consisting of
9195      words set to zero.)
9196      [Bodo Moeller]
9197
9198   *) Avoid calling abort() from within the library when problems are
9199      detected, except if preprocessor symbols have been defined
9200      (such as REF_CHECK, BN_DEBUG etc.).
9201      [Bodo Moeller]
9202
9203   *) New openssl application 'rsautl'. This utility can be
9204      used for low level RSA operations. DER public key
9205      BIO/fp routines also added.
9206      [Steve Henson]
9207
9208   *) New Configure entry and patches for compiling on QNX 4.
9209      [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
9210
9211   *) A demo state-machine implementation was sponsored by
9212      Nuron (http://www.nuron.com/) and is now available in
9213      demos/state_machine.
9214      [Ben Laurie]
9215
9216   *) New options added to the 'dgst' utility for signature
9217      generation and verification.
9218      [Steve Henson]
9219
9220   *) Unrecognized PKCS#7 content types are now handled via a
9221      catch all ASN1_TYPE structure. This allows unsupported
9222      types to be stored as a "blob" and an application can
9223      encode and decode it manually.
9224      [Steve Henson]
9225
9226   *) Fix various signed/unsigned issues to make a_strex.c
9227      compile under VC++.
9228      [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
9229
9230   *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
9231      length if passed a buffer. ASN1_INTEGER_to_BN failed
9232      if passed a NULL BN and its argument was negative.
9233      [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
9234
9235   *) Modification to PKCS#7 encoding routines to output definite
9236      length encoding. Since currently the whole structures are in
9237      memory there's not real point in using indefinite length
9238      constructed encoding. However if OpenSSL is compiled with
9239      the flag PKCS7_INDEFINITE_ENCODING the old form is used.
9240      [Steve Henson]
9241
9242   *) Added BIO_vprintf() and BIO_vsnprintf().
9243      [Richard Levitte]
9244
9245   *) Added more prefixes to parse for in the the strings written
9246      through a logging bio, to cover all the levels that are available
9247      through syslog.  The prefixes are now:
9248
9249         PANIC, EMERG, EMR       =>      LOG_EMERG
9250         ALERT, ALR              =>      LOG_ALERT
9251         CRIT, CRI               =>      LOG_CRIT
9252         ERROR, ERR              =>      LOG_ERR
9253         WARNING, WARN, WAR      =>      LOG_WARNING
9254         NOTICE, NOTE, NOT       =>      LOG_NOTICE
9255         INFO, INF               =>      LOG_INFO
9256         DEBUG, DBG              =>      LOG_DEBUG
9257
9258      and as before, if none of those prefixes are present at the
9259      beginning of the string, LOG_ERR is chosen.
9260
9261      On Win32, the LOG_* levels are mapped according to this:
9262
9263         LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
9264         LOG_WARNING                             => EVENTLOG_WARNING_TYPE
9265         LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE
9266
9267      [Richard Levitte]
9268
9269   *) Made it possible to reconfigure with just the configuration
9270      argument "reconf" or "reconfigure".  The command line arguments
9271      are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
9272      and are retrieved from there when reconfiguring.
9273      [Richard Levitte]
9274
9275   *) MD4 implemented.
9276      [Assar Westerlund <assar@sics.se>, Richard Levitte]
9277
9278   *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
9279      [Richard Levitte]
9280
9281   *) The obj_dat.pl script was messing up the sorting of object
9282      names. The reason was that it compared the quoted version
9283      of strings as a result "OCSP" > "OCSP Signing" because
9284      " > SPACE. Changed script to store unquoted versions of
9285      names and add quotes on output. It was also omitting some
9286      names from the lookup table if they were given a default
9287      value (that is if SN is missing it is given the same
9288      value as LN and vice versa), these are now added on the
9289      grounds that if an object has a name we should be able to
9290      look it up. Finally added warning output when duplicate
9291      short or long names are found.
9292      [Steve Henson]
9293
9294   *) Changes needed for Tandem NSK.
9295      [Scott Uroff <scott@xypro.com>]
9296
9297   *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
9298      RSA_padding_check_SSLv23(), special padding was never detected
9299      and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
9300      version rollback attacks was not effective.
9301
9302      In s23_clnt.c, don't use special rollback-attack detection padding
9303      (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
9304      client; similarly, in s23_srvr.c, don't do the rollback check if
9305      SSL 2.0 is the only protocol enabled in the server.
9306      [Bodo Moeller]
9307
9308   *) Make it possible to get hexdumps of unprintable data with 'openssl
9309      asn1parse'.  By implication, the functions ASN1_parse_dump() and
9310      BIO_dump_indent() are added.
9311      [Richard Levitte]
9312
9313   *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
9314      these print out strings and name structures based on various
9315      flags including RFC2253 support and proper handling of
9316      multibyte characters. Added options to the 'x509' utility
9317      to allow the various flags to be set.
9318      [Steve Henson]
9319
9320   *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
9321      Also change the functions X509_cmp_current_time() and
9322      X509_gmtime_adj() work with an ASN1_TIME structure,
9323      this will enable certificates using GeneralizedTime in validity
9324      dates to be checked.
9325      [Steve Henson]
9326
9327   *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
9328      negative public key encodings) on by default,
9329      NO_NEG_PUBKEY_BUG can be set to disable it.
9330      [Steve Henson]
9331
9332   *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
9333      content octets. An i2c_ASN1_OBJECT is unnecessary because
9334      the encoding can be trivially obtained from the structure.
9335      [Steve Henson]
9336
9337   *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
9338      not read locks (CRYPTO_r_[un]lock).
9339      [Bodo Moeller]
9340
9341   *) A first attempt at creating official support for shared
9342      libraries through configuration.  I've kept it so the
9343      default is static libraries only, and the OpenSSL programs
9344      are always statically linked for now, but there are
9345      preparations for dynamic linking in place.
9346      This has been tested on Linux and Tru64.
9347      [Richard Levitte]
9348
9349   *) Randomness polling function for Win9x, as described in:
9350      Peter Gutmann, Software Generation of Practically Strong
9351      Random Numbers.
9352      [Ulf Möller]
9353
9354   *) Fix so PRNG is seeded in req if using an already existing
9355      DSA key.
9356      [Steve Henson]
9357
9358   *) New options to smime application. -inform and -outform
9359      allow alternative formats for the S/MIME message including
9360      PEM and DER. The -content option allows the content to be
9361      specified separately. This should allow things like Netscape
9362      form signing output easier to verify.
9363      [Steve Henson]
9364
9365   *) Fix the ASN1 encoding of tags using the 'long form'.
9366      [Steve Henson]
9367
9368   *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
9369      STRING types. These convert content octets to and from the
9370      underlying type. The actual tag and length octets are
9371      already assumed to have been read in and checked. These
9372      are needed because all other string types have virtually
9373      identical handling apart from the tag. By having versions
9374      of the ASN1 functions that just operate on content octets
9375      IMPLICIT tagging can be handled properly. It also allows
9376      the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
9377      and ASN1_INTEGER are identical apart from the tag.
9378      [Steve Henson]
9379
9380   *) Change the handling of OID objects as follows:
9381
9382      - New object identifiers are inserted in objects.txt, following
9383        the syntax given in objects.README.
9384      - objects.pl is used to process obj_mac.num and create a new
9385        obj_mac.h.
9386      - obj_dat.pl is used to create a new obj_dat.h, using the data in
9387        obj_mac.h.
9388
9389      This is currently kind of a hack, and the perl code in objects.pl
9390      isn't very elegant, but it works as I intended.  The simplest way
9391      to check that it worked correctly is to look in obj_dat.h and
9392      check the array nid_objs and make sure the objects haven't moved
9393      around (this is important!).  Additions are OK, as well as
9394      consistent name changes.
9395      [Richard Levitte]
9396
9397   *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
9398      [Bodo Moeller]
9399
9400   *) Addition of the command line parameter '-rand file' to 'openssl req'.
9401      The given file adds to whatever has already been seeded into the
9402      random pool through the RANDFILE configuration file option or
9403      environment variable, or the default random state file.
9404      [Richard Levitte]
9405
9406   *) mkstack.pl now sorts each macro group into lexical order.
9407      Previously the output order depended on the order the files
9408      appeared in the directory, resulting in needless rewriting
9409      of safestack.h .
9410      [Steve Henson]
9411
9412   *) Patches to make OpenSSL compile under Win32 again. Mostly
9413      work arounds for the VC++ problem that it treats func() as
9414      func(void). Also stripped out the parts of mkdef.pl that
9415      added extra typesafe functions: these no longer exist.
9416      [Steve Henson]
9417
9418   *) Reorganisation of the stack code. The macros are now all
9419      collected in safestack.h . Each macro is defined in terms of
9420      a "stack macro" of the form SKM_<name>(type, a, b). The
9421      DEBUG_SAFESTACK is now handled in terms of function casts,
9422      this has the advantage of retaining type safety without the
9423      use of additional functions. If DEBUG_SAFESTACK is not defined
9424      then the non typesafe macros are used instead. Also modified the
9425      mkstack.pl script to handle the new form. Needs testing to see
9426      if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
9427      the default if no major problems. Similar behaviour for ASN1_SET_OF
9428      and PKCS12_STACK_OF.
9429      [Steve Henson]
9430
9431   *) When some versions of IIS use the 'NET' form of private key the
9432      key derivation algorithm is different. Normally MD5(password) is
9433      used as a 128 bit RC4 key. In the modified case
9434      MD5(MD5(password) + "SGCKEYSALT")  is used instead. Added some
9435      new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
9436      as the old Netscape_RSA functions except they have an additional
9437      'sgckey' parameter which uses the modified algorithm. Also added
9438      an -sgckey command line option to the rsa utility. Thanks to
9439      Adrian Peck <bertie@ncipher.com> for posting details of the modified
9440      algorithm to openssl-dev.
9441      [Steve Henson]
9442
9443   *) The evp_local.h macros were using 'c.##kname' which resulted in
9444      invalid expansion on some systems (SCO 5.0.5 for example).
9445      Corrected to 'c.kname'.
9446      [Phillip Porch <root@theporch.com>]
9447
9448   *) New X509_get1_email() and X509_REQ_get1_email() functions that return
9449      a STACK of email addresses from a certificate or request, these look
9450      in the subject name and the subject alternative name extensions and
9451      omit any duplicate addresses.
9452      [Steve Henson]
9453
9454   *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
9455      This makes DSA verification about 2 % faster.
9456      [Bodo Moeller]
9457
9458   *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
9459      (meaning that now 2^5 values will be precomputed, which is only 4 KB
9460      plus overhead for 1024 bit moduli).
9461      This makes exponentiations about 0.5 % faster for 1024 bit
9462      exponents (as measured by "openssl speed rsa2048").
9463      [Bodo Moeller]
9464
9465   *) Rename memory handling macros to avoid conflicts with other
9466      software:
9467           Malloc         =>  OPENSSL_malloc
9468           Malloc_locked  =>  OPENSSL_malloc_locked
9469           Realloc        =>  OPENSSL_realloc
9470           Free           =>  OPENSSL_free
9471      [Richard Levitte]
9472
9473   *) New function BN_mod_exp_mont_word for small bases (roughly 15%
9474      faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
9475      [Bodo Moeller]
9476
9477   *) CygWin32 support.
9478      [John Jarvie <jjarvie@newsguy.com>]
9479
9480   *) The type-safe stack code has been rejigged. It is now only compiled
9481      in when OpenSSL is configured with the DEBUG_SAFESTACK option and
9482      by default all type-specific stack functions are "#define"d back to
9483      standard stack functions. This results in more streamlined output
9484      but retains the type-safety checking possibilities of the original
9485      approach.
9486      [Geoff Thorpe]
9487
9488   *) The STACK code has been cleaned up, and certain type declarations
9489      that didn't make a lot of sense have been brought in line. This has
9490      also involved a cleanup of sorts in safestack.h to more correctly
9491      map type-safe stack functions onto their plain stack counterparts.
9492      This work has also resulted in a variety of "const"ifications of
9493      lots of the code, especially "_cmp" operations which should normally
9494      be prototyped with "const" parameters anyway.
9495      [Geoff Thorpe]
9496
9497   *) When generating bytes for the first time in md_rand.c, 'stir the pool'
9498      by seeding with STATE_SIZE dummy bytes (with zero entropy count).
9499      (The PRNG state consists of two parts, the large pool 'state' and 'md',
9500      where all of 'md' is used each time the PRNG is used, but 'state'
9501      is used only indexed by a cyclic counter. As entropy may not be
9502      well distributed from the beginning, 'md' is important as a
9503      chaining variable. However, the output function chains only half
9504      of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
9505      all of 'md', and seeding with STATE_SIZE dummy bytes will result
9506      in all of 'state' being rewritten, with the new values depending
9507      on virtually all of 'md'.  This overcomes the 80 bit limitation.)
9508      [Bodo Moeller]
9509
9510   *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
9511      the handshake is continued after ssl_verify_cert_chain();
9512      otherwise, if SSL_VERIFY_NONE is set, remaining error codes
9513      can lead to 'unexplainable' connection aborts later.
9514      [Bodo Moeller; problem tracked down by Lutz Jaenicke]
9515
9516   *) Major EVP API cipher revision.
9517      Add hooks for extra EVP features. This allows various cipher
9518      parameters to be set in the EVP interface. Support added for variable
9519      key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
9520      setting of RC2 and RC5 parameters.
9521
9522      Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
9523      ciphers.
9524
9525      Remove lots of duplicated code from the EVP library. For example *every*
9526      cipher init() function handles the 'iv' in the same way according to the
9527      cipher mode. They also all do nothing if the 'key' parameter is NULL and
9528      for CFB and OFB modes they zero ctx->num.
9529
9530      New functionality allows removal of S/MIME code RC2 hack.
9531
9532      Most of the routines have the same form and so can be declared in terms
9533      of macros.
9534
9535      By shifting this to the top level EVP_CipherInit() it can be removed from
9536      all individual ciphers. If the cipher wants to handle IVs or keys
9537      differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
9538      flags.
9539
9540      Change lots of functions like EVP_EncryptUpdate() to now return a
9541      value: although software versions of the algorithms cannot fail
9542      any installed hardware versions can.
9543      [Steve Henson]
9544
9545   *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
9546      this option is set, tolerate broken clients that send the negotiated
9547      protocol version number instead of the requested protocol version
9548      number.
9549      [Bodo Moeller]
9550
9551   *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
9552      i.e. non-zero for export ciphersuites, zero otherwise.
9553      Previous versions had this flag inverted, inconsistent with
9554      rsa_tmp_cb (..._TMP_RSA_CB).
9555      [Bodo Moeller; problem reported by Amit Chopra]
9556
9557   *) Add missing DSA library text string. Work around for some IIS
9558      key files with invalid SEQUENCE encoding.
9559      [Steve Henson]
9560
9561   *) Add a document (doc/standards.txt) that list all kinds of standards
9562      and so on that are implemented in OpenSSL.
9563      [Richard Levitte]
9564
9565   *) Enhance c_rehash script. Old version would mishandle certificates
9566      with the same subject name hash and wouldn't handle CRLs at all.
9567      Added -fingerprint option to crl utility, to support new c_rehash
9568      features.
9569      [Steve Henson]
9570
9571   *) Eliminate non-ANSI declarations in crypto.h and stack.h.
9572      [Ulf Möller]
9573
9574   *) Fix for SSL server purpose checking. Server checking was
9575      rejecting certificates which had extended key usage present
9576      but no ssl client purpose.
9577      [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
9578
9579   *) Make PKCS#12 code work with no password. The PKCS#12 spec
9580      is a little unclear about how a blank password is handled.
9581      Since the password in encoded as a BMPString with terminating
9582      double NULL a zero length password would end up as just the
9583      double NULL. However no password at all is different and is
9584      handled differently in the PKCS#12 key generation code. NS
9585      treats a blank password as zero length. MSIE treats it as no
9586      password on export: but it will try both on import. We now do
9587      the same: PKCS12_parse() tries zero length and no password if
9588      the password is set to "" or NULL (NULL is now a valid password:
9589      it wasn't before) as does the pkcs12 application.
