"Fossies" - the Fresh Open Source Software Archive

Member "openssl-1.1.1b/CHANGES" (26 Feb 2019, 574631 Bytes) of package /linux/misc/openssl-1.1.1b.tar.gz:


As a special service "Fossies" has tried to format the requested text file into HTML format (style: standard) with prefixed line numbers. Alternatively you can here view or download the uninterpreted source code file. See also the latest Fossies "Diffs" side-by-side code changes reports for "CHANGES": 1.1.1-pre9_vs_1.1.1 or 1.1.0i_vs_1.1.1.

    1 
    2  OpenSSL CHANGES
    3  _______________
    4 
    5  This is a high-level summary of the most important changes.
    6  For a full list of changes, see the git commit log; for example,
    7  https://github.com/openssl/openssl/commits/ and pick the appropriate
    8  release branch.
    9 
   10  Changes between 1.1.1a and 1.1.1b [26 Feb 2019]
   11 
   12   *) Added SCA hardening for modular field inversion in EC_GROUP through
   13      a new dedicated field_inv() pointer in EC_METHOD.
   14      This also addresses a leakage affecting conversions from projective
   15      to affine coordinates.
   16      [Billy Bob Brumley, Nicola Tuveri]
   17 
   18   *) Change the info callback signals for the start and end of a post-handshake
   19      message exchange in TLSv1.3. In 1.1.1/1.1.1a we used SSL_CB_HANDSHAKE_START
   20      and SSL_CB_HANDSHAKE_DONE. Experience has shown that many applications get
   21      confused by this and assume that a TLSv1.2 renegotiation has started. This
   22      can break KeyUpdate handling. Instead we no longer signal the start and end
   23      of a post handshake message exchange (although the messages themselves are
   24      still signalled). This could break some applications that were expecting
   25      the old signals. However without this KeyUpdate is not usable for many
   26      applications.
   27      [Matt Caswell]
   28 
   29   *) Fix a bug in the computation of the endpoint-pair shared secret used
   30      by DTLS over SCTP. This breaks interoperability with older versions
   31      of OpenSSL like OpenSSL 1.1.0 and OpenSSL 1.0.2. There is a runtime
   32      switch SSL_MODE_DTLS_SCTP_LABEL_LENGTH_BUG (off by default) enabling
   33      interoperability with such broken implementations. However, enabling
   34      this switch breaks interoperability with correct implementations.
   35 
   36   *) Fix a use after free bug in d2i_X509_PUBKEY when overwriting a
   37      re-used X509_PUBKEY object if the second PUBKEY is malformed.
   38      [Bernd Edlinger]
   39 
   40   *) Move strictness check from EVP_PKEY_asn1_new() to EVP_PKEY_asn1_add0().
   41      [Richard Levitte]
   42 
   43   *) Remove the 'dist' target and add a tarball building script.  The
   44      'dist' target has fallen out of use, and it shouldn't be
   45      necessary to configure just to create a source distribution.
   46      [Richard Levitte]
   47 
   48  Changes between 1.1.1 and 1.1.1a [20 Nov 2018]
   49 
   50   *) Timing vulnerability in DSA signature generation
   51 
   52      The OpenSSL DSA signature algorithm has been shown to be vulnerable to a
   53      timing side channel attack. An attacker could use variations in the signing
   54      algorithm to recover the private key.
   55 
   56      This issue was reported to OpenSSL on 16th October 2018 by Samuel Weiser.
   57      (CVE-2018-0734)
   58      [Paul Dale]
   59 
   60   *) Timing vulnerability in ECDSA signature generation
   61 
   62      The OpenSSL ECDSA signature algorithm has been shown to be vulnerable to a
   63      timing side channel attack. An attacker could use variations in the signing
   64      algorithm to recover the private key.
   65 
   66      This issue was reported to OpenSSL on 25th October 2018 by Samuel Weiser.
   67      (CVE-2018-0735)
   68      [Paul Dale]
   69 
   70   *) Added EVP_PKEY_ECDH_KDF_X9_63 and ecdh_KDF_X9_63() as replacements for
   71      the EVP_PKEY_ECDH_KDF_X9_62 KDF type and ECDH_KDF_X9_62(). The old names
   72      are retained for backwards compatibility.
   73      [Antoine Salon]
   74 
   75   *) Fixed the issue that RAND_add()/RAND_seed() silently discards random input
   76      if its length exceeds 4096 bytes. The limit has been raised to a buffer size
   77      of two gigabytes and the error handling improved.
   78 
   79      This issue was reported to OpenSSL by Dr. Falko Strenzke. It has been
   80      categorized as a normal bug, not a security issue, because the DRBG reseeds
   81      automatically and is fully functional even without additional randomness
   82      provided by the application.
   83 
   84  Changes between 1.1.0i and 1.1.1 [11 Sep 2018]
   85 
   86   *) Add a new ClientHello callback. Provides a callback interface that gives
   87      the application the ability to adjust the nascent SSL object at the
   88      earliest stage of ClientHello processing, immediately after extensions have
   89      been collected but before they have been processed. In particular, this
   90      callback can adjust the supported TLS versions in response to the contents
   91      of the ClientHello
   92      [Benjamin Kaduk]
   93 
   94   *) Add SM2 base algorithm support.
   95      [Jack Lloyd]
   96 
   97   *) s390x assembly pack: add (improved) hardware-support for the following
   98      cryptographic primitives: sha3, shake, aes-gcm, aes-ccm, aes-ctr, aes-ofb,
   99      aes-cfb/cfb8, aes-ecb.
  100      [Patrick Steuer]
  101 
  102   *) Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
  103      parameter is no longer accepted, as it leads to a corrupt table.  NULL
  104      pem_str is reserved for alias entries only.
  105      [Richard Levitte]
  106 
  107   *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
  108      step for prime curves. The new implementation is based on formulae from
  109      differential addition-and-doubling in homogeneous projective coordinates
  110      from Izu-Takagi "A fast parallel elliptic curve multiplication resistant
  111      against side channel attacks" and Brier-Joye "Weierstrass Elliptic Curves
  112      and Side-Channel Attacks" Eq. (8) for y-coordinate recovery, modified
  113      to work in projective coordinates.
  114      [Billy Bob Brumley, Nicola Tuveri]
  115 
  116   *) Change generating and checking of primes so that the error rate of not
  117      being prime depends on the intended use based on the size of the input.
  118      For larger primes this will result in more rounds of Miller-Rabin.
  119      The maximal error rate for primes with more than 1080 bits is lowered
  120      to 2^-128.
  121      [Kurt Roeckx, Annie Yousar]
  122 
  123   *) Increase the number of Miller-Rabin rounds for DSA key generating to 64.
  124      [Kurt Roeckx]
  125 
  126   *) The 'tsget' script is renamed to 'tsget.pl', to avoid confusion when
  127      moving between systems, and to avoid confusion when a Windows build is
  128      done with mingw vs with MSVC.  For POSIX installs, there's still a
  129      symlink or copy named 'tsget' to avoid that confusion as well.
  130      [Richard Levitte]
  131 
  132   *) Revert blinding in ECDSA sign and instead make problematic addition
  133      length-invariant. Switch even to fixed-length Montgomery multiplication.
  134      [Andy Polyakov]
  135 
  136   *) Use the new ec_scalar_mul_ladder scaffold to implement a specialized ladder
  137      step for binary curves. The new implementation is based on formulae from
  138      differential addition-and-doubling in mixed Lopez-Dahab projective
  139      coordinates, modified to independently blind the operands.
  140      [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri]
  141 
  142   *) Add a scaffold to optionally enhance the Montgomery ladder implementation
  143      for `ec_scalar_mul_ladder` (formerly `ec_mul_consttime`) allowing
  144      EC_METHODs to implement their own specialized "ladder step", to take
  145      advantage of more favorable coordinate systems or more efficient
  146      differential addition-and-doubling algorithms.
  147      [Billy Bob Brumley, Sohaib ul Hassan, Nicola Tuveri]
  148 
  149   *) Modified the random device based seed sources to keep the relevant
  150      file descriptors open rather than reopening them on each access.
  151      This allows such sources to operate in a chroot() jail without
  152      the associated device nodes being available. This behaviour can be
  153      controlled using RAND_keep_random_devices_open().
  154      [Paul Dale]
  155 
  156   *) Numerous side-channel attack mitigations have been applied. This may have
  157      performance impacts for some algorithms for the benefit of improved
  158      security. Specific changes are noted in this change log by their respective
  159      authors.
  160      [Matt Caswell]
  161 
  162   *) AIX shared library support overhaul. Switch to AIX "natural" way of
  163      handling shared libraries, which means collecting shared objects of
  164      different versions and bitnesses in one common archive. This allows to
  165      mitigate conflict between 1.0 and 1.1 side-by-side installations. It
  166      doesn't affect the way 3rd party applications are linked, only how
  167      multi-version installation is managed.
  168      [Andy Polyakov]
  169 
  170   *) Make ec_group_do_inverse_ord() more robust and available to other
  171      EC cryptosystems, so that irrespective of BN_FLG_CONSTTIME, SCA
  172      mitigations are applied to the fallback BN_mod_inverse().
  173      When using this function rather than BN_mod_inverse() directly, new
  174      EC cryptosystem implementations are then safer-by-default.
  175      [Billy Bob Brumley]
  176 
  177   *) Add coordinate blinding for EC_POINT and implement projective
  178      coordinate blinding for generic prime curves as a countermeasure to
  179      chosen point SCA attacks.
  180      [Sohaib ul Hassan, Nicola Tuveri, Billy Bob Brumley]
  181 
  182   *) Add blinding to ECDSA and DSA signatures to protect against side channel
  183      attacks discovered by Keegan Ryan (NCC Group).
  184      [Matt Caswell]
  185 
  186   *) Enforce checking in the pkeyutl command line app to ensure that the input
  187      length does not exceed the maximum supported digest length when performing
  188      a sign, verify or verifyrecover operation.
  189      [Matt Caswell]
  190 
  191   *) SSL_MODE_AUTO_RETRY is enabled by default. Applications that use blocking
  192      I/O in combination with something like select() or poll() will hang. This
  193      can be turned off again using SSL_CTX_clear_mode().
  194      Many applications do not properly handle non-application data records, and
  195      TLS 1.3 sends more of such records. Setting SSL_MODE_AUTO_RETRY works
  196      around the problems in those applications, but can also break some.
  197      It's recommended to read the manpages about SSL_read(), SSL_write(),
  198      SSL_get_error(), SSL_shutdown(), SSL_CTX_set_mode() and
  199      SSL_CTX_set_read_ahead() again.
  200      [Kurt Roeckx]
  201 
  202   *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we
  203      now allow empty (zero character) pass phrases.
  204      [Richard Levitte]
  205 
  206   *) Apply blinding to binary field modular inversion and remove patent
  207      pending (OPENSSL_SUN_GF2M_DIV) BN_GF2m_mod_div implementation.
  208      [Billy Bob Brumley]
  209 
  210   *) Deprecate ec2_mult.c and unify scalar multiplication code paths for
  211      binary and prime elliptic curves.
  212      [Billy Bob Brumley]
  213 
  214   *) Remove ECDSA nonce padding: EC_POINT_mul is now responsible for
  215      constant time fixed point multiplication.
  216      [Billy Bob Brumley]
  217 
  218   *) Revise elliptic curve scalar multiplication with timing attack
  219      defenses: ec_wNAF_mul redirects to a constant time implementation
  220      when computing fixed point and variable point multiplication (which
  221      in OpenSSL are mostly used with secret scalars in keygen, sign,
  222      ECDH derive operations).
  223      [Billy Bob Brumley, Nicola Tuveri, Cesar Pereida García,
  224       Sohaib ul Hassan]
  225 
  226   *) Updated CONTRIBUTING
  227      [Rich Salz]
  228 
  229   *) Updated DRBG / RAND to request nonce and additional low entropy
  230      randomness from the system.
  231      [Matthias St. Pierre]
  232 
  233   *) Updated 'openssl rehash' to use OpenSSL consistent default.
  234      [Richard Levitte]
  235 
  236   *) Moved the load of the ssl_conf module to libcrypto, which helps
  237      loading engines that libssl uses before libssl is initialised.
  238      [Matt Caswell]
  239 
  240   *) Added EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA
  241      [Matt Caswell]
  242 
  243   *) Fixed X509_NAME_ENTRY_set to get multi-valued RDNs right in all cases.
  244      [Ingo Schwarze, Rich Salz]
  245 
  246   *) Added output of accepting IP address and port for 'openssl s_server'
  247      [Richard Levitte]
  248 
  249   *) Added a new API for TLSv1.3 ciphersuites:
  250         SSL_CTX_set_ciphersuites()
  251         SSL_set_ciphersuites()
  252      [Matt Caswell]
  253 
  254   *) Memory allocation failures consistenly add an error to the error
  255      stack.
  256      [Rich Salz]
  257 
  258   *) Don't use OPENSSL_ENGINES and OPENSSL_CONF environment values
  259      in libcrypto when run as setuid/setgid.
  260      [Bernd Edlinger]
  261 
  262   *) Load any config file by default when libssl is used.
  263      [Matt Caswell]
  264 
  265   *) Added new public header file <openssl/rand_drbg.h> and documentation
  266      for the RAND_DRBG API. See manual page RAND_DRBG(7) for an overview.
  267      [Matthias St. Pierre]
  268 
  269   *) QNX support removed (cannot find contributors to get their approval
  270      for the license change).
  271      [Rich Salz]
  272 
  273   *) TLSv1.3 replay protection for early data has been implemented. See the
  274      SSL_read_early_data() man page for further details.
  275      [Matt Caswell]
  276 
  277   *) Separated TLSv1.3 ciphersuite configuration out from TLSv1.2 ciphersuite
  278      configuration. TLSv1.3 ciphersuites are not compatible with TLSv1.2 and
  279      below. Similarly TLSv1.2 ciphersuites are not compatible with TLSv1.3.
  280      In order to avoid issues where legacy TLSv1.2 ciphersuite configuration
  281      would otherwise inadvertently disable all TLSv1.3 ciphersuites the
  282      configuration has been separated out. See the ciphers man page or the
  283      SSL_CTX_set_ciphersuites() man page for more information.
  284      [Matt Caswell]
  285 
  286   *) On POSIX (BSD, Linux, ...) systems the ocsp(1) command running
  287      in responder mode now supports the new "-multi" option, which
  288      spawns the specified number of child processes to handle OCSP
  289      requests.  The "-timeout" option now also limits the OCSP
  290      responder's patience to wait to receive the full client request
  291      on a newly accepted connection. Child processes are respawned
  292      as needed, and the CA index file is automatically reloaded
  293      when changed.  This makes it possible to run the "ocsp" responder
  294      as a long-running service, making the OpenSSL CA somewhat more
  295      feature-complete.  In this mode, most diagnostic messages logged
  296      after entering the event loop are logged via syslog(3) rather than
  297      written to stderr.
  298      [Viktor Dukhovni]
  299 
  300   *) Added support for X448 and Ed448. Heavily based on original work by
  301      Mike Hamburg.
  302      [Matt Caswell]
  303 
  304   *) Extend OSSL_STORE with capabilities to search and to narrow the set of
  305      objects loaded.  This adds the functions OSSL_STORE_expect() and
  306      OSSL_STORE_find() as well as needed tools to construct searches and
  307      get the search data out of them.
  308      [Richard Levitte]
  309 
  310   *) Support for TLSv1.3 added. Note that users upgrading from an earlier
  311      version of OpenSSL should review their configuration settings to ensure
  312      that they are still appropriate for TLSv1.3. For further information see:
  313      https://wiki.openssl.org/index.php/TLS1.3
  314      [Matt Caswell]
  315 
  316   *) Grand redesign of the OpenSSL random generator
  317 
  318      The default RAND method now utilizes an AES-CTR DRBG according to
  319      NIST standard SP 800-90Ar1. The new random generator is essentially
  320      a port of the default random generator from the OpenSSL FIPS 2.0
  321      object module. It is a hybrid deterministic random bit generator
  322      using an AES-CTR bit stream and which seeds and reseeds itself
  323      automatically using trusted system entropy sources.
  324 
  325      Some of its new features are:
  326       o Support for multiple DRBG instances with seed chaining.
  327       o The default RAND method makes use of a DRBG.
  328       o There is a public and private DRBG instance.
  329       o The DRBG instances are fork-safe.
  330       o Keep all global DRBG instances on the secure heap if it is enabled.
  331       o The public and private DRBG instance are per thread for lock free
  332         operation
  333      [Paul Dale, Benjamin Kaduk, Kurt Roeckx, Rich Salz, Matthias St. Pierre]
  334 
  335   *) Changed Configure so it only says what it does and doesn't dump
  336      so much data.  Instead, ./configdata.pm should be used as a script
  337      to display all sorts of configuration data.
  338      [Richard Levitte]
  339 
  340   *) Added processing of "make variables" to Configure.
  341      [Richard Levitte]
  342 
  343   *) Added SHA512/224 and SHA512/256 algorithm support.
  344      [Paul Dale]
  345 
  346   *) The last traces of Netware support, first removed in 1.1.0, have
  347      now been removed.
  348      [Rich Salz]
  349 
  350   *) Get rid of Makefile.shared, and in the process, make the processing
  351      of certain files (rc.obj, or the .def/.map/.opt files produced from
  352      the ordinal files) more visible and hopefully easier to trace and
  353      debug (or make silent).
  354      [Richard Levitte]
  355 
  356   *) Make it possible to have environment variable assignments as
  357      arguments to config / Configure.
  358      [Richard Levitte]
  359 
  360   *) Add multi-prime RSA (RFC 8017) support.
  361      [Paul Yang]
  362 
  363   *) Add SM3 implemented according to GB/T 32905-2016
  364      [ Jack Lloyd <jack.lloyd@ribose.com>,
  365        Ronald Tse <ronald.tse@ribose.com>,
  366        Erick Borsboom <erick.borsboom@ribose.com> ]
  367 
  368   *) Add 'Maximum Fragment Length' TLS extension negotiation and support
  369      as documented in RFC6066.
  370      Based on a patch from Tomasz Moń
  371      [Filipe Raimundo da Silva]
  372 
  373   *) Add SM4 implemented according to GB/T 32907-2016.
  374      [ Jack Lloyd <jack.lloyd@ribose.com>,
  375        Ronald Tse <ronald.tse@ribose.com>,
  376        Erick Borsboom <erick.borsboom@ribose.com> ]
  377 
  378   *) Reimplement -newreq-nodes and ERR_error_string_n; the
  379      original author does not agree with the license change.
  380      [Rich Salz]
  381 
  382   *) Add ARIA AEAD TLS support.
  383      [Jon Spillett]
  384 
  385   *) Some macro definitions to support VS6 have been removed.  Visual
  386      Studio 6 has not worked since 1.1.0
  387      [Rich Salz]
  388 
  389   *) Add ERR_clear_last_mark(), to allow callers to clear the last mark
  390      without clearing the errors.
  391      [Richard Levitte]
  392 
  393   *) Add "atfork" functions.  If building on a system that without
  394      pthreads, see doc/man3/OPENSSL_fork_prepare.pod for application
  395      requirements.  The RAND facility now uses/requires this.
  396      [Rich Salz]
  397 
  398   *) Add SHA3.
  399      [Andy Polyakov]
  400 
  401   *) The UI API becomes a permanent and integral part of libcrypto, i.e.
  402      not possible to disable entirely.  However, it's still possible to
  403      disable the console reading UI method, UI_OpenSSL() (use UI_null()
  404      as a fallback).
  405 
  406      To disable, configure with 'no-ui-console'.  'no-ui' is still
  407      possible to use as an alias.  Check at compile time with the
  408      macro OPENSSL_NO_UI_CONSOLE.  The macro OPENSSL_NO_UI is still
  409      possible to check and is an alias for OPENSSL_NO_UI_CONSOLE.
  410      [Richard Levitte]
  411 
  412   *) Add a STORE module, which implements a uniform and URI based reader of
  413      stores that can contain keys, certificates, CRLs and numerous other
  414      objects.  The main API is loosely based on a few stdio functions,
  415      and includes OSSL_STORE_open, OSSL_STORE_load, OSSL_STORE_eof,
  416      OSSL_STORE_error and OSSL_STORE_close.
  417      The implementation uses backends called "loaders" to implement arbitrary
  418      URI schemes.  There is one built in "loader" for the 'file' scheme.
  419      [Richard Levitte]
  420 
  421   *) Add devcrypto engine.  This has been implemented against cryptodev-linux,
  422      then adjusted to work on FreeBSD 8.4 as well.
  423      Enable by configuring with 'enable-devcryptoeng'.  This is done by default
  424      on BSD implementations, as cryptodev.h is assumed to exist on all of them.
  425      [Richard Levitte]
  426 
  427   *) Module names can prefixed with OSSL_ or OPENSSL_.  This affects
  428      util/mkerr.pl, which is adapted to allow those prefixes, leading to
  429      error code calls like this:
  430 
  431          OSSL_FOOerr(OSSL_FOO_F_SOMETHING, OSSL_FOO_R_WHATEVER);
  432 
  433      With this change, we claim the namespaces OSSL and OPENSSL in a manner
  434      that can be encoded in C.  For the foreseeable future, this will only
  435      affect new modules.
  436      [Richard Levitte and Tim Hudson]
  437 
  438   *) Removed BSD cryptodev engine.
  439      [Rich Salz]
  440 
  441   *) Add a build target 'build_all_generated', to build all generated files
  442      and only that.  This can be used to prepare everything that requires
  443      things like perl for a system that lacks perl and then move everything
  444      to that system and do the rest of the build there.
  445      [Richard Levitte]
  446 
  447   *) In the UI interface, make it possible to duplicate the user data.  This
  448      can be used by engines that need to retain the data for a longer time
  449      than just the call where this user data is passed.
  450      [Richard Levitte]
  451 
  452   *) Ignore the '-named_curve auto' value for compatibility of applications
  453      with OpenSSL 1.0.2.
  454      [Tomas Mraz <tmraz@fedoraproject.org>]
  455 
  456   *) Fragmented SSL/TLS alerts are no longer accepted. An alert message is 2
  457      bytes long. In theory it is permissible in SSLv3 - TLSv1.2 to fragment such
  458      alerts across multiple records (some of which could be empty). In practice
  459      it make no sense to send an empty alert record, or to fragment one. TLSv1.3
  460      prohibts this altogether and other libraries (BoringSSL, NSS) do not
  461      support this at all. Supporting it adds significant complexity to the
  462      record layer, and its removal is unlikely to cause inter-operability
  463      issues.
  464      [Matt Caswell]
  465 
  466   *) Add the ASN.1 types INT32, UINT32, INT64, UINT64 and variants prefixed
  467      with Z.  These are meant to replace LONG and ZLONG and to be size safe.
  468      The use of LONG and ZLONG is discouraged and scheduled for deprecation
  469      in OpenSSL 1.2.0.
  470      [Richard Levitte]
  471 
  472   *) Add the 'z' and 'j' modifiers to BIO_printf() et al formatting string,
  473      'z' is to be used for [s]size_t, and 'j' - with [u]int64_t.
  474      [Richard Levitte, Andy Polyakov]
  475 
  476   *) Add EC_KEY_get0_engine(), which does for EC_KEY what RSA_get0_engine()
  477      does for RSA, etc.
  478      [Richard Levitte]
  479 
  480   *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
  481      platform rather than 'mingw'.
  482      [Richard Levitte]
  483 
  484   *) The functions X509_STORE_add_cert and X509_STORE_add_crl return
  485      success if they are asked to add an object which already exists
  486      in the store. This change cascades to other functions which load
  487      certificates and CRLs.
  488      [Paul Dale]
  489 
  490   *) x86_64 assembly pack: annotate code with DWARF CFI directives to
  491      facilitate stack unwinding even from assembly subroutines.
  492      [Andy Polyakov]
  493 
  494   *) Remove VAX C specific definitions of OPENSSL_EXPORT, OPENSSL_EXTERN.
  495      Also remove OPENSSL_GLOBAL entirely, as it became a no-op.
  496      [Richard Levitte]
  497 
  498   *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
  499      VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
  500      which is the minimum version we support.
  501      [Richard Levitte]
  502 
  503   *) Certificate time validation (X509_cmp_time) enforces stricter
  504      compliance with RFC 5280. Fractional seconds and timezone offsets
  505      are no longer allowed.
  506      [Emilia Käsper]
  507 
  508   *) Add support for ARIA
  509      [Paul Dale]
  510 
  511   *) s_client will now send the Server Name Indication (SNI) extension by
  512      default unless the new "-noservername" option is used. The server name is
  513      based on the host provided to the "-connect" option unless overridden by
  514      using "-servername".
  515      [Matt Caswell]
  516 
  517   *) Add support for SipHash
  518      [Todd Short]
  519 
  520   *) OpenSSL now fails if it receives an unrecognised record type in TLS1.0
  521      or TLS1.1. Previously this only happened in SSLv3 and TLS1.2. This is to
  522      prevent issues where no progress is being made and the peer continually
  523      sends unrecognised record types, using up resources processing them.
  524      [Matt Caswell]
  525 
  526   *) 'openssl passwd' can now produce SHA256 and SHA512 based output,
  527      using the algorithm defined in
  528      https://www.akkadia.org/drepper/SHA-crypt.txt
  529      [Richard Levitte]
  530 
  531   *) Heartbeat support has been removed; the ABI is changed for now.
  532      [Richard Levitte, Rich Salz]
  533 
  534   *) Support for SSL_OP_NO_ENCRYPT_THEN_MAC in SSL_CONF_cmd.
  535      [Emilia Käsper]
  536 
  537   *) The RSA "null" method, which was partially supported to avoid patent
  538      issues, has been replaced to always returns NULL.
  539      [Rich Salz]
  540 
  541 
  542  Changes between 1.1.0h and 1.1.0i [xx XXX xxxx]
  543 
  544   *) Client DoS due to large DH parameter
  545 
  546      During key agreement in a TLS handshake using a DH(E) based ciphersuite a
  547      malicious server can send a very large prime value to the client. This will
  548      cause the client to spend an unreasonably long period of time generating a
  549      key for this prime resulting in a hang until the client has finished. This
  550      could be exploited in a Denial Of Service attack.
  551 
  552      This issue was reported to OpenSSL on 5th June 2018 by Guido Vranken
  553      (CVE-2018-0732)
  554      [Guido Vranken]
  555 
  556   *) Cache timing vulnerability in RSA Key Generation
  557 
  558      The OpenSSL RSA Key generation algorithm has been shown to be vulnerable to
  559      a cache timing side channel attack. An attacker with sufficient access to
  560      mount cache timing attacks during the RSA key generation process could
  561      recover the private key.
  562 
  563      This issue was reported to OpenSSL on 4th April 2018 by Alejandro Cabrera
  564      Aldaya, Billy Brumley, Cesar Pereida Garcia and Luis Manuel Alvarez Tapia.
  565      (CVE-2018-0737)
  566      [Billy Brumley]
  567 
  568   *) Make EVP_PKEY_asn1_new() a bit stricter about its input.  A NULL pem_str
  569      parameter is no longer accepted, as it leads to a corrupt table.  NULL
  570      pem_str is reserved for alias entries only.
  571      [Richard Levitte]
  572 
  573   *) Revert blinding in ECDSA sign and instead make problematic addition
  574      length-invariant. Switch even to fixed-length Montgomery multiplication.
  575      [Andy Polyakov]
  576 
  577   *) Change generating and checking of primes so that the error rate of not
  578      being prime depends on the intended use based on the size of the input.
  579      For larger primes this will result in more rounds of Miller-Rabin.
  580      The maximal error rate for primes with more than 1080 bits is lowered
  581      to 2^-128.
  582      [Kurt Roeckx, Annie Yousar]
  583 
  584   *) Increase the number of Miller-Rabin rounds for DSA key generating to 64.
  585      [Kurt Roeckx]
  586 
  587   *) Add blinding to ECDSA and DSA signatures to protect against side channel
  588      attacks discovered by Keegan Ryan (NCC Group).
  589      [Matt Caswell]
  590 
  591   *) When unlocking a pass phrase protected PEM file or PKCS#8 container, we
  592      now allow empty (zero character) pass phrases.
  593      [Richard Levitte]
  594 
  595   *) Certificate time validation (X509_cmp_time) enforces stricter
  596      compliance with RFC 5280. Fractional seconds and timezone offsets
  597      are no longer allowed.
  598      [Emilia Käsper]
  599 
  600   *) Fixed a text canonicalisation bug in CMS
  601 
  602      Where a CMS detached signature is used with text content the text goes
  603      through a canonicalisation process first prior to signing or verifying a
  604      signature. This process strips trailing space at the end of lines, converts
  605      line terminators to CRLF and removes additional trailing line terminators
  606      at the end of a file. A bug in the canonicalisation process meant that
  607      some characters, such as form-feed, were incorrectly treated as whitespace
  608      and removed. This is contrary to the specification (RFC5485). This fix
  609      could mean that detached text data signed with an earlier version of
  610      OpenSSL 1.1.0 may fail to verify using the fixed version, or text data
  611      signed with a fixed OpenSSL may fail to verify with an earlier version of
  612      OpenSSL 1.1.0. A workaround is to only verify the canonicalised text data
  613      and use the "-binary" flag (for the "cms" command line application) or set
  614      the SMIME_BINARY/PKCS7_BINARY/CMS_BINARY flags (if using CMS_verify()).
  615      [Matt Caswell]
  616 
  617  Changes between 1.1.0g and 1.1.0h [27 Mar 2018]
  618 
  619   *) Constructed ASN.1 types with a recursive definition could exceed the stack
  620 
  621      Constructed ASN.1 types with a recursive definition (such as can be found
  622      in PKCS7) could eventually exceed the stack given malicious input with
  623      excessive recursion. This could result in a Denial Of Service attack. There
  624      are no such structures used within SSL/TLS that come from untrusted sources
  625      so this is considered safe.
  626 
  627      This issue was reported to OpenSSL on 4th January 2018 by the OSS-fuzz
  628      project.
  629      (CVE-2018-0739)
  630      [Matt Caswell]
  631 
  632   *) Incorrect CRYPTO_memcmp on HP-UX PA-RISC
  633 
  634      Because of an implementation bug the PA-RISC CRYPTO_memcmp function is
  635      effectively reduced to only comparing the least significant bit of each
  636      byte. This allows an attacker to forge messages that would be considered as
  637      authenticated in an amount of tries lower than that guaranteed by the
  638      security claims of the scheme. The module can only be compiled by the
  639      HP-UX assembler, so that only HP-UX PA-RISC targets are affected.
  640 
  641      This issue was reported to OpenSSL on 2nd March 2018 by Peter Waltenberg
  642      (IBM).
  643      (CVE-2018-0733)
  644      [Andy Polyakov]
  645 
  646   *) Add a build target 'build_all_generated', to build all generated files
  647      and only that.  This can be used to prepare everything that requires
  648      things like perl for a system that lacks perl and then move everything
  649      to that system and do the rest of the build there.
  650      [Richard Levitte]
  651 
  652   *) Backport SSL_OP_NO_RENGOTIATION
  653 
  654      OpenSSL 1.0.2 and below had the ability to disable renegotiation using the
  655      (undocumented) SSL3_FLAGS_NO_RENEGOTIATE_CIPHERS flag. Due to the opacity
  656      changes this is no longer possible in 1.1.0. Therefore the new
  657      SSL_OP_NO_RENEGOTIATION option from 1.1.1-dev has been backported to
  658      1.1.0 to provide equivalent functionality.
  659 
  660      Note that if an application built against 1.1.0h headers (or above) is run
  661      using an older version of 1.1.0 (prior to 1.1.0h) then the option will be
  662      accepted but nothing will happen, i.e. renegotiation will not be prevented.
  663      [Matt Caswell]
  664 
  665   *) Removed the OS390-Unix config target.  It relied on a script that doesn't
  666      exist.