9590      [Steve Henson]
9591
9592   *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
9593      perror when PEM_read_bio_X509_REQ fails, the error message must
9594      be obtained from the error queue.
9595      [Bodo Moeller]
9596
9597   *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
9598      it in ERR_remove_state if appropriate, and change ERR_get_state
9599      accordingly to avoid race conditions (this is necessary because
9600      thread_hash is no longer constant once set).
9601      [Bodo Moeller]
9602
9603   *) Bugfix for linux-elf makefile.one.
9604      [Ulf Möller]
9605
9606   *) RSA_get_default_method() will now cause a default
9607      RSA_METHOD to be chosen if one doesn't exist already.
9608      Previously this was only set during a call to RSA_new()
9609      or RSA_new_method(NULL) meaning it was possible for
9610      RSA_get_default_method() to return NULL.
9611      [Geoff Thorpe]
9612
9613   *) Added native name translation to the existing DSO code
9614      that will convert (if the flag to do so is set) filenames
9615      that are sufficiently small and have no path information
9616      into a canonical native form. Eg. "blah" converted to
9617      "libblah.so" or "blah.dll" etc.
9618      [Geoff Thorpe]
9619
9620   *) New function ERR_error_string_n(e, buf, len) which is like
9621      ERR_error_string(e, buf), but writes at most 'len' bytes
9622      including the 0 terminator.  For ERR_error_string_n, 'buf'
9623      may not be NULL.
9624      [Damien Miller <djm@mindrot.org>, Bodo Moeller]
9625
9626   *) CONF library reworked to become more general.  A new CONF
9627      configuration file reader "class" is implemented as well as a
9628      new functions (NCONF_*, for "New CONF") to handle it.  The now
9629      old CONF_* functions are still there, but are reimplemented to
9630      work in terms of the new functions.  Also, a set of functions
9631      to handle the internal storage of the configuration data is
9632      provided to make it easier to write new configuration file
9633      reader "classes" (I can definitely see something reading a
9634      configuration file in XML format, for example), called _CONF_*,
9635      or "the configuration storage API"...
9636
9637      The new configuration file reading functions are:
9638
9639         NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
9640         NCONF_get_section, NCONF_get_string, NCONF_get_numbre
9641
9642         NCONF_default, NCONF_WIN32
9643
9644         NCONF_dump_fp, NCONF_dump_bio
9645
9646      NCONF_default and NCONF_WIN32 are method (or "class") choosers,
9647      NCONF_new creates a new CONF object.  This works in the same way
9648      as other interfaces in OpenSSL, like the BIO interface.
9649      NCONF_dump_* dump the internal storage of the configuration file,
9650      which is useful for debugging.  All other functions take the same
9651      arguments as the old CONF_* functions wth the exception of the
9652      first that must be a `CONF *' instead of a `LHASH *'.
9653
9654      To make it easer to use the new classes with the old CONF_* functions,
9655      the function CONF_set_default_method is provided.
9656      [Richard Levitte]
9657
9658   *) Add '-tls1' option to 'openssl ciphers', which was already
9659      mentioned in the documentation but had not been implemented.
9660      (This option is not yet really useful because even the additional
9661      experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
9662      [Bodo Moeller]
9663
9664   *) Initial DSO code added into libcrypto for letting OpenSSL (and
9665      OpenSSL-based applications) load shared libraries and bind to
9666      them in a portable way.
9667      [Geoff Thorpe, with contributions from Richard Levitte]
9668
9669  Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
9670
9671   *) Make sure _lrotl and _lrotr are only used with MSVC.
9672
9673   *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
9674      (the default implementation of RAND_status).
9675
9676   *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
9677      to '-clrext' (= clear extensions), as intended and documented.
9678      [Bodo Moeller; inconsistency pointed out by Michael Attili
9679      <attili@amaxo.com>]
9680
9681   *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
9682      was larger than the MD block size.
9683      [Steve Henson, pointed out by Yost William <YostW@tce.com>]
9684
9685   *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
9686      fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
9687      using the passed key: if the passed key was a private key the result
9688      of X509_print(), for example, would be to print out all the private key
9689      components.
9690      [Steve Henson]
9691
9692   *) des_quad_cksum() byte order bug fix.
9693      [Ulf Möller, using the problem description in krb4-0.9.7, where
9694       the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
9695
9696   *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
9697      discouraged.
9698      [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
9699
9700   *) For easily testing in shell scripts whether some command
9701      'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
9702      returns with exit code 0 iff no command of the given name is available.
9703      'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
9704      the output goes to stdout and nothing is printed to stderr.
9705      Additional arguments are always ignored.
9706
9707      Since for each cipher there is a command of the same name,
9708      the 'no-cipher' compilation switches can be tested this way.
9709
9710      ('openssl no-XXX' is not able to detect pseudo-commands such
9711      as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
9712      [Bodo Moeller]
9713
9714   *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
9715      [Bodo Moeller]
9716
9717   *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
9718      is set; it will be thrown away anyway because each handshake creates
9719      its own key.
9720      ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
9721      to parameters -- in previous versions (since OpenSSL 0.9.3) the
9722      'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning
9723      you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
9724      [Bodo Moeller]
9725
9726   *) New s_client option -ign_eof: EOF at stdin is ignored, and
9727      'Q' and 'R' lose their special meanings (quit/renegotiate).
9728      This is part of what -quiet does; unlike -quiet, -ign_eof
9729      does not suppress any output.
9730      [Richard Levitte]
9731
9732   *) Add compatibility options to the purpose and trust code. The
9733      purpose X509_PURPOSE_ANY is "any purpose" which automatically
9734      accepts a certificate or CA, this was the previous behaviour,
9735      with all the associated security issues.
9736
9737      X509_TRUST_COMPAT is the old trust behaviour: only and
9738      automatically trust self signed roots in certificate store. A
9739      new trust setting X509_TRUST_DEFAULT is used to specify that
9740      a purpose has no associated trust setting and it should instead
9741      use the value in the default purpose.
9742      [Steve Henson]
9743
9744   *) Fix the PKCS#8 DSA private key code so it decodes keys again
9745      and fix a memory leak.
9746      [Steve Henson]
9747
9748   *) In util/mkerr.pl (which implements 'make errors'), preserve
9749      reason strings from the previous version of the .c file, as
9750      the default to have only downcase letters (and digits) in
9751      automatically generated reasons codes is not always appropriate.
9752      [Bodo Moeller]
9753
9754   *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
9755      using strerror.  Previously, ERR_reason_error_string() returned
9756      library names as reason strings for SYSerr; but SYSerr is a special
9757      case where small numbers are errno values, not library numbers.
9758      [Bodo Moeller]
9759
9760   *) Add '-dsaparam' option to 'openssl dhparam' application.  This
9761      converts DSA parameters into DH parameters. (When creating parameters,
9762      DSA_generate_parameters is used.)
9763      [Bodo Moeller]
9764
9765   *) Include 'length' (recommended exponent length) in C code generated
9766      by 'openssl dhparam -C'.
9767      [Bodo Moeller]
9768
9769   *) The second argument to set_label in perlasm was already being used
9770      so couldn't be used as a "file scope" flag. Moved to third argument
9771      which was free.
9772      [Steve Henson]
9773
9774   *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
9775      instead of RAND_bytes for encryption IVs and salts.
9776      [Bodo Moeller]
9777
9778   *) Include RAND_status() into RAND_METHOD instead of implementing
9779      it only for md_rand.c  Otherwise replacing the PRNG by calling
9780      RAND_set_rand_method would be impossible.
9781      [Bodo Moeller]
9782
9783   *) Don't let DSA_generate_key() enter an infinite loop if the random
9784      number generation fails.
9785      [Bodo Moeller]
9786
9787   *) New 'rand' application for creating pseudo-random output.
9788      [Bodo Moeller]
9789
9790   *) Added configuration support for Linux/IA64
9791      [Rolf Haberrecker <rolf@suse.de>]
9792
9793   *) Assembler module support for Mingw32.
9794      [Ulf Möller]
9795
9796   *) Shared library support for HPUX (in shlib/).
9797      [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
9798
9799   *) Shared library support for Solaris gcc.
9800      [Lutz Behnke <behnke@trustcenter.de>]
9801
9802  Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
9803
9804   *) PKCS7_encrypt() was adding text MIME headers twice because they
9805      were added manually and by SMIME_crlf_copy().
9806      [Steve Henson]
9807
9808   *) In bntest.c don't call BN_rand with zero bits argument.
9809      [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
9810
9811   *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
9812      case was implemented. This caused BN_div_recp() to fail occasionally.
9813      [Ulf Möller]
9814
9815   *) Add an optional second argument to the set_label() in the perl
9816      assembly language builder. If this argument exists and is set
9817      to 1 it signals that the assembler should use a symbol whose
9818      scope is the entire file, not just the current function. This
9819      is needed with MASM which uses the format label:: for this scope.
9820      [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
9821
9822   *) Change the ASN1 types so they are typedefs by default. Before
9823      almost all types were #define'd to ASN1_STRING which was causing
9824      STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
9825      for example.
9826      [Steve Henson]
9827
9828   *) Change names of new functions to the new get1/get0 naming
9829      convention: After 'get1', the caller owns a reference count
9830      and has to call ..._free; 'get0' returns a pointer to some
9831      data structure without incrementing reference counters.
9832      (Some of the existing 'get' functions increment a reference
9833      counter, some don't.)
9834      Similarly, 'set1' and 'add1' functions increase reference
9835      counters or duplicate objects.
9836      [Steve Henson]
9837
9838   *) Allow for the possibility of temp RSA key generation failure:
9839      the code used to assume it always worked and crashed on failure.
9840      [Steve Henson]
9841
9842   *) Fix potential buffer overrun problem in BIO_printf().
9843      [Ulf Möller, using public domain code by Patrick Powell; problem
9844       pointed out by David Sacerdote <das33@cornell.edu>]
9845
9846   *) Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
9847      RAND_egd() and RAND_status().  In the command line application,
9848      the EGD socket can be specified like a seed file using RANDFILE
9849      or -rand.
9850      [Ulf Möller]
9851
9852   *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
9853      Some CAs (e.g. Verisign) distribute certificates in this form.
9854      [Steve Henson]
9855
9856   *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
9857      list to exclude them. This means that no special compilation option
9858      is needed to use anonymous DH: it just needs to be included in the
9859      cipher list.
9860      [Steve Henson]
9861
9862   *) Change the EVP_MD_CTX_type macro so its meaning consistent with
9863      EVP_MD_type. The old functionality is available in a new macro called
9864      EVP_MD_md(). Change code that uses it and update docs.
9865      [Steve Henson]
9866
9867   *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
9868      where the 'void *' argument is replaced by a function pointer argument.
9869      Previously 'void *' was abused to point to functions, which works on
9870      many platforms, but is not correct.  As these functions are usually
9871      called by macros defined in OpenSSL header files, most source code
9872      should work without changes.
9873      [Richard Levitte]
9874
9875   *) <openssl/opensslconf.h> (which is created by Configure) now contains
9876      sections with information on -D... compiler switches used for
9877      compiling the library so that applications can see them.  To enable
9878      one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
9879      must be defined.  E.g.,
9880         #define OPENSSL_ALGORITHM_DEFINES
9881         #include <openssl/opensslconf.h>
9882      defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
9883      [Richard Levitte, Ulf and Bodo Möller]
9884
9885   *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
9886      record layer.
9887      [Bodo Moeller]
9888
9889   *) Change the 'other' type in certificate aux info to a STACK_OF
9890      X509_ALGOR. Although not an AlgorithmIdentifier as such it has
9891      the required ASN1 format: arbitrary types determined by an OID.
9892      [Steve Henson]
9893
9894   *) Add some PEM_write_X509_REQ_NEW() functions and a command line
9895      argument to 'req'. This is not because the function is newer or
9896      better than others it just uses the work 'NEW' in the certificate
9897      request header lines. Some software needs this.
9898      [Steve Henson]
9899
9900   *) Reorganise password command line arguments: now passwords can be
9901      obtained from various sources. Delete the PEM_cb function and make
9902      it the default behaviour: i.e. if the callback is NULL and the
9903      usrdata argument is not NULL interpret it as a null terminated pass
9904      phrase. If usrdata and the callback are NULL then the pass phrase
9905      is prompted for as usual.
9906      [Steve Henson]
9907
9908   *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
9909      the support is automatically enabled. The resulting binaries will
9910      autodetect the card and use it if present.
9911      [Ben Laurie and Compaq Inc.]
9912
9913   *) Work around for Netscape hang bug. This sends certificate request
9914      and server done in one record. Since this is perfectly legal in the
9915      SSL/TLS protocol it isn't a "bug" option and is on by default. See
9916      the bugs/SSLv3 entry for more info.
9917      [Steve Henson]
9918
9919   *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
9920      [Andy Polyakov]
9921
9922   *) Add -rand argument to smime and pkcs12 applications and read/write
9923      of seed file.
9924      [Steve Henson]
9925
9926   *) New 'passwd' tool for crypt(3) and apr1 password hashes.
9927      [Bodo Moeller]
9928
9929   *) Add command line password options to the remaining applications.
9930      [Steve Henson]
9931
9932   *) Bug fix for BN_div_recp() for numerators with an even number of
9933      bits.
9934      [Ulf Möller]
9935
9936   *) More tests in bntest.c, and changed test_bn output.
9937      [Ulf Möller]
9938
9939   *) ./config recognizes MacOS X now.
9940      [Andy Polyakov]
9941
9942   *) Bug fix for BN_div() when the first words of num and divsor are
9943      equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
9944      [Ulf Möller]
9945
9946   *) Add support for various broken PKCS#8 formats, and command line
9947      options to produce them.
9948      [Steve Henson]
9949
9950   *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
9951      get temporary BIGNUMs from a BN_CTX.
9952      [Ulf Möller]
9953
9954   *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
9955      for p == 0.
9956      [Ulf Möller]
9957
9958   *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
9959      include a #define from the old name to the new. The original intent
9960      was that statically linked binaries could for example just call
9961      SSLeay_add_all_ciphers() to just add ciphers to the table and not
9962      link with digests. This never worked because SSLeay_add_all_digests()
9963      and SSLeay_add_all_ciphers() were in the same source file so calling
9964      one would link with the other. They are now in separate source files.
9965      [Steve Henson]
9966
9967   *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
9968      [Steve Henson]
9969
9970   *) Use a less unusual form of the Miller-Rabin primality test (it used
9971      a binary algorithm for exponentiation integrated into the Miller-Rabin
9972      loop, our standard modexp algorithms are faster).
9973      [Bodo Moeller]
9974
9975   *) Support for the EBCDIC character set completed.
9976      [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
9977
9978   *) Source code cleanups: use const where appropriate, eliminate casts,
9979      use void * instead of char * in lhash.
9980      [Ulf Möller]
9981
9982   *) Bugfix: ssl3_send_server_key_exchange was not restartable
9983      (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
9984      this the server could overwrite ephemeral keys that the client
9985      has already seen).
9986      [Bodo Moeller]
9987
9988   *) Turn DSA_is_prime into a macro that calls BN_is_prime,
9989      using 50 iterations of the Rabin-Miller test.
9990
9991      DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
9992      iterations of the Rabin-Miller test as required by the appendix
9993      to FIPS PUB 186[-1]) instead of DSA_is_prime.
9994      As BN_is_prime_fasttest includes trial division, DSA parameter
9995      generation becomes much faster.
9996
9997      This implies a change for the callback functions in DSA_is_prime
9998      and DSA_generate_parameters: The callback function is called once
9999      for each positive witness in the Rabin-Miller test, not just
10000      occasionally in the inner loop; and the parameters to the
10001      callback function now provide an iteration count for the outer
10002      loop rather than for the current invocation of the inner loop.
10003      DSA_generate_parameters additionally can call the callback
10004      function with an 'iteration count' of -1, meaning that a
10005      candidate has passed the trial division test (when q is generated
10006      from an application-provided seed, trial division is skipped).
10007      [Bodo Moeller]
10008
10009   *) New function BN_is_prime_fasttest that optionally does trial
10010      division before starting the Rabin-Miller test and has
10011      an additional BN_CTX * argument (whereas BN_is_prime always
10012      has to allocate at least one BN_CTX).
10013      'callback(1, -1, cb_arg)' is called when a number has passed the
10014      trial division stage.
10015      [Bodo Moeller]
10016
10017   *) Fix for bug in CRL encoding. The validity dates weren't being handled
10018      as ASN1_TIME.
10019      [Steve Henson]
10020
10021   *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
10022      [Steve Henson]
10023
10024   *) New function BN_pseudo_rand().
10025      [Ulf Möller]
10026
10027   *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
10028      bignum version of BN_from_montgomery() with the working code from
10029      SSLeay 0.9.0 (the word based version is faster anyway), and clean up
10030      the comments.
10031      [Ulf Möller]
10032
10033   *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
10034      made it impossible to use the same SSL_SESSION data structure in
10035      SSL2 clients in multiple threads.
10036      [Bodo Moeller]
10037
10038   *) The return value of RAND_load_file() no longer counts bytes obtained
10039      by stat().  RAND_load_file(..., -1) is new and uses the complete file
10040      to seed the PRNG (previously an explicit byte count was required).
10041      [Ulf Möller, Bodo Möller]
10042
10043   *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
10044      used (char *) instead of (void *) and had casts all over the place.
10045      [Steve Henson]
10046
10047   *) Make BN_generate_prime() return NULL on error if ret!=NULL.
10048      [Ulf Möller]
10049
10050   *) Retain source code compatibility for BN_prime_checks macro:
10051      BN_is_prime(..., BN_prime_checks, ...) now uses
10052      BN_prime_checks_for_size to determine the appropriate number of
10053      Rabin-Miller iterations.
10054      [Ulf Möller]
10055
10056   *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
10057      DH_CHECK_P_NOT_SAFE_PRIME.
10058      (Check if this is true? OpenPGP calls them "strong".)
10059      [Ulf Möller]
10060
10061   *) Merge the functionality of "dh" and "gendh" programs into a new program
10062      "dhparam". The old programs are retained for now but will handle DH keys
10063      (instead of parameters) in future.
10064      [Steve Henson]
10065
10066   *) Make the ciphers, s_server and s_client programs check the return values
10067      when a new cipher list is set.
10068      [Steve Henson]
10069
10070   *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
10071      ciphers. Before when the 56bit ciphers were enabled the sorting was
10072      wrong.
10073
10074      The syntax for the cipher sorting has been extended to support sorting by
10075      cipher-strength (using the strength_bits hard coded in the tables).
10076      The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
10077
10078      Fix a bug in the cipher-command parser: when supplying a cipher command
10079      string with an "undefined" symbol (neither command nor alphanumeric
10080      [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
10081      an error is flagged.
10082
10083      Due to the strength-sorting extension, the code of the
10084      ssl_create_cipher_list() function was completely rearranged. I hope that
10085      the readability was also increased :-)
10086      [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
10087
10088   *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
10089      for the first serial number and places 2 in the serial number file. This
10090      avoids problems when the root CA is created with serial number zero and
10091      the first user certificate has the same issuer name and serial number
10092      as the root CA.
10093      [Steve Henson]
10094
10095   *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
10096      the new code. Add documentation for this stuff.
10097      [Steve Henson]
10098
10099   *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
10100      X509_*() to X509at_*() on the grounds that they don't handle X509
10101      structures and behave in an analogous way to the X509v3 functions:
10102      they shouldn't be called directly but wrapper functions should be used
10103      instead.
10104
10105      So we also now have some wrapper functions that call the X509at functions
10106      when passed certificate requests. (TO DO: similar things can be done with
10107      PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
10108      things. Some of these need some d2i or i2d and print functionality
10109      because they handle more complex structures.)
10110      [Steve Henson]
10111
10112   *) Add missing #ifndefs that caused missing symbols when building libssl
10113      as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
10114      NO_RSA in ssl/s2*.c.
10115      [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
10116
10117   *) Precautions against using the PRNG uninitialized: RAND_bytes() now
10118      has a return value which indicates the quality of the random data
10119      (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
10120      error queue. New function RAND_pseudo_bytes() generates output that is
10121      guaranteed to be unique but not unpredictable. RAND_add is like
10122      RAND_seed, but takes an extra argument for an entropy estimate
10123      (RAND_seed always assumes full entropy).
10124      [Ulf Möller]
10125
10126   *) Do more iterations of Rabin-Miller probable prime test (specifically,
10127      3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
10128      instead of only 2 for all lengths; see BN_prime_checks_for_size definition
10129      in crypto/bn/bn_prime.c for the complete table).  This guarantees a
10130      false-positive rate of at most 2^-80 for random input.
10131      [Bodo Moeller]
10132
10133   *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
10134      [Bodo Moeller]
10135
10136   *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
10137      in the 0.9.5 release), this returns the chain
10138      from an X509_CTX structure with a dup of the stack and all
10139      the X509 reference counts upped: so the stack will exist
10140      after X509_CTX_cleanup() has been called. Modify pkcs12.c
10141      to use this.
10142
10143      Also make SSL_SESSION_print() print out the verify return
10144      code.
10145      [Steve Henson]
10146
10147   *) Add manpage for the pkcs12 command. Also change the default
10148      behaviour so MAC iteration counts are used unless the new
10149      -nomaciter option is used. This improves file security and
10150      only older versions of MSIE (4.0 for example) need it.
10151      [Steve Henson]
10152
10153   *) Honor the no-xxx Configure options when creating .DEF files.
10154      [Ulf Möller]
10155
10156   *) Add PKCS#10 attributes to field table: challengePassword,
10157      unstructuredName and unstructuredAddress. These are taken from
10158      draft PKCS#9 v2.0 but are compatible with v1.2 provided no
10159      international characters are used.
10160
10161      More changes to X509_ATTRIBUTE code: allow the setting of types
10162      based on strings. Remove the 'loc' parameter when adding
10163      attributes because these will be a SET OF encoding which is sorted
10164      in ASN1 order.
10165      [Steve Henson]
10166
10167   *) Initial changes to the 'req' utility to allow request generation
10168      automation. This will allow an application to just generate a template
10169      file containing all the field values and have req construct the
10170      request.
10171
10172      Initial support for X509_ATTRIBUTE handling. Stacks of these are
10173      used all over the place including certificate requests and PKCS#7
10174      structures. They are currently handled manually where necessary with
10175      some primitive wrappers for PKCS#7. The new functions behave in a
10176      manner analogous to the X509 extension functions: they allow
10177      attributes to be looked up by NID and added.
10178
10179      Later something similar to the X509V3 code would be desirable to
10180      automatically handle the encoding, decoding and printing of the
10181      more complex types. The string types like challengePassword can
10182      be handled by the string table functions.
10183
10184      Also modified the multi byte string table handling. Now there is
10185      a 'global mask' which masks out certain types. The table itself
10186      can use the flag STABLE_NO_MASK to ignore the mask setting: this
10187      is useful when for example there is only one permissible type
10188      (as in countryName) and using the mask might result in no valid
10189      types at all.
10190      [Steve Henson]
10191
10192   *) Clean up 'Finished' handling, and add functions SSL_get_finished and
10193      SSL_get_peer_finished to allow applications to obtain the latest
10194      Finished messages sent to the peer or expected from the peer,
10195      respectively.  (SSL_get_peer_finished is usually the Finished message
10196      actually received from the peer, otherwise the protocol will be aborted.)
10197
10198      As the Finished message are message digests of the complete handshake
10199      (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
10200      be used for external authentication procedures when the authentication
10201      provided by SSL/TLS is not desired or is not enough.
10202      [Bodo Moeller]
10203
10204   *) Enhanced support for Alpha Linux is added. Now ./config checks if
10205      the host supports BWX extension and if Compaq C is present on the
10206      $PATH. Just exploiting of the BWX extension results in 20-30%
10207      performance kick for some algorithms, e.g. DES and RC4 to mention
10208      a couple. Compaq C in turn generates ~20% faster code for MD5 and
10209      SHA1.
10210      [Andy Polyakov]
10211
10212   *) Add support for MS "fast SGC". This is arguably a violation of the
10213      SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
10214      weak crypto and after checking the certificate is SGC a second one
10215      with strong crypto. MS SGC stops the first handshake after receiving
10216      the server certificate message and sends a second client hello. Since
10217      a server will typically do all the time consuming operations before
10218      expecting any further messages from the client (server key exchange
10219      is the most expensive) there is little difference between the two.
10220
10221      To get OpenSSL to support MS SGC we have to permit a second client
10222      hello message after we have sent server done. In addition we have to
10223      reset the MAC if we do get this second client hello.
10224      [Steve Henson]
10225
10226   *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
10227      if a DER encoded private key is RSA or DSA traditional format. Changed
10228      d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
10229      format DER encoded private key. Newer code should use PKCS#8 format which
10230      has the key type encoded in the ASN1 structure. Added DER private key
10231      support to pkcs8 application.
10232      [Steve Henson]
10233
10234   *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
10235      ciphersuites has been selected (as required by the SSL 3/TLS 1
10236      specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
10237      is set, we interpret this as a request to violate the specification
10238      (the worst that can happen is a handshake failure, and 'correct'
10239      behaviour would result in a handshake failure anyway).
10240      [Bodo Moeller]
10241
10242   *) In SSL_CTX_add_session, take into account that there might be multiple
10243      SSL_SESSION structures with the same session ID (e.g. when two threads
10244      concurrently obtain them from an external cache).
10245      The internal cache can handle only one SSL_SESSION with a given ID,
10246      so if there's a conflict, we now throw out the old one to achieve
10247      consistency.
10248      [Bodo Moeller]
10249
10250   *) Add OIDs for idea and blowfish in CBC mode. This will allow both
10251      to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
10252      some routines that use cipher OIDs: some ciphers do not have OIDs
10253      defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
10254      example.
10255      [Steve Henson]
10256
10257   *) Simplify the trust setting structure and code. Now we just have
10258      two sequences of OIDs for trusted and rejected settings. These will
10259      typically have values the same as the extended key usage extension
10260      and any application specific purposes.
10261
10262      The trust checking code now has a default behaviour: it will just
10263      check for an object with the same NID as the passed id. Functions can
10264      be provided to override either the default behaviour or the behaviour
10265      for a given id. SSL client, server and email already have functions
10266      in place for compatibility: they check the NID and also return "trusted"
10267      if the certificate is self signed.
10268      [Steve Henson]
10269
10270   *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
10271      traditional format into an EVP_PKEY structure.
10272      [Steve Henson]
10273
10274   *) Add a password callback function PEM_cb() which either prompts for
10275      a password if usr_data is NULL or otherwise assumes it is a null
10276      terminated password. Allow passwords to be passed on command line
10277      environment or config files in a few more utilities.
10278      [Steve Henson]
10279
10280   *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
10281      keys. Add some short names for PKCS#8 PBE algorithms and allow them
10282      to be specified on the command line for the pkcs8 and pkcs12 utilities.
10283      Update documentation.
10284      [Steve Henson]
10285
10286   *) Support for ASN1 "NULL" type. This could be handled before by using
10287      ASN1_TYPE but there wasn't any function that would try to read a NULL
10288      and produce an error if it couldn't. For compatibility we also have
10289      ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
10290      don't allocate anything because they don't need to.
10291      [Steve Henson]
10292
10293   *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
10294      for details.
10295      [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
10296
10297   *) Rebuild of the memory allocation routines used by OpenSSL code and
10298      possibly others as well.  The purpose is to make an interface that
10299      provide hooks so anyone can build a separate set of allocation and
10300      deallocation routines to be used by OpenSSL, for example memory
10301      pool implementations, or something else, which was previously hard
10302      since Malloc(), Realloc() and Free() were defined as macros having
10303      the values malloc, realloc and free, respectively (except for Win32
10304      compilations).  The same is provided for memory debugging code.
10305      OpenSSL already comes with functionality to find memory leaks, but
10306      this gives people a chance to debug other memory problems.
10307
10308      With these changes, a new set of functions and macros have appeared:
10309
10310        CRYPTO_set_mem_debug_functions()         [F]
10311        CRYPTO_get_mem_debug_functions()         [F]
10312        CRYPTO_dbg_set_options()                 [F]
10313        CRYPTO_dbg_get_options()                 [F]
10314        CRYPTO_malloc_debug_init()               [M]
10315
10316      The memory debug functions are NULL by default, unless the library
10317      is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
10318      wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
10319      gives the standard debugging functions that come with OpenSSL) or
10320      CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
10321      provided by the library user) must be used.  When the standard
10322      debugging functions are used, CRYPTO_dbg_set_options can be used to
10323      request additional information:
10324      CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
10325      the CRYPTO_MDEBUG_xxx macro when compiling the library.
10326
10327      Also, things like CRYPTO_set_mem_functions will always give the
10328      expected result (the new set of functions is used for allocation
10329      and deallocation) at all times, regardless of platform and compiler
10330      options.
10331
10332      To finish it up, some functions that were never use in any other
10333      way than through macros have a new API and new semantic:
10334
10335        CRYPTO_dbg_malloc()
10336        CRYPTO_dbg_realloc()
10337        CRYPTO_dbg_free()
10338
10339      All macros of value have retained their old syntax.
10340      [Richard Levitte and Bodo Moeller]
10341
10342   *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
10343      ordering of SMIMECapabilities wasn't in "strength order" and there
10344      was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
10345      algorithm.
10346      [Steve Henson]
10347
10348   *) Some ASN1 types with illegal zero length encoding (INTEGER,
10349      ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
10350      [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
10351
10352   *) Merge in my S/MIME library for OpenSSL. This provides a simple
10353      S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
10354      functionality to handle multipart/signed properly) and a utility
10355      called 'smime' to call all this stuff. This is based on code I
10356      originally wrote for Celo who have kindly allowed it to be
10357      included in OpenSSL.
10358      [Steve Henson]
10359
10360   *) Add variants des_set_key_checked and des_set_key_unchecked of
10361      des_set_key (aka des_key_sched).  Global variable des_check_key
10362      decides which of these is called by des_set_key; this way
10363      des_check_key behaves as it always did, but applications and
10364      the library itself, which was buggy for des_check_key == 1,
10365      have a cleaner way to pick the version they need.
10366      [Bodo Moeller]
10367
10368   *) New function PKCS12_newpass() which changes the password of a
10369      PKCS12 structure.
10370      [Steve Henson]
10371
10372   *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
10373      dynamic mix. In both cases the ids can be used as an index into the
10374      table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
10375      functions so they accept a list of the field values and the
10376      application doesn't need to directly manipulate the X509_TRUST
10377      structure.
10378      [Steve Henson]
10379
10380   *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
10381      need initialising.
10382      [Steve Henson]
10383
10384   *) Modify the way the V3 extension code looks up extensions. This now
10385      works in a similar way to the object code: we have some "standard"
10386      extensions in a static table which is searched with OBJ_bsearch()
10387      and the application can add dynamic ones if needed. The file
10388      crypto/x509v3/ext_dat.h now has the info: this file needs to be
10389      updated whenever a new extension is added to the core code and kept
10390      in ext_nid order. There is a simple program 'tabtest.c' which checks
10391      this. New extensions are not added too often so this file can readily
10392      be maintained manually.
10393
10394      There are two big advantages in doing things this way. The extensions
10395      can be looked up immediately and no longer need to be "added" using
10396      X509V3_add_standard_extensions(): this function now does nothing.
10397      [Side note: I get *lots* of email saying the extension code doesn't
10398       work because people forget to call this function]
10399      Also no dynamic allocation is done unless new extensions are added:
10400      so if we don't add custom extensions there is no need to call
10401      X509V3_EXT_cleanup().
10402      [Steve Henson]
10403
10404   *) Modify enc utility's salting as follows: make salting the default. Add a
10405      magic header, so unsalted files fail gracefully instead of just decrypting
10406      to garbage. This is because not salting is a big security hole, so people
10407      should be discouraged from doing it.