  667      [Rich Salz]
  668 
  669   *) rsaz_1024_mul_avx2 overflow bug on x86_64
  670 
  671      There is an overflow bug in the AVX2 Montgomery multiplication procedure
  672      used in exponentiation with 1024-bit moduli. No EC algorithms are affected.
  673      Analysis suggests that attacks against RSA and DSA as a result of this
  674      defect would be very difficult to perform and are not believed likely.
  675      Attacks against DH1024 are considered just feasible, because most of the
  676      work necessary to deduce information about a private key may be performed
  677      offline. The amount of resources required for such an attack would be
  678      significant. However, for an attack on TLS to be meaningful, the server
  679      would have to share the DH1024 private key among multiple clients, which is
  680      no longer an option since CVE-2016-0701.
  681 
  682      This only affects processors that support the AVX2 but not ADX extensions
  683      like Intel Haswell (4th generation).
  684 
  685      This issue was reported to OpenSSL by David Benjamin (Google). The issue
  686      was originally found via the OSS-Fuzz project.
  687      (CVE-2017-3738)
  688      [Andy Polyakov]
  689 
  690  Changes between 1.1.0f and 1.1.0g [2 Nov 2017]
  691 
  692   *) bn_sqrx8x_internal carry bug on x86_64
  693 
  694      There is a carry propagating bug in the x86_64 Montgomery squaring
  695      procedure. No EC algorithms are affected. Analysis suggests that attacks
  696      against RSA and DSA as a result of this defect would be very difficult to
  697      perform and are not believed likely. Attacks against DH are considered just
  698      feasible (although very difficult) because most of the work necessary to
  699      deduce information about a private key may be performed offline. The amount
  700      of resources required for such an attack would be very significant and
  701      likely only accessible to a limited number of attackers. An attacker would
  702      additionally need online access to an unpatched system using the target
  703      private key in a scenario with persistent DH parameters and a private
  704      key that is shared between multiple clients.
  705 
  706      This only affects processors that support the BMI1, BMI2 and ADX extensions
  707      like Intel Broadwell (5th generation) and later or AMD Ryzen.
  708 
  709      This issue was reported to OpenSSL by the OSS-Fuzz project.
  710      (CVE-2017-3736)
  711      [Andy Polyakov]
  712 
  713   *) Malformed X.509 IPAddressFamily could cause OOB read
  714 
  715      If an X.509 certificate has a malformed IPAddressFamily extension,
  716      OpenSSL could do a one-byte buffer overread. The most likely result
  717      would be an erroneous display of the certificate in text format.
  718 
  719      This issue was reported to OpenSSL by the OSS-Fuzz project.
  720      (CVE-2017-3735)
  721      [Rich Salz]
  722 
  723  Changes between 1.1.0e and 1.1.0f [25 May 2017]
  724 
  725   *) Have 'config' recognise 64-bit mingw and choose 'mingw64' as the target
  726      platform rather than 'mingw'.
  727      [Richard Levitte]
  728 
  729   *) Remove the VMS-specific reimplementation of gmtime from crypto/o_times.c.
  730      VMS C's RTL has a fully up to date gmtime() and gmtime_r() since V7.1,
  731      which is the minimum version we support.
  732      [Richard Levitte]
  733 
  734  Changes between 1.1.0d and 1.1.0e [16 Feb 2017]
  735 
  736   *) Encrypt-Then-Mac renegotiation crash
  737 
  738      During a renegotiation handshake if the Encrypt-Then-Mac extension is
  739      negotiated where it was not in the original handshake (or vice-versa) then
  740      this can cause OpenSSL to crash (dependant on ciphersuite). Both clients
  741      and servers are affected.
  742 
  743      This issue was reported to OpenSSL by Joe Orton (Red Hat).
  744      (CVE-2017-3733)
  745      [Matt Caswell]
  746 
  747  Changes between 1.1.0c and 1.1.0d [26 Jan 2017]
  748 
  749   *) Truncated packet could crash via OOB read
  750 
  751      If one side of an SSL/TLS path is running on a 32-bit host and a specific
  752      cipher is being used, then a truncated packet can cause that host to
  753      perform an out-of-bounds read, usually resulting in a crash.
  754 
  755      This issue was reported to OpenSSL by Robert Święcki of Google.
  756      (CVE-2017-3731)
  757      [Andy Polyakov]
  758 
  759   *) Bad (EC)DHE parameters cause a client crash
  760 
  761      If a malicious server supplies bad parameters for a DHE or ECDHE key
  762      exchange then this can result in the client attempting to dereference a
  763      NULL pointer leading to a client crash. This could be exploited in a Denial
  764      of Service attack.
  765 
  766      This issue was reported to OpenSSL by Guido Vranken.
  767      (CVE-2017-3730)
  768      [Matt Caswell]
  769 
  770   *) BN_mod_exp may produce incorrect results on x86_64
  771 
  772      There is a carry propagating bug in the x86_64 Montgomery squaring
  773      procedure. No EC algorithms are affected. Analysis suggests that attacks
  774      against RSA and DSA as a result of this defect would be very difficult to
  775      perform and are not believed likely. Attacks against DH are considered just
  776      feasible (although very difficult) because most of the work necessary to
  777      deduce information about a private key may be performed offline. The amount
  778      of resources required for such an attack would be very significant and
  779      likely only accessible to a limited number of attackers. An attacker would
  780      additionally need online access to an unpatched system using the target
  781      private key in a scenario with persistent DH parameters and a private
  782      key that is shared between multiple clients. For example this can occur by
  783      default in OpenSSL DHE based SSL/TLS ciphersuites. Note: This issue is very
  784      similar to CVE-2015-3193 but must be treated as a separate problem.
  785 
  786      This issue was reported to OpenSSL by the OSS-Fuzz project.
  787      (CVE-2017-3732)
  788      [Andy Polyakov]
  789 
  790  Changes between 1.1.0b and 1.1.0c [10 Nov 2016]
  791 
  792   *) ChaCha20/Poly1305 heap-buffer-overflow
  793 
  794      TLS connections using *-CHACHA20-POLY1305 ciphersuites are susceptible to
  795      a DoS attack by corrupting larger payloads. This can result in an OpenSSL
  796      crash. This issue is not considered to be exploitable beyond a DoS.
  797 
  798      This issue was reported to OpenSSL by Robert Święcki (Google Security Team)
  799      (CVE-2016-7054)
  800      [Richard Levitte]
  801 
  802   *) CMS Null dereference
  803 
  804      Applications parsing invalid CMS structures can crash with a NULL pointer
  805      dereference. This is caused by a bug in the handling of the ASN.1 CHOICE
  806      type in OpenSSL 1.1.0 which can result in a NULL value being passed to the
  807      structure callback if an attempt is made to free certain invalid encodings.
  808      Only CHOICE structures using a callback which do not handle NULL value are
  809      affected.
  810 
  811      This issue was reported to OpenSSL by Tyler Nighswander of ForAllSecure.
  812      (CVE-2016-7053)
  813      [Stephen Henson]
  814 
  815   *) Montgomery multiplication may produce incorrect results
  816 
  817      There is a carry propagating bug in the Broadwell-specific Montgomery
  818      multiplication procedure that handles input lengths divisible by, but
  819      longer than 256 bits. Analysis suggests that attacks against RSA, DSA
  820      and DH private keys are impossible. This is because the subroutine in
  821      question is not used in operations with the private key itself and an input
  822      of the attacker's direct choice. Otherwise the bug can manifest itself as
  823      transient authentication and key negotiation failures or reproducible
  824      erroneous outcome of public-key operations with specially crafted input.
  825      Among EC algorithms only Brainpool P-512 curves are affected and one
  826      presumably can attack ECDH key negotiation. Impact was not analyzed in
  827      detail, because pre-requisites for attack are considered unlikely. Namely
  828      multiple clients have to choose the curve in question and the server has to
  829      share the private key among them, neither of which is default behaviour.
  830      Even then only clients that chose the curve will be affected.
  831 
  832      This issue was publicly reported as transient failures and was not
  833      initially recognized as a security issue. Thanks to Richard Morgan for
  834      providing reproducible case.
  835      (CVE-2016-7055)
  836      [Andy Polyakov]
  837 
  838   *) Removed automatic addition of RPATH in shared libraries and executables,
  839      as this was a remainder from OpenSSL 1.0.x and isn't needed any more.
  840      [Richard Levitte]
  841 
  842  Changes between 1.1.0a and 1.1.0b [26 Sep 2016]
  843 
  844   *) Fix Use After Free for large message sizes
  845 
  846      The patch applied to address CVE-2016-6307 resulted in an issue where if a
  847      message larger than approx 16k is received then the underlying buffer to
  848      store the incoming message is reallocated and moved. Unfortunately a
  849      dangling pointer to the old location is left which results in an attempt to
  850      write to the previously freed location. This is likely to result in a
  851      crash, however it could potentially lead to execution of arbitrary code.
  852 
  853      This issue only affects OpenSSL 1.1.0a.
  854 
  855      This issue was reported to OpenSSL by Robert Święcki.
  856      (CVE-2016-6309)
  857      [Matt Caswell]
  858 
  859  Changes between 1.1.0 and 1.1.0a [22 Sep 2016]
  860 
  861   *) OCSP Status Request extension unbounded memory growth
  862 
  863      A malicious client can send an excessively large OCSP Status Request
  864      extension. If that client continually requests renegotiation, sending a
  865      large OCSP Status Request extension each time, then there will be unbounded
  866      memory growth on the server. This will eventually lead to a Denial Of
  867      Service attack through memory exhaustion. Servers with a default
  868      configuration are vulnerable even if they do not support OCSP. Builds using
  869      the "no-ocsp" build time option are not affected.
  870 
  871      This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
  872      (CVE-2016-6304)
  873      [Matt Caswell]
  874 
  875   *) SSL_peek() hang on empty record
  876 
  877      OpenSSL 1.1.0 SSL/TLS will hang during a call to SSL_peek() if the peer
  878      sends an empty record. This could be exploited by a malicious peer in a
  879      Denial Of Service attack.
  880 
  881      This issue was reported to OpenSSL by Alex Gaynor.
  882      (CVE-2016-6305)
  883      [Matt Caswell]
  884 
  885   *) Excessive allocation of memory in tls_get_message_header() and
  886      dtls1_preprocess_fragment()
  887 
  888      A (D)TLS message includes 3 bytes for its length in the header for the
  889      message. This would allow for messages up to 16Mb in length. Messages of
  890      this length are excessive and OpenSSL includes a check to ensure that a
  891      peer is sending reasonably sized messages in order to avoid too much memory
  892      being consumed to service a connection. A flaw in the logic of version
  893      1.1.0 means that memory for the message is allocated too early, prior to
  894      the excessive message length check. Due to way memory is allocated in
  895      OpenSSL this could mean an attacker could force up to 21Mb to be allocated
  896      to service a connection. This could lead to a Denial of Service through
  897      memory exhaustion. However, the excessive message length check still takes
  898      place, and this would cause the connection to immediately fail. Assuming
  899      that the application calls SSL_free() on the failed connection in a timely
  900      manner then the 21Mb of allocated memory will then be immediately freed
  901      again. Therefore the excessive memory allocation will be transitory in
  902      nature. This then means that there is only a security impact if:
  903 
  904      1) The application does not call SSL_free() in a timely manner in the event
  905      that the connection fails
  906      or
  907      2) The application is working in a constrained environment where there is
  908      very little free memory
  909      or
  910      3) The attacker initiates multiple connection attempts such that there are
  911      multiple connections in a state where memory has been allocated for the
  912      connection; SSL_free() has not yet been called; and there is insufficient
  913      memory to service the multiple requests.
  914 
  915      Except in the instance of (1) above any Denial Of Service is likely to be
  916      transitory because as soon as the connection fails the memory is
  917      subsequently freed again in the SSL_free() call. However there is an
  918      increased risk during this period of application crashes due to the lack of
  919      memory - which would then mean a more serious Denial of Service.
  920 
  921      This issue was reported to OpenSSL by Shi Lei (Gear Team, Qihoo 360 Inc.)
  922      (CVE-2016-6307 and CVE-2016-6308)
  923      [Matt Caswell]
  924 
  925   *) solaris-x86-cc, i.e. 32-bit configuration with vendor compiler,
  926      had to be removed. Primary reason is that vendor assembler can't
  927      assemble our modules with -KPIC flag. As result it, assembly
  928      support, was not even available as option. But its lack means
  929      lack of side-channel resistant code, which is incompatible with
  930      security by todays standards. Fortunately gcc is readily available
  931      prepackaged option, which we firmly point at...
  932      [Andy Polyakov]
  933 
  934  Changes between 1.0.2h and 1.1.0  [25 Aug 2016]
  935 
  936   *) Windows command-line tool supports UTF-8 opt-in option for arguments
  937      and console input. Setting OPENSSL_WIN32_UTF8 environment variable
  938      (to any value) allows Windows user to access PKCS#12 file generated
  939      with Windows CryptoAPI and protected with non-ASCII password, as well
  940      as files generated under UTF-8 locale on Linux also protected with
  941      non-ASCII password.
  942      [Andy Polyakov]
  943 
  944   *) To mitigate the SWEET32 attack (CVE-2016-2183), 3DES cipher suites
  945      have been disabled by default and removed from DEFAULT, just like RC4.
  946      See the RC4 item below to re-enable both.
  947      [Rich Salz]
  948 
  949   *) The method for finding the storage location for the Windows RAND seed file
  950      has changed. First we check %RANDFILE%. If that is not set then we check
  951      the directories %HOME%, %USERPROFILE% and %SYSTEMROOT% in that order. If
  952      all else fails we fall back to C:\.
  953      [Matt Caswell]
  954 
  955   *) The EVP_EncryptUpdate() function has had its return type changed from void
  956      to int. A return of 0 indicates and error while a return of 1 indicates
  957      success.
  958      [Matt Caswell]
  959 
  960   *) The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
  961      DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
  962      off the constant time implementation for RSA, DSA and DH have been made
  963      no-ops and deprecated.
  964      [Matt Caswell]
  965 
  966   *) Windows RAND implementation was simplified to only get entropy by
  967      calling CryptGenRandom(). Various other RAND-related tickets
  968      were also closed.
  969      [Joseph Wylie Yandle, Rich Salz]
  970 
  971   *) The stack and lhash API's were renamed to start with OPENSSL_SK_
  972      and OPENSSL_LH_, respectively.  The old names are available
  973      with API compatibility.  They new names are now completely documented.
  974      [Rich Salz]
  975 
  976   *) Unify TYPE_up_ref(obj) methods signature.
  977      SSL_CTX_up_ref(), SSL_up_ref(), X509_up_ref(), EVP_PKEY_up_ref(),
  978      X509_CRL_up_ref(), X509_OBJECT_up_ref_count() methods are now returning an
  979      int (instead of void) like all others TYPE_up_ref() methods.
  980      So now these methods also check the return value of CRYPTO_atomic_add(),
  981      and the validity of object reference counter.
  982      [fdasilvayy@gmail.com]
  983 
  984   *) With Windows Visual Studio builds, the .pdb files are installed
  985      alongside the installed libraries and executables.  For a static
  986      library installation, ossl_static.pdb is the associate compiler
  987      generated .pdb file to be used when linking programs.
  988      [Richard Levitte]
  989 
  990   *) Remove openssl.spec.  Packaging files belong with the packagers.
  991      [Richard Levitte]
  992 
  993   *) Automatic Darwin/OSX configuration has had a refresh, it will now
  994      recognise x86_64 architectures automatically.  You can still decide
  995      to build for a different bitness with the environment variable
  996      KERNEL_BITS (can be 32 or 64), for example:
  997 
  998          KERNEL_BITS=32 ./config
  999 
 1000      [Richard Levitte]
 1001 
 1002   *) Change default algorithms in pkcs8 utility to use PKCS#5 v2.0,
 1003      256 bit AES and HMAC with SHA256.
 1004      [Steve Henson]
 1005 
 1006   *) Remove support for MIPS o32 ABI on IRIX (and IRIX only).
 1007      [Andy Polyakov]
 1008 
 1009   *) Triple-DES ciphers have been moved from HIGH to MEDIUM.
 1010      [Rich Salz]
 1011 
 1012   *) To enable users to have their own config files and build file templates,
 1013      Configure looks in the directory indicated by the environment variable
 1014      OPENSSL_LOCAL_CONFIG_DIR as well as the in-source Configurations/
 1015      directory.  On VMS, OPENSSL_LOCAL_CONFIG_DIR is expected to be a logical
 1016      name and is used as is.
 1017      [Richard Levitte]
 1018 
 1019   *) The following datatypes were made opaque: X509_OBJECT, X509_STORE_CTX,
 1020      X509_STORE, X509_LOOKUP, and X509_LOOKUP_METHOD.  The unused type
 1021      X509_CERT_FILE_CTX was removed.
 1022      [Rich Salz]
 1023 
 1024   *) "shared" builds are now the default. To create only static libraries use
 1025      the "no-shared" Configure option.
 1026      [Matt Caswell]
 1027 
 1028   *) Remove the no-aes, no-hmac, no-rsa, no-sha and no-md5 Configure options.
 1029      All of these option have not worked for some while and are fundamental
 1030      algorithms.
 1031      [Matt Caswell]
 1032 
 1033   *) Make various cleanup routines no-ops and mark them as deprecated. Most
 1034      global cleanup functions are no longer required because they are handled
 1035      via auto-deinit (see OPENSSL_init_crypto and OPENSSL_init_ssl man pages).
 1036      Explicitly de-initing can cause problems (e.g. where a library that uses
 1037      OpenSSL de-inits, but an application is still using it). The affected
 1038      functions are CONF_modules_free(), ENGINE_cleanup(), OBJ_cleanup(),
 1039      EVP_cleanup(), BIO_sock_cleanup(), CRYPTO_cleanup_all_ex_data(),
 1040      RAND_cleanup(), SSL_COMP_free_compression_methods(), ERR_free_strings() and
 1041      COMP_zlib_cleanup().
 1042      [Matt Caswell]
 1043 
 1044   *) --strict-warnings no longer enables runtime debugging options
 1045      such as REF_DEBUG. Instead, debug options are automatically
 1046      enabled with '--debug' builds.
 1047      [Andy Polyakov, Emilia Käsper]
 1048 
 1049   *) Made DH and DH_METHOD opaque. The structures for managing DH objects
 1050      have been moved out of the public header files. New functions for managing
 1051      these have been added.
 1052      [Matt Caswell]
 1053 
 1054   *) Made RSA and RSA_METHOD opaque. The structures for managing RSA
 1055      objects have been moved out of the public header files. New
 1056      functions for managing these have been added.
 1057      [Richard Levitte]
 1058 
 1059   *) Made DSA and DSA_METHOD opaque. The structures for managing DSA objects
 1060      have been moved out of the public header files. New functions for managing
 1061      these have been added.
 1062      [Matt Caswell]
 1063 
 1064   *) Made BIO and BIO_METHOD opaque. The structures for managing BIOs have been
 1065      moved out of the public header files. New functions for managing these
 1066      have been added.
 1067      [Matt Caswell]
 1068 
 1069   *) Removed no-rijndael as a config option. Rijndael is an old name for AES.
 1070      [Matt Caswell]
 1071 
 1072   *) Removed the mk1mf build scripts.
 1073      [Richard Levitte]
 1074 
 1075   *) Headers are now wrapped, if necessary, with OPENSSL_NO_xxx, so
 1076      it is always safe to #include a header now.
 1077      [Rich Salz]
 1078 
 1079   *) Removed the aged BC-32 config and all its supporting scripts
 1080      [Richard Levitte]
 1081 
 1082   *) Removed support for Ultrix, Netware, and OS/2.
 1083      [Rich Salz]
 1084 
 1085   *) Add support for HKDF.
 1086      [Alessandro Ghedini]
 1087 
 1088   *) Add support for blake2b and blake2s
 1089      [Bill Cox]
 1090 
 1091   *) Added support for "pipelining". Ciphers that have the
 1092      EVP_CIPH_FLAG_PIPELINE flag set have a capability to process multiple
 1093      encryptions/decryptions simultaneously. There are currently no built-in
 1094      ciphers with this property but the expectation is that engines will be able
 1095      to offer it to significantly improve throughput. Support has been extended
 1096      into libssl so that multiple records for a single connection can be
 1097      processed in one go (for >=TLS 1.1).
 1098      [Matt Caswell]
 1099 
 1100   *) Added the AFALG engine. This is an async capable engine which is able to
 1101      offload work to the Linux kernel. In this initial version it only supports
 1102      AES128-CBC. The kernel must be version 4.1.0 or greater.
 1103      [Catriona Lucey]
 1104 
 1105   *) OpenSSL now uses a new threading API. It is no longer necessary to
 1106      set locking callbacks to use OpenSSL in a multi-threaded environment. There
 1107      are two supported threading models: pthreads and windows threads. It is
 1108      also possible to configure OpenSSL at compile time for "no-threads". The
 1109      old threading API should no longer be used. The functions have been
 1110      replaced with "no-op" compatibility macros.
 1111      [Alessandro Ghedini, Matt Caswell]
 1112 
 1113   *) Modify behavior of ALPN to invoke callback after SNI/servername
 1114      callback, such that updates to the SSL_CTX affect ALPN.
 1115      [Todd Short]
 1116 
 1117   *) Add SSL_CIPHER queries for authentication and key-exchange.
 1118      [Todd Short]
 1119 
 1120   *) Changes to the DEFAULT cipherlist:
 1121        - Prefer (EC)DHE handshakes over plain RSA.
 1122        - Prefer AEAD ciphers over legacy ciphers.
 1123        - Prefer ECDSA over RSA when both certificates are available.
 1124        - Prefer TLSv1.2 ciphers/PRF.
 1125        - Remove DSS, SEED, IDEA, CAMELLIA, and AES-CCM from the
 1126          default cipherlist.
 1127      [Emilia Käsper]
 1128 
 1129   *) Change the ECC default curve list to be this, in order: x25519,
 1130      secp256r1, secp521r1, secp384r1.
 1131      [Rich Salz]
 1132 
 1133   *) RC4 based libssl ciphersuites are now classed as "weak" ciphers and are
 1134      disabled by default. They can be re-enabled using the
 1135      enable-weak-ssl-ciphers option to Configure.
 1136      [Matt Caswell]
 1137 
 1138   *) If the server has ALPN configured, but supports no protocols that the
 1139      client advertises, send a fatal "no_application_protocol" alert.
 1140      This behaviour is SHALL in RFC 7301, though it isn't universally
 1141      implemented by other servers.
 1142      [Emilia Käsper]
 1143 
 1144   *) Add X25519 support.
 1145      Add ASN.1 and EVP_PKEY methods for X25519. This includes support
 1146      for public and private key encoding using the format documented in
 1147      draft-ietf-curdle-pkix-02. The corresponding EVP_PKEY method supports
 1148      key generation and key derivation.
 1149 
 1150      TLS support complies with draft-ietf-tls-rfc4492bis-08 and uses
 1151      X25519(29).
 1152      [Steve Henson]
 1153 
 1154   *) Deprecate SRP_VBASE_get_by_user.
 1155      SRP_VBASE_get_by_user had inconsistent memory management behaviour.
 1156      In order to fix an unavoidable memory leak (CVE-2016-0798),
 1157      SRP_VBASE_get_by_user was changed to ignore the "fake user" SRP
 1158      seed, even if the seed is configured.
 1159 
 1160      Users should use SRP_VBASE_get1_by_user instead. Note that in
 1161      SRP_VBASE_get1_by_user, caller must free the returned value. Note
 1162      also that even though configuring the SRP seed attempts to hide
 1163      invalid usernames by continuing the handshake with fake
 1164      credentials, this behaviour is not constant time and no strong
 1165      guarantees are made that the handshake is indistinguishable from
 1166      that of a valid user.
 1167      [Emilia Käsper]
 1168 
 1169   *) Configuration change; it's now possible to build dynamic engines
 1170      without having to build shared libraries and vice versa.  This
 1171      only applies to the engines in engines/, those in crypto/engine/
 1172      will always be built into libcrypto (i.e. "static").
 1173 
 1174      Building dynamic engines is enabled by default; to disable, use
 1175      the configuration option "disable-dynamic-engine".
 1176 
 1177      The only requirements for building dynamic engines are the
 1178      presence of the DSO module and building with position independent
 1179      code, so they will also automatically be disabled if configuring
 1180      with "disable-dso" or "disable-pic".
 1181 
 1182      The macros OPENSSL_NO_STATIC_ENGINE and OPENSSL_NO_DYNAMIC_ENGINE
 1183      are also taken away from openssl/opensslconf.h, as they are
 1184      irrelevant.
 1185      [Richard Levitte]
 1186 
 1187   *) Configuration change; if there is a known flag to compile
 1188      position independent code, it will always be applied on the
 1189      libcrypto and libssl object files, and never on the application
 1190      object files.  This means other libraries that use routines from
 1191      libcrypto / libssl can be made into shared libraries regardless
 1192      of how OpenSSL was configured.
 1193 
 1194      If this isn't desirable, the configuration options "disable-pic"
 1195      or "no-pic" can be used to disable the use of PIC.  This will
 1196      also disable building shared libraries and dynamic engines.
 1197      [Richard Levitte]
 1198 
 1199   *) Removed JPAKE code.  It was experimental and has no wide use.
 1200      [Rich Salz]
 1201 
 1202   *) The INSTALL_PREFIX Makefile variable has been renamed to
 1203      DESTDIR.  That makes for less confusion on what this variable
 1204      is for.  Also, the configuration option --install_prefix is
 1205      removed.
 1206      [Richard Levitte]
 1207 
 1208   *) Heartbeat for TLS has been removed and is disabled by default
 1209      for DTLS; configure with enable-heartbeats.  Code that uses the
 1210      old #define's might need to be updated.
 1211      [Emilia Käsper, Rich Salz]
 1212 
 1213   *) Rename REF_CHECK to REF_DEBUG.
 1214      [Rich Salz]
 1215 
 1216   *) New "unified" build system
 1217 
 1218      The "unified" build system is aimed to be a common system for all
 1219      platforms we support.  With it comes new support for VMS.
 1220 
 1221      This system builds supports building in a different directory tree
 1222      than the source tree.  It produces one Makefile (for unix family
 1223      or lookalikes), or one descrip.mms (for VMS).
 1224 
 1225      The source of information to make the Makefile / descrip.mms is
 1226      small files called 'build.info', holding the necessary
 1227      information for each directory with source to compile, and a
 1228      template in Configurations, like unix-Makefile.tmpl or
 1229      descrip.mms.tmpl.
 1230 
 1231      With this change, the library names were also renamed on Windows
 1232      and on VMS.  They now have names that are closer to the standard
 1233      on Unix, and include the major version number, and in certain
 1234      cases, the architecture they are built for.  See "Notes on shared
 1235      libraries" in INSTALL.
 1236 
 1237      We rely heavily on the perl module Text::Template.
 1238      [Richard Levitte]
 1239 
 1240   *) Added support for auto-initialisation and de-initialisation of the library.
 1241      OpenSSL no longer requires explicit init or deinit routines to be called,
 1242      except in certain circumstances. See the OPENSSL_init_crypto() and
 1243      OPENSSL_init_ssl() man pages for further information.
 1244      [Matt Caswell]
 1245 
 1246   *) The arguments to the DTLSv1_listen function have changed. Specifically the
 1247      "peer" argument is now expected to be a BIO_ADDR object.
 1248 
 1249   *) Rewrite of BIO networking library. The BIO library lacked consistent
 1250      support of IPv6, and adding it required some more extensive
 1251      modifications.  This introduces the BIO_ADDR and BIO_ADDRINFO types,
 1252      which hold all types of addresses and chains of address information.
 1253      It also introduces a new API, with functions like BIO_socket,
 1254      BIO_connect, BIO_listen, BIO_lookup and a rewrite of BIO_accept.
 1255      The source/sink BIOs BIO_s_connect, BIO_s_accept and BIO_s_datagram
 1256      have been adapted accordingly.
 1257      [Richard Levitte]
 1258 
 1259   *) RSA_padding_check_PKCS1_type_1 now accepts inputs with and without
 1260      the leading 0-byte.
 1261      [Emilia Käsper]
 1262 
 1263   *) CRIME protection: disable compression by default, even if OpenSSL is
 1264      compiled with zlib enabled. Applications can still enable compression
 1265      by calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by
 1266      using the SSL_CONF library to configure compression.
 1267      [Emilia Käsper]
 1268 
 1269   *) The signature of the session callback configured with
 1270      SSL_CTX_sess_set_get_cb was changed. The read-only input buffer
 1271      was explicitly marked as 'const unsigned char*' instead of
 1272      'unsigned char*'.
 1273      [Emilia Käsper]
 1274 
 1275   *) Always DPURIFY. Remove the use of uninitialized memory in the
 1276      RNG, and other conditional uses of DPURIFY. This makes -DPURIFY a no-op.
 1277      [Emilia Käsper]
 1278 
 1279   *) Removed many obsolete configuration items, including
 1280         DES_PTR, DES_RISC1, DES_RISC2, DES_INT
 1281         MD2_CHAR, MD2_INT, MD2_LONG
 1282         BF_PTR, BF_PTR2
 1283         IDEA_SHORT, IDEA_LONG
 1284         RC2_SHORT, RC2_LONG, RC4_LONG, RC4_CHUNK, RC4_INDEX
 1285      [Rich Salz, with advice from Andy Polyakov]
 1286 
 1287   *) Many BN internals have been moved to an internal header file.
 1288      [Rich Salz with help from Andy Polyakov]
 1289 
 1290   *) Configuration and writing out the results from it has changed.
 1291      Files such as Makefile include/openssl/opensslconf.h and are now
 1292      produced through general templates, such as Makefile.in and
 1293      crypto/opensslconf.h.in and some help from the perl module
 1294      Text::Template.
 1295 
 1296      Also, the center of configuration information is no longer
 1297      Makefile.  Instead, Configure produces a perl module in
 1298      configdata.pm which holds most of the config data (in the hash
 1299      table %config), the target data that comes from the target
 1300      configuration in one of the Configurations/*.conf files (in
 1301      %target).
 1302      [Richard Levitte]
 1303 
 1304   *) To clarify their intended purposes, the Configure options
 1305      --prefix and --openssldir change their semantics, and become more
 1306      straightforward and less interdependent.
 1307 
 1308      --prefix shall be used exclusively to give the location INSTALLTOP
 1309      where programs, scripts, libraries, include files and manuals are
 1310      going to be installed.  The default is now /usr/local.
 1311 
 1312      --openssldir shall be used exclusively to give the default
 1313      location OPENSSLDIR where certificates, private keys, CRLs are
 1314      managed.  This is also where the default openssl.cnf gets
 1315      installed.
 1316      If the directory given with this option is a relative path, the
 1317      values of both the --prefix value and the --openssldir value will
 1318      be combined to become OPENSSLDIR.
 1319      The default for --openssldir is INSTALLTOP/ssl.
 1320 
 1321      Anyone who uses --openssldir to specify where OpenSSL is to be
 1322      installed MUST change to use --prefix instead.
 1323      [Richard Levitte]
 1324 
 1325   *) The GOST engine was out of date and therefore it has been removed. An up
 1326      to date GOST engine is now being maintained in an external repository.
 1327      See: https://wiki.openssl.org/index.php/Binaries. Libssl still retains
 1328      support for GOST ciphersuites (these are only activated if a GOST engine
 1329      is present).
 1330      [Matt Caswell]
 1331 
 1332   *) EGD is no longer supported by default; use enable-egd when
 1333      configuring.
 1334      [Ben Kaduk and Rich Salz]
 1335 
 1336   *) The distribution now has Makefile.in files, which are used to
 1337      create Makefile's when Configure is run.  *Configure must be run
 1338      before trying to build now.*
 1339      [Rich Salz]
 1340 
 1341   *) The return value for SSL_CIPHER_description() for error conditions
 1342      has changed.
 1343      [Rich Salz]
 1344 
 1345   *) Support for RFC6698/RFC7671 DANE TLSA peer authentication.
 1346 
 1347      Obtaining and performing DNSSEC validation of TLSA records is
 1348      the application's responsibility.  The application provides
 1349      the TLSA records of its choice to OpenSSL, and these are then
 1350      used to authenticate the peer.
 1351 
 1352      The TLSA records need not even come from DNS.  They can, for
 1353      example, be used to implement local end-entity certificate or
 1354      trust-anchor "pinning", where the "pin" data takes the form
 1355      of TLSA records, which can augment or replace verification
 1356      based on the usual WebPKI public certification authorities.
 1357      [Viktor Dukhovni]
 1358 
 1359   *) Revert default OPENSSL_NO_DEPRECATED setting.  Instead OpenSSL
 1360      continues to support deprecated interfaces in default builds.
 1361      However, applications are strongly advised to compile their
 1362      source files with -DOPENSSL_API_COMPAT=0x10100000L, which hides
 1363      the declarations of all interfaces deprecated in 0.9.8, 1.0.0
 1364      or the 1.1.0 releases.
 1365 
 1366      In environments in which all applications have been ported to
 1367      not use any deprecated interfaces OpenSSL's Configure script
 1368      should be used with the --api=1.1.0 option to entirely remove
 1369      support for the deprecated features from the library and
 1370      unconditionally disable them in the installed headers.
 1371      Essentially the same effect can be achieved with the "no-deprecated"
 1372      argument to Configure, except that this will always restrict
 1373      the build to just the latest API, rather than a fixed API
 1374      version.
 1375 
 1376      As applications are ported to future revisions of the API,
 1377      they should update their compile-time OPENSSL_API_COMPAT define
 1378      accordingly, but in most cases should be able to continue to
 1379      compile with later releases.
 1380 
 1381      The OPENSSL_API_COMPAT versions for 1.0.0, and 0.9.8 are
 1382      0x10000000L and 0x00908000L, respectively.  However those
 1383      versions did not support the OPENSSL_API_COMPAT feature, and
 1384      so applications are not typically tested for explicit support
 1385      of just the undeprecated features of either release.
 1386      [Viktor Dukhovni]
 1387 
 1388   *) Add support for setting the minimum and maximum supported protocol.
 1389      It can bet set via the SSL_set_min_proto_version() and
 1390      SSL_set_max_proto_version(), or via the SSL_CONF's MinProtocol and
 1391      MaxProtocol.  It's recommended to use the new APIs to disable
 1392      protocols instead of disabling individual protocols using
 1393      SSL_set_options() or SSL_CONF's Protocol.  This change also
 1394      removes support for disabling TLS 1.2 in the OpenSSL TLS
 1395      client at compile time by defining OPENSSL_NO_TLS1_2_CLIENT.
 1396      [Kurt Roeckx]
 1397 
 1398   *) Support for ChaCha20 and Poly1305 added to libcrypto and libssl.
 1399      [Andy Polyakov]
 1400 
 1401   *) New EC_KEY_METHOD, this replaces the older ECDSA_METHOD and ECDH_METHOD
 1402      and integrates ECDSA and ECDH functionality into EC. Implementations can
 1403      now redirect key generation and no longer need to convert to or from
 1404      ECDSA_SIG format.
 1405 
 1406      Note: the ecdsa.h and ecdh.h headers are now no longer needed and just
 1407      include the ec.h header file instead.
 1408      [Steve Henson]
 1409 
 1410   *) Remove support for all 40 and 56 bit ciphers.  This includes all the export
 1411      ciphers who are no longer supported and drops support the ephemeral RSA key
 1412      exchange. The LOW ciphers currently doesn't have any ciphers in it.
 1413      [Kurt Roeckx]
 1414 
 1415   *) Made EVP_MD_CTX, EVP_MD, EVP_CIPHER_CTX, EVP_CIPHER and HMAC_CTX
 1416      opaque.  For HMAC_CTX, the following constructors and destructors
 1417      were added:
 1418 
 1419         HMAC_CTX *HMAC_CTX_new(void);
 1420         void HMAC_CTX_free(HMAC_CTX *ctx);
 1421 
 1422      For EVP_MD and EVP_CIPHER, complete APIs to create, fill and
 1423      destroy such methods has been added.  See EVP_MD_meth_new(3) and
 1424      EVP_CIPHER_meth_new(3) for documentation.
 1425 
 1426      Additional changes:
 1427      1) EVP_MD_CTX_cleanup(), EVP_CIPHER_CTX_cleanup() and
 1428         HMAC_CTX_cleanup() were removed.  HMAC_CTX_reset() and
 1429         EVP_MD_CTX_reset() should be called instead to reinitialise
 1430         an already created structure.
 1431      2) For consistency with the majority of our object creators and
 1432         destructors, EVP_MD_CTX_(create|destroy) were renamed to
 1433         EVP_MD_CTX_(new|free).  The old names are retained as macros
 1434         for deprecated builds.
 1435      [Richard Levitte]
 1436 
 1437   *) Added ASYNC support. Libcrypto now includes the async sub-library to enable
 1438      cryptographic operations to be performed asynchronously as long as an
 1439      asynchronous capable engine is used. See the ASYNC_start_job() man page for
 1440      further details. Libssl has also had this capability integrated with the
 1441      introduction of the new mode SSL_MODE_ASYNC and associated error
 1442      SSL_ERROR_WANT_ASYNC. See the SSL_CTX_set_mode() and SSL_get_error() man
 1443      pages. This work was developed in partnership with Intel Corp.
 1444      [Matt Caswell]
 1445 
 1446   *) SSL_{CTX_}set_ecdh_auto() has been removed and ECDH is support is
 1447      always enabled now.  If you want to disable the support you should
 1448      exclude it using the list of supported ciphers. This also means that the
 1449      "-no_ecdhe" option has been removed from s_server.
 1450      [Kurt Roeckx]
 1451 
 1452   *) SSL_{CTX}_set_tmp_ecdh() which can set 1 EC curve now internally calls
 1453      SSL_{CTX_}set1_curves() which can set a list.
 1454      [Kurt Roeckx]
 1455 
 1456   *) Remove support for SSL_{CTX_}set_tmp_ecdh_callback().  You should set the
 1457      curve you want to support using SSL_{CTX_}set1_curves().
 1458      [Kurt Roeckx]
 1459 
 1460   *) State machine rewrite. The state machine code has been significantly
 1461      refactored in order to remove much duplication of code and solve issues
 1462      with the old code (see ssl/statem/README for further details). This change
 1463      does have some associated API changes. Notably the SSL_state() function
 1464      has been removed and replaced by SSL_get_state which now returns an
 1465      "OSSL_HANDSHAKE_STATE" instead of an int. SSL_set_state() has been removed
 1466      altogether. The previous handshake states defined in ssl.h and ssl3.h have
 1467      also been removed.
 1468      [Matt Caswell]
 1469 
 1470   *) All instances of the string "ssleay" in the public API were replaced
 1471      with OpenSSL (case-matching; e.g., OPENSSL_VERSION for #define's)
 1472      Some error codes related to internal RSA_eay API's were renamed.
 1473      [Rich Salz]
 1474 
 1475   *) The demo files in crypto/threads were moved to demo/threads.
 1476      [Rich Salz]
 1477 
 1478   *) Removed obsolete engines: 4758cca, aep, atalla, cswift, nuron, gmp,
 1479      sureware and ubsec.
 1480      [Matt Caswell, Rich Salz]
 1481 
 1482   *) New ASN.1 embed macro.
 1483 
 1484      New ASN.1 macro ASN1_EMBED. This is the same as ASN1_SIMPLE except the
 1485      structure is not allocated: it is part of the parent. That is instead of
 1486 
 1487      FOO *x;
 1488 
 1489      it must be:
 1490 
 1491      FOO x;
 1492 
 1493      This reduces memory fragmentation and make it impossible to accidentally
 1494      set a mandatory field to NULL.
 1495 
 1496      This currently only works for some fields specifically a SEQUENCE, CHOICE,
 1497      or ASN1_STRING type which is part of a parent SEQUENCE. Since it is
 1498      equivalent to ASN1_SIMPLE it cannot be tagged, OPTIONAL, SET OF or
 1499      SEQUENCE OF.
 1500      [Steve Henson]
 1501 
 1502   *) Remove EVP_CHECK_DES_KEY, a compile-time option that never compiled.
 1503      [Emilia Käsper]
 1504 
 1505   *) Removed DES and RC4 ciphersuites from DEFAULT. Also removed RC2 although
 1506      in 1.0.2 EXPORT was already removed and the only RC2 ciphersuite is also
 1507      an EXPORT one. COMPLEMENTOFDEFAULT has been updated accordingly to add
 1508      DES and RC4 ciphersuites.
 1509      [Matt Caswell]
 1510 
 1511   *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
 1512      This changes the decoding behaviour for some invalid messages,
 1513      though the change is mostly in the more lenient direction, and
 1514      legacy behaviour is preserved as much as possible.
 1515      [Emilia Käsper]
 1516 
 1517   *) Fix no-stdio build.
 1518     [ David Woodhouse <David.Woodhouse@intel.com> and also
 1519       Ivan Nestlerode <ivan.nestlerode@sonos.com> ]
 1520 
 1521   *) New testing framework
 1522      The testing framework has been largely rewritten and is now using
 1523      perl and the perl modules Test::Harness and an extended variant of
 1524      Test::More called OpenSSL::Test to do its work.  All test scripts in
 1525      test/ have been rewritten into test recipes, and all direct calls to
 1526      executables in test/Makefile have become individual recipes using the
 1527      simplified testing OpenSSL::Test::Simple.
 1528 
 1529      For documentation on our testing modules, do:
 1530 
 1531         perldoc test/testlib/OpenSSL/Test/Simple.pm
 1532         perldoc test/testlib/OpenSSL/Test.pm
 1533 
 1534      [Richard Levitte]
 1535 
 1536   *) Revamped memory debug; only -DCRYPTO_MDEBUG and -DCRYPTO_MDEBUG_ABORT
 1537      are used; the latter aborts on memory leaks (usually checked on exit).
 1538      Some undocumented "set malloc, etc., hooks" functions were removed
 1539      and others were changed.  All are now documented.
 1540      [Rich Salz]
 1541 
 1542   *) In DSA_generate_parameters_ex, if the provided seed is too short,
 1543      return an error
 1544      [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
 1545 
 1546   *) Rewrite PSK to support ECDHE_PSK, DHE_PSK and RSA_PSK. Add ciphersuites
 1547      from RFC4279, RFC4785, RFC5487, RFC5489.
 1548 
 1549      Thanks to Christian J. Dietrich and Giuseppe D'Angelo for the
 1550      original RSA_PSK patch.
 1551      [Steve Henson]
 1552 
 1553   *) Dropped support for the SSL3_FLAGS_DELAY_CLIENT_FINISHED flag. This SSLeay
 1554      era flag was never set throughout the codebase (only read). Also removed
 1555      SSL3_FLAGS_POP_BUFFER which was only used if
 1556      SSL3_FLAGS_DELAY_CLIENT_FINISHED was also set.
 1557      [Matt Caswell]
 1558 
 1559   *) Changed the default name options in the "ca", "crl", "req" and "x509"
 1560      to be "oneline" instead of "compat".
 1561      [Richard Levitte]
 1562 
 1563   *) Remove SSL_OP_TLS_BLOCK_PADDING_BUG. This is SSLeay legacy, we're
 1564      not aware of clients that still exhibit this bug, and the workaround
 1565      hasn't been working properly for a while.
 1566      [Emilia Käsper]
 1567 
 1568   *) The return type of BIO_number_read() and BIO_number_written() as well as
 1569      the corresponding num_read and num_write members in the BIO structure has
 1570      changed from unsigned long to uint64_t. On platforms where an unsigned
 1571      long is 32 bits (e.g. Windows) these counters could overflow if >4Gb is
 1572      transferred.
 1573      [Matt Caswell]
 1574 
 1575   *) Given the pervasive nature of TLS extensions it is inadvisable to run
 1576      OpenSSL without support for them. It also means that maintaining
 1577      the OPENSSL_NO_TLSEXT option within the code is very invasive (and probably
 1578      not well tested). Therefore the OPENSSL_NO_TLSEXT option has been removed.
 1579      [Matt Caswell]
 1580 
 1581   *) Removed support for the two export grade static DH ciphersuites
 1582      EXP-DH-RSA-DES-CBC-SHA and EXP-DH-DSS-DES-CBC-SHA. These two ciphersuites
 1583      were newly added (along with a number of other static DH ciphersuites) to
 1584      1.0.2. However the two export ones have *never* worked since they were
 1585      introduced. It seems strange in any case to be adding new export
 1586      ciphersuites, and given "logjam" it also does not seem correct to fix them.
 1587      [Matt Caswell]
 1588 
 1589   *) Version negotiation has been rewritten. In particular SSLv23_method(),
 1590      SSLv23_client_method() and SSLv23_server_method() have been deprecated,
 1591      and turned into macros which simply call the new preferred function names
 1592      TLS_method(), TLS_client_method() and TLS_server_method(). All new code
 1593      should use the new names instead. Also as part of this change the ssl23.h
 1594      header file has been removed.
 1595      [Matt Caswell]
 1596 
 1597   *) Support for Kerberos ciphersuites in TLS (RFC2712) has been removed. This
 1598      code and the associated standard is no longer considered fit-for-purpose.
 1599      [Matt Caswell]
 1600 
 1601   *) RT2547 was closed.  When generating a private key, try to make the
 1602      output file readable only by the owner.  This behavior change might
 1603      be noticeable when interacting with other software.
 1604 
 1605   *) Documented all exdata functions.  Added CRYPTO_free_ex_index.
 1606      Added a test.
 1607      [Rich Salz]
 1608 
 1609   *) Added HTTP GET support to the ocsp command.
 1610      [Rich Salz]
 1611 
 1612   *) Changed default digest for the dgst and enc commands from MD5 to
 1613      sha256
 1614      [Rich Salz]
 1615 
 1616   *) RAND_pseudo_bytes has been deprecated. Users should use RAND_bytes instead.
 1617      [Matt Caswell]
 1618 
 1619   *) Added support for TLS extended master secret from
 1620      draft-ietf-tls-session-hash-03.txt. Thanks for Alfredo Pironti for an
 1621      initial patch which was a great help during development.
 1622      [Steve Henson]
 1623 
 1624   *) All libssl internal structures have been removed from the public header
 1625      files, and the OPENSSL_NO_SSL_INTERN option has been removed (since it is
 1626      now redundant). Users should not attempt to access internal structures
 1627      directly. Instead they should use the provided API functions.
 1628      [Matt Caswell]
 1629 
 1630   *) config has been changed so that by default OPENSSL_NO_DEPRECATED is used.
 1631      Access to deprecated functions can be re-enabled by running config with
 1632      "enable-deprecated". In addition applications wishing to use deprecated
 1633      functions must define OPENSSL_USE_DEPRECATED. Note that this new behaviour
 1634      will, by default, disable some transitive includes that previously existed
 1635      in the header files (e.g. ec.h will no longer, by default, include bn.h)
 1636      [Matt Caswell]
 1637 
 1638   *) Added support for OCB mode. OpenSSL has been granted a patent license
 1639      compatible with the OpenSSL license for use of OCB. Details are available
 1640      at https://www.openssl.org/source/OCB-patent-grant-OpenSSL.pdf. Support
 1641      for OCB can be removed by calling config with no-ocb.
 1642      [Matt Caswell]
 1643 
 1644   *) SSLv2 support has been removed.  It still supports receiving a SSLv2
 1645      compatible client hello.
 1646      [Kurt Roeckx]
 1647 
 1648   *) Increased the minimal RSA keysize from 256 to 512 bits [Rich Salz],
 1649      done while fixing the error code for the key-too-small case.
 1650      [Annie Yousar <a.yousar@informatik.hu-berlin.de>]
 1651 
 1652   *) CA.sh has been removed; use CA.pl instead.
 1653      [Rich Salz]
 1654 
 1655   *) Removed old DES API.
 1656      [Rich Salz]
 1657 
 1658   *) Remove various unsupported platforms:
 1659         Sony NEWS4
 1660         BEOS and BEOS_R5
 1661         NeXT
 1662         SUNOS
 1663         MPE/iX
 1664         Sinix/ReliantUNIX RM400
 1665         DGUX
 1666         NCR
 1667         Tandem
 1668         Cray
 1669         16-bit platforms such as WIN16
 1670      [Rich Salz]
 1671 
 1672   *) Clean up OPENSSL_NO_xxx #define's
 1673         Use setbuf() and remove OPENSSL_NO_SETVBUF_IONBF
 1674         Rename OPENSSL_SYSNAME_xxx to OPENSSL_SYS_xxx
 1675         OPENSSL_NO_EC{DH,DSA} merged into OPENSSL_NO_EC
 1676         OPENSSL_NO_RIPEMD160, OPENSSL_NO_RIPEMD merged into OPENSSL_NO_RMD160
 1677         OPENSSL_NO_FP_API merged into OPENSSL_NO_STDIO
 1678         Remove OPENSSL_NO_BIO OPENSSL_NO_BUFFER OPENSSL_NO_CHAIN_VERIFY
 1679         OPENSSL_NO_EVP OPENSSL_NO_FIPS_ERR OPENSSL_NO_HASH_COMP
 1680         OPENSSL_NO_LHASH OPENSSL_NO_OBJECT OPENSSL_NO_SPEED OPENSSL_NO_STACK
 1681         OPENSSL_NO_X509 OPENSSL_NO_X509_VERIFY
 1682         Remove MS_STATIC; it's a relic from platforms <32 bits.
 1683      [Rich Salz]
 1684 
 1685   *) Cleaned up dead code
 1686         Remove all but one '#ifdef undef' which is to be looked at.
 1687      [Rich Salz]
 1688 
 1689   *) Clean up calling of xxx_free routines.
 1690         Just like free(), fix most of the xxx_free routines to accept
 1691         NULL.  Remove the non-null checks from callers.  Save much code.
 1692      [Rich Salz]
 1693 
 1694   *) Add secure heap for storage of private keys (when possible).
 1695      Add BIO_s_secmem(), CBIGNUM, etc.
 1696      Contributed by Akamai Technologies under our Corporate CLA.
 1697      [Rich Salz]
 1698 
 1699   *) Experimental support for a new, fast, unbiased prime candidate generator,
 1700      bn_probable_prime_dh_coprime(). Not currently used by any prime generator.
 1701      [Felix Laurie von Massenbach <felix@erbridge.co.uk>]
 1702 
 1703   *) New output format NSS in the sess_id command line tool. This allows
 1704      exporting the session id and the master key in NSS keylog format.
 1705      [Martin Kaiser <martin@kaiser.cx>]
 1706 
 1707   *) Harmonize version and its documentation. -f flag is used to display
 1708      compilation flags.
 1709      [mancha <mancha1@zoho.com>]
 1710 
 1711   *) Fix eckey_priv_encode so it immediately returns an error upon a failure
 1712      in i2d_ECPrivateKey.  Thanks to Ted Unangst for feedback on this issue.
 1713      [mancha <mancha1@zoho.com>]
 1714 
 1715   *) Fix some double frees. These are not thought to be exploitable.
 1716      [mancha <mancha1@zoho.com>]
 1717 
 1718   *) A missing bounds check in the handling of the TLS heartbeat extension
 1719      can be used to reveal up to 64k of memory to a connected client or
 1720      server.
 1721 
 1722      Thanks for Neel Mehta of Google Security for discovering this bug and to
 1723      Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
 1724      preparing the fix (CVE-2014-0160)
 1725      [Adam Langley, Bodo Moeller]
 1726 
 1727   *) Fix for the attack described in the paper "Recovering OpenSSL
 1728      ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
 1729      by Yuval Yarom and Naomi Benger. Details can be obtained from:
 1730      http://eprint.iacr.org/2014/140
 1731 
 1732      Thanks to Yuval Yarom and Naomi Benger for discovering this
 1733      flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
 1734      [Yuval Yarom and Naomi Benger]
 1735 
 1736   *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
 1737      this fixes a limitation in previous versions of OpenSSL.
 1738      [Steve Henson]
 1739 
 1740   *) Experimental encrypt-then-mac support.
 1741 
 1742      Experimental support for encrypt then mac from
 1743      draft-gutmann-tls-encrypt-then-mac-02.txt
 1744 
 1745      To enable it set the appropriate extension number (0x42 for the test
 1746      server) using e.g. -DTLSEXT_TYPE_encrypt_then_mac=0x42
 1747 
 1748      For non-compliant peers (i.e. just about everything) this should have no
 1749      effect.
 1750 
 1751      WARNING: EXPERIMENTAL, SUBJECT TO CHANGE.
 1752 
 1753      [Steve Henson]
 1754 
 1755   *) Add EVP support for key wrapping algorithms, to avoid problems with
 1756      existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
 1757      the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
 1758      algorithms and include tests cases.
 1759      [Steve Henson]
 1760 
 1761   *) Extend CMS code to support RSA-PSS signatures and RSA-OAEP for
 1762      enveloped data.
 1763      [Steve Henson]
 1764 
 1765   *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
 1766      MGF1 digest and OAEP label.
 1767      [Steve Henson]
 1768 
 1769   *) Make openssl verify return errors.
 1770      [Chris Palmer <palmer@google.com> and Ben Laurie]
 1771 
 1772   *) New function ASN1_TIME_diff to calculate the difference between two
 1773      ASN1_TIME structures or one structure and the current time.
 1774      [Steve Henson]
 1775 
 1776   *) Update fips_test_suite to support multiple command line options. New
 1777      test to induce all self test errors in sequence and check expected
 1778      failures.
 1779      [Steve Henson]
 1780 
 1781   *) Add FIPS_{rsa,dsa,ecdsa}_{sign,verify} functions which digest and
 1782      sign or verify all in one operation.
 1783      [Steve Henson]
 1784 
 1785   *) Add fips_algvs: a multicall fips utility incorporating all the algorithm
 1786      test programs and fips_test_suite. Includes functionality to parse
 1787      the minimal script output of fipsalgest.pl directly.
 1788      [Steve Henson]
 1789 
 1790   *) Add authorisation parameter to FIPS_module_mode_set().
 1791      [Steve Henson]
 1792 
 1793   *) Add FIPS selftest for ECDH algorithm using P-224 and B-233 curves.
 1794      [Steve Henson]
 1795 
 1796   *) Use separate DRBG fields for internal and external flags. New function
 1797      FIPS_drbg_health_check() to perform on demand health checking. Add
 1798      generation tests to fips_test_suite with reduced health check interval to
 1799      demonstrate periodic health checking. Add "nodh" option to
 1800      fips_test_suite to skip very slow DH test.
 1801      [Steve Henson]
 1802 
 1803   *) New function FIPS_get_cipherbynid() to lookup FIPS supported ciphers
 1804      based on NID.
 1805      [Steve Henson]
 1806 
 1807   *) More extensive health check for DRBG checking many more failure modes.
 1808      New function FIPS_selftest_drbg_all() to handle every possible DRBG
 1809      combination: call this in fips_test_suite.
 1810      [Steve Henson]
 1811 
 1812   *) Add support for canonical generation of DSA parameter 'g'. See
 1813      FIPS 186-3 A.2.3.
 1814 
 1815   *) Add support for HMAC DRBG from SP800-90. Update DRBG algorithm test and
 1816      POST to handle HMAC cases.
 1817      [Steve Henson]
 1818 
 1819   *) Add functions FIPS_module_version() and FIPS_module_version_text()
 1820      to return numerical and string versions of the FIPS module number.
 1821      [Steve Henson]
 1822 
 1823   *) Rename FIPS_mode_set and FIPS_mode to FIPS_module_mode_set and
 1824      FIPS_module_mode. FIPS_mode and FIPS_mode_set will be implemented
 1825      outside the validated module in the FIPS capable OpenSSL.
 1826      [Steve Henson]
 1827 
 1828   *) Minor change to DRBG entropy callback semantics. In some cases
 1829      there is no multiple of the block length between min_len and
 1830      max_len. Allow the callback to return more than max_len bytes
 1831      of entropy but discard any extra: it is the callback's responsibility
 1832      to ensure that the extra data discarded does not impact the
 1833      requested amount of entropy.
 1834      [Steve Henson]
 1835 
 1836   *) Add PRNG security strength checks to RSA, DSA and ECDSA using
 1837      information in FIPS186-3, SP800-57 and SP800-131A.
 1838      [Steve Henson]
 1839 
 1840   *) CCM support via EVP. Interface is very similar to GCM case except we
 1841      must supply all data in one chunk (i.e. no update, final) and the
 1842      message length must be supplied if AAD is used. Add algorithm test
 1843      support.
 1844      [Steve Henson]
 1845 
 1846   *) Initial version of POST overhaul. Add POST callback to allow the status
 1847      of POST to be monitored and/or failures induced. Modify fips_test_suite
 1848      to use callback. Always run all selftests even if one fails.
 1849      [Steve Henson]
 1850 
 1851   *) XTS support including algorithm test driver in the fips_gcmtest program.
 1852      Note: this does increase the maximum key length from 32 to 64 bytes but
 1853      there should be no binary compatibility issues as existing applications
 1854      will never use XTS mode.
 1855      [Steve Henson]
 1856 
 1857   *) Extensive reorganisation of FIPS PRNG behaviour. Remove all dependencies
 1858      to OpenSSL RAND code and replace with a tiny FIPS RAND API which also
 1859      performs algorithm blocking for unapproved PRNG types. Also do not
 1860      set PRNG type in FIPS_mode_set(): leave this to the application.
 1861      Add default OpenSSL DRBG handling: sets up FIPS PRNG and seeds with
 1862      the standard OpenSSL PRNG: set additional data to a date time vector.
 1863      [Steve Henson]
 1864 
 1865   *) Rename old X9.31 PRNG functions of the form FIPS_rand* to FIPS_x931*.
 1866      This shouldn't present any incompatibility problems because applications
 1867      shouldn't be using these directly and any that are will need to rethink
 1868      anyway as the X9.31 PRNG is now deprecated by FIPS 140-2
 1869      [Steve Henson]
 1870 
 1871   *) Extensive self tests and health checking required by SP800-90 DRBG.
 1872      Remove strength parameter from FIPS_drbg_instantiate and always
 1873      instantiate at maximum supported strength.
 1874      [Steve Henson]
 1875 
 1876   *) Add ECDH code to fips module and fips_ecdhvs for primitives only testing.
 1877      [Steve Henson]
 1878 
 1879   *) New algorithm test program fips_dhvs to handle DH primitives only testing.
 1880      [Steve Henson]
 1881 
 1882   *) New function DH_compute_key_padded() to compute a DH key and pad with
 1883      leading zeroes if needed: this complies with SP800-56A et al.
 1884      [Steve Henson]
 1885 
 1886   *) Initial implementation of SP800-90 DRBGs for Hash and CTR. Not used by
 1887      anything, incomplete, subject to change and largely untested at present.
 1888      [Steve Henson]
 1889 
 1890   *) Modify fipscanisteronly build option to only build the necessary object
 1891      files by filtering FIPS_EX_OBJ through a perl script in crypto/Makefile.
 1892      [Steve Henson]
 1893 
 1894   *) Add experimental option FIPSSYMS to give all symbols in
 1895      fipscanister.o and FIPS or fips prefix. This will avoid
 1896      conflicts with future versions of OpenSSL. Add perl script
 1897      util/fipsas.pl to preprocess assembly language source files
 1898      and rename any affected symbols.
 1899      [Steve Henson]
 1900 
 1901   *) Add selftest checks and algorithm block of non-fips algorithms in
 1902      FIPS mode. Remove DES2 from selftests.
 1903      [Steve Henson]
 1904 
 1905   *) Add ECDSA code to fips module. Add tiny fips_ecdsa_check to just
 1906      return internal method without any ENGINE dependencies. Add new
 1907      tiny fips sign and verify functions.
 1908      [Steve Henson]
 1909 
 1910   *) New build option no-ec2m to disable characteristic 2 code.
 1911      [Steve Henson]
 1912 
 1913   *) New build option "fipscanisteronly". This only builds fipscanister.o
 1914      and (currently) associated fips utilities. Uses the file Makefile.fips
 1915      instead of Makefile.org as the prototype.
 1916      [Steve Henson]
 1917 
 1918   *) Add some FIPS mode restrictions to GCM. Add internal IV generator.
 1919      Update fips_gcmtest to use IV generator.
 1920      [Steve Henson]
 1921 
 1922   *) Initial, experimental EVP support for AES-GCM. AAD can be input by
 1923      setting output buffer to NULL. The *Final function must be
 1924      called although it will not retrieve any additional data. The tag
 1925      can be set or retrieved with a ctrl. The IV length is by default 12
 1926      bytes (96 bits) but can be set to an alternative value. If the IV
 1927      length exceeds the maximum IV length (currently 16 bytes) it cannot be
 1928      set before the key.
 1929      [Steve Henson]
 1930 
 1931   *) New flag in ciphers: EVP_CIPH_FLAG_CUSTOM_CIPHER. This means the
 1932      underlying do_cipher function handles all cipher semantics itself
 1933      including padding and finalisation. This is useful if (for example)
 1934      an ENGINE cipher handles block padding itself. The behaviour of
 1935      do_cipher is subtly changed if this flag is set: the return value
 1936      is the number of characters written to the output buffer (zero is
 1937      no longer an error code) or a negative error code. Also if the
 1938      input buffer is NULL and length 0 finalisation should be performed.
 1939      [Steve Henson]
 1940 
 1941   *) If a candidate issuer certificate is already part of the constructed
 1942      path ignore it: new debug notification X509_V_ERR_PATH_LOOP for this case.
 1943      [Steve Henson]
 1944 
 1945   *) Improve forward-security support: add functions
 1946 
 1947        void SSL_CTX_set_not_resumable_session_callback(SSL_CTX *ctx, int (*cb)(SSL *ssl, int is_forward_secure))
 1948        void SSL_set_not_resumable_session_callback(SSL *ssl, int (*cb)(SSL *ssl, int is_forward_secure))
 1949 
 1950      for use by SSL/TLS servers; the callback function will be called whenever a
 1951      new session is created, and gets to decide whether the session may be
 1952      cached to make it resumable (return 0) or not (return 1).  (As by the
 1953      SSL/TLS protocol specifications, the session_id sent by the server will be
 1954      empty to indicate that the session is not resumable; also, the server will
 1955      not generate RFC 4507 (RFC 5077) session tickets.)
 1956 
 1957      A simple reasonable callback implementation is to return is_forward_secure.
 1958      This parameter will be set to 1 or 0 depending on the ciphersuite selected
 1959      by the SSL/TLS server library, indicating whether it can provide forward
 1960      security.
 1961      [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)]
 1962 
 1963   *) New -verify_name option in command line utilities to set verification
 1964      parameters by name.
 1965      [Steve Henson]
 1966 
 1967   *) Initial CMAC implementation. WARNING: EXPERIMENTAL, API MAY CHANGE.
 1968      Add CMAC pkey methods.
 1969      [Steve Henson]
 1970 
 1971   *) Experimental renegotiation in s_server -www mode. If the client
 1972      browses /reneg connection is renegotiated. If /renegcert it is
 1973      renegotiated requesting a certificate.
 1974      [Steve Henson]
 1975 
 1976   *) Add an "external" session cache for debugging purposes to s_server. This
 1977      should help trace issues which normally are only apparent in deployed
 1978      multi-process servers.