10408      [Ben Laurie]
10409
10410   *) Fixes and enhancements to the 'x509' utility. It allowed a message
10411      digest to be passed on the command line but it only used this
10412      parameter when signing a certificate. Modified so all relevant
10413      operations are affected by the digest parameter including the
10414      -fingerprint and -x509toreq options. Also -x509toreq choked if a
10415      DSA key was used because it didn't fix the digest.
10416      [Steve Henson]
10417
10418   *) Initial certificate chain verify code. Currently tests the untrusted
10419      certificates for consistency with the verify purpose (which is set
10420      when the X509_STORE_CTX structure is set up) and checks the pathlength.
10421
10422      There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
10423      this is because it will reject chains with invalid extensions whereas
10424      every previous version of OpenSSL and SSLeay made no checks at all.
10425
10426      Trust code: checks the root CA for the relevant trust settings. Trust
10427      settings have an initial value consistent with the verify purpose: e.g.
10428      if the verify purpose is for SSL client use it expects the CA to be
10429      trusted for SSL client use. However the default value can be changed to
10430      permit custom trust settings: one example of this would be to only trust
10431      certificates from a specific "secure" set of CAs.
10432
10433      Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
10434      which should be used for version portability: especially since the
10435      verify structure is likely to change more often now.
10436
10437      SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
10438      to set them. If not set then assume SSL clients will verify SSL servers
10439      and vice versa.
10440
10441      Two new options to the verify program: -untrusted allows a set of
10442      untrusted certificates to be passed in and -purpose which sets the
10443      intended purpose of the certificate. If a purpose is set then the
10444      new chain verify code is used to check extension consistency.
10445      [Steve Henson]
10446
10447   *) Support for the authority information access extension.
10448      [Steve Henson]
10449
10450   *) Modify RSA and DSA PEM read routines to transparently handle
10451      PKCS#8 format private keys. New *_PUBKEY_* functions that handle
10452      public keys in a format compatible with certificate
10453      SubjectPublicKeyInfo structures. Unfortunately there were already
10454      functions called *_PublicKey_* which used various odd formats so
10455      these are retained for compatibility: however the DSA variants were
10456      never in a public release so they have been deleted. Changed dsa/rsa
10457      utilities to handle the new format: note no releases ever handled public
10458      keys so we should be OK.
10459
10460      The primary motivation for this change is to avoid the same fiasco
10461      that dogs private keys: there are several incompatible private key
10462      formats some of which are standard and some OpenSSL specific and
10463      require various evil hacks to allow partial transparent handling and
10464      even then it doesn't work with DER formats. Given the option anything
10465      other than PKCS#8 should be dumped: but the other formats have to
10466      stay in the name of compatibility.
10467
10468      With public keys and the benefit of hindsight one standard format
10469      is used which works with EVP_PKEY, RSA or DSA structures: though
10470      it clearly returns an error if you try to read the wrong kind of key.
10471
10472      Added a -pubkey option to the 'x509' utility to output the public key.
10473      Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
10474      (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
10475      EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
10476      that do the same as the EVP_PKEY_assign_*() except they up the
10477      reference count of the added key (they don't "swallow" the
10478      supplied key).
10479      [Steve Henson]
10480
10481   *) Fixes to crypto/x509/by_file.c the code to read in certificates and
10482      CRLs would fail if the file contained no certificates or no CRLs:
10483      added a new function to read in both types and return the number
10484      read: this means that if none are read it will be an error. The
10485      DER versions of the certificate and CRL reader would always fail
10486      because it isn't possible to mix certificates and CRLs in DER format
10487      without choking one or the other routine. Changed this to just read
10488      a certificate: this is the best we can do. Also modified the code
10489      in apps/verify.c to take notice of return codes: it was previously
10490      attempting to read in certificates from NULL pointers and ignoring
10491      any errors: this is one reason why the cert and CRL reader seemed
10492      to work. It doesn't check return codes from the default certificate
10493      routines: these may well fail if the certificates aren't installed.
10494      [Steve Henson]
10495
10496   *) Code to support otherName option in GeneralName.
10497      [Steve Henson]
10498
10499   *) First update to verify code. Change the verify utility
10500      so it warns if it is passed a self signed certificate:
10501      for consistency with the normal behaviour. X509_verify
10502      has been modified to it will now verify a self signed
10503      certificate if *exactly* the same certificate appears
10504      in the store: it was previously impossible to trust a
10505      single self signed certificate. This means that:
10506      openssl verify ss.pem
10507      now gives a warning about a self signed certificate but
10508      openssl verify -CAfile ss.pem ss.pem
10509      is OK.
10510      [Steve Henson]
10511
10512   *) For servers, store verify_result in SSL_SESSION data structure
10513      (and add it to external session representation).
10514      This is needed when client certificate verifications fails,
10515      but an application-provided verification callback (set by
10516      SSL_CTX_set_cert_verify_callback) allows accepting the session
10517      anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
10518      but returns 1): When the session is reused, we have to set
10519      ssl->verify_result to the appropriate error code to avoid
10520      security holes.
10521      [Bodo Moeller, problem pointed out by Lutz Jaenicke]
10522
10523   *) Fix a bug in the new PKCS#7 code: it didn't consider the
10524      case in PKCS7_dataInit() where the signed PKCS7 structure
10525      didn't contain any existing data because it was being created.
10526      [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
10527
10528   *) Add a salt to the key derivation routines in enc.c. This
10529      forms the first 8 bytes of the encrypted file. Also add a
10530      -S option to allow a salt to be input on the command line.
10531      [Steve Henson]
10532
10533   *) New function X509_cmp(). Oddly enough there wasn't a function
10534      to compare two certificates. We do this by working out the SHA1
10535      hash and comparing that. X509_cmp() will be needed by the trust
10536      code.
10537      [Steve Henson]
10538
10539   *) SSL_get1_session() is like SSL_get_session(), but increments
10540      the reference count in the SSL_SESSION returned.
10541      [Geoff Thorpe <geoff@eu.c2.net>]
10542
10543   *) Fix for 'req': it was adding a null to request attributes.
10544      Also change the X509_LOOKUP and X509_INFO code to handle
10545      certificate auxiliary information.
10546      [Steve Henson]
10547
10548   *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
10549      the 'enc' command.
10550      [Steve Henson]
10551
10552   *) Add the possibility to add extra information to the memory leak
10553      detecting output, to form tracebacks, showing from where each
10554      allocation was originated: CRYPTO_push_info("constant string") adds
10555      the string plus current file name and line number to a per-thread
10556      stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
10557      is like calling CYRPTO_pop_info() until the stack is empty.
10558      Also updated memory leak detection code to be multi-thread-safe.
10559      [Richard Levitte]
10560
10561   *) Add options -text and -noout to pkcs7 utility and delete the
10562      encryption options which never did anything. Update docs.
10563      [Steve Henson]
10564
10565   *) Add options to some of the utilities to allow the pass phrase
10566      to be included on either the command line (not recommended on
10567      OSes like Unix) or read from the environment. Update the
10568      manpages and fix a few bugs.
10569      [Steve Henson]
10570
10571   *) Add a few manpages for some of the openssl commands.
10572      [Steve Henson]
10573
10574   *) Fix the -revoke option in ca. It was freeing up memory twice,
10575      leaking and not finding already revoked certificates.
10576      [Steve Henson]
10577
10578   *) Extensive changes to support certificate auxiliary information.
10579      This involves the use of X509_CERT_AUX structure and X509_AUX
10580      functions. An X509_AUX function such as PEM_read_X509_AUX()
10581      can still read in a certificate file in the usual way but it
10582      will also read in any additional "auxiliary information". By
10583      doing things this way a fair degree of compatibility can be
10584      retained: existing certificates can have this information added
10585      using the new 'x509' options.
10586
10587      Current auxiliary information includes an "alias" and some trust
10588      settings. The trust settings will ultimately be used in enhanced
10589      certificate chain verification routines: currently a certificate
10590      can only be trusted if it is self signed and then it is trusted
10591      for all purposes.
10592      [Steve Henson]
10593
10594   *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
10595      The problem was that one of the replacement routines had not been working
10596      since SSLeay releases.  For now the offending routine has been replaced
10597      with non-optimised assembler.  Even so, this now gives around 95%
10598      performance improvement for 1024 bit RSA signs.
10599      [Mark Cox]
10600
10601   *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
10602      handling. Most clients have the effective key size in bits equal to
10603      the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
10604      A few however don't do this and instead use the size of the decrypted key
10605      to determine the RC2 key length and the AlgorithmIdentifier to determine
10606      the effective key length. In this case the effective key length can still
10607      be 40 bits but the key length can be 168 bits for example. This is fixed
10608      by manually forcing an RC2 key into the EVP_PKEY structure because the
10609      EVP code can't currently handle unusual RC2 key sizes: it always assumes
10610      the key length and effective key length are equal.
10611      [Steve Henson]
10612
10613   *) Add a bunch of functions that should simplify the creation of
10614      X509_NAME structures. Now you should be able to do:
10615      X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
10616      and have it automatically work out the correct field type and fill in
10617      the structures. The more adventurous can try:
10618      X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
10619      and it will (hopefully) work out the correct multibyte encoding.
10620      [Steve Henson]
10621
10622   *) Change the 'req' utility to use the new field handling and multibyte
10623      copy routines. Before the DN field creation was handled in an ad hoc
10624      way in req, ca, and x509 which was rather broken and didn't support
10625      BMPStrings or UTF8Strings. Since some software doesn't implement
10626      BMPStrings or UTF8Strings yet, they can be enabled using the config file
10627      using the dirstring_type option. See the new comment in the default
10628      openssl.cnf for more info.
10629      [Steve Henson]
10630
10631   *) Make crypto/rand/md_rand.c more robust:
10632      - Assure unique random numbers after fork().
10633      - Make sure that concurrent threads access the global counter and
10634        md serializably so that we never lose entropy in them
10635        or use exactly the same state in multiple threads.
10636        Access to the large state is not always serializable because
10637        the additional locking could be a performance killer, and
10638        md should be large enough anyway.
10639      [Bodo Moeller]
10640
10641   *) New file apps/app_rand.c with commonly needed functionality
10642      for handling the random seed file.
10643
10644      Use the random seed file in some applications that previously did not:
10645           ca,
10646           dsaparam -genkey (which also ignored its '-rand' option),
10647           s_client,
10648           s_server,
10649           x509 (when signing).
10650      Except on systems with /dev/urandom, it is crucial to have a random
10651      seed file at least for key creation, DSA signing, and for DH exchanges;
10652      for RSA signatures we could do without one.
10653
10654      gendh and gendsa (unlike genrsa) used to read only the first byte
10655      of each file listed in the '-rand' option.  The function as previously
10656      found in genrsa is now in app_rand.c and is used by all programs
10657      that support '-rand'.
10658      [Bodo Moeller]
10659
10660   *) In RAND_write_file, use mode 0600 for creating files;
10661      don't just chmod when it may be too late.
10662      [Bodo Moeller]
10663
10664   *) Report an error from X509_STORE_load_locations
10665      when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
10666      [Bill Perry]
10667
10668   *) New function ASN1_mbstring_copy() this copies a string in either
10669      ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
10670      into an ASN1_STRING type. A mask of permissible types is passed
10671      and it chooses the "minimal" type to use or an error if not type
10672      is suitable.
10673      [Steve Henson]
10674
10675   *) Add function equivalents to the various macros in asn1.h. The old
10676      macros are retained with an M_ prefix. Code inside the library can
10677      use the M_ macros. External code (including the openssl utility)
10678      should *NOT* in order to be "shared library friendly".
10679      [Steve Henson]
10680
10681   *) Add various functions that can check a certificate's extensions
10682      to see if it usable for various purposes such as SSL client,
10683      server or S/MIME and CAs of these types. This is currently
10684      VERY EXPERIMENTAL but will ultimately be used for certificate chain
10685      verification. Also added a -purpose flag to x509 utility to
10686      print out all the purposes.
10687      [Steve Henson]
10688
10689   *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
10690      functions.
10691      [Steve Henson]
10692
10693   *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
10694      for, obtain and decode and extension and obtain its critical flag.
10695      This allows all the necessary extension code to be handled in a
10696      single function call.
10697      [Steve Henson]
10698
10699   *) RC4 tune-up featuring 30-40% performance improvement on most RISC
10700      platforms. See crypto/rc4/rc4_enc.c for further details.
10701      [Andy Polyakov]
10702
10703   *) New -noout option to asn1parse. This causes no output to be produced
10704      its main use is when combined with -strparse and -out to extract data
10705      from a file (which may not be in ASN.1 format).
10706      [Steve Henson]
10707
10708   *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
10709      when producing the local key id.
10710      [Richard Levitte <levitte@stacken.kth.se>]
10711
10712   *) New option -dhparam in s_server. This allows a DH parameter file to be
10713      stated explicitly. If it is not stated then it tries the first server
10714      certificate file. The previous behaviour hard coded the filename
10715      "server.pem".
10716      [Steve Henson]
10717
10718   *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
10719      a public key to be input or output. For example:
10720      openssl rsa -in key.pem -pubout -out pubkey.pem
10721      Also added necessary DSA public key functions to handle this.
10722      [Steve Henson]
10723
10724   *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
10725      in the message. This was handled by allowing
10726      X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
10727      [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
10728
10729   *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
10730      to the end of the strings whereas this didn't. This would cause problems
10731      if strings read with d2i_ASN1_bytes() were later modified.
10732      [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
10733
10734   *) Fix for base64 decode bug. When a base64 bio reads only one line of
10735      data and it contains EOF it will end up returning an error. This is
10736      caused by input 46 bytes long. The cause is due to the way base64
10737      BIOs find the start of base64 encoded data. They do this by trying a
10738      trial decode on each line until they find one that works. When they
10739      do a flag is set and it starts again knowing it can pass all the
10740      data directly through the decoder. Unfortunately it doesn't reset
10741      the context it uses. This means that if EOF is reached an attempt
10742      is made to pass two EOFs through the context and this causes the
10743      resulting error. This can also cause other problems as well. As is
10744      usual with these problems it takes *ages* to find and the fix is
10745      trivial: move one line.
10746      [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
10747
10748   *) Ugly workaround to get s_client and s_server working under Windows. The
10749      old code wouldn't work because it needed to select() on sockets and the
10750      tty (for keypresses and to see if data could be written). Win32 only
10751      supports select() on sockets so we select() with a 1s timeout on the
10752      sockets and then see if any characters are waiting to be read, if none
10753      are present then we retry, we also assume we can always write data to
10754      the tty. This isn't nice because the code then blocks until we've
10755      received a complete line of data and it is effectively polling the
10756      keyboard at 1s intervals: however it's quite a bit better than not
10757      working at all :-) A dedicated Windows application might handle this
10758      with an event loop for example.
10759      [Steve Henson]
10760
10761   *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
10762      and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
10763      will be called when RSA_sign() and RSA_verify() are used. This is useful
10764      if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
10765      For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
10766      should *not* be used: RSA_sign() and RSA_verify() must be used instead.
10767      This necessitated the support of an extra signature type NID_md5_sha1
10768      for SSL signatures and modifications to the SSL library to use it instead
10769      of calling RSA_public_decrypt() and RSA_private_encrypt().
10770      [Steve Henson]
10771
10772   *) Add new -verify -CAfile and -CApath options to the crl program, these
10773      will lookup a CRL issuers certificate and verify the signature in a
10774      similar way to the verify program. Tidy up the crl program so it
10775      no longer accesses structures directly. Make the ASN1 CRL parsing a bit
10776      less strict. It will now permit CRL extensions even if it is not
10777      a V2 CRL: this will allow it to tolerate some broken CRLs.
10778      [Steve Henson]
10779
10780   *) Initialize all non-automatic variables each time one of the openssl
10781      sub-programs is started (this is necessary as they may be started
10782      multiple times from the "OpenSSL>" prompt).