 1979      [Steve Henson]
 1980 
 1981   *) Extensive audit of libcrypto with DEBUG_UNUSED. Fix many cases where
 1982      return value is ignored. NB. The functions RAND_add(), RAND_seed(),
 1983      BIO_set_cipher() and some obscure PEM functions were changed so they
 1984      can now return an error. The RAND changes required a change to the
 1985      RAND_METHOD structure.
 1986      [Steve Henson]
 1987 
 1988   *) New macro __owur for "OpenSSL Warn Unused Result". This makes use of
 1989      a gcc attribute to warn if the result of a function is ignored. This
 1990      is enable if DEBUG_UNUSED is set. Add to several functions in evp.h
 1991      whose return value is often ignored.
 1992      [Steve Henson]
 1993 
 1994   *) New -noct, -requestct, -requirect and -ctlogfile options for s_client.
 1995      These allow SCTs (signed certificate timestamps) to be requested and
 1996      validated when establishing a connection.
 1997      [Rob Percival <robpercival@google.com>]
 1998 
 1999  Changes between 1.0.2g and 1.0.2h [3 May 2016]
 2000 
 2001   *) Prevent padding oracle in AES-NI CBC MAC check
 2002 
 2003      A MITM attacker can use a padding oracle attack to decrypt traffic
 2004      when the connection uses an AES CBC cipher and the server support
 2005      AES-NI.
 2006 
 2007      This issue was introduced as part of the fix for Lucky 13 padding
 2008      attack (CVE-2013-0169). The padding check was rewritten to be in
 2009      constant time by making sure that always the same bytes are read and
 2010      compared against either the MAC or padding bytes. But it no longer
 2011      checked that there was enough data to have both the MAC and padding
 2012      bytes.
 2013 
 2014      This issue was reported by Juraj Somorovsky using TLS-Attacker.
 2015      (CVE-2016-2107)
 2016      [Kurt Roeckx]
 2017 
 2018   *) Fix EVP_EncodeUpdate overflow
 2019 
 2020      An overflow can occur in the EVP_EncodeUpdate() function which is used for
 2021      Base64 encoding of binary data. If an attacker is able to supply very large
 2022      amounts of input data then a length check can overflow resulting in a heap
 2023      corruption.
 2024 
 2025      Internally to OpenSSL the EVP_EncodeUpdate() function is primarily used by
 2026      the PEM_write_bio* family of functions. These are mainly used within the
 2027      OpenSSL command line applications, so any application which processes data
 2028      from an untrusted source and outputs it as a PEM file should be considered
 2029      vulnerable to this issue. User applications that call these APIs directly
 2030      with large amounts of untrusted data may also be vulnerable.
 2031 
 2032      This issue was reported by Guido Vranken.
 2033      (CVE-2016-2105)
 2034      [Matt Caswell]
 2035 
 2036   *) Fix EVP_EncryptUpdate overflow
 2037 
 2038      An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
 2039      is able to supply very large amounts of input data after a previous call to
 2040      EVP_EncryptUpdate() with a partial block then a length check can overflow
 2041      resulting in a heap corruption. Following an analysis of all OpenSSL
 2042      internal usage of the EVP_EncryptUpdate() function all usage is one of two
 2043      forms. The first form is where the EVP_EncryptUpdate() call is known to be
 2044      the first called function after an EVP_EncryptInit(), and therefore that
 2045      specific call must be safe. The second form is where the length passed to
 2046      EVP_EncryptUpdate() can be seen from the code to be some small value and
 2047      therefore there is no possibility of an overflow. Since all instances are
 2048      one of these two forms, it is believed that there can be no overflows in
 2049      internal code due to this problem. It should be noted that
 2050      EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
 2051      Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
 2052      of these calls have also been analysed too and it is believed there are no
 2053      instances in internal usage where an overflow could occur.
 2054 
 2055      This issue was reported by Guido Vranken.
 2056      (CVE-2016-2106)
 2057      [Matt Caswell]
 2058 
 2059   *) Prevent ASN.1 BIO excessive memory allocation
 2060 
 2061      When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
 2062      a short invalid encoding can cause allocation of large amounts of memory
 2063      potentially consuming excessive resources or exhausting memory.
 2064 
 2065      Any application parsing untrusted data through d2i BIO functions is
 2066      affected. The memory based functions such as d2i_X509() are *not* affected.
 2067      Since the memory based functions are used by the TLS library, TLS
 2068      applications are not affected.
 2069 
 2070      This issue was reported by Brian Carpenter.
 2071      (CVE-2016-2109)
 2072      [Stephen Henson]
 2073 
 2074   *) EBCDIC overread
 2075 
 2076      ASN1 Strings that are over 1024 bytes can cause an overread in applications
 2077      using the X509_NAME_oneline() function on EBCDIC systems. This could result
 2078      in arbitrary stack data being returned in the buffer.
 2079 
 2080      This issue was reported by Guido Vranken.
 2081      (CVE-2016-2176)
 2082      [Matt Caswell]
 2083 
 2084   *) Modify behavior of ALPN to invoke callback after SNI/servername
 2085      callback, such that updates to the SSL_CTX affect ALPN.
 2086      [Todd Short]
 2087 
 2088   *) Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
 2089      default.
 2090      [Kurt Roeckx]
 2091 
 2092   *) Only remove the SSLv2 methods with the no-ssl2-method option. When the
 2093      methods are enabled and ssl2 is disabled the methods return NULL.
 2094      [Kurt Roeckx]
 2095 
 2096  Changes between 1.0.2f and 1.0.2g [1 Mar 2016]
 2097 
 2098   * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
 2099     Builds that are not configured with "enable-weak-ssl-ciphers" will not
 2100     provide any "EXPORT" or "LOW" strength ciphers.
 2101     [Viktor Dukhovni]
 2102 
 2103   * Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
 2104     is by default disabled at build-time.  Builds that are not configured with
 2105     "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
 2106     users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
 2107     will need to explicitly call either of:
 2108 
 2109         SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
 2110     or
 2111         SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
 2112 
 2113     as appropriate.  Even if either of those is used, or the application
 2114     explicitly uses the version-specific SSLv2_method() or its client and
 2115     server variants, SSLv2 ciphers vulnerable to exhaustive search key
 2116     recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
 2117     ciphers, and SSLv2 56-bit DES are no longer available.
 2118     (CVE-2016-0800)
 2119     [Viktor Dukhovni]
 2120 
 2121   *) Fix a double-free in DSA code
 2122 
 2123      A double free bug was discovered when OpenSSL parses malformed DSA private
 2124      keys and could lead to a DoS attack or memory corruption for applications
 2125      that receive DSA private keys from untrusted sources.  This scenario is
 2126      considered rare.
 2127 
 2128      This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
 2129      libFuzzer.
 2130      (CVE-2016-0705)
 2131      [Stephen Henson]
 2132 
 2133   *) Disable SRP fake user seed to address a server memory leak.
 2134 
 2135      Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
 2136 
 2137      SRP_VBASE_get_by_user had inconsistent memory management behaviour.
 2138      In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
 2139      was changed to ignore the "fake user" SRP seed, even if the seed
 2140      is configured.
 2141 
 2142      Users should use SRP_VBASE_get1_by_user instead. Note that in
 2143      SRP_VBASE_get1_by_user, caller must free the returned value. Note
 2144      also that even though configuring the SRP seed attempts to hide
 2145      invalid usernames by continuing the handshake with fake
 2146      credentials, this behaviour is not constant time and no strong
 2147      guarantees are made that the handshake is indistinguishable from
 2148      that of a valid user.
 2149      (CVE-2016-0798)
 2150      [Emilia Käsper]
 2151 
 2152   *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
 2153 
 2154      In the BN_hex2bn function the number of hex digits is calculated using an
 2155      int value |i|. Later |bn_expand| is called with a value of |i * 4|. For
 2156      large values of |i| this can result in |bn_expand| not allocating any
 2157      memory because |i * 4| is negative. This can leave the internal BIGNUM data
 2158      field as NULL leading to a subsequent NULL ptr deref. For very large values
 2159      of |i|, the calculation |i * 4| could be a positive value smaller than |i|.
 2160      In this case memory is allocated to the internal BIGNUM data field, but it
 2161      is insufficiently sized leading to heap corruption. A similar issue exists
 2162      in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
 2163      is ever called by user applications with very large untrusted hex/dec data.
 2164      This is anticipated to be a rare occurrence.
 2165 
 2166      All OpenSSL internal usage of these functions use data that is not expected
 2167      to be untrusted, e.g. config file data or application command line
 2168      arguments. If user developed applications generate config file data based
 2169      on untrusted data then it is possible that this could also lead to security
 2170      consequences. This is also anticipated to be rare.
 2171 
 2172      This issue was reported to OpenSSL by Guido Vranken.
 2173      (CVE-2016-0797)
 2174      [Matt Caswell]
 2175 
 2176   *) Fix memory issues in BIO_*printf functions
 2177 
 2178      The internal |fmtstr| function used in processing a "%s" format string in
 2179      the BIO_*printf functions could overflow while calculating the length of a
 2180      string and cause an OOB read when printing very long strings.
 2181 
 2182      Additionally the internal |doapr_outch| function can attempt to write to an
 2183      OOB memory location (at an offset from the NULL pointer) in the event of a
 2184      memory allocation failure. In 1.0.2 and below this could be caused where
 2185      the size of a buffer to be allocated is greater than INT_MAX. E.g. this
 2186      could be in processing a very long "%s" format string. Memory leaks can
 2187      also occur.
 2188 
 2189      The first issue may mask the second issue dependent on compiler behaviour.
 2190      These problems could enable attacks where large amounts of untrusted data
 2191      is passed to the BIO_*printf functions. If applications use these functions
 2192      in this way then they could be vulnerable. OpenSSL itself uses these
 2193      functions when printing out human-readable dumps of ASN.1 data. Therefore
 2194      applications that print this data could be vulnerable if the data is from
 2195      untrusted sources. OpenSSL command line applications could also be
 2196      vulnerable where they print out ASN.1 data, or if untrusted data is passed
 2197      as command line arguments.
 2198 
 2199      Libssl is not considered directly vulnerable. Additionally certificates etc
 2200      received via remote connections via libssl are also unlikely to be able to
 2201      trigger these issues because of message size limits enforced within libssl.
 2202 
 2203      This issue was reported to OpenSSL Guido Vranken.
 2204      (CVE-2016-0799)
 2205      [Matt Caswell]
 2206 
 2207   *) Side channel attack on modular exponentiation
 2208 
 2209      A side-channel attack was found which makes use of cache-bank conflicts on
 2210      the Intel Sandy-Bridge microarchitecture which could lead to the recovery
 2211      of RSA keys.  The ability to exploit this issue is limited as it relies on
 2212      an attacker who has control of code in a thread running on the same
 2213      hyper-threaded core as the victim thread which is performing decryptions.
 2214 
 2215      This issue was reported to OpenSSL by Yuval Yarom, The University of
 2216      Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
 2217      Nadia Heninger, University of Pennsylvania with more information at
 2218      http://cachebleed.info.
 2219      (CVE-2016-0702)
 2220      [Andy Polyakov]
 2221 
 2222   *) Change the req app to generate a 2048-bit RSA/DSA key by default,
 2223      if no keysize is specified with default_bits. This fixes an
 2224      omission in an earlier change that changed all RSA/DSA key generation
 2225      apps to use 2048 bits by default.
 2226      [Emilia Käsper]
 2227 
 2228  Changes between 1.0.2e and 1.0.2f [28 Jan 2016]
 2229   *) DH small subgroups
 2230 
 2231      Historically OpenSSL only ever generated DH parameters based on "safe"
 2232      primes. More recently (in version 1.0.2) support was provided for
 2233      generating X9.42 style parameter files such as those required for RFC 5114
 2234      support. The primes used in such files may not be "safe". Where an
 2235      application is using DH configured with parameters based on primes that are
 2236      not "safe" then an attacker could use this fact to find a peer's private
 2237      DH exponent. This attack requires that the attacker complete multiple
 2238      handshakes in which the peer uses the same private DH exponent. For example
 2239      this could be used to discover a TLS server's private DH exponent if it's
 2240      reusing the private DH exponent or it's using a static DH ciphersuite.
 2241 
 2242      OpenSSL provides the option SSL_OP_SINGLE_DH_USE for ephemeral DH (DHE) in
 2243      TLS. It is not on by default. If the option is not set then the server
 2244      reuses the same private DH exponent for the life of the server process and
 2245      would be vulnerable to this attack. It is believed that many popular
 2246      applications do set this option and would therefore not be at risk.
 2247 
 2248      The fix for this issue adds an additional check where a "q" parameter is
 2249      available (as is the case in X9.42 based parameters). This detects the
 2250      only known attack, and is the only possible defense for static DH
 2251      ciphersuites. This could have some performance impact.
 2252 
 2253      Additionally the SSL_OP_SINGLE_DH_USE option has been switched on by
 2254      default and cannot be disabled. This could have some performance impact.
 2255 
 2256      This issue was reported to OpenSSL by Antonio Sanso (Adobe).
 2257      (CVE-2016-0701)
 2258      [Matt Caswell]
 2259 
 2260   *) SSLv2 doesn't block disabled ciphers
 2261 
 2262      A malicious client can negotiate SSLv2 ciphers that have been disabled on
 2263      the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
 2264      been disabled, provided that the SSLv2 protocol was not also disabled via
 2265      SSL_OP_NO_SSLv2.
 2266 
 2267      This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
 2268      and Sebastian Schinzel.
 2269      (CVE-2015-3197)
 2270      [Viktor Dukhovni]
 2271 
 2272  Changes between 1.0.2d and 1.0.2e [3 Dec 2015]
 2273 
 2274   *) BN_mod_exp may produce incorrect results on x86_64
 2275 
 2276      There is a carry propagating bug in the x86_64 Montgomery squaring
 2277      procedure. No EC algorithms are affected. Analysis suggests that attacks
 2278      against RSA and DSA as a result of this defect would be very difficult to
 2279      perform and are not believed likely. Attacks against DH are considered just
 2280      feasible (although very difficult) because most of the work necessary to
 2281      deduce information about a private key may be performed offline. The amount
 2282      of resources required for such an attack would be very significant and
 2283      likely only accessible to a limited number of attackers. An attacker would
 2284      additionally need online access to an unpatched system using the target
 2285      private key in a scenario with persistent DH parameters and a private
 2286      key that is shared between multiple clients. For example this can occur by
 2287      default in OpenSSL DHE based SSL/TLS ciphersuites.
 2288 
 2289      This issue was reported to OpenSSL by Hanno Böck.
 2290      (CVE-2015-3193)
 2291      [Andy Polyakov]
 2292 
 2293   *) Certificate verify crash with missing PSS parameter
 2294 
 2295      The signature verification routines will crash with a NULL pointer
 2296      dereference if presented with an ASN.1 signature using the RSA PSS
 2297      algorithm and absent mask generation function parameter. Since these
 2298      routines are used to verify certificate signature algorithms this can be
 2299      used to crash any certificate verification operation and exploited in a
 2300      DoS attack. Any application which performs certificate verification is
 2301      vulnerable including OpenSSL clients and servers which enable client
 2302      authentication.
 2303 
 2304      This issue was reported to OpenSSL by Loïc Jonas Etienne (Qnective AG).
 2305      (CVE-2015-3194)
 2306      [Stephen Henson]
 2307 
 2308   *) X509_ATTRIBUTE memory leak
 2309 
 2310      When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
 2311      memory. This structure is used by the PKCS#7 and CMS routines so any
 2312      application which reads PKCS#7 or CMS data from untrusted sources is
 2313      affected. SSL/TLS is not affected.
 2314 
 2315      This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
 2316      libFuzzer.
 2317      (CVE-2015-3195)
 2318      [Stephen Henson]
 2319 
 2320   *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
 2321      This changes the decoding behaviour for some invalid messages,
 2322      though the change is mostly in the more lenient direction, and
 2323      legacy behaviour is preserved as much as possible.
 2324      [Emilia Käsper]
 2325 
 2326   *) In DSA_generate_parameters_ex, if the provided seed is too short,
 2327      return an error
 2328      [Rich Salz and Ismo Puustinen <ismo.puustinen@intel.com>]
 2329 
 2330  Changes between 1.0.2c and 1.0.2d [9 Jul 2015]
 2331 
 2332   *) Alternate chains certificate forgery
 2333 
 2334      During certificate verification, OpenSSL will attempt to find an
 2335      alternative certificate chain if the first attempt to build such a chain
 2336      fails. An error in the implementation of this logic can mean that an
 2337      attacker could cause certain checks on untrusted certificates to be
 2338      bypassed, such as the CA flag, enabling them to use a valid leaf
 2339      certificate to act as a CA and "issue" an invalid certificate.
 2340 
 2341      This issue was reported to OpenSSL by Adam Langley/David Benjamin
 2342      (Google/BoringSSL).
 2343      [Matt Caswell]
 2344 
 2345  Changes between 1.0.2b and 1.0.2c [12 Jun 2015]
 2346 
 2347   *) Fix HMAC ABI incompatibility. The previous version introduced an ABI
 2348      incompatibility in the handling of HMAC. The previous ABI has now been
 2349      restored.
 2350      [Matt Caswell]
 2351 
 2352  Changes between 1.0.2a and 1.0.2b [11 Jun 2015]
 2353 
 2354   *) Malformed ECParameters causes infinite loop
 2355 
 2356      When processing an ECParameters structure OpenSSL enters an infinite loop
 2357      if the curve specified is over a specially malformed binary polynomial
 2358      field.
 2359 
 2360      This can be used to perform denial of service against any
 2361      system which processes public keys, certificate requests or
 2362      certificates.  This includes TLS clients and TLS servers with
 2363      client authentication enabled.
 2364 
 2365      This issue was reported to OpenSSL by Joseph Barr-Pixton.
 2366      (CVE-2015-1788)
 2367      [Andy Polyakov]
 2368 
 2369   *) Exploitable out-of-bounds read in X509_cmp_time
 2370 
 2371      X509_cmp_time does not properly check the length of the ASN1_TIME
 2372      string and can read a few bytes out of bounds. In addition,
 2373      X509_cmp_time accepts an arbitrary number of fractional seconds in the
 2374      time string.
 2375 
 2376      An attacker can use this to craft malformed certificates and CRLs of
 2377      various sizes and potentially cause a segmentation fault, resulting in
 2378      a DoS on applications that verify certificates or CRLs. TLS clients
 2379      that verify CRLs are affected. TLS clients and servers with client
 2380      authentication enabled may be affected if they use custom verification
 2381      callbacks.
 2382 
 2383      This issue was reported to OpenSSL by Robert Swiecki (Google), and
 2384      independently by Hanno Böck.
 2385      (CVE-2015-1789)
 2386      [Emilia Käsper]
 2387 
 2388   *) PKCS7 crash with missing EnvelopedContent
 2389 
 2390      The PKCS#7 parsing code does not handle missing inner EncryptedContent
 2391      correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
 2392      with missing content and trigger a NULL pointer dereference on parsing.
 2393 
 2394      Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
 2395      structures from untrusted sources are affected. OpenSSL clients and
 2396      servers are not affected.
 2397 
 2398      This issue was reported to OpenSSL by Michal Zalewski (Google).
 2399      (CVE-2015-1790)
 2400      [Emilia Käsper]
 2401 
 2402   *) CMS verify infinite loop with unknown hash function
 2403 
 2404      When verifying a signedData message the CMS code can enter an infinite loop
 2405      if presented with an unknown hash function OID. This can be used to perform
 2406      denial of service against any system which verifies signedData messages using
 2407      the CMS code.
 2408      This issue was reported to OpenSSL by Johannes Bauer.
 2409      (CVE-2015-1792)
 2410      [Stephen Henson]
 2411 
 2412   *) Race condition handling NewSessionTicket
 2413 
 2414      If a NewSessionTicket is received by a multi-threaded client when attempting to
 2415      reuse a previous ticket then a race condition can occur potentially leading to
 2416      a double free of the ticket data.
 2417      (CVE-2015-1791)
 2418      [Matt Caswell]
 2419 
 2420   *) Only support 256-bit or stronger elliptic curves with the
 2421      'ecdh_auto' setting (server) or by default (client). Of supported
 2422      curves, prefer P-256 (both).
 2423      [Emilia Kasper]
 2424 
 2425  Changes between 1.0.2 and 1.0.2a [19 Mar 2015]
 2426 
 2427   *) ClientHello sigalgs DoS fix
 2428 
 2429      If a client connects to an OpenSSL 1.0.2 server and renegotiates with an
 2430      invalid signature algorithms extension a NULL pointer dereference will
 2431      occur. This can be exploited in a DoS attack against the server.
 2432 
 2433      This issue was was reported to OpenSSL by David Ramos of Stanford
 2434      University.
 2435      (CVE-2015-0291)
 2436      [Stephen Henson and Matt Caswell]
 2437 
 2438   *) Multiblock corrupted pointer fix
 2439 
 2440      OpenSSL 1.0.2 introduced the "multiblock" performance improvement. This
 2441      feature only applies on 64 bit x86 architecture platforms that support AES
 2442      NI instructions. A defect in the implementation of "multiblock" can cause
 2443      OpenSSL's internal write buffer to become incorrectly set to NULL when
 2444      using non-blocking IO. Typically, when the user application is using a
 2445      socket BIO for writing, this will only result in a failed connection.
 2446      However if some other BIO is used then it is likely that a segmentation
 2447      fault will be triggered, thus enabling a potential DoS attack.
 2448 
 2449      This issue was reported to OpenSSL by Daniel Danner and Rainer Mueller.
 2450      (CVE-2015-0290)
 2451      [Matt Caswell]
 2452 
 2453   *) Segmentation fault in DTLSv1_listen fix
 2454 
 2455      The DTLSv1_listen function is intended to be stateless and processes the
 2456      initial ClientHello from many peers. It is common for user code to loop
 2457      over the call to DTLSv1_listen until a valid ClientHello is received with
 2458      an associated cookie. A defect in the implementation of DTLSv1_listen means
 2459      that state is preserved in the SSL object from one invocation to the next
 2460      that can lead to a segmentation fault. Errors processing the initial
 2461      ClientHello can trigger this scenario. An example of such an error could be
 2462      that a DTLS1.0 only client is attempting to connect to a DTLS1.2 only
 2463      server.
 2464 
 2465      This issue was reported to OpenSSL by Per Allansson.
 2466      (CVE-2015-0207)
 2467      [Matt Caswell]
 2468 
 2469   *) Segmentation fault in ASN1_TYPE_cmp fix
 2470 
 2471      The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
 2472      made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
 2473      certificate signature algorithm consistency this can be used to crash any
 2474      certificate verification operation and exploited in a DoS attack. Any
 2475      application which performs certificate verification is vulnerable including
 2476      OpenSSL clients and servers which enable client authentication.
 2477      (CVE-2015-0286)
 2478      [Stephen Henson]
 2479 
 2480   *) Segmentation fault for invalid PSS parameters fix
 2481 
 2482      The signature verification routines will crash with a NULL pointer
 2483      dereference if presented with an ASN.1 signature using the RSA PSS
 2484      algorithm and invalid parameters. Since these routines are used to verify
 2485      certificate signature algorithms this can be used to crash any
 2486      certificate verification operation and exploited in a DoS attack. Any
 2487      application which performs certificate verification is vulnerable including
 2488      OpenSSL clients and servers which enable client authentication.
 2489 
 2490      This issue was was reported to OpenSSL by Brian Carpenter.
 2491      (CVE-2015-0208)
 2492      [Stephen Henson]
 2493 
 2494   *) ASN.1 structure reuse memory corruption fix
 2495 
 2496      Reusing a structure in ASN.1 parsing may allow an attacker to cause
 2497      memory corruption via an invalid write. Such reuse is and has been
 2498      strongly discouraged and is believed to be rare.
 2499 
 2500      Applications that parse structures containing CHOICE or ANY DEFINED BY
 2501      components may be affected. Certificate parsing (d2i_X509 and related
 2502      functions) are however not affected. OpenSSL clients and servers are
 2503      not affected.
 2504      (CVE-2015-0287)
 2505      [Stephen Henson]
 2506 
 2507   *) PKCS7 NULL pointer dereferences fix
 2508 
 2509      The PKCS#7 parsing code does not handle missing outer ContentInfo
 2510      correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
 2511      missing content and trigger a NULL pointer dereference on parsing.
 2512 
 2513      Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
 2514      otherwise parse PKCS#7 structures from untrusted sources are
 2515      affected. OpenSSL clients and servers are not affected.
 2516 
 2517      This issue was reported to OpenSSL by Michal Zalewski (Google).
 2518      (CVE-2015-0289)
 2519      [Emilia Käsper]
 2520 
 2521   *) DoS via reachable assert in SSLv2 servers fix
 2522 
 2523      A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
 2524      servers that both support SSLv2 and enable export cipher suites by sending
 2525      a specially crafted SSLv2 CLIENT-MASTER-KEY message.
 2526 
 2527      This issue was discovered by Sean Burford (Google) and Emilia Käsper
 2528      (OpenSSL development team).
 2529      (CVE-2015-0293)
 2530      [Emilia Käsper]
 2531 
 2532   *) Empty CKE with client auth and DHE fix
 2533 
 2534      If client auth is used then a server can seg fault in the event of a DHE
 2535      ciphersuite being selected and a zero length ClientKeyExchange message
 2536      being sent by the client. This could be exploited in a DoS attack.
 2537      (CVE-2015-1787)
 2538      [Matt Caswell]
 2539 
 2540   *) Handshake with unseeded PRNG fix
 2541 
 2542      Under certain conditions an OpenSSL 1.0.2 client can complete a handshake
 2543      with an unseeded PRNG. The conditions are:
 2544      - The client is on a platform where the PRNG has not been seeded
 2545      automatically, and the user has not seeded manually
 2546      - A protocol specific client method version has been used (i.e. not
 2547      SSL_client_methodv23)
 2548      - A ciphersuite is used that does not require additional random data from
 2549      the PRNG beyond the initial ClientHello client random (e.g. PSK-RC4-SHA).
 2550 
 2551      If the handshake succeeds then the client random that has been used will
 2552      have been generated from a PRNG with insufficient entropy and therefore the
 2553      output may be predictable.
 2554 
 2555      For example using the following command with an unseeded openssl will
 2556      succeed on an unpatched platform:
 2557 
 2558      openssl s_client -psk 1a2b3c4d -tls1_2 -cipher PSK-RC4-SHA
 2559      (CVE-2015-0285)
 2560      [Matt Caswell]
 2561 
 2562   *) Use After Free following d2i_ECPrivatekey error fix
 2563 
 2564      A malformed EC private key file consumed via the d2i_ECPrivateKey function
 2565      could cause a use after free condition. This, in turn, could cause a double
 2566      free in several private key parsing functions (such as d2i_PrivateKey
 2567      or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
 2568      for applications that receive EC private keys from untrusted
 2569      sources. This scenario is considered rare.
 2570 
 2571      This issue was discovered by the BoringSSL project and fixed in their
 2572      commit 517073cd4b.
 2573      (CVE-2015-0209)
 2574      [Matt Caswell]
 2575 
 2576   *) X509_to_X509_REQ NULL pointer deref fix
 2577 
 2578      The function X509_to_X509_REQ will crash with a NULL pointer dereference if
 2579      the certificate key is invalid. This function is rarely used in practice.
 2580 
 2581      This issue was discovered by Brian Carpenter.
 2582      (CVE-2015-0288)
 2583      [Stephen Henson]
 2584 
 2585   *) Removed the export ciphers from the DEFAULT ciphers
 2586      [Kurt Roeckx]
 2587 
 2588  Changes between 1.0.1l and 1.0.2 [22 Jan 2015]
 2589 
 2590   *) Facilitate "universal" ARM builds targeting range of ARM ISAs, e.g.
 2591      ARMv5 through ARMv8, as opposite to "locking" it to single one.
 2592      So far those who have to target multiple platforms would compromise
 2593      and argue that binary targeting say ARMv5 would still execute on
 2594      ARMv8. "Universal" build resolves this compromise by providing
 2595      near-optimal performance even on newer platforms.
 2596      [Andy Polyakov]
 2597 
 2598   *) Accelerated NIST P-256 elliptic curve implementation for x86_64
 2599      (other platforms pending).
 2600      [Shay Gueron & Vlad Krasnov (Intel Corp), Andy Polyakov]
 2601 
 2602   *) Add support for the SignedCertificateTimestampList certificate and
 2603      OCSP response extensions from RFC6962.
 2604      [Rob Stradling]
 2605 
 2606   *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
 2607      for corner cases. (Certain input points at infinity could lead to
 2608      bogus results, with non-infinity inputs mapped to infinity too.)
 2609      [Bodo Moeller]
 2610 
 2611   *) Initial support for PowerISA 2.0.7, first implemented in POWER8.
 2612      This covers AES, SHA256/512 and GHASH. "Initial" means that most
 2613      common cases are optimized and there still is room for further
 2614      improvements. Vector Permutation AES for Altivec is also added.
 2615      [Andy Polyakov]
 2616 
 2617   *) Add support for little-endian ppc64 Linux target.
 2618      [Marcelo Cerri (IBM)]
 2619 
 2620   *) Initial support for AMRv8 ISA crypto extensions. This covers AES,
 2621      SHA1, SHA256 and GHASH. "Initial" means that most common cases
 2622      are optimized and there still is room for further improvements.
 2623      Both 32- and 64-bit modes are supported.
 2624      [Andy Polyakov, Ard Biesheuvel (Linaro)]
 2625 
 2626   *) Improved ARMv7 NEON support.
 2627      [Andy Polyakov]
 2628 
 2629   *) Support for SPARC Architecture 2011 crypto extensions, first
 2630      implemented in SPARC T4. This covers AES, DES, Camellia, SHA1,
 2631      SHA256/512, MD5, GHASH and modular exponentiation.
 2632      [Andy Polyakov, David Miller]
 2633 
 2634   *) Accelerated modular exponentiation for Intel processors, a.k.a.
 2635      RSAZ.
 2636      [Shay Gueron & Vlad Krasnov (Intel Corp)]
 2637 
 2638   *) Support for new and upcoming Intel processors, including AVX2,
 2639      BMI and SHA ISA extensions. This includes additional "stitched"
 2640      implementations, AESNI-SHA256 and GCM, and multi-buffer support
 2641      for TLS encrypt.
 2642 
 2643      This work was sponsored by Intel Corp.
 2644      [Andy Polyakov]
 2645 
 2646   *) Support for DTLS 1.2. This adds two sets of DTLS methods: DTLS_*_method()
 2647      supports both DTLS 1.2 and 1.0 and should use whatever version the peer
 2648      supports and DTLSv1_2_*_method() which supports DTLS 1.2 only.
 2649      [Steve Henson]
 2650 
 2651   *) Use algorithm specific chains in SSL_CTX_use_certificate_chain_file():
 2652      this fixes a limitation in previous versions of OpenSSL.
 2653      [Steve Henson]
 2654 
 2655   *) Extended RSA OAEP support via EVP_PKEY API. Options to specify digest,
 2656      MGF1 digest and OAEP label.
 2657      [Steve Henson]
 2658 
 2659   *) Add EVP support for key wrapping algorithms, to avoid problems with
 2660      existing code the flag EVP_CIPHER_CTX_WRAP_ALLOW has to be set in
 2661      the EVP_CIPHER_CTX or an error is returned. Add AES and DES3 wrap
 2662      algorithms and include tests cases.
 2663      [Steve Henson]
 2664 
 2665   *) Add functions to allocate and set the fields of an ECDSA_METHOD
 2666      structure.
 2667      [Douglas E. Engert, Steve Henson]
 2668 
 2669   *) New functions OPENSSL_gmtime_diff and ASN1_TIME_diff to find the
 2670      difference in days and seconds between two tm or ASN1_TIME structures.
 2671      [Steve Henson]
 2672 
 2673   *) Add -rev test option to s_server to just reverse order of characters
 2674      received by client and send back to server. Also prints an abbreviated
 2675      summary of the connection parameters.
 2676      [Steve Henson]
 2677 
 2678   *) New option -brief for s_client and s_server to print out a brief summary
 2679      of connection parameters.
 2680      [Steve Henson]
 2681 
 2682   *) Add callbacks for arbitrary TLS extensions.
 2683      [Trevor Perrin <trevp@trevp.net> and Ben Laurie]
 2684 
 2685   *) New option -crl_download in several openssl utilities to download CRLs
 2686      from CRLDP extension in certificates.
 2687      [Steve Henson]
 2688 
 2689   *) New options -CRL and -CRLform for s_client and s_server for CRLs.
 2690      [Steve Henson]
 2691 
 2692   *) New function X509_CRL_diff to generate a delta CRL from the difference
 2693      of two full CRLs. Add support to "crl" utility.
 2694      [Steve Henson]
 2695 
 2696   *) New functions to set lookup_crls function and to retrieve
 2697      X509_STORE from X509_STORE_CTX.
 2698      [Steve Henson]
 2699 
 2700   *) Print out deprecated issuer and subject unique ID fields in
 2701      certificates.
 2702      [Steve Henson]
 2703 
 2704   *) Extend OCSP I/O functions so they can be used for simple general purpose
 2705      HTTP as well as OCSP. New wrapper function which can be used to download
 2706      CRLs using the OCSP API.
 2707      [Steve Henson]
 2708 
 2709   *) Delegate command line handling in s_client/s_server to SSL_CONF APIs.
 2710      [Steve Henson]
 2711 
 2712   *) SSL_CONF* functions. These provide a common framework for application
 2713      configuration using configuration files or command lines.
 2714      [Steve Henson]
 2715 
 2716   *) SSL/TLS tracing code. This parses out SSL/TLS records using the
 2717      message callback and prints the results. Needs compile time option
 2718      "enable-ssl-trace". New options to s_client and s_server to enable
 2719      tracing.
 2720      [Steve Henson]
 2721 
 2722   *) New ctrl and macro to retrieve supported points extensions.
 2723      Print out extension in s_server and s_client.
 2724      [Steve Henson]
 2725 
 2726   *) New functions to retrieve certificate signature and signature
 2727      OID NID.
 2728      [Steve Henson]
 2729 
 2730   *) Add functions to retrieve and manipulate the raw cipherlist sent by a
 2731      client to OpenSSL.
 2732      [Steve Henson]
 2733 
 2734   *) New Suite B modes for TLS code. These use and enforce the requirements
 2735      of RFC6460: restrict ciphersuites, only permit Suite B algorithms and
 2736      only use Suite B curves. The Suite B modes can be set by using the
 2737      strings "SUITEB128", "SUITEB192" or "SUITEB128ONLY" for the cipherstring.
 2738      [Steve Henson]
 2739 
 2740   *) New chain verification flags for Suite B levels of security. Check
 2741      algorithms are acceptable when flags are set in X509_verify_cert.
 2742      [Steve Henson]
 2743 
 2744   *) Make tls1_check_chain return a set of flags indicating checks passed
 2745      by a certificate chain. Add additional tests to handle client
 2746      certificates: checks for matching certificate type and issuer name
 2747      comparison.
 2748      [Steve Henson]
 2749 
 2750   *) If an attempt is made to use a signature algorithm not in the peer
 2751      preference list abort the handshake. If client has no suitable
 2752      signature algorithms in response to a certificate request do not
 2753      use the certificate.
 2754      [Steve Henson]
 2755 
 2756   *) If server EC tmp key is not in client preference list abort handshake.
 2757      [Steve Henson]
 2758 
 2759   *) Add support for certificate stores in CERT structure. This makes it
 2760      possible to have different stores per SSL structure or one store in
 2761      the parent SSL_CTX. Include distinct stores for certificate chain
 2762      verification and chain building. New ctrl SSL_CTRL_BUILD_CERT_CHAIN
 2763      to build and store a certificate chain in CERT structure: returning
 2764      an error if the chain cannot be built: this will allow applications
 2765      to test if a chain is correctly configured.
 2766 
 2767      Note: if the CERT based stores are not set then the parent SSL_CTX
 2768      store is used to retain compatibility with existing behaviour.
 2769 
 2770      [Steve Henson]
 2771 
 2772   *) New function ssl_set_client_disabled to set a ciphersuite disabled
 2773      mask based on the current session, check mask when sending client
 2774      hello and checking the requested ciphersuite.
 2775      [Steve Henson]
 2776 
 2777   *) New ctrls to retrieve and set certificate types in a certificate
 2778      request message. Print out received values in s_client. If certificate
 2779      types is not set with custom values set sensible values based on
 2780      supported signature algorithms.
 2781      [Steve Henson]
 2782 
 2783   *) Support for distinct client and server supported signature algorithms.
 2784      [Steve Henson]
 2785 
 2786   *) Add certificate callback. If set this is called whenever a certificate
 2787      is required by client or server. An application can decide which
 2788      certificate chain to present based on arbitrary criteria: for example
 2789      supported signature algorithms. Add very simple example to s_server.
 2790      This fixes many of the problems and restrictions of the existing client
 2791      certificate callback: for example you can now clear an existing
 2792      certificate and specify the whole chain.
 2793      [Steve Henson]
 2794 
 2795   *) Add new "valid_flags" field to CERT_PKEY structure which determines what
 2796      the certificate can be used for (if anything). Set valid_flags field
 2797      in new tls1_check_chain function. Simplify ssl_set_cert_masks which used
 2798      to have similar checks in it.
 2799 
 2800      Add new "cert_flags" field to CERT structure and include a "strict mode".
 2801      This enforces some TLS certificate requirements (such as only permitting
 2802      certificate signature algorithms contained in the supported algorithms
 2803      extension) which some implementations ignore: this option should be used
 2804      with caution as it could cause interoperability issues.
 2805      [Steve Henson]
 2806 
 2807   *) Update and tidy signature algorithm extension processing. Work out
 2808      shared signature algorithms based on preferences and peer algorithms
 2809      and print them out in s_client and s_server. Abort handshake if no
 2810      shared signature algorithms.
 2811      [Steve Henson]
 2812 
 2813   *) Add new functions to allow customised supported signature algorithms
 2814      for SSL and SSL_CTX structures. Add options to s_client and s_server
 2815      to support them.
 2816      [Steve Henson]
 2817 
 2818   *) New function SSL_certs_clear() to delete all references to certificates
 2819      from an SSL structure. Before this once a certificate had been added
 2820      it couldn't be removed.
 2821      [Steve Henson]
 2822 
 2823   *) Integrate hostname, email address and IP address checking with certificate
 2824      verification. New verify options supporting checking in openssl utility.
 2825      [Steve Henson]
 2826 
 2827   *) Fixes and wildcard matching support to hostname and email checking
 2828      functions. Add manual page.
 2829      [Florian Weimer (Red Hat Product Security Team)]
 2830 
 2831   *) New functions to check a hostname email or IP address against a
 2832      certificate. Add options x509 utility to print results of checks against
 2833      a certificate.
 2834      [Steve Henson]
 2835 
 2836   *) Fix OCSP checking.
 2837      [Rob Stradling <rob.stradling@comodo.com> and Ben Laurie]
 2838 
 2839   *) Initial experimental support for explicitly trusted non-root CAs.
 2840      OpenSSL still tries to build a complete chain to a root but if an
 2841      intermediate CA has a trust setting included that is used. The first
 2842      setting is used: whether to trust (e.g., -addtrust option to the x509
 2843      utility) or reject.
 2844      [Steve Henson]
 2845 
 2846   *) Add -trusted_first option which attempts to find certificates in the
 2847      trusted store even if an untrusted chain is also supplied.
 2848      [Steve Henson]
 2849 
 2850   *) MIPS assembly pack updates: support for MIPS32r2 and SmartMIPS ASE,
 2851      platform support for Linux and Android.
 2852      [Andy Polyakov]
 2853 
 2854   *) Support for linux-x32, ILP32 environment in x86_64 framework.
 2855      [Andy Polyakov]
 2856 
 2857   *) Experimental multi-implementation support for FIPS capable OpenSSL.
 2858      When in FIPS mode the approved implementations are used as normal,
 2859      when not in FIPS mode the internal unapproved versions are used instead.
 2860      This means that the FIPS capable OpenSSL isn't forced to use the
 2861      (often lower performance) FIPS implementations outside FIPS mode.
 2862      [Steve Henson]
 2863 
 2864   *) Transparently support X9.42 DH parameters when calling
 2865      PEM_read_bio_DHparameters. This means existing applications can handle
 2866      the new parameter format automatically.
 2867      [Steve Henson]
 2868 
 2869   *) Initial experimental support for X9.42 DH parameter format: mainly
 2870      to support use of 'q' parameter for RFC5114 parameters.
 2871      [Steve Henson]
 2872 
 2873   *) Add DH parameters from RFC5114 including test data to dhtest.
 2874      [Steve Henson]
 2875 
 2876   *) Support for automatic EC temporary key parameter selection. If enabled
 2877      the most preferred EC parameters are automatically used instead of
 2878      hardcoded fixed parameters. Now a server just has to call:
 2879      SSL_CTX_set_ecdh_auto(ctx, 1) and the server will automatically
 2880      support ECDH and use the most appropriate parameters.
 2881      [Steve Henson]
 2882 
 2883   *) Enhance and tidy EC curve and point format TLS extension code. Use
 2884      static structures instead of allocation if default values are used.
 2885      New ctrls to set curves we wish to support and to retrieve shared curves.
 2886      Print out shared curves in s_server. New options to s_server and s_client
 2887      to set list of supported curves.
 2888      [Steve Henson]
 2889 
 2890   *) New ctrls to retrieve supported signature algorithms and
 2891      supported curve values as an array of NIDs. Extend openssl utility
 2892      to print out received values.
 2893      [Steve Henson]
 2894 
 2895   *) Add new APIs EC_curve_nist2nid and EC_curve_nid2nist which convert
 2896      between NIDs and the more common NIST names such as "P-256". Enhance
 2897      ecparam utility and ECC method to recognise the NIST names for curves.
 2898      [Steve Henson]
 2899 
 2900   *) Enhance SSL/TLS certificate chain handling to support different
 2901      chains for each certificate instead of one chain in the parent SSL_CTX.
 2902      [Steve Henson]
 2903 
 2904   *) Support for fixed DH ciphersuite client authentication: where both
 2905      server and client use DH certificates with common parameters.
 2906      [Steve Henson]
 2907 
 2908   *) Support for fixed DH ciphersuites: those requiring DH server
 2909      certificates.
 2910      [Steve Henson]
 2911 
 2912   *) New function i2d_re_X509_tbs for re-encoding the TBS portion of
 2913      the certificate.
 2914      Note: Related 1.0.2-beta specific macros X509_get_cert_info,
 2915      X509_CINF_set_modified, X509_CINF_get_issuer, X509_CINF_get_extensions and
 2916      X509_CINF_get_signature were reverted post internal team review.
 2917 
 2918  Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
 2919 
 2920   *) Build fixes for the Windows and OpenVMS platforms
 2921      [Matt Caswell and Richard Levitte]
 2922 
 2923  Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
 2924 
 2925   *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
 2926      message can cause a segmentation fault in OpenSSL due to a NULL pointer
 2927      dereference. This could lead to a Denial Of Service attack. Thanks to
 2928      Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
 2929      (CVE-2014-3571)
 2930      [Steve Henson]
 2931 
 2932   *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
 2933      dtls1_buffer_record function under certain conditions. In particular this
 2934      could occur if an attacker sent repeated DTLS records with the same
 2935      sequence number but for the next epoch. The memory leak could be exploited
 2936      by an attacker in a Denial of Service attack through memory exhaustion.
 2937      Thanks to Chris Mueller for reporting this issue.
 2938      (CVE-2015-0206)
 2939      [Matt Caswell]
 2940 
 2941   *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
 2942      built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
 2943      method would be set to NULL which could later result in a NULL pointer
 2944      dereference. Thanks to Frank Schmirler for reporting this issue.
 2945      (CVE-2014-3569)
 2946      [Kurt Roeckx]
 2947 
 2948   *) Abort handshake if server key exchange message is omitted for ephemeral
 2949      ECDH ciphersuites.
 2950 
 2951      Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
 2952      reporting this issue.
 2953      (CVE-2014-3572)
 2954      [Steve Henson]
 2955 
 2956   *) Remove non-export ephemeral RSA code on client and server. This code
 2957      violated the TLS standard by allowing the use of temporary RSA keys in
 2958      non-export ciphersuites and could be used by a server to effectively
 2959      downgrade the RSA key length used to a value smaller than the server
 2960      certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
 2961      INRIA or reporting this issue.
 2962      (CVE-2015-0204)
 2963      [Steve Henson]
 2964 
 2965   *) Fixed issue where DH client certificates are accepted without verification.
 2966      An OpenSSL server will accept a DH certificate for client authentication
 2967      without the certificate verify message. This effectively allows a client to
 2968      authenticate without the use of a private key. This only affects servers
 2969      which trust a client certificate authority which issues certificates
 2970      containing DH keys: these are extremely rare and hardly ever encountered.
 2971      Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
 2972      this issue.
 2973      (CVE-2015-0205)
 2974      [Steve Henson]
 2975 
 2976   *) Ensure that the session ID context of an SSL is updated when its
 2977      SSL_CTX is updated via SSL_set_SSL_CTX.
 2978 
 2979      The session ID context is typically set from the parent SSL_CTX,
 2980      and can vary with the CTX.
 2981      [Adam Langley]
 2982 
 2983   *) Fix various certificate fingerprint issues.
 2984 
 2985      By using non-DER or invalid encodings outside the signed portion of a
 2986      certificate the fingerprint can be changed without breaking the signature.
 2987      Although no details of the signed portion of the certificate can be changed
 2988      this can cause problems with some applications: e.g. those using the
 2989      certificate fingerprint for blacklists.
 2990 
 2991      1. Reject signatures with non zero unused bits.
 2992 
 2993      If the BIT STRING containing the signature has non zero unused bits reject
 2994      the signature. All current signature algorithms require zero unused bits.
 2995 
 2996      2. Check certificate algorithm consistency.
 2997 
 2998      Check the AlgorithmIdentifier inside TBS matches the one in the
 2999      certificate signature. NB: this will result in signature failure
 3000      errors for some broken certificates.
 3001 
 3002      Thanks to Konrad Kraszewski from Google for reporting this issue.
 3003 
 3004      3. Check DSA/ECDSA signatures use DER.
 3005 
 3006      Re-encode DSA/ECDSA signatures and compare with the original received
 3007      signature. Return an error if there is a mismatch.
 3008 
 3009      This will reject various cases including garbage after signature
 3010      (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
 3011      program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
 3012      (negative or with leading zeroes).
 3013 
 3014      Further analysis was conducted and fixes were developed by Stephen Henson
 3015      of the OpenSSL core team.
 3016 
 3017      (CVE-2014-8275)
 3018      [Steve Henson]
 3019 
 3020    *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
 3021       results on some platforms, including x86_64. This bug occurs at random
 3022       with a very low probability, and is not known to be exploitable in any
 3023       way, though its exact impact is difficult to determine. Thanks to Pieter
 3024       Wuille (Blockstream) who reported this issue and also suggested an initial
 3025       fix. Further analysis was conducted by the OpenSSL development team and
 3026       Adam Langley of Google. The final fix was developed by Andy Polyakov of
 3027       the OpenSSL core team.
 3028       (CVE-2014-3570)
 3029       [Andy Polyakov]
 3030 
 3031    *) Do not resume sessions on the server if the negotiated protocol
 3032       version does not match the session's version. Resuming with a different
 3033       version, while not strictly forbidden by the RFC, is of questionable
 3034       sanity and breaks all known clients.
 3035       [David Benjamin, Emilia Käsper]
 3036 
 3037    *) Tighten handling of the ChangeCipherSpec (CCS) message: reject
 3038       early CCS messages during renegotiation. (Note that because
 3039       renegotiation is encrypted, this early CCS was not exploitable.)
 3040       [Emilia Käsper]
 3041 
 3042    *) Tighten client-side session ticket handling during renegotiation:
 3043       ensure that the client only accepts a session ticket if the server sends
 3044       the extension anew in the ServerHello. Previously, a TLS client would
 3045       reuse the old extension state and thus accept a session ticket if one was
 3046       announced in the initial ServerHello.
 3047 
 3048       Similarly, ensure that the client requires a session ticket if one
 3049       was advertised in the ServerHello. Previously, a TLS client would
 3050       ignore a missing NewSessionTicket message.
 3051       [Emilia Käsper]
 3052 
 3053  Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
 3054 
 3055   *) SRTP Memory Leak.
 3056 
 3057      A flaw in the DTLS SRTP extension parsing code allows an attacker, who
 3058      sends a carefully crafted handshake message, to cause OpenSSL to fail
 3059      to free up to 64k of memory causing a memory leak. This could be
 3060      exploited in a Denial Of Service attack. This issue affects OpenSSL
 3061      1.0.1 server implementations for both SSL/TLS and DTLS regardless of
 3062      whether SRTP is used or configured. Implementations of OpenSSL that
 3063      have been compiled with OPENSSL_NO_SRTP defined are not affected.
 3064 
 3065      The fix was developed by the OpenSSL team.
 3066      (CVE-2014-3513)
 3067      [OpenSSL team]
 3068 
 3069   *) Session Ticket Memory Leak.
 3070 
 3071      When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
 3072      integrity of that ticket is first verified. In the event of a session
 3073      ticket integrity check failing, OpenSSL will fail to free memory
 3074      causing a memory leak. By sending a large number of invalid session
 3075      tickets an attacker could exploit this issue in a Denial Of Service
 3076      attack.
 3077      (CVE-2014-3567)
 3078      [Steve Henson]
 3079 
 3080   *) Build option no-ssl3 is incomplete.
 3081 
 3082      When OpenSSL is configured with "no-ssl3" as a build option, servers
 3083      could accept and complete a SSL 3.0 handshake, and clients could be
 3084      configured to send them.
 3085      (CVE-2014-3568)
 3086      [Akamai and the OpenSSL team]
 3087 
 3088   *) Add support for TLS_FALLBACK_SCSV.
 3089      Client applications doing fallback retries should call
 3090      SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
 3091      (CVE-2014-3566)
 3092      [Adam Langley, Bodo Moeller]
 3093 
 3094   *) Add additional DigestInfo checks.
 3095 
 3096      Re-encode DigestInto in DER and check against the original when
 3097      verifying RSA signature: this will reject any improperly encoded
 3098      DigestInfo structures.
 3099 
 3100      Note: this is a precautionary measure and no attacks are currently known.
 3101 
 3102      [Steve Henson]
 3103 
 3104  Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
 3105 
 3106   *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
 3107      SRP code can be overrun an internal buffer. Add sanity check that
 3108      g, A, B < N to SRP code.
 3109 
 3110      Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
 3111      Group for discovering this issue.
 3112      (CVE-2014-3512)
 3113      [Steve Henson]
 3114 
 3115   *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
 3116      TLS 1.0 instead of higher protocol versions when the ClientHello message
 3117      is badly fragmented. This allows a man-in-the-middle attacker to force a
 3118      downgrade to TLS 1.0 even if both the server and the client support a
 3119      higher protocol version, by modifying the client's TLS records.
 3120 
 3121      Thanks to David Benjamin and Adam Langley (Google) for discovering and
 3122      researching this issue.
 3123      (CVE-2014-3511)
 3124      [David Benjamin]
 3125 
 3126   *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
 3127      to a denial of service attack. A malicious server can crash the client
 3128      with a null pointer dereference (read) by specifying an anonymous (EC)DH
 3129      ciphersuite and sending carefully crafted handshake messages.
 3130 
 3131      Thanks to Felix Gröbert (Google) for discovering and researching this
 3132      issue.
 3133      (CVE-2014-3510)
 3134      [Emilia Käsper]
 3135 
 3136   *) By sending carefully crafted DTLS packets an attacker could cause openssl
 3137      to leak memory. This can be exploited through a Denial of Service attack.
 3138      Thanks to Adam Langley for discovering and researching this issue.
 3139      (CVE-2014-3507)
 3140      [Adam Langley]
 3141 
 3142   *) An attacker can force openssl to consume large amounts of memory whilst
 3143      processing DTLS handshake messages. This can be exploited through a
 3144      Denial of Service attack.
 3145      Thanks to Adam Langley for discovering and researching this issue.
 3146      (CVE-2014-3506)
 3147      [Adam Langley]
 3148 
 3149   *) An attacker can force an error condition which causes openssl to crash
 3150      whilst processing DTLS packets due to memory being freed twice. This
 3151      can be exploited through a Denial of Service attack.
 3152      Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
 3153      this issue.
 3154      (CVE-2014-3505)
 3155      [Adam Langley]
 3156 
 3157   *) If a multithreaded client connects to a malicious server using a resumed
 3158      session and the server sends an ec point format extension it could write
 3159      up to 255 bytes to freed memory.
 3160 
 3161      Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
 3162      issue.
 3163      (CVE-2014-3509)
 3164      [Gabor Tyukasz]
 3165 
 3166   *) A malicious server can crash an OpenSSL client with a null pointer
 3167      dereference (read) by specifying an SRP ciphersuite even though it was not
 3168      properly negotiated with the client. This can be exploited through a
 3169      Denial of Service attack.
 3170 
 3171      Thanks to Joonas Kuorilehto and Riku Hietamäki (Codenomicon) for
 3172      discovering and researching this issue.
 3173      (CVE-2014-5139)
 3174      [Steve Henson]
 3175 
 3176   *) A flaw in OBJ_obj2txt may cause pretty printing functions such as
 3177      X509_name_oneline, X509_name_print_ex et al. to leak some information
 3178      from the stack. Applications may be affected if they echo pretty printing
 3179      output to the attacker.
 3180 
 3181      Thanks to Ivan Fratric (Google) for discovering this issue.
 3182      (CVE-2014-3508)
 3183      [Emilia Käsper, and Steve Henson]
 3184 
 3185   *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
 3186      for corner cases. (Certain input points at infinity could lead to
 3187      bogus results, with non-infinity inputs mapped to infinity too.)
 3188      [Bodo Moeller]
 3189 
 3190  Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
 3191 
 3192   *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
 3193      handshake can force the use of weak keying material in OpenSSL
 3194      SSL/TLS clients and servers.
 3195 
 3196      Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
 3197      researching this issue. (CVE-2014-0224)
 3198      [KIKUCHI Masashi, Steve Henson]
 3199 
 3200   *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
 3201      OpenSSL DTLS client the code can be made to recurse eventually crashing
 3202      in a DoS attack.
 3203 
 3204      Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
 3205      (CVE-2014-0221)
 3206      [Imre Rad, Steve Henson]
 3207 
 3208   *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
 3209      be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
 3210      client or server. This is potentially exploitable to run arbitrary
 3211      code on a vulnerable client or server.
 3212 
 3213      Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195)
 3214      [Jüri Aedla, Steve Henson]
 3215 
 3216   *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
 3217      are subject to a denial of service attack.
 3218 
 3219      Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
 3220      this issue. (CVE-2014-3470)
 3221      [Felix Gröbert, Ivan Fratric, Steve Henson]
 3222 
 3223   *) Harmonize version and its documentation. -f flag is used to display
 3224      compilation flags.
 3225      [mancha <mancha1@zoho.com>]
 3226 
 3227   *) Fix eckey_priv_encode so it immediately returns an error upon a failure
 3228      in i2d_ECPrivateKey.
 3229      [mancha <mancha1@zoho.com>]
 3230 
 3231   *) Fix some double frees. These are not thought to be exploitable.
 3232      [mancha <mancha1@zoho.com>]
 3233 
 3234  Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
 3235 
 3236   *) A missing bounds check in the handling of the TLS heartbeat extension
 3237      can be used to reveal up to 64k of memory to a connected client or
 3238      server.
 3239 
 3240      Thanks for Neel Mehta of Google Security for discovering this bug and to
 3241      Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
 3242      preparing the fix (CVE-2014-0160)
 3243      [Adam Langley, Bodo Moeller]
 3244 
 3245   *) Fix for the attack described in the paper "Recovering OpenSSL
 3246      ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
 3247      by Yuval Yarom and Naomi Benger. Details can be obtained from:
 3248      http://eprint.iacr.org/2014/140
 3249 
 3250      Thanks to Yuval Yarom and Naomi Benger for discovering this
 3251      flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
 3252      [Yuval Yarom and Naomi Benger]
 3253 
 3254   *) TLS pad extension: draft-agl-tls-padding-03
 3255 
 3256      Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
 3257      TLS client Hello record length value would otherwise be > 255 and
 3258      less that 512 pad with a dummy extension containing zeroes so it
 3259      is at least 512 bytes long.
 3260 
 3261      [Adam Langley, Steve Henson]
 3262 
 3263  Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
 3264 
 3265   *) Fix for TLS record tampering bug. A carefully crafted invalid
 3266      handshake could crash OpenSSL with a NULL pointer exception.
 3267      Thanks to Anton Johansson for reporting this issues.
 3268      (CVE-2013-4353)
 3269 
 3270   *) Keep original DTLS digest and encryption contexts in retransmission
 3271      structures so we can use the previous session parameters if they need
 3272      to be resent. (CVE-2013-6450)
 3273      [Steve Henson]
 3274 
 3275   *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
 3276      avoids preferring ECDHE-ECDSA ciphers when the client appears to be
 3277      Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
 3278      several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
 3279      is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
 3280      10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
 3281      [Rob Stradling, Adam Langley]
 3282 
 3283  Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
 3284 
 3285   *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI
 3286      supporting platforms or when small records were transferred.
 3287      [Andy Polyakov, Steve Henson]
 3288 
 3289  Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
 3290 
 3291   *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
 3292 
 3293      This addresses the flaw in CBC record processing discovered by
 3294      Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
 3295      at: http://www.isg.rhul.ac.uk/tls/
 3296 
 3297      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
 3298      Security Group at Royal Holloway, University of London
 3299      (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
 3300      Emilia Käsper for the initial patch.
 3301      (CVE-2013-0169)
 3302      [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
 3303 
 3304   *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
 3305      ciphersuites which can be exploited in a denial of service attack.
 3306      Thanks go to and to Adam Langley <agl@chromium.org> for discovering
 3307      and detecting this bug and to Wolfgang Ettlinger
 3308      <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
 3309      (CVE-2012-2686)
 3310      [Adam Langley]
 3311 
 3312   *) Return an error when checking OCSP signatures when key is NULL.
 3313      This fixes a DoS attack. (CVE-2013-0166)
 3314      [Steve Henson]
 3315 
 3316   *) Make openssl verify return errors.
 3317      [Chris Palmer <palmer@google.com> and Ben Laurie]
 3318 
 3319   *) Call OCSP Stapling callback after ciphersuite has been chosen, so
 3320      the right response is stapled. Also change SSL_get_certificate()
 3321      so it returns the certificate actually sent.
 3322      See http://rt.openssl.org/Ticket/Display.html?id=2836.
 3323      [Rob Stradling <rob.stradling@comodo.com>]
 3324 
 3325   *) Fix possible deadlock when decoding public keys.
 3326      [Steve Henson]
 3327 
 3328   *) Don't use TLS 1.0 record version number in initial client hello
 3329      if renegotiating.
 3330      [Steve Henson]
 3331 
 3332  Changes between 1.0.1b and 1.0.1c [10 May 2012]
 3333 
 3334   *) Sanity check record length before skipping explicit IV in TLS
 3335      1.2, 1.1 and DTLS to fix DoS attack.
 3336 
 3337      Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
 3338      fuzzing as a service testing platform.
 3339      (CVE-2012-2333)
 3340      [Steve Henson]
 3341 
 3342   *) Initialise tkeylen properly when encrypting CMS messages.
 3343      Thanks to Solar Designer of Openwall for reporting this issue.
 3344      [Steve Henson]
 3345 
 3346   *) In FIPS mode don't try to use composite ciphers as they are not
 3347      approved.
 3348      [Steve Henson]
 3349 
 3350  Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
 3351 
 3352   *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
 3353      1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
 3354      mean any application compiled against OpenSSL 1.0.0 headers setting
 3355      SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disabling
 3356      TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
 3357      0x10000000L Any application which was previously compiled against
 3358      OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
 3359      will need to be recompiled as a result. Letting be results in
 3360      inability to disable specifically TLS 1.1 and in client context,
 3361      in unlike event, limit maximum offered version to TLS 1.0 [see below].
 3362      [Steve Henson]
 3363 
 3364   *) In order to ensure interoperability SSL_OP_NO_protocolX does not
 3365      disable just protocol X, but all protocols above X *if* there are
 3366      protocols *below* X still enabled. In more practical terms it means
 3367      that if application wants to disable TLS1.0 in favor of TLS1.1 and
 3368      above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass
 3369      SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to
 3370      client side.
 3371      [Andy Polyakov]
 3372 
 3373  Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
 3374 
 3375   *) Check for potentially exploitable overflows in asn1_d2i_read_bio
 3376      BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
 3377      in CRYPTO_realloc_clean.
 3378 
 3379      Thanks to Tavis Ormandy, Google Security Team, for discovering this
 3380      issue and to Adam Langley <agl@chromium.org> for fixing it.
 3381      (CVE-2012-2110)
 3382      [Adam Langley (Google), Tavis Ormandy, Google Security Team]
 3383 
 3384   *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
 3385      [Adam Langley]
 3386 
 3387   *) Workarounds for some broken servers that "hang" if a client hello
 3388      record length exceeds 255 bytes.
 3389 
 3390      1. Do not use record version number > TLS 1.0 in initial client
 3391         hello: some (but not all) hanging servers will now work.
 3392      2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
 3393         the number of ciphers sent in the client hello. This should be
 3394         set to an even number, such as 50, for example by passing:
 3395         -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
 3396         Most broken servers should now work.
 3397      3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
 3398         TLS 1.2 client support entirely.
 3399      [Steve Henson]
 3400 
 3401   *) Fix SEGV in Vector Permutation AES module observed in OpenSSH.
 3402      [Andy Polyakov]
 3403 
 3404  Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
 3405 
 3406   *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET
 3407      STRING form instead of a DigestInfo.
 3408      [Steve Henson]
 3409 
 3410   *) The format used for MDC2 RSA signatures is inconsistent between EVP
 3411      and the RSA_sign/RSA_verify functions. This was made more apparent when
 3412      OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
 3413      those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect
 3414      the correct format in RSA_verify so both forms transparently work.
 3415      [Steve Henson]
 3416 
 3417   *) Some servers which support TLS 1.0 can choke if we initially indicate
 3418      support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
 3419      encrypted premaster secret. As a workaround use the maximum permitted
 3420      client version in client hello, this should keep such servers happy
 3421      and still work with previous versions of OpenSSL.
 3422      [Steve Henson]
 3423 
 3424   *) Add support for TLS/DTLS heartbeats.
 3425      [Robin Seggelmann <seggelmann@fh-muenster.de>]
 3426 
 3427   *) Add support for SCTP.
 3428      [Robin Seggelmann <seggelmann@fh-muenster.de>]
 3429 
 3430   *) Improved PRNG seeding for VOS.
 3431      [Paul Green <Paul.Green@stratus.com>]
 3432 
 3433   *) Extensive assembler packs updates, most notably:
 3434 
 3435         - x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
 3436         - x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
 3437         - x86_64:       bit-sliced AES implementation;
 3438         - ARM:          NEON support, contemporary platforms optimizations;
 3439         - s390x:        z196 support;
 3440         - *:            GHASH and GF(2^m) multiplication implementations;
 3441 
 3442      [Andy Polyakov]
 3443 
 3444   *) Make TLS-SRP code conformant with RFC 5054 API cleanup
 3445      (removal of unnecessary code)
 3446      [Peter Sylvester <peter.sylvester@edelweb.fr>]
 3447 
 3448   *) Add TLS key material exporter from RFC 5705.
 3449      [Eric Rescorla]
 3450 
 3451   *) Add DTLS-SRTP negotiation from RFC 5764.
 3452      [Eric Rescorla]
 3453 
 3454   *) Add Next Protocol Negotiation,
 3455      http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
 3456      disabled with a no-npn flag to config or Configure. Code donated
 3457      by Google.