10783      [Lennart Bang, Bodo Moeller]
10784
10785   *) Preliminary compilation option RSA_NULL which disables RSA crypto without
10786      removing all other RSA functionality (this is what NO_RSA does). This
10787      is so (for example) those in the US can disable those operations covered
10788      by the RSA patent while allowing storage and parsing of RSA keys and RSA
10789      key generation.
10790      [Steve Henson]
10791
10792   *) Non-copying interface to BIO pairs.
10793      (still largely untested)
10794      [Bodo Moeller]
10795
10796   *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
10797      ASCII string. This was handled independently in various places before.
10798      [Steve Henson]
10799
10800   *) New functions UTF8_getc() and UTF8_putc() that parse and generate
10801      UTF8 strings a character at a time.
10802      [Steve Henson]
10803
10804   *) Use client_version from client hello to select the protocol
10805      (s23_srvr.c) and for RSA client key exchange verification
10806      (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
10807      [Bodo Moeller]
10808
10809   *) Add various utility functions to handle SPKACs, these were previously
10810      handled by poking round in the structure internals. Added new function
10811      NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
10812      print, verify and generate SPKACs. Based on an original idea from
10813      Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
10814      [Steve Henson]
10815
10816   *) RIPEMD160 is operational on all platforms and is back in 'make test'.
10817      [Andy Polyakov]
10818
10819   *) Allow the config file extension section to be overwritten on the
10820      command line. Based on an original idea from Massimiliano Pala
10821      <madwolf@comune.modena.it>. The new option is called -extensions
10822      and can be applied to ca, req and x509. Also -reqexts to override
10823      the request extensions in req and -crlexts to override the crl extensions
10824      in ca.
10825      [Steve Henson]
10826
10827   *) Add new feature to the SPKAC handling in ca.  Now you can include
10828      the same field multiple times by preceding it by "XXXX." for example:
10829      1.OU="Unit name 1"
10830      2.OU="Unit name 2"
10831      this is the same syntax as used in the req config file.
10832      [Steve Henson]
10833
10834   *) Allow certificate extensions to be added to certificate requests. These
10835      are specified in a 'req_extensions' option of the req section of the
10836      config file. They can be printed out with the -text option to req but
10837      are otherwise ignored at present.
10838      [Steve Henson]
10839
10840   *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
10841      data read consists of only the final block it would not decrypted because
10842      EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
10843      A misplaced 'break' also meant the decrypted final block might not be
10844      copied until the next read.
10845      [Steve Henson]
10846
10847   *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
10848      a few extra parameters to the DH structure: these will be useful if
10849      for example we want the value of 'q' or implement X9.42 DH.
10850      [Steve Henson]
10851
10852   *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
10853      provides hooks that allow the default DSA functions or functions on a
10854      "per key" basis to be replaced. This allows hardware acceleration and
10855      hardware key storage to be handled without major modification to the
10856      library. Also added low level modexp hooks and CRYPTO_EX structure and
10857      associated functions.
10858      [Steve Henson]
10859
10860   *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
10861      as "read only": it can't be written to and the buffer it points to will
10862      not be freed. Reading from a read only BIO is much more efficient than
10863      a normal memory BIO. This was added because there are several times when
10864      an area of memory needs to be read from a BIO. The previous method was
10865      to create a memory BIO and write the data to it, this results in two
10866      copies of the data and an O(n^2) reading algorithm. There is a new
10867      function BIO_new_mem_buf() which creates a read only memory BIO from
10868      an area of memory. Also modified the PKCS#7 routines to use read only
10869      memory BIOs.
10870      [Steve Henson]
10871
10872   *) Bugfix: ssl23_get_client_hello did not work properly when called in
10873      state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
10874      a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
10875      but a retry condition occurred while trying to read the rest.
10876      [Bodo Moeller]
10877
10878   *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
10879      NID_pkcs7_encrypted by default: this was wrong since this should almost
10880      always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
10881      the encrypted data type: this is a more sensible place to put it and it
10882      allows the PKCS#12 code to be tidied up that duplicated this
10883      functionality.
10884      [Steve Henson]
10885
10886   *) Changed obj_dat.pl script so it takes its input and output files on
10887      the command line. This should avoid shell escape redirection problems
10888      under Win32.
10889      [Steve Henson]
10890
10891   *) Initial support for certificate extension requests, these are included
10892      in things like Xenroll certificate requests. Included functions to allow
10893      extensions to be obtained and added.
10894      [Steve Henson]
10895
10896   *) -crlf option to s_client and s_server for sending newlines as
10897      CRLF (as required by many protocols).
10898      [Bodo Moeller]
10899
10900  Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
10901
10902   *) Install libRSAglue.a when OpenSSL is built with RSAref.
10903      [Ralf S. Engelschall]
10904
10905   *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
10906      [Andrija Antonijevic <TheAntony2@bigfoot.com>]
10907
10908   *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
10909      program.
10910      [Steve Henson]
10911
10912   *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
10913      DH parameters/keys (q is lost during that conversion, but the resulting
10914      DH parameters contain its length).
10915
10916      For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
10917      much faster than DH_generate_parameters (which creates parameters
10918      where p = 2*q + 1), and also the smaller q makes DH computations
10919      much more efficient (160-bit exponentiation instead of 1024-bit
10920      exponentiation); so this provides a convenient way to support DHE
10921      ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
10922      utter importance to use
10923          SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
10924      or
10925          SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
10926      when such DH parameters are used, because otherwise small subgroup
10927      attacks may become possible!
10928      [Bodo Moeller]
10929
10930   *) Avoid memory leak in i2d_DHparams.
10931      [Bodo Moeller]
10932
10933   *) Allow the -k option to be used more than once in the enc program:
10934      this allows the same encrypted message to be read by multiple recipients.
10935      [Steve Henson]
10936
10937   *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
10938      an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
10939      it will always use the numerical form of the OID, even if it has a short
10940      or long name.
10941      [Steve Henson]
10942
10943   *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
10944      method only got called if p,q,dmp1,dmq1,iqmp components were present,
10945      otherwise bn_mod_exp was called. In the case of hardware keys for example
10946      no private key components need be present and it might store extra data
10947      in the RSA structure, which cannot be accessed from bn_mod_exp.
10948      By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
10949      private key operations.
10950      [Steve Henson]
10951
10952   *) Added support for SPARC Linux.
10953      [Andy Polyakov]
10954
10955   *) pem_password_cb function type incompatibly changed from
10956           typedef int pem_password_cb(char *buf, int size, int rwflag);
10957      to
10958           ....(char *buf, int size, int rwflag, void *userdata);
10959      so that applications can pass data to their callbacks:
10960      The PEM[_ASN1]_{read,write}... functions and macros now take an
10961      additional void * argument, which is just handed through whenever
10962      the password callback is called.
10963      [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
10964
10965      New function SSL_CTX_set_default_passwd_cb_userdata.
10966
10967      Compatibility note: As many C implementations push function arguments
10968      onto the stack in reverse order, the new library version is likely to
10969      interoperate with programs that have been compiled with the old
10970      pem_password_cb definition (PEM_whatever takes some data that
10971      happens to be on the stack as its last argument, and the callback
10972      just ignores this garbage); but there is no guarantee whatsoever that
10973      this will work.
10974
10975   *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
10976      (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
10977      problems not only on Windows, but also on some Unix platforms.
10978      To avoid problematic command lines, these definitions are now in an
10979      auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
10980      for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
10981      [Bodo Moeller]
10982
10983   *) MIPS III/IV assembler module is reimplemented.
10984      [Andy Polyakov]
10985
10986   *) More DES library cleanups: remove references to srand/rand and
10987      delete an unused file.
10988      [Ulf Möller]
10989
10990   *) Add support for the the free Netwide assembler (NASM) under Win32,
10991      since not many people have MASM (ml) and it can be hard to obtain.
10992      This is currently experimental but it seems to work OK and pass all
10993      the tests. Check out INSTALL.W32 for info.
10994      [Steve Henson]
10995
10996   *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
10997      without temporary keys kept an extra copy of the server key,
10998      and connections with temporary keys did not free everything in case
10999      of an error.
11000      [Bodo Moeller]
11001
11002   *) New function RSA_check_key and new openssl rsa option -check
11003      for verifying the consistency of RSA keys.
11004      [Ulf Moeller, Bodo Moeller]
11005
11006   *) Various changes to make Win32 compile work:
11007      1. Casts to avoid "loss of data" warnings in p5_crpt2.c
11008      2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
11009         comparison" warnings.
11010      3. Add sk_<TYPE>_sort to DEF file generator and do make update.
11011      [Steve Henson]
11012
11013   *) Add a debugging option to PKCS#5 v2 key generation function: when
11014      you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
11015      derived keys are printed to stderr.
11016      [Steve Henson]
11017
11018   *) Copy the flags in ASN1_STRING_dup().
11019      [Roman E. Pavlov <pre@mo.msk.ru>]
11020
11021   *) The x509 application mishandled signing requests containing DSA
11022      keys when the signing key was also DSA and the parameters didn't match.
11023
11024      It was supposed to omit the parameters when they matched the signing key:
11025      the verifying software was then supposed to automatically use the CA's
11026      parameters if they were absent from the end user certificate.
11027
11028      Omitting parameters is no longer recommended. The test was also
11029      the wrong way round! This was probably due to unusual behaviour in
11030      EVP_cmp_parameters() which returns 1 if the parameters match.
11031      This meant that parameters were omitted when they *didn't* match and
11032      the certificate was useless. Certificates signed with 'ca' didn't have
11033      this bug.
11034      [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
11035
11036   *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
11037      The interface is as follows:
11038      Applications can use
11039          CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
11040          CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
11041      "off" is now the default.
11042      The library internally uses
11043          CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
11044          CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
11045      to disable memory-checking temporarily.
11046
11047      Some inconsistent states that previously were possible (and were
11048      even the default) are now avoided.
11049
11050      -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
11051      with each memory chunk allocated; this is occasionally more helpful
11052      than just having a counter.
11053
11054      -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
11055
11056      -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
11057      extensions.
11058      [Bodo Moeller]
11059
11060   *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
11061      which largely parallels "options", but is for changing API behaviour,
11062      whereas "options" are about protocol behaviour.
11063      Initial "mode" flags are:
11064
11065      SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
11066                                      a single record has been written.
11067      SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
11068                                      retries use the same buffer location.
11069                                      (But all of the contents must be
11070                                      copied!)
11071      [Bodo Moeller]
11072
11073   *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
11074      worked.
11075
11076   *) Fix problems with no-hmac etc.
11077      [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
11078
11079   *) New functions RSA_get_default_method(), RSA_set_method() and
11080      RSA_get_method(). These allows replacement of RSA_METHODs without having
11081      to mess around with the internals of an RSA structure.
11082      [Steve Henson]
11083
11084   *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
11085      Also really enable memory leak checks in openssl.c and in some
11086      test programs.
11087      [Chad C. Mulligan, Bodo Moeller]
11088
11089   *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
11090      up the length of negative integers. This has now been simplified to just
11091      store the length when it is first determined and use it later, rather
11092      than trying to keep track of where data is copied and updating it to
11093      point to the end.
11094      [Steve Henson, reported by Brien Wheeler
11095       <bwheeler@authentica-security.com>]
11096
11097   *) Add a new function PKCS7_signatureVerify. This allows the verification
11098      of a PKCS#7 signature but with the signing certificate passed to the
11099      function itself. This contrasts with PKCS7_dataVerify which assumes the
11100      certificate is present in the PKCS#7 structure. This isn't always the
11101      case: certificates can be omitted from a PKCS#7 structure and be
11102      distributed by "out of band" means (such as a certificate database).
11103      [Steve Henson]
11104
11105   *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
11106      function prototypes in pem.h, also change util/mkdef.pl to add the
11107      necessary function names.
11108      [Steve Henson]
11109
11110   *) mk1mf.pl (used by Windows builds) did not properly read the
11111      options set by Configure in the top level Makefile, and Configure
11112      was not even able to write more than one option correctly.
11113      Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
11114      [Bodo Moeller]
11115
11116   *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
11117      file to be loaded from a BIO or FILE pointer. The BIO version will
11118      for example allow memory BIOs to contain config info.
11119      [Steve Henson]
11120
11121   *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
11122      Whoever hopes to achieve shared-library compatibility across versions
11123      must use this, not the compile-time macro.
11124      (Exercise 0.9.4: Which is the minimum library version required by
11125      such programs?)
11126      Note: All this applies only to multi-threaded programs, others don't
11127      need locks.
11128      [Bodo Moeller]
11129
11130   *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
11131      through a BIO pair triggered the default case, i.e.
11132      SSLerr(...,SSL_R_UNKNOWN_STATE).
11133      [Bodo Moeller]
11134
11135   *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
11136      can use the SSL library even if none of the specific BIOs is
11137      appropriate.
11138      [Bodo Moeller]
11139
11140   *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
11141      for the encoded length.
11142      [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
11143
11144   *) Add initial documentation of the X509V3 functions.
11145      [Steve Henson]
11146
11147   *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and
11148      PEM_write_bio_PKCS8PrivateKey() that are equivalent to
11149      PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
11150      secure PKCS#8 private key format with a high iteration count.
11151      [Steve Henson]
11152
11153   *) Fix determination of Perl interpreter: A perl or perl5
11154      _directory_ in $PATH was also accepted as the interpreter.
11155      [Ralf S. Engelschall]
11156
11157   *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
11158      wrong with it but it was very old and did things like calling
11159      PEM_ASN1_read() directly and used MD5 for the hash not to mention some
11160      unusual formatting.
11161      [Steve Henson]
11162
11163   *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
11164      to use the new extension code.
11165      [Steve Henson]
11166
11167   *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
11168      with macros. This should make it easier to change their form, add extra
11169      arguments etc. Fix a few PEM prototypes which didn't have cipher as a
11170      constant.
11171      [Steve Henson]
11172
11173   *) Add to configuration table a new entry that can specify an alternative
11174      name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
11175      according to Mark Crispin <MRC@Panda.COM>.
11176      [Bodo Moeller]
11177
11178 #if 0
11179   *) DES CBC did not update the IV. Weird.
11180      [Ben Laurie]
11181 #else
11182      des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
11183      Changing the behaviour of the former might break existing programs --
11184      where IV updating is needed, des_ncbc_encrypt can be used.
11185 #endif
11186
11187   *) When bntest is run from "make test" it drives bc to check its
11188      calculations, as well as internally checking them. If an internal check
11189      fails, it needs to cause bc to give a non-zero result or make test carries
11190      on without noticing the failure. Fixed.
11191      [Ben Laurie]
11192
11193   *) DES library cleanups.
11194      [Ulf Möller]
11195
11196   *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
11197      used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
11198      ciphers. NOTE: although the key derivation function has been verified
11199      against some published test vectors it has not been extensively tested
11200      yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
11201      of v2.0.
11202      [Steve Henson]
11203
11204   *) Instead of "mkdir -p", which is not fully portable, use new
11205      Perl script "util/mkdir-p.pl".
11206      [Bodo Moeller]
11207
11208   *) Rewrite the way password based encryption (PBE) is handled. It used to
11209      assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
11210      structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
11211      but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
11212      the 'parameter' field of the AlgorithmIdentifier is passed to the
11213      underlying key generation function so it must do its own ASN1 parsing.
11214      This has also changed the EVP_PBE_CipherInit() function which now has a
11215      'parameter' argument instead of literal salt and iteration count values
11216      and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
11217      [Steve Henson]
11218
11219   *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
11220      and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
11221      Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
11222      KEY" because this clashed with PKCS#8 unencrypted string. Since this
11223      value was just used as a "magic string" and not used directly its
11224      value doesn't matter.
11225      [Steve Henson]
11226
11227   *) Introduce some semblance of const correctness to BN. Shame C doesn't
11228      support mutable.
11229      [Ben Laurie]
11230
11231   *) "linux-sparc64" configuration (ultrapenguin).
11232      [Ray Miller <ray.miller@oucs.ox.ac.uk>]
11233      "linux-sparc" configuration.
11234      [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
11235
11236   *) config now generates no-xxx options for missing ciphers.
11237      [Ulf Möller]
11238
11239   *) Support the EBCDIC character set (work in progress).
11240      File ebcdic.c not yet included because it has a different license.
11241      [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
11242
11243   *) Support BS2000/OSD-POSIX.
11244      [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
11245
11246   *) Make callbacks for key generation use void * instead of char *.
11247      [Ben Laurie]
11248
11249   *) Make S/MIME samples compile (not yet tested).
11250      [Ben Laurie]
11251
11252   *) Additional typesafe stacks.
11253      [Ben Laurie]
11254
11255   *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
11256      [Bodo Moeller]
11257
11258
11259  Changes between 0.9.3 and 0.9.3a  [29 May 1999]
11260
11261   *) New configuration variant "sco5-gcc".
11262
11263   *) Updated some demos.
11264      [Sean O Riordain, Wade Scholine]
11265
11266   *) Add missing BIO_free at exit of pkcs12 application.
11267      [Wu Zhigang]
11268
11269   *) Fix memory leak in conf.c.
11270      [Steve Henson]
11271
11272   *) Updates for Win32 to assembler version of MD5.
11273      [Steve Henson]
11274
11275   *) Set #! path to perl in apps/der_chop to where we found it
11276      instead of using a fixed path.
11277      [Bodo Moeller]
11278
11279   *) SHA library changes for irix64-mips4-cc.
11280      [Andy Polyakov]
11281
11282   *) Improvements for VMS support.
11283      [Richard Levitte]
11284
11285
11286  Changes between 0.9.2b and 0.9.3  [24 May 1999]
11287
11288   *) Bignum library bug fix. IRIX 6 passes "make test" now!
11289      This also avoids the problems with SC4.2 and unpatched SC5.
11290      [Andy Polyakov <appro@fy.chalmers.se>]
11291
11292   *) New functions sk_num, sk_value and sk_set to replace the previous macros.
11293      These are required because of the typesafe stack would otherwise break
11294      existing code. If old code used a structure member which used to be STACK
11295      and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
11296      sk_num or sk_value it would produce an error because the num, data members
11297      are not present in STACK_OF. Now it just produces a warning. sk_set
11298      replaces the old method of assigning a value to sk_value
11299      (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
11300      that does this will no longer work (and should use sk_set instead) but
11301      this could be regarded as a "questionable" behaviour anyway.
11302      [Steve Henson]
11303
11304   *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
11305      correctly handle encrypted S/MIME data.
11306      [Steve Henson]
11307
11308   *) Change type of various DES function arguments from des_cblock
11309      (which means, in function argument declarations, pointer to char)
11310      to des_cblock * (meaning pointer to array with 8 char elements),
11311      which allows the compiler to do more typechecking; it was like
11312      that back in SSLeay, but with lots of ugly casts.
11313
11314      Introduce new type const_des_cblock.
11315      [Bodo Moeller]
11316
11317   *) Reorganise the PKCS#7 library and get rid of some of the more obvious
11318      problems: find RecipientInfo structure that matches recipient certificate
11319      and initialise the ASN1 structures properly based on passed cipher.
11320      [Steve Henson]
11321
11322   *) Belatedly make the BN tests actually check the results.
11323      [Ben Laurie]
11324
11325   *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
11326      to and from BNs: it was completely broken. New compilation option
11327      NEG_PUBKEY_BUG to allow for some broken certificates that encode public
11328      key elements as negative integers.
11329      [Steve Henson]
11330
11331   *) Reorganize and speed up MD5.
11332      [Andy Polyakov <appro@fy.chalmers.se>]
11333
11334   *) VMS support.
11335      [Richard Levitte <richard@levitte.org>]
11336
11337   *) New option -out to asn1parse to allow the parsed structure to be
11338      output to a file. This is most useful when combined with the -strparse
11339      option to examine the output of things like OCTET STRINGS.
11340      [Steve Henson]
11341
11342   *) Make SSL library a little more fool-proof by not requiring any longer
11343      that SSL_set_{accept,connect}_state be called before
11344      SSL_{accept,connect} may be used (SSL_set_..._state is omitted
11345      in many applications because usually everything *appeared* to work as
11346      intended anyway -- now it really works as intended).
11347      [Bodo Moeller]
11348
11349   *) Move openssl.cnf out of lib/.
11350      [Ulf Möller]
11351
11352   *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
11353      -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
11354      -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+
11355      [Ralf S. Engelschall]
11356
11357   *) Various fixes to the EVP and PKCS#7 code. It may now be able to
11358      handle PKCS#7 enveloped data properly.
11359      [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
11360
11361   *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
11362      copying pointers.  The cert_st handling is changed by this in
11363      various ways (and thus what used to be known as ctx->default_cert
11364      is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
11365      any longer when s->cert does not give us what we need).
11366      ssl_cert_instantiate becomes obsolete by this change.
11367      As soon as we've got the new code right (possibly it already is?),
11368      we have solved a couple of bugs of the earlier code where s->cert
11369      was used as if it could not have been shared with other SSL structures.
11370
11371      Note that using the SSL API in certain dirty ways now will result
11372      in different behaviour than observed with earlier library versions:
11373      Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
11374      does not influence s as it used to.
11375
11376      In order to clean up things more thoroughly, inside SSL_SESSION
11377      we don't use CERT any longer, but a new structure SESS_CERT
11378      that holds per-session data (if available); currently, this is
11379      the peer's certificate chain and, for clients, the server's certificate
11380      and temporary key.  CERT holds only those values that can have
11381      meaningful defaults in an SSL_CTX.
11382      [Bodo Moeller]
11383
11384   *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
11385      from the internal representation. Various PKCS#7 fixes: remove some
11386      evil casts and set the enc_dig_alg field properly based on the signing
11387      key type.
11388      [Steve Henson]
11389
11390   *) Allow PKCS#12 password to be set from the command line or the
11391      environment. Let 'ca' get its config file name from the environment
11392      variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
11393      and 'x509').
11394      [Steve Henson]
11395
11396   *) Allow certificate policies extension to use an IA5STRING for the
11397      organization field. This is contrary to the PKIX definition but
11398      VeriSign uses it and IE5 only recognises this form. Document 'x509'
11399      extension option.
11400      [Steve Henson]
11401
11402   *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
11403      without disallowing inline assembler and the like for non-pedantic builds.
11404      [Ben Laurie]
11405
11406   *) Support Borland C++ builder.
11407      [Janez Jere <jj@void.si>, modified by Ulf Möller]
11408
11409   *) Support Mingw32.
11410      [Ulf Möller]
11411
11412   *) SHA-1 cleanups and performance enhancements.
11413      [Andy Polyakov <appro@fy.chalmers.se>]
11414
11415   *) Sparc v8plus assembler for the bignum library.
11416      [Andy Polyakov <appro@fy.chalmers.se>]
11417
11418   *) Accept any -xxx and +xxx compiler options in Configure.
11419      [Ulf Möller]
11420
11421   *) Update HPUX configuration.
11422      [Anonymous]
11423
11424   *) Add missing sk_<type>_unshift() function to safestack.h
11425      [Ralf S. Engelschall]
11426
11427   *) New function SSL_CTX_use_certificate_chain_file that sets the
11428      "extra_cert"s in addition to the certificate.  (This makes sense
11429      only for "PEM" format files, as chains as a whole are not
11430      DER-encoded.)
11431      [Bodo Moeller]
11432
11433   *) Support verify_depth from the SSL API.
11434      x509_vfy.c had what can be considered an off-by-one-error:
11435      Its depth (which was not part of the external interface)
11436      was actually counting the number of certificates in a chain;
11437      now it really counts the depth.
11438      [Bodo Moeller]
11439
11440   *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
11441      instead of X509err, which often resulted in confusing error
11442      messages since the error codes are not globally unique
11443      (e.g. an alleged error in ssl3_accept when a certificate
11444      didn't match the private key).
11445
11446   *) New function SSL_CTX_set_session_id_context that allows to set a default
11447      value (so that you don't need SSL_set_session_id_context for each
11448      connection using the SSL_CTX).
11449      [Bodo Moeller]
11450
11451   *) OAEP decoding bug fix.
11452      [Ulf Möller]
11453
11454   *) Support INSTALL_PREFIX for package builders, as proposed by
11455      David Harris.
11456      [Bodo Moeller]
11457
11458   *) New Configure options "threads" and "no-threads".  For systems
11459      where the proper compiler options are known (currently Solaris
11460      and Linux), "threads" is the default.
11461      [Bodo Moeller]
11462
11463   *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
11464      [Bodo Moeller]
11465
11466   *) Install various scripts to $(OPENSSLDIR)/misc, not to
11467      $(INSTALLTOP)/bin -- they shouldn't clutter directories
11468      such as /usr/local/bin.
11469      [Bodo Moeller]
11470
11471   *) "make linux-shared" to build shared libraries.
11472      [Niels Poppe <niels@netbox.org>]
11473
11474   *) New Configure option no-<cipher> (rsa, idea, rc5, ...).
11475      [Ulf Möller]
11476
11477   *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
11478      extension adding in x509 utility.
11479      [Steve Henson]
11480
11481   *) Remove NOPROTO sections and error code comments.
11482      [Ulf Möller]
11483
11484   *) Partial rewrite of the DEF file generator to now parse the ANSI
11485      prototypes.
11486      [Steve Henson]
11487
11488   *) New Configure options --prefix=DIR and --openssldir=DIR.
11489      [Ulf Möller]
11490
11491   *) Complete rewrite of the error code script(s). It is all now handled
11492      by one script at the top level which handles error code gathering,
11493      header rewriting and C source file generation. It should be much better
11494      than the old method: it now uses a modified version of Ulf's parser to
11495      read the ANSI prototypes in all header files (thus the old K&R definitions
11496      aren't needed for error creation any more) and do a better job of
11497      translating function codes into names. The old 'ASN1 error code imbedded
11498      in a comment' is no longer necessary and it doesn't use .err files which
11499      have now been deleted. Also the error code call doesn't have to appear all
11500      on one line (which resulted in some large lines...).
11501      [Steve Henson]
11502
11503   *) Change #include filenames from <foo.h> to <openssl/foo.h>.
11504      [Bodo Moeller]
11505
11506   *) Change behaviour of ssl2_read when facing length-0 packets: Don't return
11507      0 (which usually indicates a closed connection), but continue reading.
11508      [Bodo Moeller]
11509
11510   *) Fix some race conditions.
11511      [Bodo Moeller]
11512
11513   *) Add support for CRL distribution points extension. Add Certificate
11514      Policies and CRL distribution points documentation.
11515      [Steve Henson]
11516
11517   *) Move the autogenerated header file parts to crypto/opensslconf.h.
11518      [Ulf Möller]
11519
11520   *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
11521      8 of keying material. Merlin has also confirmed interop with this fix
11522      between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
11523      [Merlin Hughes <merlin@baltimore.ie>]
11524
11525   *) Fix lots of warnings.
11526      [Richard Levitte <levitte@stacken.kth.se>]
11527
11528   *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
11529      the directory spec didn't end with a LIST_SEPARATOR_CHAR.
11530      [Richard Levitte <levitte@stacken.kth.se>]
11531
11532   *) Fix problems with sizeof(long) == 8.
11533      [Andy Polyakov <appro@fy.chalmers.se>]
11534
11535   *) Change functions to ANSI C.
11536      [Ulf Möller]
11537
11538   *) Fix typos in error codes.
11539      [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller]
11540
11541   *) Remove defunct assembler files from Configure.
11542      [Ulf Möller]
11543
11544   *) SPARC v8 assembler BIGNUM implementation.
11545      [Andy Polyakov <appro@fy.chalmers.se>]
11546
11547   *) Support for Certificate Policies extension: both print and set.
11548      Various additions to support the r2i method this uses.
11549      [Steve Henson]
11550
11551   *) A lot of constification, and fix a bug in X509_NAME_oneline() that could
11552      return a const string when you are expecting an allocated buffer.
11553      [Ben Laurie]
11554
11555   *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
11556      types DirectoryString and DisplayText.
11557      [Steve Henson]
11558
11559   *) Add code to allow r2i extensions to access the configuration database,
11560      add an LHASH database driver and add several ctx helper functions.
11561      [Steve Henson]
11562
11563   *) Fix an evil bug in bn_expand2() which caused various BN functions to
11564      fail when they extended the size of a BIGNUM.
11565      [Steve Henson]
11566
11567   *) Various utility functions to handle SXNet extension. Modify mkdef.pl to
11568      support typesafe stack.
11569      [Steve Henson]
11570
11571   *) Fix typo in SSL_[gs]et_options().
11572      [Nils Frostberg <nils@medcom.se>]
11573
11574   *) Delete various functions and files that belonged to the (now obsolete)
11575      old X509V3 handling code.
11576      [Steve Henson]
11577
11578   *) New Configure option "rsaref".
11579      [Ulf Möller]
11580
11581   *) Don't auto-generate pem.h.
11582      [Bodo Moeller]
11583
11584   *) Introduce type-safe ASN.1 SETs.
11585      [Ben Laurie]
11586
11587   *) Convert various additional casted stacks to type-safe STACK_OF() variants.
11588      [Ben Laurie, Ralf S. Engelschall, Steve Henson]
11589
11590   *) Introduce type-safe STACKs. This will almost certainly break lots of code
11591      that links with OpenSSL (well at least cause lots of warnings), but fear
11592      not: the conversion is trivial, and it eliminates loads of evil casts. A
11593      few STACKed things have been converted already. Feel free to convert more.
11594      In the fullness of time, I'll do away with the STACK type altogether.
11595      [Ben Laurie]
11596
11597   *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate
11598      specified in <certfile> by updating the entry in the index.txt file.
11599      This way one no longer has to edit the index.txt file manually for
11600      revoking a certificate. The -revoke option does the gory details now.
11601      [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall]
11602
11603   *) Fix `openssl crl -noout -text' combination where `-noout' killed the
11604      `-text' option at all and this way the `-noout -text' combination was
11605      inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'.
11606      [Ralf S. Engelschall]
11607
11608   *) Make sure a corresponding plain text error message exists for the
11609      X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
11610      verify callback function determined that a certificate was revoked.
11611      [Ralf S. Engelschall]
11612
11613   *) Bugfix: In test/testenc, don't test "openssl <cipher>" for
11614      ciphers that were excluded, e.g. by -DNO_IDEA.  Also, test
11615      all available cipers including rc5, which was forgotten until now.
11616      In order to let the testing shell script know which algorithms
11617      are available, a new (up to now undocumented) command
11618      "openssl list-cipher-commands" is used.
11619      [Bodo Moeller]
11620
11621   *) Bugfix: s_client occasionally would sleep in select() when
11622      it should have checked SSL_pending() first.
11623      [Bodo Moeller]
11624
11625   *) New functions DSA_do_sign and DSA_do_verify to provide access to
11626      the raw DSA values prior to ASN.1 encoding.
11627      [Ulf Möller]
11628
11629   *) Tweaks to Configure
11630      [Niels Poppe <niels@netbox.org>]
11631
11632   *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
11633      yet...
11634      [Steve Henson]
11635
11636   *) New variables $(RANLIB) and $(PERL) in the Makefiles.
11637      [Ulf Möller]
11638
11639   *) New config option to avoid instructions that are illegal on the 80386.
11640      The default code is faster, but requires at least a 486.
11641      [Ulf Möller]
11642
11643   *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
11644      SSL2_SERVER_VERSION (not used at all) macros, which are now the
11645      same as SSL2_VERSION anyway.
11646      [Bodo Moeller]
11647
11648   *) New "-showcerts" option for s_client.
11649      [Bodo Moeller]
11650
11651   *) Still more PKCS#12 integration. Add pkcs12 application to openssl
11652      application. Various cleanups and fixes.
11653      [Steve Henson]
11654
11655   *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
11656      modify error routines to work internally. Add error codes and PBE init
11657      to library startup routines.
11658      [Steve Henson]
11659
11660   *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and
11661      packing functions to asn1 and evp. Changed function names and error
11662      codes along the way.
11663      [Steve Henson]
11664
11665   *) PKCS12 integration: and so it begins... First of several patches to
11666      slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
11667      objects to objects.h
11668      [Steve Henson]
11669
11670   *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1
11671      and display support for Thawte strong extranet extension.