 3458      [Adam Langley <agl@google.com> and Ben Laurie]
 3459 
 3460   *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
 3461      NIST-P256, NIST-P521, with constant-time single point multiplication on
 3462      typical inputs. Compiler support for the nonstandard type __uint128_t is
 3463      required to use this (present in gcc 4.4 and later, for 64-bit builds).
 3464      Code made available under Apache License version 2.0.
 3465 
 3466      Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
 3467      line to include this in your build of OpenSSL, and run "make depend" (or
 3468      "make update"). This enables the following EC_METHODs:
 3469 
 3470          EC_GFp_nistp224_method()
 3471          EC_GFp_nistp256_method()
 3472          EC_GFp_nistp521_method()
 3473 
 3474      EC_GROUP_new_by_curve_name() will automatically use these (while
 3475      EC_GROUP_new_curve_GFp() currently prefers the more flexible
 3476      implementations).
 3477      [Emilia Käsper, Adam Langley, Bodo Moeller (Google)]
 3478 
 3479   *) Use type ossl_ssize_t instad of ssize_t which isn't available on
 3480      all platforms. Move ssize_t definition from e_os.h to the public
 3481      header file e_os2.h as it now appears in public header file cms.h
 3482      [Steve Henson]
 3483 
 3484   *) New -sigopt option to the ca, req and x509 utilities. Additional
 3485      signature parameters can be passed using this option and in
 3486      particular PSS.
 3487      [Steve Henson]
 3488 
 3489   *) Add RSA PSS signing function. This will generate and set the
 3490      appropriate AlgorithmIdentifiers for PSS based on those in the
 3491      corresponding EVP_MD_CTX structure. No application support yet.
 3492      [Steve Henson]
 3493 
 3494   *) Support for companion algorithm specific ASN1 signing routines.
 3495      New function ASN1_item_sign_ctx() signs a pre-initialised
 3496      EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
 3497      the appropriate parameters.
 3498      [Steve Henson]
 3499 
 3500   *) Add new algorithm specific ASN1 verification initialisation function
 3501      to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
 3502      handling will be the same no matter what EVP_PKEY_METHOD is used.
 3503      Add a PSS handler to support verification of PSS signatures: checked
 3504      against a number of sample certificates.
 3505      [Steve Henson]
 3506 
 3507   *) Add signature printing for PSS. Add PSS OIDs.
 3508      [Steve Henson, Martin Kaiser <lists@kaiser.cx>]
 3509 
 3510   *) Add algorithm specific signature printing. An individual ASN1 method
 3511      can now print out signatures instead of the standard hex dump.
 3512 
 3513      More complex signatures (e.g. PSS) can print out more meaningful
 3514      information. Include DSA version that prints out the signature
 3515      parameters r, s.
 3516      [Steve Henson]
 3517 
 3518   *) Password based recipient info support for CMS library: implementing
 3519      RFC3211.
 3520      [Steve Henson]
 3521 
 3522   *) Split password based encryption into PBES2 and PBKDF2 functions. This
 3523      neatly separates the code into cipher and PBE sections and is required
 3524      for some algorithms that split PBES2 into separate pieces (such as
 3525      password based CMS).
 3526      [Steve Henson]
 3527 
 3528   *) Session-handling fixes:
 3529      - Fix handling of connections that are resuming with a session ID,
 3530        but also support Session Tickets.
 3531      - Fix a bug that suppressed issuing of a new ticket if the client
 3532        presented a ticket with an expired session.
 3533      - Try to set the ticket lifetime hint to something reasonable.
 3534      - Make tickets shorter by excluding irrelevant information.
 3535      - On the client side, don't ignore renewed tickets.
 3536      [Adam Langley, Bodo Moeller (Google)]
 3537 
 3538   *) Fix PSK session representation.
 3539      [Bodo Moeller]
 3540 
 3541   *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
 3542 
 3543      This work was sponsored by Intel.
 3544      [Andy Polyakov]
 3545 
 3546   *) Add GCM support to TLS library. Some custom code is needed to split
 3547      the IV between the fixed (from PRF) and explicit (from TLS record)
 3548      portions. This adds all GCM ciphersuites supported by RFC5288 and
 3549      RFC5289. Generalise some AES* cipherstrings to include GCM and
 3550      add a special AESGCM string for GCM only.
 3551      [Steve Henson]
 3552 
 3553   *) Expand range of ctrls for AES GCM. Permit setting invocation
 3554      field on decrypt and retrieval of invocation field only on encrypt.
 3555      [Steve Henson]
 3556 
 3557   *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
 3558      As required by RFC5289 these ciphersuites cannot be used if for
 3559      versions of TLS earlier than 1.2.
 3560      [Steve Henson]
 3561 
 3562   *) For FIPS capable OpenSSL interpret a NULL default public key method
 3563      as unset and return the appropriate default but do *not* set the default.
 3564      This means we can return the appropriate method in applications that
 3565      switch between FIPS and non-FIPS modes.
 3566      [Steve Henson]
 3567 
 3568   *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
 3569      ENGINE is used then we cannot handle that in the FIPS module so we
 3570      keep original code iff non-FIPS operations are allowed.
 3571      [Steve Henson]
 3572 
 3573   *) Add -attime option to openssl utilities.
 3574      [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson]
 3575 
 3576   *) Redirect DSA and DH operations to FIPS module in FIPS mode.
 3577      [Steve Henson]
 3578 
 3579   *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
 3580      FIPS EC methods unconditionally for now.
 3581      [Steve Henson]
 3582 
 3583   *) New build option no-ec2m to disable characteristic 2 code.
 3584      [Steve Henson]
 3585 
 3586   *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
 3587      all cases can be covered as some introduce binary incompatibilities.
 3588      [Steve Henson]
 3589 
 3590   *) Redirect RSA operations to FIPS module including keygen,
 3591      encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
 3592      [Steve Henson]
 3593 
 3594   *) Add similar low level API blocking to ciphers.
 3595      [Steve Henson]
 3596 
 3597   *) Low level digest APIs are not approved in FIPS mode: any attempt
 3598      to use these will cause a fatal error. Applications that *really* want
 3599      to use them can use the private_* version instead.
 3600      [Steve Henson]
 3601 
 3602   *) Redirect cipher operations to FIPS module for FIPS builds.
 3603      [Steve Henson]
 3604 
 3605   *) Redirect digest operations to FIPS module for FIPS builds.
 3606      [Steve Henson]
 3607 
 3608   *) Update build system to add "fips" flag which will link in fipscanister.o
 3609      for static and shared library builds embedding a signature if needed.
 3610      [Steve Henson]
 3611 
 3612   *) Output TLS supported curves in preference order instead of numerical
 3613      order. This is currently hardcoded for the highest order curves first.
 3614      This should be configurable so applications can judge speed vs strength.
 3615      [Steve Henson]
 3616 
 3617   *) Add TLS v1.2 server support for client authentication.
 3618      [Steve Henson]
 3619 
 3620   *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
 3621      and enable MD5.
 3622      [Steve Henson]
 3623 
 3624   *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
 3625      FIPS modules versions.
 3626      [Steve Henson]
 3627 
 3628   *) Add TLS v1.2 client side support for client authentication. Keep cache
 3629      of handshake records longer as we don't know the hash algorithm to use
 3630      until after the certificate request message is received.
 3631      [Steve Henson]
 3632 
 3633   *) Initial TLS v1.2 client support. Add a default signature algorithms
 3634      extension including all the algorithms we support. Parse new signature
 3635      format in client key exchange. Relax some ECC signing restrictions for
 3636      TLS v1.2 as indicated in RFC5246.
 3637      [Steve Henson]
 3638 
 3639   *) Add server support for TLS v1.2 signature algorithms extension. Switch
 3640      to new signature format when needed using client digest preference.
 3641      All server ciphersuites should now work correctly in TLS v1.2. No client
 3642      support yet and no support for client certificates.
 3643      [Steve Henson]
 3644 
 3645   *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
 3646      to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
 3647      ciphersuites. At present only RSA key exchange ciphersuites work with
 3648      TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
 3649      SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
 3650      and version checking.
 3651      [Steve Henson]
 3652 
 3653   *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
 3654      with this defined it will not be affected by any changes to ssl internal
 3655      structures. Add several utility functions to allow openssl application
 3656      to work with OPENSSL_NO_SSL_INTERN defined.
 3657      [Steve Henson]
 3658 
 3659   *) A long standing patch to add support for SRP from EdelWeb (Peter
 3660      Sylvester and Christophe Renou) was integrated.
 3661      [Christophe Renou <christophe.renou@edelweb.fr>, Peter Sylvester
 3662      <peter.sylvester@edelweb.fr>, Tom Wu <tjw@cs.stanford.edu>, and
 3663      Ben Laurie]
 3664 
 3665   *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
 3666      [Steve Henson]
 3667 
 3668   *) Permit abbreviated handshakes when renegotiating using the function
 3669      SSL_renegotiate_abbreviated().
 3670      [Robin Seggelmann <seggelmann@fh-muenster.de>]
 3671 
 3672   *) Add call to ENGINE_register_all_complete() to
 3673      ENGINE_load_builtin_engines(), so some implementations get used
 3674      automatically instead of needing explicit application support.
 3675      [Steve Henson]
 3676 
 3677   *) Add support for TLS key exporter as described in RFC5705.
 3678      [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson]
 3679 
 3680   *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
 3681      a few changes are required:
 3682 
 3683        Add SSL_OP_NO_TLSv1_1 flag.
 3684        Add TLSv1_1 methods.
 3685        Update version checking logic to handle version 1.1.
 3686        Add explicit IV handling (ported from DTLS code).
 3687        Add command line options to s_client/s_server.
 3688      [Steve Henson]
 3689 
 3690  Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
 3691 
 3692   *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
 3693      in CMS and PKCS7 code. When RSA decryption fails use a random key for
 3694      content decryption and always return the same error. Note: this attack
 3695      needs on average 2^20 messages so it only affects automated senders. The
 3696      old behaviour can be re-enabled in the CMS code by setting the
 3697      CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
 3698      an MMA defence is not necessary.
 3699      Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
 3700      this issue. (CVE-2012-0884)
 3701      [Steve Henson]
 3702 
 3703   *) Fix CVE-2011-4619: make sure we really are receiving a
 3704      client hello before rejecting multiple SGC restarts. Thanks to
 3705      Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
 3706      [Steve Henson]
 3707 
 3708  Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
 3709 
 3710   *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
 3711      Thanks to Antonio Martin, Enterprise Secure Access Research and
 3712      Development, Cisco Systems, Inc. for discovering this bug and
 3713      preparing a fix. (CVE-2012-0050)
 3714      [Antonio Martin]
 3715 
 3716  Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
 3717 
 3718   *) Nadhem Alfardan and Kenny Paterson have discovered an extension
 3719      of the Vaudenay padding oracle attack on CBC mode encryption
 3720      which enables an efficient plaintext recovery attack against
 3721      the OpenSSL implementation of DTLS. Their attack exploits timing
 3722      differences arising during decryption processing. A research
 3723      paper describing this attack can be found at:
 3724                   http://www.isg.rhul.ac.uk/~kp/dtls.pdf
 3725      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
 3726      Security Group at Royal Holloway, University of London
 3727      (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
 3728      <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
 3729      for preparing the fix. (CVE-2011-4108)
 3730      [Robin Seggelmann, Michael Tuexen]
 3731 
 3732   *) Clear bytes used for block padding of SSL 3.0 records.
 3733      (CVE-2011-4576)
 3734      [Adam Langley (Google)]
 3735 
 3736   *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
 3737      Kadianakis <desnacked@gmail.com> for discovering this issue and
 3738      Adam Langley for preparing the fix. (CVE-2011-4619)
 3739      [Adam Langley (Google)]
 3740 
 3741   *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027)
 3742      [Andrey Kulikov <amdeich@gmail.com>]
 3743 
 3744   *) Prevent malformed RFC3779 data triggering an assertion failure.
 3745      Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
 3746      and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577)
 3747      [Rob Austein <sra@hactrn.net>]
 3748 
 3749   *) Improved PRNG seeding for VOS.
 3750      [Paul Green <Paul.Green@stratus.com>]
 3751 
 3752   *) Fix ssl_ciph.c set-up race.
 3753      [Adam Langley (Google)]
 3754 
 3755   *) Fix spurious failures in ecdsatest.c.
 3756      [Emilia Käsper (Google)]
 3757 
 3758   *) Fix the BIO_f_buffer() implementation (which was mixing different
 3759      interpretations of the '..._len' fields).
 3760      [Adam Langley (Google)]
 3761 
 3762   *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
 3763      BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
 3764      threads won't reuse the same blinding coefficients.
 3765 
 3766      This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
 3767      lock to call BN_BLINDING_invert_ex, and avoids one use of
 3768      BN_BLINDING_update for each BN_BLINDING structure (previously,
 3769      the last update always remained unused).
 3770      [Emilia Käsper (Google)]
 3771 
 3772   *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
 3773      [Bob Buckholz (Google)]
 3774 
 3775  Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
 3776 
 3777   *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
 3778      by initialising X509_STORE_CTX properly. (CVE-2011-3207)
 3779      [Kaspar Brand <ossl@velox.ch>]
 3780 
 3781   *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
 3782      for multi-threaded use of ECDH. (CVE-2011-3210)
 3783      [Adam Langley (Google)]
 3784 
 3785   *) Fix x509_name_ex_d2i memory leak on bad inputs.
 3786      [Bodo Moeller]
 3787 
 3788   *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
 3789      signature public key algorithm by using OID xref utilities instead.
 3790      Before this you could only use some ECC ciphersuites with SHA1 only.
 3791      [Steve Henson]
 3792 
 3793   *) Add protection against ECDSA timing attacks as mentioned in the paper
 3794      by Billy Bob Brumley and Nicola Tuveri, see:
 3795 
 3796         http://eprint.iacr.org/2011/232.pdf
 3797 
 3798      [Billy Bob Brumley and Nicola Tuveri]
 3799 
 3800  Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
 3801 
 3802   *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
 3803      [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
 3804 
 3805   *) Fix bug in string printing code: if *any* escaping is enabled we must
 3806      escape the escape character (backslash) or the resulting string is
 3807      ambiguous.
 3808      [Steve Henson]
 3809 
 3810  Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
 3811 
 3812   *) Disable code workaround for ancient and obsolete Netscape browsers
 3813      and servers: an attacker can use it in a ciphersuite downgrade attack.
 3814      Thanks to Martin Rex for discovering this bug. CVE-2010-4180
 3815      [Steve Henson]
 3816 
 3817   *) Fixed J-PAKE implementation error, originally discovered by
 3818      Sebastien Martini, further info and confirmation from Stefan
 3819      Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
 3820      [Ben Laurie]
 3821 
 3822  Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
 3823 
 3824   *) Fix extension code to avoid race conditions which can result in a buffer
 3825      overrun vulnerability: resumed sessions must not be modified as they can
 3826      be shared by multiple threads. CVE-2010-3864
 3827      [Steve Henson]
 3828 
 3829   *) Fix WIN32 build system to correctly link an ENGINE directory into
 3830      a DLL.
 3831      [Steve Henson]
 3832 
 3833  Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
 3834 
 3835   *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover
 3836      (CVE-2010-1633)
 3837      [Steve Henson, Peter-Michael Hager <hager@dortmund.net>]
 3838 
 3839  Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
 3840 
 3841   *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
 3842      context. The operation can be customised via the ctrl mechanism in
 3843      case ENGINEs want to include additional functionality.
 3844      [Steve Henson]
 3845 
 3846   *) Tolerate yet another broken PKCS#8 key format: private key value negative.
 3847      [Steve Henson]
 3848 
 3849   *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
 3850      output hashes compatible with older versions of OpenSSL.
 3851      [Willy Weisz <weisz@vcpc.univie.ac.at>]
 3852 
 3853   *) Fix compression algorithm handling: if resuming a session use the
 3854      compression algorithm of the resumed session instead of determining
 3855      it from client hello again. Don't allow server to change algorithm.
 3856      [Steve Henson]
 3857 
 3858   *) Add load_crls() function to apps tidying load_certs() too. Add option
 3859      to verify utility to allow additional CRLs to be included.
 3860      [Steve Henson]
 3861 
 3862   *) Update OCSP request code to permit adding custom headers to the request:
 3863      some responders need this.
 3864      [Steve Henson]
 3865 
 3866   *) The function EVP_PKEY_sign() returns <=0 on error: check return code
 3867      correctly.
 3868      [Julia Lawall <julia@diku.dk>]
 3869 
 3870   *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
 3871      needlessly dereferenced structures, used obsolete functions and
 3872      didn't handle all updated verify codes correctly.
 3873      [Steve Henson]
 3874 
 3875   *) Disable MD2 in the default configuration.
 3876      [Steve Henson]
 3877 
 3878   *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
 3879      indicate the initial BIO being pushed or popped. This makes it possible
 3880      to determine whether the BIO is the one explicitly called or as a result
 3881      of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
 3882      it handles reference counts correctly and doesn't zero out the I/O bio
 3883      when it is not being explicitly popped. WARNING: applications which
 3884      included workarounds for the old buggy behaviour will need to be modified
 3885      or they could free up already freed BIOs.
 3886      [Steve Henson]
 3887 
 3888   *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
 3889      renaming to all platforms (within the 0.9.8 branch, this was
 3890      done conditionally on Netware platforms to avoid a name clash).
 3891      [Guenter <lists@gknw.net>]
 3892 
 3893   *) Add ECDHE and PSK support to DTLS.
 3894      [Michael Tuexen <tuexen@fh-muenster.de>]
 3895 
 3896   *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
 3897      be used on C++.
 3898      [Steve Henson]
 3899 
 3900   *) Add "missing" function EVP_MD_flags() (without this the only way to
 3901      retrieve a digest flags is by accessing the structure directly. Update
 3902      EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
 3903      or cipher is registered as in the "from" argument. Print out all
 3904      registered digests in the dgst usage message instead of manually
 3905      attempting to work them out.
 3906      [Steve Henson]
 3907 
 3908   *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
 3909      this allows the use of compression and extensions. Change default cipher
 3910      string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
 3911      by default unless an application cipher string requests it.
 3912      [Steve Henson]
 3913 
 3914   *) Alter match criteria in PKCS12_parse(). It used to try to use local
 3915      key ids to find matching certificates and keys but some PKCS#12 files
 3916      don't follow the (somewhat unwritten) rules and this strategy fails.
 3917      Now just gather all certificates together and the first private key
 3918      then look for the first certificate that matches the key.
 3919      [Steve Henson]
 3920 
 3921   *) Support use of registered digest and cipher names for dgst and cipher
 3922      commands instead of having to add each one as a special case. So now
 3923      you can do:
 3924 
 3925         openssl sha256 foo
 3926 
 3927      as well as:
 3928 
 3929         openssl dgst -sha256 foo
 3930 
 3931      and this works for ENGINE based algorithms too.
 3932 
 3933      [Steve Henson]
 3934 
 3935   *) Update Gost ENGINE to support parameter files.
 3936      [Victor B. Wagner <vitus@cryptocom.ru>]
 3937 
 3938   *) Support GeneralizedTime in ca utility.
 3939      [Oliver Martin <oliver@volatilevoid.net>, Steve Henson]
 3940 
 3941   *) Enhance the hash format used for certificate directory links. The new
 3942      form uses the canonical encoding (meaning equivalent names will work
 3943      even if they aren't identical) and uses SHA1 instead of MD5. This form
 3944      is incompatible with the older format and as a result c_rehash should
 3945      be used to rebuild symbolic links.
 3946      [Steve Henson]
 3947 
 3948   *) Make PKCS#8 the default write format for private keys, replacing the
 3949      traditional format. This form is standardised, more secure and doesn't
 3950      include an implicit MD5 dependency.
 3951      [Steve Henson]
 3952 
 3953   *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
 3954      committed to OpenSSL should pass this lot as a minimum.
 3955      [Steve Henson]
 3956 
 3957   *) Add session ticket override functionality for use by EAP-FAST.
 3958      [Jouni Malinen <j@w1.fi>]
 3959 
 3960   *) Modify HMAC functions to return a value. Since these can be implemented
 3961      in an ENGINE errors can occur.
 3962      [Steve Henson]
 3963 
 3964   *) Type-checked OBJ_bsearch_ex.
 3965      [Ben Laurie]
 3966 
 3967   *) Type-checked OBJ_bsearch. Also some constification necessitated
 3968      by type-checking.  Still to come: TXT_DB, bsearch(?),
 3969      OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
 3970      CONF_VALUE.
 3971      [Ben Laurie]
 3972 
 3973   *) New function OPENSSL_gmtime_adj() to add a specific number of days and
 3974      seconds to a tm structure directly, instead of going through OS
 3975      specific date routines. This avoids any issues with OS routines such
 3976      as the year 2038 bug. New *_adj() functions for ASN1 time structures
 3977      and X509_time_adj_ex() to cover the extended range. The existing
 3978      X509_time_adj() is still usable and will no longer have any date issues.
 3979      [Steve Henson]
 3980 
 3981   *) Delta CRL support. New use deltas option which will attempt to locate
 3982      and search any appropriate delta CRLs available.
 3983 
 3984      This work was sponsored by Google.
 3985      [Steve Henson]
 3986 
 3987   *) Support for CRLs partitioned by reason code. Reorganise CRL processing
 3988      code and add additional score elements. Validate alternate CRL paths
 3989      as part of the CRL checking and indicate a new error "CRL path validation
 3990      error" in this case. Applications wanting additional details can use
 3991      the verify callback and check the new "parent" field. If this is not
 3992      NULL CRL path validation is taking place. Existing applications won't
 3993      see this because it requires extended CRL support which is off by
 3994      default.
 3995 
 3996      This work was sponsored by Google.
 3997      [Steve Henson]
 3998 
 3999   *) Support for freshest CRL extension.
 4000 
 4001      This work was sponsored by Google.
 4002      [Steve Henson]
 4003 
 4004   *) Initial indirect CRL support. Currently only supported in the CRLs
 4005      passed directly and not via lookup. Process certificate issuer
 4006      CRL entry extension and lookup CRL entries by bother issuer name
 4007      and serial number. Check and process CRL issuer entry in IDP extension.
 4008 
 4009      This work was sponsored by Google.
 4010      [Steve Henson]
 4011 
 4012   *) Add support for distinct certificate and CRL paths. The CRL issuer
 4013      certificate is validated separately in this case. Only enabled if
 4014      an extended CRL support flag is set: this flag will enable additional
 4015      CRL functionality in future.
 4016 
 4017      This work was sponsored by Google.
 4018      [Steve Henson]
 4019 
 4020   *) Add support for policy mappings extension.
 4021 
 4022      This work was sponsored by Google.
 4023      [Steve Henson]
 4024 
 4025   *) Fixes to pathlength constraint, self issued certificate handling,
 4026      policy processing to align with RFC3280 and PKITS tests.
 4027 
 4028      This work was sponsored by Google.
 4029      [Steve Henson]
 4030 
 4031   *) Support for name constraints certificate extension. DN, email, DNS
 4032      and URI types are currently supported.
 4033 
 4034      This work was sponsored by Google.
 4035      [Steve Henson]
 4036 
 4037   *) To cater for systems that provide a pointer-based thread ID rather
 4038      than numeric, deprecate the current numeric thread ID mechanism and
 4039      replace it with a structure and associated callback type. This
 4040      mechanism allows a numeric "hash" to be extracted from a thread ID in
 4041      either case, and on platforms where pointers are larger than 'long',
 4042      mixing is done to help ensure the numeric 'hash' is usable even if it
 4043      can't be guaranteed unique. The default mechanism is to use "&errno"
 4044      as a pointer-based thread ID to distinguish between threads.
 4045 
 4046      Applications that want to provide their own thread IDs should now use
 4047      CRYPTO_THREADID_set_callback() to register a callback that will call
 4048      either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
 4049 
 4050      Note that ERR_remove_state() is now deprecated, because it is tied
 4051      to the assumption that thread IDs are numeric.  ERR_remove_state(0)
 4052      to free the current thread's error state should be replaced by
 4053      ERR_remove_thread_state(NULL).
 4054 
 4055      (This new approach replaces the functions CRYPTO_set_idptr_callback(),
 4056      CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
 4057      OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
 4058      application was previously providing a numeric thread callback that
 4059      was inappropriate for distinguishing threads, then uniqueness might
 4060      have been obtained with &errno that happened immediately in the
 4061      intermediate development versions of OpenSSL; this is no longer the
 4062      case, the numeric thread callback will now override the automatic use
 4063      of &errno.)
 4064      [Geoff Thorpe, with help from Bodo Moeller]
 4065 
 4066   *) Initial support for different CRL issuing certificates. This covers a
 4067      simple case where the self issued certificates in the chain exist and
 4068      the real CRL issuer is higher in the existing chain.
 4069 
 4070      This work was sponsored by Google.
 4071      [Steve Henson]
 4072 
 4073   *) Removed effectively defunct crypto/store from the build.
 4074      [Ben Laurie]
 4075 
 4076   *) Revamp of STACK to provide stronger type-checking. Still to come:
 4077      TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
 4078      ASN1_STRING, CONF_VALUE.
 4079      [Ben Laurie]
 4080 
 4081   *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
 4082      RAM on SSL connections.  This option can save about 34k per idle SSL.
 4083      [Nick Mathewson]
 4084 
 4085   *) Revamp of LHASH to provide stronger type-checking. Still to come:
 4086      STACK, TXT_DB, bsearch, qsort.
 4087      [Ben Laurie]
 4088 
 4089   *) Initial support for Cryptographic Message Syntax (aka CMS) based
 4090      on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
 4091      support for data, signedData, compressedData, digestedData and
 4092      encryptedData, envelopedData types included. Scripts to check against
 4093      RFC4134 examples draft and interop and consistency checks of many
 4094      content types and variants.
 4095      [Steve Henson]
 4096 
 4097   *) Add options to enc utility to support use of zlib compression BIO.
 4098      [Steve Henson]
 4099 
 4100   *) Extend mk1mf to support importing of options and assembly language
 4101      files from Configure script, currently only included in VC-WIN32.
 4102      The assembly language rules can now optionally generate the source
 4103      files from the associated perl scripts.
 4104      [Steve Henson]
 4105 
 4106   *) Implement remaining functionality needed to support GOST ciphersuites.
 4107      Interop testing has been performed using CryptoPro implementations.
 4108      [Victor B. Wagner <vitus@cryptocom.ru>]
 4109 
 4110   *) s390x assembler pack.
 4111      [Andy Polyakov]
 4112 
 4113   *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
 4114      "family."
 4115      [Andy Polyakov]
 4116 
 4117   *) Implement Opaque PRF Input TLS extension as specified in
 4118      draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
 4119      official specification yet and no extension type assignment by
 4120      IANA exists, this extension (for now) will have to be explicitly
 4121      enabled when building OpenSSL by providing the extension number
 4122      to use.  For example, specify an option
 4123 
 4124          -DTLSEXT_TYPE_opaque_prf_input=0x9527
 4125 
 4126      to the "config" or "Configure" script to enable the extension,
 4127      assuming extension number 0x9527 (which is a completely arbitrary
 4128      and unofficial assignment based on the MD5 hash of the Internet
 4129      Draft).  Note that by doing so, you potentially lose
 4130      interoperability with other TLS implementations since these might
 4131      be using the same extension number for other purposes.
 4132 
 4133      SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
 4134      opaque PRF input value to use in the handshake.  This will create
 4135      an internal copy of the length-'len' string at 'src', and will
 4136      return non-zero for success.
 4137 
 4138      To get more control and flexibility, provide a callback function
 4139      by using
 4140 
 4141           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
 4142           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
 4143 
 4144      where
 4145 
 4146           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
 4147           void *arg;
 4148 
 4149      Callback function 'cb' will be called in handshakes, and is
 4150      expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
 4151      Argument 'arg' is for application purposes (the value as given to
 4152      SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
 4153      be provided to the callback function).  The callback function
 4154      has to return non-zero to report success: usually 1 to use opaque
 4155      PRF input just if possible, or 2 to enforce use of the opaque PRF
 4156      input.  In the latter case, the library will abort the handshake
 4157      if opaque PRF input is not successfully negotiated.
 4158 
 4159      Arguments 'peerinput' and 'len' given to the callback function
 4160      will always be NULL and 0 in the case of a client.  A server will
 4161      see the client's opaque PRF input through these variables if
 4162      available (NULL and 0 otherwise).  Note that if the server
 4163      provides an opaque PRF input, the length must be the same as the
 4164      length of the client's opaque PRF input.
 4165 
 4166      Note that the callback function will only be called when creating
 4167      a new session (session resumption can resume whatever was
 4168      previously negotiated), and will not be called in SSL 2.0
 4169      handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
 4170      SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
 4171      for applications that need to enforce opaque PRF input.
 4172 
 4173      [Bodo Moeller]
 4174 
 4175   *) Update ssl code to support digests other than SHA1+MD5 for handshake
 4176      MAC.
 4177 
 4178      [Victor B. Wagner <vitus@cryptocom.ru>]
 4179 
 4180   *) Add RFC4507 support to OpenSSL. This includes the corrections in
 4181      RFC4507bis. The encrypted ticket format is an encrypted encoded
 4182      SSL_SESSION structure, that way new session features are automatically
 4183      supported.
 4184 
 4185      If a client application caches session in an SSL_SESSION structure
 4186      support is transparent because tickets are now stored in the encoded
 4187      SSL_SESSION.
 4188 
 4189      The SSL_CTX structure automatically generates keys for ticket
 4190      protection in servers so again support should be possible
 4191      with no application modification.
 4192 
 4193      If a client or server wishes to disable RFC4507 support then the option
 4194      SSL_OP_NO_TICKET can be set.
 4195 
 4196      Add a TLS extension debugging callback to allow the contents of any client
 4197      or server extensions to be examined.
 4198 
 4199      This work was sponsored by Google.
 4200      [Steve Henson]
 4201 
 4202   *) Final changes to avoid use of pointer pointer casts in OpenSSL.
 4203      OpenSSL should now compile cleanly on gcc 4.2
 4204      [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson]
 4205 
 4206   *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
 4207      support including streaming MAC support: this is required for GOST
 4208      ciphersuite support.
 4209      [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson]
 4210 
 4211   *) Add option -stream to use PKCS#7 streaming in smime utility. New
 4212      function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
 4213      to output in BER and PEM format.
 4214      [Steve Henson]
 4215 
 4216   *) Experimental support for use of HMAC via EVP_PKEY interface. This
 4217      allows HMAC to be handled via the EVP_DigestSign*() interface. The
 4218      EVP_PKEY "key" in this case is the HMAC key, potentially allowing
 4219      ENGINE support for HMAC keys which are unextractable. New -mac and
 4220      -macopt options to dgst utility.
 4221      [Steve Henson]
 4222 
 4223   *) New option -sigopt to dgst utility. Update dgst to use
 4224      EVP_Digest{Sign,Verify}*. These two changes make it possible to use
 4225      alternative signing parameters such as X9.31 or PSS in the dgst
 4226      utility.
 4227      [Steve Henson]
 4228 
 4229   *) Change ssl_cipher_apply_rule(), the internal function that does
 4230      the work each time a ciphersuite string requests enabling
 4231      ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
 4232      removing ("!foo+bar") a class of ciphersuites: Now it maintains
 4233      the order of disabled ciphersuites such that those ciphersuites
 4234      that most recently went from enabled to disabled not only stay
 4235      in order with respect to each other, but also have higher priority
 4236      than other disabled ciphersuites the next time ciphersuites are
 4237      enabled again.
 4238 
 4239      This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
 4240      the same ciphersuites as with "HIGH" alone, but in a specific
 4241      order where the PSK ciphersuites come first (since they are the
 4242      most recently disabled ciphersuites when "HIGH" is parsed).