11672      [Steve Henson]
11673
11674   *) Add LinuxPPC support.
11675      [Jeff Dubrule <igor@pobox.org>]
11676
11677   *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
11678      bn_div_words in alpha.s.
11679      [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie]
11680
11681   *) Make sure the RSA OAEP test is skipped under -DRSAref because
11682      OAEP isn't supported when OpenSSL is built with RSAref.
11683      [Ulf Moeller <ulf@fitug.de>]
11684
11685   *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
11686      so they no longer are missing under -DNOPROTO.
11687      [Soren S. Jorvang <soren@t.dk>]
11688
11689
11690  Changes between 0.9.1c and 0.9.2b  [22 Mar 1999]
11691
11692   *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
11693      doesn't work when the session is reused. Coming soon!
11694      [Ben Laurie]
11695
11696   *) Fix a security hole, that allows sessions to be reused in the wrong
11697      context thus bypassing client cert protection! All software that uses
11698      client certs and session caches in multiple contexts NEEDS PATCHING to
11699      allow session reuse! A fuller solution is in the works.
11700      [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)]
11701
11702   *) Some more source tree cleanups (removed obsolete files
11703      crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
11704      permission on "config" script to be executable) and a fix for the INSTALL
11705      document.
11706      [Ulf Moeller <ulf@fitug.de>]
11707
11708   *) Remove some legacy and erroneous uses of malloc, free instead of
11709      Malloc, Free.
11710      [Lennart Bang <lob@netstream.se>, with minor changes by Steve]
11711
11712   *) Make rsa_oaep_test return non-zero on error.
11713      [Ulf Moeller <ulf@fitug.de>]
11714
11715   *) Add support for native Solaris shared libraries. Configure
11716      solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
11717      if someone would make that last step automatic.
11718      [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>]
11719
11720   *) ctx_size was not built with the right compiler during "make links". Fixed.
11721      [Ben Laurie]
11722
11723   *) Change the meaning of 'ALL' in the cipher list. It now means "everything
11724      except NULL ciphers". This means the default cipher list will no longer
11725      enable NULL ciphers. They need to be specifically enabled e.g. with
11726      the string "DEFAULT:eNULL".
11727      [Steve Henson]
11728
11729   *) Fix to RSA private encryption routines: if p < q then it would
11730      occasionally produce an invalid result. This will only happen with
11731      externally generated keys because OpenSSL (and SSLeay) ensure p > q.
11732      [Steve Henson]
11733
11734   *) Be less restrictive and allow also `perl util/perlpath.pl
11735      /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin',
11736      because this way one can also use an interpreter named `perl5' (which is
11737      usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
11738      installed as `perl').
11739      [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
11740
11741   *) Let util/clean-depend.pl work also with older Perl 5.00x versions.
11742      [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
11743
11744   *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
11745      advapi32.lib to Win32 build and change the pem test comparison
11746      to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
11747      suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
11748      and crypto/des/ede_cbcm_enc.c.
11749      [Steve Henson]
11750
11751   *) DES quad checksum was broken on big-endian architectures. Fixed.
11752      [Ben Laurie]
11753
11754   *) Comment out two functions in bio.h that aren't implemented. Fix up the
11755      Win32 test batch file so it (might) work again. The Win32 test batch file
11756      is horrible: I feel ill....
11757      [Steve Henson]
11758
11759   *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
11760      in e_os.h. Audit of header files to check ANSI and non ANSI
11761      sections: 10 functions were absent from non ANSI section and not exported
11762      from Windows DLLs. Fixed up libeay.num for new functions.
11763      [Steve Henson]
11764
11765   *) Make `openssl version' output lines consistent.
11766      [Ralf S. Engelschall]
11767
11768   *) Fix Win32 symbol export lists for BIO functions: Added
11769      BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
11770      to ms/libeay{16,32}.def.
11771      [Ralf S. Engelschall]
11772
11773   *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
11774      fine under Unix and passes some trivial tests I've now added. But the
11775      whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
11776      added to make sure no one expects that this stuff really works in the
11777      OpenSSL 0.9.2 release.  Additionally I've started to clean the XS sources
11778      up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
11779      openssl_bio.xs.
11780      [Ralf S. Engelschall]
11781
11782   *) Fix the generation of two part addresses in perl.
11783      [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie]
11784
11785   *) Add config entry for Linux on MIPS.
11786      [John Tobey <jtobey@channel1.com>]
11787
11788   *) Make links whenever Configure is run, unless we are on Windoze.
11789      [Ben Laurie]
11790
11791   *) Permit extensions to be added to CRLs using crl_section in openssl.cnf.
11792      Currently only issuerAltName and AuthorityKeyIdentifier make any sense
11793      in CRLs.
11794      [Steve Henson]
11795
11796   *) Add a useful kludge to allow package maintainers to specify compiler and
11797      other platforms details on the command line without having to patch the
11798      Configure script everytime: One now can use ``perl Configure
11799      <id>:<details>'', i.e. platform ids are allowed to have details appended
11800      to them (separated by colons). This is treated as there would be a static
11801      pre-configured entry in Configure's %table under key <id> with value
11802      <details> and ``perl Configure <id>'' is called.  So, when you want to
11803      perform a quick test-compile under FreeBSD 3.1 with pgcc and without
11804      assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"''
11805      now, which overrides the FreeBSD-elf entry on-the-fly.
11806      [Ralf S. Engelschall]
11807
11808   *) Disable new TLS1 ciphersuites by default: they aren't official yet.
11809      [Ben Laurie]
11810
11811   *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
11812      on the `perl Configure ...' command line. This way one can compile
11813      OpenSSL libraries with Position Independent Code (PIC) which is needed
11814      for linking it into DSOs.
11815      [Ralf S. Engelschall]
11816
11817   *) Remarkably, export ciphers were totally broken and no-one had noticed!
11818      Fixed.
11819      [Ben Laurie]
11820
11821   *) Cleaned up the LICENSE document: The official contact for any license
11822      questions now is the OpenSSL core team under openssl-core@openssl.org.
11823      And add a paragraph about the dual-license situation to make sure people
11824      recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
11825      to the OpenSSL toolkit.
11826      [Ralf S. Engelschall]
11827
11828   *) General source tree makefile cleanups: Made `making xxx in yyy...'
11829      display consistent in the source tree and replaced `/bin/rm' by `rm'.
11830      Additionally cleaned up the `make links' target: Remove unnecessary
11831      semicolons, subsequent redundant removes, inline point.sh into mklink.sh
11832      to speed processing and no longer clutter the display with confusing
11833      stuff. Instead only the actually done links are displayed.
11834      [Ralf S. Engelschall]
11835
11836   *) Permit null encryption ciphersuites, used for authentication only. It used
11837      to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
11838      It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
11839      encryption.
11840      [Ben Laurie]
11841
11842   *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
11843      signed attributes when verifying signatures (this would break them),
11844      the detached data encoding was wrong and public keys obtained using
11845      X509_get_pubkey() weren't freed.
11846      [Steve Henson]
11847
11848   *) Add text documentation for the BUFFER functions. Also added a work around
11849      to a Win95 console bug. This was triggered by the password read stuff: the
11850      last character typed gets carried over to the next fread(). If you were
11851      generating a new cert request using 'req' for example then the last
11852      character of the passphrase would be CR which would then enter the first
11853      field as blank.
11854      [Steve Henson]
11855
11856   *) Added the new `Includes OpenSSL Cryptography Software' button as
11857      doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
11858      button and can be used by applications based on OpenSSL to show the
11859      relationship to the OpenSSL project.
11860      [Ralf S. Engelschall]
11861
11862   *) Remove confusing variables in function signatures in files
11863      ssl/ssl_lib.c and ssl/ssl.h.
11864      [Lennart Bong <lob@kulthea.stacken.kth.se>]
11865
11866   *) Don't install bss_file.c under PREFIX/include/
11867      [Lennart Bong <lob@kulthea.stacken.kth.se>]
11868
11869   *) Get the Win32 compile working again. Modify mkdef.pl so it can handle
11870      functions that return function pointers and has support for NT specific
11871      stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
11872      #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
11873      unsigned to signed types: this was killing the Win32 compile.
11874      [Steve Henson]
11875
11876   *) Add new certificate file to stack functions,
11877      SSL_add_dir_cert_subjects_to_stack() and
11878      SSL_add_file_cert_subjects_to_stack().  These largely supplant
11879      SSL_load_client_CA_file(), and can be used to add multiple certs easily
11880      to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
11881      This means that Apache-SSL and similar packages don't have to mess around
11882      to add as many CAs as they want to the preferred list.
11883      [Ben Laurie]
11884
11885   *) Experiment with doxygen documentation. Currently only partially applied to
11886      ssl/ssl_lib.c.
11887      See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with
11888      openssl.doxy as the configuration file.
11889      [Ben Laurie]
11890
11891   *) Get rid of remaining C++-style comments which strict C compilers hate.
11892      [Ralf S. Engelschall, pointed out by Carlos Amengual]
11893
11894   *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
11895      compiled in by default: it has problems with large keys.
11896      [Steve Henson]
11897
11898   *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
11899      DH private keys and/or callback functions which directly correspond to
11900      their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
11901      is needed for applications which have to configure certificates on a
11902      per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
11903      (e.g. s_server).
11904         For the RSA certificate situation is makes no difference, but
11905      for the DSA certificate situation this fixes the "no shared cipher"
11906      problem where the OpenSSL cipher selection procedure failed because the
11907      temporary keys were not overtaken from the context and the API provided
11908      no way to reconfigure them.
11909         The new functions now let applications reconfigure the stuff and they
11910      are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
11911      SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback.  Additionally a new
11912      non-public-API function ssl_cert_instantiate() is used as a helper
11913      function and also to reduce code redundancy inside ssl_rsa.c.
11914      [Ralf S. Engelschall]
11915
11916   *) Move s_server -dcert and -dkey options out of the undocumented feature
11917      area because they are useful for the DSA situation and should be
11918      recognized by the users.
11919      [Ralf S. Engelschall]
11920
11921   *) Fix the cipher decision scheme for export ciphers: the export bits are
11922      *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
11923      SSL_EXP_MASK.  So, the original variable has to be used instead of the
11924      already masked variable.
11925      [Richard Levitte <levitte@stacken.kth.se>]
11926
11927   *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c
11928      [Richard Levitte <levitte@stacken.kth.se>]
11929
11930   *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
11931      from `int' to `unsigned int' because it's a length and initialized by
11932      EVP_DigestFinal() which expects an `unsigned int *'.
11933      [Richard Levitte <levitte@stacken.kth.se>]
11934
11935   *) Don't hard-code path to Perl interpreter on shebang line of Configure
11936      script. Instead use the usual Shell->Perl transition trick.
11937      [Ralf S. Engelschall]
11938
11939   *) Make `openssl x509 -noout -modulus' functional also for DSA certificates
11940      (in addition to RSA certificates) to match the behaviour of `openssl dsa
11941      -noout -modulus' as it's already the case for `openssl rsa -noout
11942      -modulus'.  For RSA the -modulus is the real "modulus" while for DSA
11943      currently the public key is printed (a decision which was already done by
11944      `openssl dsa -modulus' in the past) which serves a similar purpose.
11945      Additionally the NO_RSA no longer completely removes the whole -modulus
11946      option; it now only avoids using the RSA stuff. Same applies to NO_DSA
11947      now, too.
11948      [Ralf S.  Engelschall]
11949
11950   *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
11951      BIO. See the source (crypto/evp/bio_ok.c) for more info.
11952      [Arne Ansper <arne@ats.cyber.ee>]
11953
11954   *) Dump the old yucky req code that tried (and failed) to allow raw OIDs
11955      to be added. Now both 'req' and 'ca' can use new objects defined in the
11956      config file.
11957      [Steve Henson]
11958
11959   *) Add cool BIO that does syslog (or event log on NT).
11960      [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie]
11961
11962   *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
11963      TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
11964      TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
11965      Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
11966      [Ben Laurie]
11967
11968   *) Add preliminary config info for new extension code.
11969      [Steve Henson]
11970
11971   *) Make RSA_NO_PADDING really use no padding.
11972      [Ulf Moeller <ulf@fitug.de>]
11973
11974   *) Generate errors when private/public key check is done.
11975      [Ben Laurie]
11976
11977   *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
11978      for some CRL extensions and new objects added.
11979      [Steve Henson]
11980
11981   *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private
11982      key usage extension and fuller support for authority key id.
11983      [Steve Henson]
11984
11985   *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
11986      padding method for RSA, which is recommended for new applications in PKCS
11987      #1 v2.0 (RFC 2437, October 1998).
11988      OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
11989      foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
11990      against Bleichbacher's attack on RSA.
11991      [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
11992       Ben Laurie]
11993
11994   *) Updates to the new SSL compression code
11995      [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
11996
11997   *) Fix so that the version number in the master secret, when passed
11998      via RSA, checks that if TLS was proposed, but we roll back to SSLv3
11999      (because the server will not accept higher), that the version number
12000      is 0x03,0x01, not 0x03,0x00
12001      [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
12002
12003   *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory
12004      leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes
12005      in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c
12006      [Steve Henson]
12007
12008   *) Support for RAW extensions where an arbitrary extension can be
12009      created by including its DER encoding. See apps/openssl.cnf for
12010      an example.
12011      [Steve Henson]
12012
12013   *) Make sure latest Perl versions don't interpret some generated C array
12014      code as Perl array code in the crypto/err/err_genc.pl script.
12015      [Lars Weber <3weber@informatik.uni-hamburg.de>]
12016
12017   *) Modify ms/do_ms.bat to not generate assembly language makefiles since
12018      not many people have the assembler. Various Win32 compilation fixes and
12019      update to the INSTALL.W32 file with (hopefully) more accurate Win32
12020      build instructions.
12021      [Steve Henson]
12022
12023   *) Modify configure script 'Configure' to automatically create crypto/date.h
12024      file under Win32 and also build pem.h from pem.org. New script
12025      util/mkfiles.pl to create the MINFO file on environments that can't do a
12026      'make files': perl util/mkfiles.pl >MINFO should work.
12027      [Steve Henson]
12028
12029   *) Major rework of DES function declarations, in the pursuit of correctness
12030      and purity. As a result, many evil casts evaporated, and some weirdness,
12031      too. You may find this causes warnings in your code. Zapping your evil
12032      casts will probably fix them. Mostly.
12033      [Ben Laurie]
12034
12035   *) Fix for a typo in asn1.h. Bug fix to object creation script
12036      obj_dat.pl. It considered a zero in an object definition to mean
12037      "end of object": none of the objects in objects.h have any zeros
12038      so it wasn't spotted.
12039      [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>]
12040
12041   *) Add support for Triple DES Cipher Block Chaining with Output Feedback
12042      Masking (CBCM). In the absence of test vectors, the best I have been able
12043      to do is check that the decrypt undoes the encrypt, so far. Send me test
12044      vectors if you have them.
12045      [Ben Laurie]
12046
12047   *) Correct calculation of key length for export ciphers (too much space was
12048      allocated for null ciphers). This has not been tested!
12049      [Ben Laurie]
12050
12051   *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage
12052      message is now correct (it understands "crypto" and "ssl" on its
12053      command line). There is also now an "update" option. This will update
12054      the util/ssleay.num and util/libeay.num files with any new functions.
12055      If you do a:
12056      perl util/mkdef.pl crypto ssl update
12057      it will update them.
12058      [Steve Henson]
12059
12060   *) Overhauled the Perl interface (perl/*):
12061      - ported BN stuff to OpenSSL's different BN library
12062      - made the perl/ source tree CVS-aware
12063      - renamed the package from SSLeay to OpenSSL (the files still contain
12064        their history because I've copied them in the repository)
12065      - removed obsolete files (the test scripts will be replaced
12066        by better Test::Harness variants in the future)
12067      [Ralf S. Engelschall]
12068
12069   *) First cut for a very conservative source tree cleanup:
12070      1. merge various obsolete readme texts into doc/ssleay.txt
12071      where we collect the old documents and readme texts.
12072      2. remove the first part of files where I'm already sure that we no
12073      longer need them because of three reasons: either they are just temporary
12074      files which were left by Eric or they are preserved original files where
12075      I've verified that the diff is also available in the CVS via "cvs diff
12076      -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
12077      the crypto/md/ stuff).
12078      [Ralf S. Engelschall]
12079
12080   *) More extension code. Incomplete support for subject and issuer alt
12081      name, issuer and authority key id. Change the i2v function parameters
12082      and add an extra 'crl' parameter in the X509V3_CTX structure: guess
12083      what that's for :-) Fix to ASN1 macro which messed up
12084      IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
12085      [Steve Henson]
12086
12087   *) Preliminary support for ENUMERATED type. This is largely copied from the
12088      INTEGER code.
12089      [Steve Henson]
12090
12091   *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
12092      [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
12093
12094   *) Make sure `make rehash' target really finds the `openssl' program.
12095      [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
12096
12097   *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
12098      like to hear about it if this slows down other processors.
12099      [Ben Laurie]
12100
12101   *) Add CygWin32 platform information to Configure script.
12102      [Alan Batie <batie@aahz.jf.intel.com>]
12103
12104   *) Fixed ms/32all.bat script: `no_asm' -> `no-asm'
12105      [Rainer W. Gerling <gerling@mpg-gv.mpg.de>]
12106
12107   *) New program nseq to manipulate netscape certificate sequences
12108      [Steve Henson]
12109
12110   *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
12111      few typos.
12112      [Steve Henson]
12113
12114   *) Fixes to BN code.  Previously the default was to define BN_RECURSION
12115      but the BN code had some problems that would cause failures when
12116      doing certificate verification and some other functions.
12117      [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
12118
12119   *) Add ASN1 and PEM code to support netscape certificate sequences.
12120      [Steve Henson]
12121
12122   *) Add ASN1 and PEM code to support netscape certificate sequences.
12123      [Steve Henson]
12124
12125   *) Add several PKIX and private extended key usage OIDs.
12126      [Steve Henson]
12127
12128   *) Modify the 'ca' program to handle the new extension code. Modify
12129      openssl.cnf for new extension format, add comments.
12130      [Steve Henson]
12131
12132   *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
12133      and add a sample to openssl.cnf so req -x509 now adds appropriate
12134      CA extensions.
12135      [Steve Henson]
12136
12137   *) Continued X509 V3 changes. Add to other makefiles, integrate with the
12138      error code, add initial support to X509_print() and x509 application.
12139      [Steve Henson]
12140
12141   *) Takes a deep breath and start adding X509 V3 extension support code. Add
12142      files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
12143      stuff is currently isolated and isn't even compiled yet.
12144      [Steve Henson]
12145
12146   *) Continuing patches for GeneralizedTime. Fix up certificate and CRL
12147      ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
12148      Removed the versions check from X509 routines when loading extensions:
12149      this allows certain broken certificates that don't set the version
12150      properly to be processed.
12151      [Steve Henson]
12152
12153   *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
12154      Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
12155      can still be regenerated with "make depend".
12156      [Ben Laurie]
12157
12158   *) Spelling mistake in C version of CAST-128.
12159      [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>]
12160
12161   *) Changes to the error generation code. The perl script err-code.pl
12162      now reads in the old error codes and retains the old numbers, only
12163      adding new ones if necessary. It also only changes the .err files if new
12164      codes are added. The makefiles have been modified to only insert errors
12165      when needed (to avoid needlessly modifying header files). This is done
12166      by only inserting errors if the .err file is newer than the auto generated
12167      C file. To rebuild all the error codes from scratch (the old behaviour)
12168      either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
12169      or delete all the .err files.
12170      [Steve Henson]
12171
12172   *) CAST-128 was incorrectly implemented for short keys. The C version has
12173      been fixed, but is untested. The assembler versions are also fixed, but
12174      new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
12175      to regenerate it if needed.
12176      [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
12177       Hagino <itojun@kame.net>]
12178
12179   *) File was opened incorrectly in randfile.c.
12180      [Ulf Möller <ulf@fitug.de>]
12181
12182   *) Beginning of support for GeneralizedTime. d2i, i2d, check and print
12183      functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
12184      GeneralizedTime. ASN1_TIME is the proper type used in certificates et
12185      al: it's just almost always a UTCTime. Note this patch adds new error
12186      codes so do a "make errors" if there are problems.
12187      [Steve Henson]
12188
12189   *) Correct Linux 1 recognition in config.
12190      [Ulf Möller <ulf@fitug.de>]
12191
12192   *) Remove pointless MD5 hash when using DSA keys in ca.
12193      [Anonymous <nobody@replay.com>]
12194
12195   *) Generate an error if given an empty string as a cert directory. Also
12196      generate an error if handed NULL (previously returned 0 to indicate an
12197      error, but didn't set one).
12198      [Ben Laurie, reported by Anonymous <nobody@replay.com>]
12199
12200   *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
12201      [Ben Laurie]
12202
12203   *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
12204      parameters. This was causing a warning which killed off the Win32 compile.
12205      [Steve Henson]
12206
12207   *) Remove C++ style comments from crypto/bn/bn_local.h.
12208      [Neil Costigan <neil.costigan@celocom.com>]
12209
12210   *) The function OBJ_txt2nid was broken. It was supposed to return a nid
12211      based on a text string, looking up short and long names and finally
12212      "dot" format. The "dot" format stuff didn't work. Added new function
12213      OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
12214      OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
12215      OID is not part of the table.
12216      [Steve Henson]
12217
12218   *) Add prototypes to X509 lookup/verify methods, fixing a bug in
12219      X509_LOOKUP_by_alias().
12220      [Ben Laurie]
12221
12222   *) Sort openssl functions by name.
12223      [Ben Laurie]
12224
12225   *) Get the gendsa program working (hopefully) and add it to app list. Remove
12226      encryption from sample DSA keys (in case anyone is interested the password
12227      was "1234").
12228      [Steve Henson]
12229
12230   *) Make _all_ *_free functions accept a NULL pointer.
12231      [Frans Heymans <fheymans@isaserver.be>]
12232
12233   *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use
12234      NULL pointers.
12235      [Anonymous <nobody@replay.com>]
12236
12237   *) s_server should send the CAfile as acceptable CAs, not its own cert.
12238      [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
12239
12240   *) Don't blow it for numeric -newkey arguments to apps/req.
12241      [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
12242
12243   *) Temp key "for export" tests were wrong in s3_srvr.c.
12244      [Anonymous <nobody@replay.com>]
12245
12246   *) Add prototype for temp key callback functions
12247      SSL_CTX_set_tmp_{rsa,dh}_callback().
12248      [Ben Laurie]
12249
12250   *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
12251      DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
12252      [Steve Henson]
12253
12254   *) X509_name_add_entry() freed the wrong thing after an error.
12255      [Arne Ansper <arne@ats.cyber.ee>]
12256
12257   *) rsa_eay.c would attempt to free a NULL context.
12258      [Arne Ansper <arne@ats.cyber.ee>]
12259
12260   *) BIO_s_socket() had a broken should_retry() on Windoze.
12261      [Arne Ansper <arne@ats.cyber.ee>]
12262
12263   *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
12264      [Arne Ansper <arne@ats.cyber.ee>]
12265
12266   *) Make sure the already existing X509_STORE->depth variable is initialized
12267      in X509_STORE_new(), but document the fact that this variable is still
12268      unused in the certificate verification process.
12269      [Ralf S. Engelschall]
12270
12271   *) Fix the various library and apps files to free up pkeys obtained from
12272      X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
12273      [Steve Henson]
12274
12275   *) Fix reference counting in X509_PUBKEY_get(). This makes
12276      demos/maurice/example2.c work, amongst others, probably.
12277      [Steve Henson and Ben Laurie]
12278
12279   *) First cut of a cleanup for apps/. First the `ssleay' program is now named
12280      `openssl' and second, the shortcut symlinks for the `openssl <command>'
12281      are no longer created. This way we have a single and consistent command
12282      line interface `openssl <command>', similar to `cvs <command>'.
12283      [Ralf S. Engelschall, Paul Sutton and Ben Laurie]
12284
12285   *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
12286      BIT STRING wrapper always have zero unused bits.
12287      [Steve Henson]
12288
12289   *) Add CA.pl, perl version of CA.sh, add extended key usage OID.
12290      [Steve Henson]
12291
12292   *) Make the top-level INSTALL documentation easier to understand.
12293      [Paul Sutton]
12294
12295   *) Makefiles updated to exit if an error occurs in a sub-directory
12296      make (including if user presses ^C) [Paul Sutton]
12297
12298   *) Make Montgomery context stuff explicit in RSA data structure.
12299      [Ben Laurie]
12300
12301   *) Fix build order of pem and err to allow for generated pem.h.
12302      [Ben Laurie]
12303
12304   *) Fix renumbering bug in X509_NAME_delete_entry().
12305      [Ben Laurie]
12306
12307   *) Enhanced the err-ins.pl script so it makes the error library number
12308      global and can add a library name. This is needed for external ASN1 and
12309      other error libraries.
12310      [Steve Henson]
12311
12312   *) Fixed sk_insert which never worked properly.
12313      [Steve Henson]
12314
12315   *) Fix ASN1 macros so they can handle indefinite length constructed
12316      EXPLICIT tags. Some non standard certificates use these: they can now
12317      be read in.
12318      [Steve Henson]
12319
12320   *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
12321      into a single doc/ssleay.txt bundle. This way the information is still
12322      preserved but no longer messes up this directory. Now it's new room for
12323      the new set of documentation files.
12324      [Ralf S. Engelschall]
12325
12326   *) SETs were incorrectly DER encoded. This was a major pain, because they
12327      shared code with SEQUENCEs, which aren't coded the same. This means that
12328      almost everything to do with SETs or SEQUENCEs has either changed name or
12329      number of arguments.
12330      [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>]
12331
12332   *) Fix test data to work with the above.
12333      [Ben Laurie]
12334
12335   *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
12336      was already fixed by Eric for 0.9.1 it seems.
12337      [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>]
12338
12339   *) Autodetect FreeBSD3.
12340      [Ben Laurie]
12341
12342   *) Fix various bugs in Configure. This affects the following platforms:
12343      nextstep
12344      ncr-scde
12345      unixware-2.0
12346      unixware-2.0-pentium
12347      sco5-cc.
12348      [Ben Laurie]
12349
12350   *) Eliminate generated files from CVS. Reorder tests to regenerate files
12351      before they are needed.
12352      [Ben Laurie]
12353
12354   *) Generate Makefile.ssl from Makefile.org (to keep CVS happy).
12355      [Ben Laurie]
12356
12357
12358  Changes between 0.9.1b and 0.9.1c  [23-Dec-1998]
12359
12360   *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
12361      changed SSLeay to OpenSSL in version strings.
12362      [Ralf S. Engelschall]
12363
12364   *) Some fixups to the top-level documents.
12365      [Paul Sutton]
12366
12367   *) Fixed the nasty bug where rsaref.h was not found under compile-time
12368      because the symlink to include/ was missing.
12369      [Ralf S. Engelschall]
12370
12371   *) Incorporated the popular no-RSA/DSA-only patches
12372      which allow to compile a RSA-free SSLeay.
12373      [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall]
12374
12375   *) Fixed nasty rehash problem under `make -f Makefile.ssl links'
12376      when "ssleay" is still not found.
12377      [Ralf S. Engelschall]
12378
12379   *) Added more platforms to Configure: Cray T3E, HPUX 11,
12380      [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>]
12381
12382   *) Updated the README file.
12383      [Ralf S. Engelschall]
12384
12385   *) Added various .cvsignore files in the CVS repository subdirs
12386      to make a "cvs update" really silent.
12387      [Ralf S. Engelschall]
12388
12389   *) Recompiled the error-definition header files and added
12390      missing symbols to the Win32 linker tables.
12391      [Ralf S. Engelschall]
12392
12393   *) Cleaned up the top-level documents;
12394      o new files: CHANGES and LICENSE
12395      o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
12396      o merged COPYRIGHT into LICENSE
12397      o removed obsolete TODO file
12398      o renamed MICROSOFT to INSTALL.W32
12399      [Ralf S. Engelschall]
12400
12401   *) Removed dummy files from the 0.9.1b source tree:
12402      crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
12403      crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
12404      crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
12405      crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
12406      util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
12407      [Ralf S. Engelschall]
12408
12409   *) Added various platform portability fixes.
12410      [Mark J. Cox]
12411
12412   *) The Genesis of the OpenSSL rpject:
12413      We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
12414      Young and Tim J. Hudson created while they were working for C2Net until
12415      summer 1998.
12416      [The OpenSSL Project]
12417
12418
12419  Changes between 0.9.0b and 0.9.1b  [not released]
12420
12421   *) Updated a few CA certificates under certs/
12422      [Eric A. Young]
12423
12424   *) Changed some BIGNUM api stuff.
12425      [Eric A. Young]
12426
12427   *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
12428      DGUX x86, Linux Alpha, etc.
12429      [Eric A. Young]
12430
12431   *) New COMP library [crypto/comp/] for SSL Record Layer Compression:
12432      RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
12433      available).
12434      [Eric A. Young]
12435
12436   *) Add -strparse option to asn1pars program which parses nested
12437      binary structures
12438      [Dr Stephen Henson <shenson@bigfoot.com>]
12439
12440   *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
12441      [Eric A. Young]
12442
12443   *) DSA fix for "ca" program.
12444      [Eric A. Young]
12445
12446   *) Added "-genkey" option to "dsaparam" program.
12447      [Eric A. Young]
12448
12449   *) Added RIPE MD160 (rmd160) message digest.
12450      [Eric A. Young]
12451
12452   *) Added -a (all) option to "ssleay version" command.
12453      [Eric A. Young]
12454
12455   *) Added PLATFORM define which is the id given to Configure.
12456      [Eric A. Young]
12457
12458   *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
12459      [Eric A. Young]
12460
12461   *) Extended the ASN.1 parser routines.
12462      [Eric A. Young]
12463
12464   *) Extended BIO routines to support REUSEADDR, seek, tell, etc.
12465      [Eric A. Young]
12466
12467   *) Added a BN_CTX to the BN library.
12468      [Eric A. Young]
12469
12470   *) Fixed the weak key values in DES library
12471      [Eric A. Young]
12472
12473   *) Changed API in EVP library for cipher aliases.
12474      [Eric A. Young]
12475
12476   *) Added support for RC2/64bit cipher.
12477      [Eric A. Young]
12478
12479   *) Converted the lhash library to the crypto/mem.c functions.
12480      [Eric A. Young]
12481
12482   *) Added more recognized ASN.1 object ids.
12483      [Eric A. Young]
12484
12485   *) Added more RSA padding checks for SSL/TLS.
12486      [Eric A. Young]
12487
12488   *) Added BIO proxy/filter functionality.
12489      [Eric A. Young]
12490
12491   *) Added extra_certs to SSL_CTX which can be used
12492      send extra CA certificates to the client in the CA cert chain sending
12493      process. It can be configured with SSL_CTX_add_extra_chain_cert().
12494      [Eric A. Young]
12495
12496   *) Now Fortezza is denied in the authentication phase because
12497      this is key exchange mechanism is not supported by SSLeay at all.
12498      [Eric A. Young]
12499
12500   *) Additional PKCS1 checks.
12501      [Eric A. Young]
12502
12503   *) Support the string "TLSv1" for all TLS v1 ciphers.
12504      [Eric A. Young]
12505
12506   *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
12507      ex_data index of the SSL context in the X509_STORE_CTX ex_data.
12508      [Eric A. Young]
12509
12510   *) Fixed a few memory leaks.
12511      [Eric A. Young]
12512
12513   *) Fixed various code and comment typos.
12514      [Eric A. Young]
12515
12516   *) A minor bug in ssl/s3_clnt.c where there would always be 4 0
12517      bytes sent in the client random.
12518      [Edward Bishop <ebishop@spyglass.com>]
12519