 4243 
 4244      Also, change ssl_create_cipher_list() (using this new
 4245      functionality) such that between otherwise identical
 4246      ciphersuites, ephemeral ECDH is preferred over ephemeral DH in
 4247      the default order.
 4248      [Bodo Moeller]
 4249 
 4250   *) Change ssl_create_cipher_list() so that it automatically
 4251      arranges the ciphersuites in reasonable order before starting
 4252      to process the rule string.  Thus, the definition for "DEFAULT"
 4253      (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
 4254      remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH".
 4255      This makes it much easier to arrive at a reasonable default order
 4256      in applications for which anonymous ciphers are OK (meaning
 4257      that you can't actually use DEFAULT).
 4258      [Bodo Moeller; suggested by Victor Duchovni]
 4259 
 4260   *) Split the SSL/TLS algorithm mask (as used for ciphersuite string
 4261      processing) into multiple integers instead of setting
 4262      "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
 4263      "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
 4264      (These masks as well as the individual bit definitions are hidden
 4265      away into the non-exported interface ssl/ssl_locl.h, so this
 4266      change to the definition of the SSL_CIPHER structure shouldn't
 4267      affect applications.)  This give us more bits for each of these
 4268      categories, so there is no longer a need to coagulate AES128 and
 4269      AES256 into a single algorithm bit, and to coagulate Camellia128
 4270      and Camellia256 into a single algorithm bit, which has led to all
 4271      kinds of kludges.
 4272 
 4273      Thus, among other things, the kludge introduced in 0.9.7m and
 4274      0.9.8e for masking out AES256 independently of AES128 or masking
 4275      out Camellia256 independently of AES256 is not needed here in 0.9.9.
 4276 
 4277      With the change, we also introduce new ciphersuite aliases that
 4278      so far were missing: "AES128", "AES256", "CAMELLIA128", and
 4279      "CAMELLIA256".
 4280      [Bodo Moeller]
 4281 
 4282   *) Add support for dsa-with-SHA224 and dsa-with-SHA256.
 4283      Use the leftmost N bytes of the signature input if the input is
 4284      larger than the prime q (with N being the size in bytes of q).
 4285      [Nils Larsch]
 4286 
 4287   *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
 4288      it yet and it is largely untested.
 4289      [Steve Henson]
 4290 
 4291   *) Add support for the ecdsa-with-SHA224/256/384/512 signature types.
 4292      [Nils Larsch]
 4293 
 4294   *) Initial incomplete changes to avoid need for function casts in OpenSSL
 4295      some compilers (gcc 4.2 and later) reject their use. Safestack is
 4296      reimplemented.  Update ASN1 to avoid use of legacy functions.
 4297      [Steve Henson]
 4298 
 4299   *) Win32/64 targets are linked with Winsock2.
 4300      [Andy Polyakov]
 4301 
 4302   *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
 4303      to external functions. This can be used to increase CRL handling
 4304      efficiency especially when CRLs are very large by (for example) storing
 4305      the CRL revoked certificates in a database.
 4306      [Steve Henson]
 4307 
 4308   *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so
 4309      new CRLs added to a directory can be used. New command line option
 4310      -verify_return_error to s_client and s_server. This causes real errors
 4311      to be returned by the verify callback instead of carrying on no matter
 4312      what. This reflects the way a "real world" verify callback would behave.
 4313      [Steve Henson]
 4314 
 4315   *) GOST engine, supporting several GOST algorithms and public key formats.
 4316      Kindly donated by Cryptocom.
 4317      [Cryptocom]
 4318 
 4319   *) Partial support for Issuing Distribution Point CRL extension. CRLs
 4320      partitioned by DP are handled but no indirect CRL or reason partitioning
 4321      (yet). Complete overhaul of CRL handling: now the most suitable CRL is
 4322      selected via a scoring technique which handles IDP and AKID in CRLs.
 4323      [Steve Henson]
 4324 
 4325   *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
 4326      will ultimately be used for all verify operations: this will remove the
 4327      X509_STORE dependency on certificate verification and allow alternative
 4328      lookup methods.  X509_STORE based implementations of these two callbacks.
 4329      [Steve Henson]
 4330 
 4331   *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
 4332      Modify get_crl() to find a valid (unexpired) CRL if possible.
 4333      [Steve Henson]
 4334 
 4335   *) New function X509_CRL_match() to check if two CRLs are identical. Normally
 4336      this would be called X509_CRL_cmp() but that name is already used by
 4337      a function that just compares CRL issuer names. Cache several CRL
 4338      extensions in X509_CRL structure and cache CRLDP in X509.
 4339      [Steve Henson]
 4340 
 4341   *) Store a "canonical" representation of X509_NAME structure (ASN1 Name)
 4342      this maps equivalent X509_NAME structures into a consistent structure.
 4343      Name comparison can then be performed rapidly using memcmp().
 4344      [Steve Henson]
 4345 
 4346   *) Non-blocking OCSP request processing. Add -timeout option to ocsp
 4347      utility.
 4348      [Steve Henson]
 4349 
 4350   *) Allow digests to supply their own micalg string for S/MIME type using
 4351      the ctrl EVP_MD_CTRL_MICALG.
 4352      [Steve Henson]
 4353 
 4354   *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the
 4355      EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
 4356      ctrl. It can then customise the structure before and/or after signing
 4357      if necessary.
 4358      [Steve Henson]
 4359 
 4360   *) New function OBJ_add_sigid() to allow application defined signature OIDs
 4361      to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
 4362      to free up any added signature OIDs.
 4363      [Steve Henson]
 4364 
 4365   *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
 4366      EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
 4367      digest and cipher tables. New options added to openssl utility:
 4368      list-message-digest-algorithms and list-cipher-algorithms.
 4369      [Steve Henson]
 4370 
 4371   *) Change the array representation of binary polynomials: the list
 4372      of degrees of non-zero coefficients is now terminated with -1.
 4373      Previously it was terminated with 0, which was also part of the
 4374      value; thus, the array representation was not applicable to
 4375      polynomials where t^0 has coefficient zero.  This change makes
 4376      the array representation useful in a more general context.
 4377      [Douglas Stebila]
 4378 
 4379   *) Various modifications and fixes to SSL/TLS cipher string
 4380      handling.  For ECC, the code now distinguishes between fixed ECDH
 4381      with RSA certificates on the one hand and with ECDSA certificates
 4382      on the other hand, since these are separate ciphersuites.  The
 4383      unused code for Fortezza ciphersuites has been removed.
 4384 
 4385      For consistency with EDH, ephemeral ECDH is now called "EECDH"
 4386      (not "ECDHE").  For consistency with the code for DH
 4387      certificates, use of ECDH certificates is now considered ECDH
 4388      authentication, not RSA or ECDSA authentication (the latter is
 4389      merely the CA's signing algorithm and not actively used in the
 4390      protocol).
 4391 
 4392      The temporary ciphersuite alias "ECCdraft" is no longer
 4393      available, and ECC ciphersuites are no longer excluded from "ALL"
 4394      and "DEFAULT".  The following aliases now exist for RFC 4492
 4395      ciphersuites, most of these by analogy with the DH case:
 4396 
 4397          kECDHr   - ECDH cert, signed with RSA
 4398          kECDHe   - ECDH cert, signed with ECDSA
 4399          kECDH    - ECDH cert (signed with either RSA or ECDSA)
 4400          kEECDH   - ephemeral ECDH
 4401          ECDH     - ECDH cert or ephemeral ECDH
 4402 
 4403          aECDH    - ECDH cert
 4404          aECDSA   - ECDSA cert
 4405          ECDSA    - ECDSA cert
 4406 
 4407          AECDH    - anonymous ECDH
 4408          EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
 4409 
 4410      [Bodo Moeller]
 4411 
 4412   *) Add additional S/MIME capabilities for AES and GOST ciphers if supported.
 4413      Use correct micalg parameters depending on digest(s) in signed message.
 4414      [Steve Henson]
 4415 
 4416   *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
 4417      an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
 4418      [Steve Henson]
 4419 
 4420   *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
 4421      an engine to register a method. Add ENGINE lookups for methods and
 4422      functional reference processing.
 4423      [Steve Henson]
 4424 
 4425   *) New functions EVP_Digest{Sign,Verify)*. These are enhanced versions of
 4426      EVP_{Sign,Verify}* which allow an application to customise the signature
 4427      process.
 4428      [Steve Henson]
 4429 
 4430   *) New -resign option to smime utility. This adds one or more signers
 4431      to an existing PKCS#7 signedData structure. Also -md option to use an
 4432      alternative message digest algorithm for signing.
 4433      [Steve Henson]
 4434 
 4435   *) Tidy up PKCS#7 routines and add new functions to make it easier to
 4436      create PKCS7 structures containing multiple signers. Update smime
 4437      application to support multiple signers.
 4438      [Steve Henson]
 4439 
 4440   *) New -macalg option to pkcs12 utility to allow setting of an alternative
 4441      digest MAC.
 4442      [Steve Henson]
 4443 
 4444   *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
 4445      Reorganize PBE internals to lookup from a static table using NIDs,
 4446      add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
 4447      EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
 4448      PRF which will be automatically used with PBES2.
 4449      [Steve Henson]
 4450 
 4451   *) Replace the algorithm specific calls to generate keys in "req" with the
 4452      new API.
 4453      [Steve Henson]
 4454 
 4455   *) Update PKCS#7 enveloped data routines to use new API. This is now
 4456      supported by any public key method supporting the encrypt operation. A
 4457      ctrl is added to allow the public key algorithm to examine or modify
 4458      the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
 4459      a no op.
 4460      [Steve Henson]
 4461 
 4462   *) Add a ctrl to asn1 method to allow a public key algorithm to express
 4463      a default digest type to use. In most cases this will be SHA1 but some
 4464      algorithms (such as GOST) need to specify an alternative digest. The
 4465      return value indicates how strong the preference is 1 means optional and
 4466      2 is mandatory (that is it is the only supported type). Modify
 4467      ASN1_item_sign() to accept a NULL digest argument to indicate it should
 4468      use the default md. Update openssl utilities to use the default digest
 4469      type for signing if it is not explicitly indicated.
 4470      [Steve Henson]
 4471 
 4472   *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New
 4473      EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
 4474      signing method from the key type. This effectively removes the link
 4475      between digests and public key types.
 4476      [Steve Henson]
 4477 
 4478   *) Add an OID cross reference table and utility functions. Its purpose is to
 4479      translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
 4480      rsaEncryption. This will allow some of the algorithm specific hackery
 4481      needed to use the correct OID to be removed.
 4482      [Steve Henson]
 4483 
 4484   *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
 4485      structures for PKCS7_sign(). They are now set up by the relevant public
 4486      key ASN1 method.
 4487      [Steve Henson]
 4488 
 4489   *) Add provisional EC pkey method with support for ECDSA and ECDH.
 4490      [Steve Henson]
 4491 
 4492   *) Add support for key derivation (agreement) in the API, DH method and
 4493      pkeyutl.
 4494      [Steve Henson]
 4495 
 4496   *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
 4497      public and private key formats. As a side effect these add additional
 4498      command line functionality not previously available: DSA signatures can be
 4499      generated and verified using pkeyutl and DH key support and generation in
 4500      pkey, genpkey.
 4501      [Steve Henson]
 4502 
 4503   *) BeOS support.
 4504      [Oliver Tappe <zooey@hirschkaefer.de>]
 4505 
 4506   *) New make target "install_html_docs" installs HTML renditions of the
 4507      manual pages.
 4508      [Oliver Tappe <zooey@hirschkaefer.de>]
 4509 
 4510   *) New utility "genpkey" this is analogous to "genrsa" etc except it can
 4511      generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
 4512      support key and parameter generation and add initial key generation
 4513      functionality for RSA.
 4514      [Steve Henson]
 4515 
 4516   *) Add functions for main EVP_PKEY_method operations. The undocumented
 4517      functions EVP_PKEY_{encrypt,decrypt} have been renamed to
 4518      EVP_PKEY_{encrypt,decrypt}_old.
 4519      [Steve Henson]
 4520 
 4521   *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public
 4522      key API, doesn't do much yet.
 4523      [Steve Henson]
 4524 
 4525   *) New function EVP_PKEY_asn1_get0_info() to retrieve information about
 4526      public key algorithms. New option to openssl utility:
 4527      "list-public-key-algorithms" to print out info.
 4528      [Steve Henson]
 4529 
 4530   *) Implement the Supported Elliptic Curves Extension for
 4531      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
 4532      [Douglas Stebila]
 4533 
 4534   *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
 4535      EVP_CIPHER structures to avoid later problems in EVP_cleanup().
 4536      [Steve Henson]
 4537 
 4538   *) New utilities pkey and pkeyparam. These are similar to algorithm specific
 4539      utilities such as rsa, dsa, dsaparam etc except they process any key
 4540      type.
 4541      [Steve Henson]
 4542 
 4543   *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New
 4544      functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
 4545      EVP_PKEY_print_param() to print public key data from an EVP_PKEY
 4546      structure.
 4547      [Steve Henson]
 4548 
 4549   *) Initial support for pluggable public key ASN1.
 4550      De-spaghettify the public key ASN1 handling. Move public and private
 4551      key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
 4552      algorithm specific handling to a single module within the relevant
 4553      algorithm directory. Add functions to allow (near) opaque processing
 4554      of public and private key structures.
 4555      [Steve Henson]
 4556 
 4557   *) Implement the Supported Point Formats Extension for
 4558      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
 4559      [Douglas Stebila]
 4560 
 4561   *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
 4562      for the psk identity [hint] and the psk callback functions to the
 4563      SSL_SESSION, SSL and SSL_CTX structure.
 4564 
 4565      New ciphersuites:
 4566          PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
 4567          PSK-AES256-CBC-SHA
 4568 
 4569      New functions:
 4570          SSL_CTX_use_psk_identity_hint
 4571          SSL_get_psk_identity_hint
 4572          SSL_get_psk_identity
 4573          SSL_use_psk_identity_hint
 4574 
 4575      [Mika Kousa and Pasi Eronen of Nokia Corporation]
 4576 
 4577   *) Add RFC 3161 compliant time stamp request creation, response generation
 4578      and response verification functionality.
 4579      [Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project]
 4580 
 4581   *) Add initial support for TLS extensions, specifically for the server_name
 4582      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
 4583      have new members for a host name.  The SSL data structure has an
 4584      additional member SSL_CTX *initial_ctx so that new sessions can be
 4585      stored in that context to allow for session resumption, even after the
 4586      SSL has been switched to a new SSL_CTX in reaction to a client's
 4587      server_name extension.
 4588 
 4589      New functions (subject to change):
 4590 
 4591          SSL_get_servername()
 4592          SSL_get_servername_type()
 4593          SSL_set_SSL_CTX()
 4594 
 4595      New CTRL codes and macros (subject to change):
 4596 
 4597          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
 4598                                  - SSL_CTX_set_tlsext_servername_callback()
 4599          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
 4600                                       - SSL_CTX_set_tlsext_servername_arg()
 4601          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
 4602 
 4603      openssl s_client has a new '-servername ...' option.
 4604 
 4605      openssl s_server has new options '-servername_host ...', '-cert2 ...',
 4606      '-key2 ...', '-servername_fatal' (subject to change).  This allows
 4607      testing the HostName extension for a specific single host name ('-cert'
 4608      and '-key' remain fallbacks for handshakes without HostName
 4609      negotiation).  If the unrecognized_name alert has to be sent, this by
 4610      default is a warning; it becomes fatal with the '-servername_fatal'
 4611      option.
 4612 
 4613      [Peter Sylvester,  Remy Allais, Christophe Renou]
 4614 
 4615   *) Whirlpool hash implementation is added.
 4616      [Andy Polyakov]
 4617 
 4618   *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
 4619      bn(64,32). Because of instruction set limitations it doesn't have
 4620      any negative impact on performance. This was done mostly in order
 4621      to make it possible to share assembler modules, such as bn_mul_mont
 4622      implementations, between 32- and 64-bit builds without hassle.
 4623      [Andy Polyakov]
 4624 
 4625   *) Move code previously exiled into file crypto/ec/ec2_smpt.c
 4626      to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
 4627      macro.
 4628      [Bodo Moeller]
 4629 
 4630   *) New candidate for BIGNUM assembler implementation, bn_mul_mont,
 4631      dedicated Montgomery multiplication procedure, is introduced.
 4632      BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
 4633      "64-bit" performance on certain 32-bit targets.
 4634      [Andy Polyakov]
 4635 
 4636   *) New option SSL_OP_NO_COMP to disable use of compression selectively
 4637      in SSL structures. New SSL ctrl to set maximum send fragment size.
 4638      Save memory by setting the I/O buffer sizes dynamically instead of
 4639      using the maximum available value.
 4640      [Steve Henson]
 4641 
 4642   *) New option -V for 'openssl ciphers'. This prints the ciphersuite code
 4643      in addition to the text details.
 4644      [Bodo Moeller]
 4645 
 4646   *) Very, very preliminary EXPERIMENTAL support for printing of general
 4647      ASN1 structures. This currently produces rather ugly output and doesn't
 4648      handle several customised structures at all.
 4649      [Steve Henson]
 4650 
 4651   *) Integrated support for PVK file format and some related formats such
 4652      as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
 4653      these in the 'rsa' and 'dsa' utilities.
 4654      [Steve Henson]
 4655 
 4656   *) Support for PKCS#1 RSAPublicKey format on rsa utility command line.
 4657      [Steve Henson]
 4658 
 4659   *) Remove the ancient ASN1_METHOD code. This was only ever used in one
 4660      place for the (very old) "NETSCAPE" format certificates which are now
 4661      handled using new ASN1 code equivalents.
 4662      [Steve Henson]
 4663 
 4664   *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
 4665      pointer and make the SSL_METHOD parameter in SSL_CTX_new,
 4666      SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
 4667      [Nils Larsch]
 4668 
 4669   *) Modify CRL distribution points extension code to print out previously
 4670      unsupported fields. Enhance extension setting code to allow setting of
 4671      all fields.
 4672      [Steve Henson]
 4673 
 4674   *) Add print and set support for Issuing Distribution Point CRL extension.
 4675      [Steve Henson]
 4676 
 4677   *) Change 'Configure' script to enable Camellia by default.
 4678      [NTT]
 4679 
 4680  Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
 4681 
 4682   *) When rejecting SSL/TLS records due to an incorrect version number, never
 4683      update s->server with a new major version number.  As of
 4684      - OpenSSL 0.9.8m if 'short' is a 16-bit type,
 4685      - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
 4686      the previous behavior could result in a read attempt at NULL when
 4687      receiving specific incorrect SSL/TLS records once record payload
 4688      protection is active.  (CVE-2010-0740)
 4689      [Bodo Moeller, Adam Langley <agl@chromium.org>]
 4690 
 4691   *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL
 4692      could be crashed if the relevant tables were not present (e.g. chrooted).
 4693      [Tomas Hoger <thoger@redhat.com>]
 4694 
 4695  Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
 4696 
 4697   *) Always check bn_wexpand() return values for failure.  (CVE-2009-3245)
 4698      [Martin Olsson, Neel Mehta]
 4699 
 4700   *) Fix X509_STORE locking: Every 'objs' access requires a lock (to
 4701      accommodate for stack sorting, always a write lock!).
 4702      [Bodo Moeller]
 4703 
 4704   *) On some versions of WIN32 Heap32Next is very slow. This can cause
 4705      excessive delays in the RAND_poll(): over a minute. As a workaround
 4706      include a time check in the inner Heap32Next loop too.
 4707      [Steve Henson]
 4708 
 4709   *) The code that handled flushing of data in SSL/TLS originally used the
 4710      BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
 4711      the problem outlined in PR#1949. The fix suggested there however can
 4712      trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
 4713      of Apache). So instead simplify the code to flush unconditionally.
 4714      This should be fine since flushing with no data to flush is a no op.
 4715      [Steve Henson]
 4716 
 4717   *) Handle TLS versions 2.0 and later properly and correctly use the
 4718      highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
 4719      off ancient servers have a habit of sticking around for a while...
 4720      [Steve Henson]
 4721 
 4722   *) Modify compression code so it frees up structures without using the
 4723      ex_data callbacks. This works around a problem where some applications
 4724      call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
 4725      restarting) then use compression (e.g. SSL with compression) later.
 4726      This results in significant per-connection memory leaks and
 4727      has caused some security issues including CVE-2008-1678 and
 4728      CVE-2009-4355.
 4729      [Steve Henson]
 4730 
 4731   *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
 4732      change when encrypting or decrypting.
 4733      [Bodo Moeller]
 4734 
 4735   *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
 4736      connect and renegotiate with servers which do not support RI.
 4737      Until RI is more widely deployed this option is enabled by default.
 4738      [Steve Henson]
 4739 
 4740   *) Add "missing" ssl ctrls to clear options and mode.
 4741      [Steve Henson]
 4742 
 4743   *) If client attempts to renegotiate and doesn't support RI respond with
 4744      a no_renegotiation alert as required by RFC5746.  Some renegotiating
 4745      TLS clients will continue a connection gracefully when they receive
 4746      the alert. Unfortunately OpenSSL mishandled this alert and would hang
 4747      waiting for a server hello which it will never receive. Now we treat a
 4748      received no_renegotiation alert as a fatal error. This is because
 4749      applications requesting a renegotiation might well expect it to succeed
 4750      and would have no code in place to handle the server denying it so the
 4751      only safe thing to do is to terminate the connection.
 4752      [Steve Henson]
 4753 
 4754   *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
 4755      peer supports secure renegotiation and 0 otherwise. Print out peer
 4756      renegotiation support in s_client/s_server.
 4757      [Steve Henson]
 4758 
 4759   *) Replace the highly broken and deprecated SPKAC certification method with
 4760      the updated NID creation version. This should correctly handle UTF8.
 4761      [Steve Henson]
 4762 
 4763   *) Implement RFC5746. Re-enable renegotiation but require the extension
 4764      as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
 4765      turns out to be a bad idea. It has been replaced by
 4766      SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
 4767      SSL_CTX_set_options(). This is really not recommended unless you
 4768      know what you are doing.
 4769      [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson]
 4770 
 4771   *) Fixes to stateless session resumption handling. Use initial_ctx when
 4772      issuing and attempting to decrypt tickets in case it has changed during
 4773      servername handling. Use a non-zero length session ID when attempting
 4774      stateless session resumption: this makes it possible to determine if
 4775      a resumption has occurred immediately after receiving server hello
 4776      (several places in OpenSSL subtly assume this) instead of later in
 4777      the handshake.
 4778      [Steve Henson]
 4779 
 4780   *) The functions ENGINE_ctrl(), OPENSSL_isservice(),
 4781      CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
 4782      fixes for a few places where the return code is not checked
 4783      correctly.
 4784      [Julia Lawall <julia@diku.dk>]
 4785 
 4786   *) Add --strict-warnings option to Configure script to include devteam
 4787      warnings in other configurations.
 4788      [Steve Henson]
 4789 
 4790   *) Add support for --libdir option and LIBDIR variable in makefiles. This
 4791      makes it possible to install openssl libraries in locations which
 4792      have names other than "lib", for example "/usr/lib64" which some
 4793      systems need.
 4794      [Steve Henson, based on patch from Jeremy Utley]
 4795 
 4796   *) Don't allow the use of leading 0x80 in OIDs. This is a violation of
 4797      X690 8.9.12 and can produce some misleading textual output of OIDs.
 4798      [Steve Henson, reported by Dan Kaminsky]
 4799 
 4800   *) Delete MD2 from algorithm tables. This follows the recommendation in
 4801      several standards that it is not used in new applications due to
 4802      several cryptographic weaknesses. For binary compatibility reasons
 4803      the MD2 API is still compiled in by default.
 4804      [Steve Henson]
 4805 
 4806   *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
 4807      and restored.
 4808      [Steve Henson]
 4809 
 4810   *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
 4811      OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
 4812      clash.
 4813      [Guenter <lists@gknw.net>]
 4814 
 4815   *) Fix the server certificate chain building code to use X509_verify_cert(),
 4816      it used to have an ad-hoc builder which was unable to cope with anything
 4817      other than a simple chain.
 4818      [David Woodhouse <dwmw2@infradead.org>, Steve Henson]
 4819 
 4820   *) Don't check self signed certificate signatures in X509_verify_cert()
 4821      by default (a flag can override this): it just wastes time without
 4822      adding any security. As a useful side effect self signed root CAs
 4823      with non-FIPS digests are now usable in FIPS mode.
 4824      [Steve Henson]
 4825 
 4826   *) In dtls1_process_out_of_seq_message() the check if the current message
 4827      is already buffered was missing. For every new message was memory
 4828      allocated, allowing an attacker to perform an denial of service attack
 4829      with sending out of seq handshake messages until there is no memory
 4830      left. Additionally every future message was buffered, even if the
 4831      sequence number made no sense and would be part of another handshake.
 4832      So only messages with sequence numbers less than 10 in advance will be
 4833      buffered.  (CVE-2009-1378)
 4834      [Robin Seggelmann, discovered by Daniel Mentz]
 4835 
 4836   *) Records are buffered if they arrive with a future epoch to be
 4837      processed after finishing the corresponding handshake. There is
 4838      currently no limitation to this buffer allowing an attacker to perform
 4839      a DOS attack with sending records with future epochs until there is no
 4840      memory left. This patch adds the pqueue_size() function to determine
 4841      the size of a buffer and limits the record buffer to 100 entries.
 4842      (CVE-2009-1377)
 4843      [Robin Seggelmann, discovered by Daniel Mentz]
 4844 
 4845   *) Keep a copy of frag->msg_header.frag_len so it can be used after the
 4846      parent structure is freed.  (CVE-2009-1379)
 4847      [Daniel Mentz]
 4848 
 4849   *) Handle non-blocking I/O properly in SSL_shutdown() call.
 4850      [Darryl Miles <darryl-mailinglists@netbauds.net>]
 4851 
 4852   *) Add 2.5.4.* OIDs
 4853      [Ilya O. <vrghost@gmail.com>]
 4854 
 4855  Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
 4856 
 4857   *) Disable renegotiation completely - this fixes a severe security
 4858      problem (CVE-2009-3555) at the cost of breaking all
 4859      renegotiation. Renegotiation can be re-enabled by setting
 4860      SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
 4861      run-time. This is really not recommended unless you know what
 4862      you're doing.
 4863      [Ben Laurie]
 4864 
 4865  Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
 4866 
 4867   *) Don't set val to NULL when freeing up structures, it is freed up by
 4868      underlying code. If sizeof(void *) > sizeof(long) this can result in
 4869      zeroing past the valid field. (CVE-2009-0789)
 4870      [Paolo Ganci <Paolo.Ganci@AdNovum.CH>]
 4871 
 4872   *) Fix bug where return value of CMS_SignerInfo_verify_content() was not
 4873      checked correctly. This would allow some invalid signed attributes to
 4874      appear to verify correctly. (CVE-2009-0591)
 4875      [Ivan Nestlerode <inestlerode@us.ibm.com>]
 4876 
 4877   *) Reject UniversalString and BMPString types with invalid lengths. This
 4878      prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
 4879      a legal length. (CVE-2009-0590)
 4880      [Steve Henson]
 4881 
 4882   *) Set S/MIME signing as the default purpose rather than setting it
 4883      unconditionally. This allows applications to override it at the store
 4884      level.
 4885      [Steve Henson]
 4886 
 4887   *) Permit restricted recursion of ASN1 strings. This is needed in practice
 4888      to handle some structures.
 4889      [Steve Henson]
 4890 
 4891   *) Improve efficiency of mem_gets: don't search whole buffer each time
 4892      for a '\n'
 4893      [Jeremy Shapiro <jnshapir@us.ibm.com>]
 4894 
 4895   *) New -hex option for openssl rand.
 4896      [Matthieu Herrb]
 4897 
 4898   *) Print out UTF8String and NumericString when parsing ASN1.
 4899      [Steve Henson]
 4900 
 4901   *) Support NumericString type for name components.
 4902      [Steve Henson]
 4903 
 4904   *) Allow CC in the environment to override the automatically chosen
 4905      compiler. Note that nothing is done to ensure flags work with the
 4906      chosen compiler.
 4907      [Ben Laurie]
 4908 
 4909  Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
 4910 
 4911   *) Properly check EVP_VerifyFinal() and similar return values
 4912      (CVE-2008-5077).
 4913      [Ben Laurie, Bodo Moeller, Google Security Team]
 4914 
 4915   *) Enable TLS extensions by default.
 4916      [Ben Laurie]
 4917 
 4918   *) Allow the CHIL engine to be loaded, whether the application is
 4919      multithreaded or not. (This does not release the developer from the
 4920      obligation to set up the dynamic locking callbacks.)
 4921      [Sander Temme <sander@temme.net>]
 4922 
 4923   *) Use correct exit code if there is an error in dgst command.
 4924      [Steve Henson; problem pointed out by Roland Dirlewanger]
 4925 
 4926   *) Tweak Configure so that you need to say "experimental-jpake" to enable
 4927      JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
 4928      [Bodo Moeller]
 4929 
 4930   *) Add experimental JPAKE support, including demo authentication in
 4931      s_client and s_server.
 4932      [Ben Laurie]
 4933 
 4934   *) Set the comparison function in v3_addr_canonize().
 4935      [Rob Austein <sra@hactrn.net>]
 4936 
 4937   *) Add support for XMPP STARTTLS in s_client.
 4938      [Philip Paeps <philip@freebsd.org>]
 4939 
 4940   *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
 4941      to ensure that even with this option, only ciphersuites in the
 4942      server's preference list will be accepted.  (Note that the option
 4943      applies only when resuming a session, so the earlier behavior was
 4944      just about the algorithm choice for symmetric cryptography.)
 4945      [Bodo Moeller]
 4946 
 4947  Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
 4948 
 4949   *) Fix NULL pointer dereference if a DTLS server received
 4950      ChangeCipherSpec as first record (CVE-2009-1386).
 4951      [PR #1679]
 4952 
 4953   *) Fix a state transition in s3_srvr.c and d1_srvr.c
 4954      (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
 4955      [Nagendra Modadugu]
 4956 
 4957   *) The fix in 0.9.8c that supposedly got rid of unsafe
 4958      double-checked locking was incomplete for RSA blinding,
 4959      addressing just one layer of what turns out to have been
 4960      doubly unsafe triple-checked locking.
 4961 
 4962      So now fix this for real by retiring the MONT_HELPER macro
 4963      in crypto/rsa/rsa_eay.c.
 4964 
 4965      [Bodo Moeller; problem pointed out by Marius Schilder]
 4966 
 4967   *) Various precautionary measures:
 4968 
 4969      - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
 4970 
 4971      - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
 4972        (NB: This would require knowledge of the secret session ticket key
 4973        to exploit, in which case you'd be SOL either way.)
 4974 
 4975      - Change bn_nist.c so that it will properly handle input BIGNUMs
 4976        outside the expected range.
 4977 
 4978      - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
 4979        builds.
 4980 
 4981      [Neel Mehta, Bodo Moeller]
 4982 
 4983   *) Allow engines to be "soft loaded" - i.e. optionally don't die if
 4984      the load fails. Useful for distros.
 4985      [Ben Laurie and the FreeBSD team]
 4986 
 4987   *) Add support for Local Machine Keyset attribute in PKCS#12 files.
 4988      [Steve Henson]
 4989 
 4990   *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
 4991      [Huang Ying]
 4992 
 4993   *) Expand ENGINE to support engine supplied SSL client certificate functions.
 4994 
 4995      This work was sponsored by Logica.
 4996      [Steve Henson]
 4997 
 4998   *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
 4999      keystores. Support for SSL/TLS client authentication too.
 5000      Not compiled unless enable-capieng specified to Configure.
 5001 
 5002      This work was sponsored by Logica.
 5003      [Steve Henson]
 5004 
 5005   *) Fix bug in X509_ATTRIBUTE creation: don't set attribute using
 5006      ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
 5007      attribute creation routines such as certificate requests and PKCS#12
 5008      files.
 5009      [Steve Henson]
 5010 
 5011  Changes between 0.9.8g and 0.9.8h  [28 May 2008]
 5012 
 5013   *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
 5014      handshake which could lead to a client crash as found using the
 5015      Codenomicon TLS test suite (CVE-2008-1672)
 5016      [Steve Henson, Mark Cox]
 5017 
 5018   *) Fix double free in TLS server name extensions which could lead to
 5019      a remote crash found by Codenomicon TLS test suite (CVE-2008-0891)
 5020      [Joe Orton]
 5021 
 5022   *) Clear error queue in SSL_CTX_use_certificate_chain_file()
 5023 
 5024      Clear the error queue to ensure that error entries left from
 5025      older function calls do not interfere with the correct operation.
 5026      [Lutz Jaenicke, Erik de Castro Lopo]
 5027 
 5028   *) Remove root CA certificates of commercial CAs:
 5029 
 5030      The OpenSSL project does not recommend any specific CA and does not
 5031      have any policy with respect to including or excluding any CA.
 5032      Therefore it does not make any sense to ship an arbitrary selection
 5033      of root CA certificates with the OpenSSL software.
 5034      [Lutz Jaenicke]
 5035 
 5036   *) RSA OAEP patches to fix two separate invalid memory reads.
 5037      The first one involves inputs when 'lzero' is greater than
 5038      'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
 5039      before the beginning of from). The second one involves inputs where
 5040      the 'db' section contains nothing but zeroes (there is a one-byte
 5041      invalid read after the end of 'db').
 5042      [Ivan Nestlerode <inestlerode@us.ibm.com>]
 5043 
 5044   *) Partial backport from 0.9.9-dev:
 5045 
 5046      Introduce bn_mul_mont (dedicated Montgomery multiplication
 5047      procedure) as a candidate for BIGNUM assembler implementation.
 5048      While 0.9.9-dev uses assembler for various architectures, only
 5049      x86_64 is available by default here in the 0.9.8 branch, and
 5050      32-bit x86 is available through a compile-time setting.
 5051 
 5052      To try the 32-bit x86 assembler implementation, use Configure
 5053      option "enable-montasm" (which exists only for this backport).
 5054 
 5055      As "enable-montasm" for 32-bit x86 disclaims code stability
 5056      anyway, in this constellation we activate additional code
 5057      backported from 0.9.9-dev for further performance improvements,
 5058      namely BN_from_montgomery_word.  (To enable this otherwise,
 5059      e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
 5060 
 5061      [Andy Polyakov (backport partially by Bodo Moeller)]
 5062 
 5063   *) Add TLS session ticket callback. This allows an application to set
 5064      TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
 5065      values. This is useful for key rollover for example where several key
 5066      sets may exist with different names.
 5067      [Steve Henson]
 5068 
 5069   *) Reverse ENGINE-internal logic for caching default ENGINE handles.
 5070      This was broken until now in 0.9.8 releases, such that the only way
 5071      a registered ENGINE could be used (assuming it initialises
 5072      successfully on the host) was to explicitly set it as the default
 5073      for the relevant algorithms. This is in contradiction with 0.9.7
 5074      behaviour and the documentation. With this fix, when an ENGINE is
 5075      registered into a given algorithm's table of implementations, the
 5076      'uptodate' flag is reset so that auto-discovery will be used next
 5077      time a new context for that algorithm attempts to select an
 5078      implementation.
 5079      [Ian Lister (tweaked by Geoff Thorpe)]
 5080 
 5081   *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
 5082      implementation in the following ways:
 5083 
 5084      Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
 5085      hard coded.
 5086 
 5087      Lack of BER streaming support means one pass streaming processing is
 5088      only supported if data is detached: setting the streaming flag is
 5089      ignored for embedded content.
 5090 
 5091      CMS support is disabled by default and must be explicitly enabled
 5092      with the enable-cms configuration option.
 5093      [Steve Henson]
 5094 
 5095   *) Update the GMP engine glue to do direct copies between BIGNUM and
 5096      mpz_t when openssl and GMP use the same limb size. Otherwise the
 5097      existing "conversion via a text string export" trick is still used.
 5098      [Paul Sheer <paulsheer@gmail.com>]
 5099 
 5100   *) Zlib compression BIO. This is a filter BIO which compressed and
 5101      uncompresses any data passed through it.
 5102      [Steve Henson]
 5103 
 5104   *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
 5105      RFC3394 compatible AES key wrapping.
 5106      [Steve Henson]
 5107 
 5108   *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
 5109      sets string data without copying. X509_ALGOR_set0() and
 5110      X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
 5111      data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
 5112      from an X509_ATTRIBUTE structure optionally checking it occurs only
 5113      once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
 5114      data.
 5115      [Steve Henson]
 5116 
 5117   *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
 5118      to get the expected BN_FLG_CONSTTIME behavior.
 5119      [Bodo Moeller (Google)]
 5120 
 5121   *) Netware support:
 5122 
 5123      - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
 5124      - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
 5125      - added some more tests to do_tests.pl
 5126      - fixed RunningProcess usage so that it works with newer LIBC NDKs too
 5127      - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
 5128      - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
 5129        netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
 5130      - various changes to netware.pl to enable gcc-cross builds on Win32
 5131        platform
 5132      - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
 5133      - various changes to fix missing prototype warnings
 5134      - fixed x86nasm.pl to create correct asm files for NASM COFF output
 5135      - added AES, WHIRLPOOL and CPUID assembler code to build files
 5136      - added missing AES assembler make rules to mk1mf.pl
 5137      - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
 5138      [Guenter Knauf <eflash@gmx.net>]
 5139 
 5140   *) Implement certificate status request TLS extension defined in RFC3546.
 5141      A client can set the appropriate parameters and receive the encoded
 5142      OCSP response via a callback. A server can query the supplied parameters
 5143      and set the encoded OCSP response in the callback. Add simplified examples
 5144      to s_client and s_server.
 5145      [Steve Henson]
 5146 
 5147  Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
 5148 
 5149   *) Fix various bugs:
 5150      + Binary incompatibility of ssl_ctx_st structure
 5151      + DTLS interoperation with non-compliant servers
 5152      + Don't call get_session_cb() without proposed session
 5153      + Fix ia64 assembler code
 5154      [Andy Polyakov, Steve Henson]
 5155 
 5156  Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
 5157 
 5158   *) DTLS Handshake overhaul. There were longstanding issues with
 5159      OpenSSL DTLS implementation, which were making it impossible for
 5160      RFC 4347 compliant client to communicate with OpenSSL server.
 5161      Unfortunately just fixing these incompatibilities would "cut off"
 5162      pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
 5163      server keeps tolerating non RFC compliant syntax. The opposite is
 5164      not true, 0.9.8f client can not communicate with earlier server.
 5165      This update even addresses CVE-2007-4995.
 5166      [Andy Polyakov]
 5167 
 5168   *) Changes to avoid need for function casts in OpenSSL: some compilers
 5169      (gcc 4.2 and later) reject their use.
 5170      [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
 5171       Steve Henson]
 5172 
 5173   *) Add RFC4507 support to OpenSSL. This includes the corrections in
 5174      RFC4507bis. The encrypted ticket format is an encrypted encoded
 5175      SSL_SESSION structure, that way new session features are automatically
 5176      supported.
 5177 
 5178      If a client application caches session in an SSL_SESSION structure
 5179      support is transparent because tickets are now stored in the encoded
 5180      SSL_SESSION.
 5181 
 5182      The SSL_CTX structure automatically generates keys for ticket
 5183      protection in servers so again support should be possible
 5184      with no application modification.
 5185 
 5186      If a client or server wishes to disable RFC4507 support then the option
 5187      SSL_OP_NO_TICKET can be set.
 5188 
 5189      Add a TLS extension debugging callback to allow the contents of any client
 5190      or server extensions to be examined.
 5191 
 5192      This work was sponsored by Google.
 5193      [Steve Henson]
 5194 
 5195   *) Add initial support for TLS extensions, specifically for the server_name
 5196      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
 5197      have new members for a host name.  The SSL data structure has an
 5198      additional member SSL_CTX *initial_ctx so that new sessions can be
 5199      stored in that context to allow for session resumption, even after the
 5200      SSL has been switched to a new SSL_CTX in reaction to a client's
 5201      server_name extension.
 5202 
 5203      New functions (subject to change):
 5204 
 5205          SSL_get_servername()
 5206          SSL_get_servername_type()
 5207          SSL_set_SSL_CTX()
 5208 
 5209      New CTRL codes and macros (subject to change):
 5210 
 5211          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
 5212                                  - SSL_CTX_set_tlsext_servername_callback()
 5213          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
 5214                                       - SSL_CTX_set_tlsext_servername_arg()
 5215          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
 5216 
 5217      openssl s_client has a new '-servername ...' option.
 5218 
 5219      openssl s_server has new options '-servername_host ...', '-cert2 ...',
 5220      '-key2 ...', '-servername_fatal' (subject to change).  This allows
 5221      testing the HostName extension for a specific single host name ('-cert'
 5222      and '-key' remain fallbacks for handshakes without HostName
 5223      negotiation).  If the unrecognized_name alert has to be sent, this by
 5224      default is a warning; it becomes fatal with the '-servername_fatal'
 5225      option.
 5226 
 5227      [Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson]
 5228 
 5229   *) Add AES and SSE2 assembly language support to VC++ build.
 5230      [Steve Henson]
 5231 
 5232   *) Mitigate attack on final subtraction in Montgomery reduction.
 5233      [Andy Polyakov]
 5234 
 5235   *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
 5236      (which previously caused an internal error).
 5237      [Bodo Moeller]
 5238 
 5239   *) Squeeze another 10% out of IGE mode when in != out.
 5240      [Ben Laurie]
 5241 
 5242   *) AES IGE mode speedup.
 5243      [Dean Gaudet (Google)]
 5244 
 5245   *) Add the Korean symmetric 128-bit cipher SEED (see
 5246      http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
 5247      add SEED ciphersuites from RFC 4162:
 5248 
 5249         TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
 5250         TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
 5251         TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
 5252         TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
 5253 
 5254      To minimize changes between patchlevels in the OpenSSL 0.9.8
 5255      series, SEED remains excluded from compilation unless OpenSSL
 5256      is configured with 'enable-seed'.
 5257      [KISA, Bodo Moeller]
 5258 
 5259   *) Mitigate branch prediction attacks, which can be practical if a
 5260      single processor is shared, allowing a spy process to extract
 5261      information.  For detailed background information, see
 5262      http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
 5263      J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
 5264      and Necessary Software Countermeasures").  The core of the change
 5265      are new versions BN_div_no_branch() and
 5266      BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
 5267      respectively, which are slower, but avoid the security-relevant
 5268      conditional branches.  These are automatically called by BN_div()
 5269      and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
 5270      of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
 5271      remove a conditional branch.
 5272 
 5273      BN_FLG_CONSTTIME is the new name for the previous
 5274      BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
 5275      modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
 5276      in the exponent causes BN_mod_exp_mont() to use the alternative
 5277      implementation in BN_mod_exp_mont_consttime().)  The old name
 5278      remains as a deprecated alias.
 5279 
 5280      Similarly, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
 5281      RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
 5282      constant-time implementations for more than just exponentiation.
 5283      Here too the old name is kept as a deprecated alias.
 5284 
 5285      BN_BLINDING_new() will now use BN_dup() for the modulus so that
 5286      the BN_BLINDING structure gets an independent copy of the
 5287      modulus.  This means that the previous "BIGNUM *m" argument to
 5288      BN_BLINDING_new() and to BN_BLINDING_create_param() now
 5289      essentially becomes "const BIGNUM *m", although we can't actually
 5290      change this in the header file before 0.9.9.  It allows
 5291      RSA_setup_blinding() to use BN_with_flags() on the modulus to
 5292      enable BN_FLG_CONSTTIME.
 5293 
 5294      [Matthew D Wood (Intel Corp)]
 5295 
 5296   *) In the SSL/TLS server implementation, be strict about session ID
 5297      context matching (which matters if an application uses a single
 5298      external cache for different purposes).  Previously,
 5299      out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
 5300      set.  This did ensure strict client verification, but meant that,
 5301      with applications using a single external cache for quite
 5302      different requirements, clients could circumvent ciphersuite
 5303      restrictions for a given session ID context by starting a session
 5304      in a different context.
 5305      [Bodo Moeller]
 5306 
 5307   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
 5308      a ciphersuite string such as "DEFAULT:RSA" cannot enable
 5309      authentication-only ciphersuites.
 5310      [Bodo Moeller]
 5311 
 5312   *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
 5313      not complete and could lead to a possible single byte overflow
 5314      (CVE-2007-5135) [Ben Laurie]
 5315 
 5316  Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
 5317 
 5318   *) Since AES128 and AES256 (and similarly Camellia128 and
 5319      Camellia256) share a single mask bit in the logic of
 5320      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
 5321      kludge to work properly if AES128 is available and AES256 isn't
 5322      (or if Camellia128 is available and Camellia256 isn't).
 5323      [Victor Duchovni]
 5324 
 5325   *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
 5326      (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
 5327      When a point or a seed is encoded in a BIT STRING, we need to
 5328      prevent the removal of trailing zero bits to get the proper DER
 5329      encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
 5330      of a NamedBitList, for which trailing 0 bits need to be removed.)
 5331      [Bodo Moeller]
 5332 
 5333   *) Have SSL/TLS server implementation tolerate "mismatched" record
 5334      protocol version while receiving ClientHello even if the
 5335      ClientHello is fragmented.  (The server can't insist on the
 5336      particular protocol version it has chosen before the ServerHello
 5337      message has informed the client about his choice.)
 5338      [Bodo Moeller]
 5339 
 5340   *) Add RFC 3779 support.
 5341      [Rob Austein for ARIN, Ben Laurie]
 5342 
 5343   *) Load error codes if they are not already present instead of using a
 5344      static variable. This allows them to be cleanly unloaded and reloaded.
 5345      Improve header file function name parsing.
 5346      [Steve Henson]
 5347 
 5348   *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
 5349      or CAPABILITY handshake as required by RFCs.
 5350      [Goetz Babin-Ebell]
 5351 
 5352  Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
 5353 
 5354   *) Introduce limits to prevent malicious keys being able to
 5355      cause a denial of service.  (CVE-2006-2940)
 5356      [Steve Henson, Bodo Moeller]
 5357 
 5358   *) Fix ASN.1 parsing of certain invalid structures that can result
 5359      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
 5360 
 5361   *) Fix buffer overflow in SSL_get_shared_ciphers() function.
 5362      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
 5363 
 5364   *) Fix SSL client code which could crash if connecting to a
 5365      malicious SSLv2 server.  (CVE-2006-4343)
 5366      [Tavis Ormandy and Will Drewry, Google Security Team]
 5367 
 5368   *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
 5369      match only those.  Before that, "AES256-SHA" would be interpreted
 5370      as a pattern and match "AES128-SHA" too (since AES128-SHA got
 5371      the same strength classification in 0.9.7h) as we currently only
 5372      have a single AES bit in the ciphersuite description bitmap.
 5373      That change, however, also applied to ciphersuite strings such as
 5374      "RC4-MD5" that intentionally matched multiple ciphersuites --
 5375      namely, SSL 2.0 ciphersuites in addition to the more common ones
 5376      from SSL 3.0/TLS 1.0.
 5377 
 5378      So we change the selection algorithm again: Naming an explicit
 5379      ciphersuite selects this one ciphersuite, and any other similar
 5380      ciphersuite (same bitmap) from *other* protocol versions.
 5381      Thus, "RC4-MD5" again will properly select both the SSL 2.0
 5382      ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
 5383 
 5384      Since SSL 2.0 does not have any ciphersuites for which the
 5385      128/256 bit distinction would be relevant, this works for now.
 5386      The proper fix will be to use different bits for AES128 and
 5387      AES256, which would have avoided the problems from the beginning;
 5388      however, bits are scarce, so we can only do this in a new release
 5389      (not just a patchlevel) when we can change the SSL_CIPHER
 5390      definition to split the single 'unsigned long mask' bitmap into
 5391      multiple values to extend the available space.
 5392 
 5393      [Bodo Moeller]
 5394 
 5395  Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
 5396 
 5397   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
 5398      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
 5399 
 5400   *) Add AES IGE and biIGE modes.
 5401      [Ben Laurie]
 5402 
 5403   *) Change the Unix randomness entropy gathering to use poll() when
 5404      possible instead of select(), since the latter has some
 5405      undesirable limitations.
 5406      [Darryl Miles via Richard Levitte and Bodo Moeller]
 5407 
 5408   *) Disable "ECCdraft" ciphersuites more thoroughly.  Now special
 5409      treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
 5410      cannot be implicitly activated as part of, e.g., the "AES" alias.
 5411      However, please upgrade to OpenSSL 0.9.9[-dev] for
 5412      non-experimental use of the ECC ciphersuites to get TLS extension
 5413      support, which is required for curve and point format negotiation
 5414      to avoid potential handshake problems.
 5415      [Bodo Moeller]
 5416 
 5417   *) Disable rogue ciphersuites:
 5418 
 5419       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
 5420       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
 5421       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
 5422 
 5423      The latter two were purportedly from
 5424      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
 5425      appear there.
 5426 
 5427      Also deactivate the remaining ciphersuites from
 5428      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
 5429      unofficial, and the ID has long expired.
 5430      [Bodo Moeller]
 5431 
 5432   *) Fix RSA blinding Heisenbug (problems sometimes occurred on
 5433      dual-core machines) and other potential thread-safety issues.
 5434      [Bodo Moeller]
 5435 
 5436   *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
 5437      versions), which is now available for royalty-free use
 5438      (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
 5439      Also, add Camellia TLS ciphersuites from RFC 4132.
 5440 
 5441      To minimize changes between patchlevels in the OpenSSL 0.9.8
 5442      series, Camellia remains excluded from compilation unless OpenSSL
 5443      is configured with 'enable-camellia'.
 5444      [NTT]
 5445 
 5446   *) Disable the padding bug check when compression is in use. The padding
 5447      bug check assumes the first packet is of even length, this is not
 5448      necessarily true if compression is enabled and can result in false
 5449      positives causing handshake failure. The actual bug test is ancient
 5450      code so it is hoped that implementations will either have fixed it by
 5451      now or any which still have the bug do not support compression.
 5452      [Steve Henson]
 5453 
 5454  Changes between 0.9.8a and 0.9.8b  [04 May 2006]
 5455 
 5456   *) When applying a cipher rule check to see if string match is an explicit
 5457      cipher suite and only match that one cipher suite if it is.
 5458      [Steve Henson]
 5459 
 5460   *) Link in manifests for VC++ if needed.
 5461      [Austin Ziegler <halostatue@gmail.com>]
 5462 
 5463   *) Update support for ECC-based TLS ciphersuites according to
 5464      draft-ietf-tls-ecc-12.txt with proposed changes (but without
 5465      TLS extensions, which are supported starting with the 0.9.9
 5466      branch, not in the OpenSSL 0.9.8 branch).
 5467      [Douglas Stebila]
 5468 
 5469   *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
 5470      opaque EVP_CIPHER_CTX handling.
 5471      [Steve Henson]
 5472 
 5473   *) Fixes and enhancements to zlib compression code. We now only use
 5474      "zlib1.dll" and use the default __cdecl calling convention on Win32
 5475      to conform with the standards mentioned here:
 5476            http://www.zlib.net/DLL_FAQ.txt
 5477      Static zlib linking now works on Windows and the new --with-zlib-include
 5478      --with-zlib-lib options to Configure can be used to supply the location
 5479      of the headers and library. Gracefully handle case where zlib library
 5480      can't be loaded.
 5481      [Steve Henson]
 5482 
 5483   *) Several fixes and enhancements to the OID generation code. The old code
 5484      sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
 5485      handle numbers larger than ULONG_MAX, truncated printing and had a
 5486      non standard OBJ_obj2txt() behaviour.
 5487      [Steve Henson]
 5488 
 5489   *) Add support for building of engines under engine/ as shared libraries
 5490      under VC++ build system.
 5491      [Steve Henson]
 5492 
 5493   *) Corrected the numerous bugs in the Win32 path splitter in DSO.
 5494      Hopefully, we will not see any false combination of paths any more.
 5495      [Richard Levitte]
 5496 
 5497  Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
 5498 
 5499   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
 5500      (part of SSL_OP_ALL).  This option used to disable the
 5501      countermeasure against man-in-the-middle protocol-version
 5502      rollback in the SSL 2.0 server implementation, which is a bad
 5503      idea.  (CVE-2005-2969)
 5504 
 5505      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
 5506      for Information Security, National Institute of Advanced Industrial
 5507      Science and Technology [AIST], Japan)]
 5508 
 5509   *) Add two function to clear and return the verify parameter flags.
 5510      [Steve Henson]
 5511 
 5512   *) Keep cipherlists sorted in the source instead of sorting them at
 5513      runtime, thus removing the need for a lock.
 5514      [Nils Larsch]
 5515 
 5516   *) Avoid some small subgroup attacks in Diffie-Hellman.
 5517      [Nick Mathewson and Ben Laurie]
 5518 
 5519   *) Add functions for well-known primes.
 5520      [Nick Mathewson]
 5521 
 5522   *) Extended Windows CE support.
 5523      [Satoshi Nakamura and Andy Polyakov]
 5524 
 5525   *) Initialize SSL_METHOD structures at compile time instead of during
 5526      runtime, thus removing the need for a lock.
 5527      [Steve Henson]
 5528 
 5529   *) Make PKCS7_decrypt() work even if no certificate is supplied by
 5530      attempting to decrypt each encrypted key in turn. Add support to
 5531      smime utility.
 5532      [Steve Henson]
 5533 
 5534  Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
 5535 
 5536   [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
 5537   OpenSSL 0.9.8.]
 5538 
 5539   *) Add libcrypto.pc and libssl.pc for those who feel they need them.
 5540      [Richard Levitte]
 5541 
 5542   *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
 5543      key into the same file any more.
 5544      [Richard Levitte]
 5545 
 5546   *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
 5547      [Andy Polyakov]
 5548 
 5549   *) Add -utf8 command line and config file option to 'ca'.
 5550      [Stefan <stf@udoma.org]
 5551 
 5552   *) Removed the macro des_crypt(), as it seems to conflict with some
 5553      libraries.  Use DES_crypt().
 5554      [Richard Levitte]
 5555 
 5556   *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
 5557      involves renaming the source and generated shared-libs for
 5558      both. The engines will accept the corrected or legacy ids
 5559      ('ncipher' and '4758_cca' respectively) when binding. NB,
 5560      this only applies when building 'shared'.
 5561      [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe]
 5562 
 5563   *) Add attribute functions to EVP_PKEY structure. Modify
 5564      PKCS12_create() to recognize a CSP name attribute and
 5565      use it. Make -CSP option work again in pkcs12 utility.
 5566      [Steve Henson]
 5567 
 5568   *) Add new functionality to the bn blinding code:
 5569      - automatic re-creation of the BN_BLINDING parameters after
 5570        a fixed number of uses (currently 32)
 5571      - add new function for parameter creation
 5572      - introduce flags to control the update behaviour of the
 5573        BN_BLINDING parameters
 5574      - hide BN_BLINDING structure
 5575      Add a second BN_BLINDING slot to the RSA structure to improve
 5576      performance when a single RSA object is shared among several
 5577      threads.
 5578      [Nils Larsch]
 5579 
 5580   *) Add support for DTLS.
 5581      [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie]
 5582 
 5583   *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
 5584      to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
 5585      [Walter Goulet]
 5586 
 5587   *) Remove buggy and incomplete DH cert support from
 5588      ssl/ssl_rsa.c and ssl/s3_both.c
 5589      [Nils Larsch]
 5590 
 5591   *) Use SHA-1 instead of MD5 as the default digest algorithm for
 5592      the apps/openssl applications.
 5593      [Nils Larsch]
 5594 
 5595   *) Compile clean with "-Wall -Wmissing-prototypes
 5596      -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
 5597      DEBUG_SAFESTACK must also be set.
 5598      [Ben Laurie]
 5599 
 5600   *) Change ./Configure so that certain algorithms can be disabled by default.
 5601      The new counterpiece to "no-xxx" is "enable-xxx".
 5602 
 5603      The patented RC5 and MDC2 algorithms will now be disabled unless
 5604      "enable-rc5" and "enable-mdc2", respectively, are specified.
 5605 
 5606      (IDEA remains enabled despite being patented.  This is because IDEA
 5607      is frequently required for interoperability, and there is no license
 5608      fee for non-commercial use.  As before, "no-idea" can be used to
 5609      avoid this algorithm.)
 5610 
 5611      [Bodo Moeller]
 5612 
 5613   *) Add processing of proxy certificates (see RFC 3820).  This work was
 5614      sponsored by KTH (The Royal Institute of Technology in Stockholm) and
 5615      EGEE (Enabling Grids for E-science in Europe).
 5616      [Richard Levitte]
 5617 
 5618   *) RC4 performance overhaul on modern architectures/implementations, such
 5619      as Intel P4, IA-64 and AMD64.
 5620      [Andy Polyakov]
 5621 
 5622   *) New utility extract-section.pl. This can be used specify an alternative
 5623      section number in a pod file instead of having to treat each file as
 5624      a separate case in Makefile. This can be done by adding two lines to the
 5625      pod file:
 5626 
 5627      =for comment openssl_section:XXX
 5628 
 5629      The blank line is mandatory.
 5630 
 5631      [Steve Henson]
 5632 
 5633   *) New arguments -certform, -keyform and -pass for s_client and s_server
 5634      to allow alternative format key and certificate files and passphrase
 5635      sources.
 5636      [Steve Henson]
 5637 
 5638   *) New structure X509_VERIFY_PARAM which combines current verify parameters,
 5639      update associated structures and add various utility functions.
 5640 
 5641      Add new policy related verify parameters, include policy checking in
 5642      standard verify code. Enhance 'smime' application with extra parameters
 5643      to support policy checking and print out.
 5644      [Steve Henson]
 5645 
 5646   *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
 5647      Nehemiah processors. These extensions support AES encryption in hardware
 5648      as well as RNG (though RNG support is currently disabled).
 5649      [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
 5650 
 5651   *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
 5652      [Geoff Thorpe]
 5653 
 5654   *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
 5655      [Andy Polyakov and a number of other people]
 5656 
 5657   *) Improved PowerPC platform support. Most notably BIGNUM assembler
 5658      implementation contributed by IBM.
 5659      [Suresh Chari, Peter Waltenberg, Andy Polyakov]
 5660 
 5661   *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
 5662      exponent rather than 'unsigned long'. There is a corresponding change to
 5663      the new 'rsa_keygen' element of the RSA_METHOD structure.
 5664      [Jelte Jansen, Geoff Thorpe]
 5665 
 5666   *) Functionality for creating the initial serial number file is now
 5667      moved from CA.pl to the 'ca' utility with a new option -create_serial.
 5668 
 5669      (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
 5670      number file to 1, which is bound to cause problems.  To avoid
 5671      the problems while respecting compatibility between different 0.9.7
 5672      patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
 5673      CA.pl for serial number initialization.  With the new release 0.9.8,
 5674      we can fix the problem directly in the 'ca' utility.)
 5675      [Steve Henson]
 5676 
 5677   *) Reduced header interdependencies by declaring more opaque objects in
 5678      ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
 5679      give fewer recursive includes, which could break lazy source code - so
 5680      this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
 5681      developers should define this symbol when building and using openssl to
 5682      ensure they track the recommended behaviour, interfaces, [etc], but
 5683      backwards-compatible behaviour prevails when this isn't defined.
 5684      [Geoff Thorpe]
 5685 
 5686   *) New function X509_POLICY_NODE_print() which prints out policy nodes.
 5687      [Steve Henson]
 5688 
 5689   *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
 5690      This will generate a random key of the appropriate length based on the
 5691      cipher context. The EVP_CIPHER can provide its own random key generation
 5692      routine to support keys of a specific form. This is used in the des and
 5693      3des routines to generate a key of the correct parity. Update S/MIME
 5694      code to use new functions and hence generate correct parity DES keys.
 5695      Add EVP_CHECK_DES_KEY #define to return an error if the key is not
 5696      valid (weak or incorrect parity).
 5697      [Steve Henson]
 5698 
 5699   *) Add a local set of CRLs that can be used by X509_verify_cert() as well
 5700      as looking them up. This is useful when the verified structure may contain
 5701      CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
 5702      present unless the new PKCS7_NO_CRL flag is asserted.
 5703      [Steve Henson]
 5704 
 5705   *) Extend ASN1 oid configuration module. It now additionally accepts the
 5706      syntax:
 5707 
 5708      shortName = some long name, 1.2.3.4
 5709      [Steve Henson]
 5710 
 5711   *) Reimplemented the BN_CTX implementation. There is now no more static
 5712      limitation on the number of variables it can handle nor the depth of the
 5713      "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
 5714      information can now expand as required, and rather than having a single
 5715      static array of bignums, BN_CTX now uses a linked-list of such arrays
 5716      allowing it to expand on demand whilst maintaining the usefulness of
 5717      BN_CTX's "bundling".
 5718      [Geoff Thorpe]
 5719 
 5720   *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
 5721      to allow all RSA operations to function using a single BN_CTX.
 5722      [Geoff Thorpe]
 5723 
 5724   *) Preliminary support for certificate policy evaluation and checking. This
 5725      is initially intended to pass the tests outlined in "Conformance Testing
 5726      of Relying Party Client Certificate Path Processing Logic" v1.07.
 5727      [Steve Henson]
 5728 
 5729   *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
 5730      remained unused and not that useful. A variety of other little bignum
 5731      tweaks and fixes have also been made continuing on from the audit (see
 5732      below).
 5733      [Geoff Thorpe]
 5734 
 5735   *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
 5736      associated ASN1, EVP and SSL functions and old ASN1 macros.
 5737      [Richard Levitte]
 5738 
 5739   *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
 5740      and this should never fail. So the return value from the use of
 5741      BN_set_word() (which can fail due to needless expansion) is now deprecated;
 5742      if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
 5743      [Geoff Thorpe]
 5744 
 5745   *) BN_CTX_get() should return zero-valued bignums, providing the same
 5746      initialised value as BN_new().
 5747      [Geoff Thorpe, suggested by Ulf Möller]
 5748 
 5749   *) Support for inhibitAnyPolicy certificate extension.
 5750      [Steve Henson]
 5751 
 5752   *) An audit of the BIGNUM code is underway, for which debugging code is
 5753      enabled when BN_DEBUG is defined. This makes stricter enforcements on what
 5754      is considered valid when processing BIGNUMs, and causes execution to
 5755      assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
 5756      further steps are taken to deliberately pollute unused data in BIGNUM
 5757      structures to try and expose faulty code further on. For now, openssl will
 5758      (in its default mode of operation) continue to tolerate the inconsistent
 5759      forms that it has tolerated in the past, but authors and packagers should
 5760      consider trying openssl and their own applications when compiled with
 5761      these debugging symbols defined. It will help highlight potential bugs in
 5762      their own code, and will improve the test coverage for OpenSSL itself. At
 5763      some point, these tighter rules will become openssl's default to improve
 5764      maintainability, though the assert()s and other overheads will remain only
 5765      in debugging configurations. See bn.h for more details.
 5766      [Geoff Thorpe, Nils Larsch, Ulf Möller]
 5767 
 5768   *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
 5769      that can only be obtained through BN_CTX_new() (which implicitly
 5770      initialises it). The presence of this function only made it possible
 5771      to overwrite an existing structure (and cause memory leaks).
 5772      [Geoff Thorpe]
 5773 
 5774   *) Because of the callback-based approach for implementing LHASH as a
 5775      template type, lh_insert() adds opaque objects to hash-tables and
 5776      lh_doall() or lh_doall_arg() are typically used with a destructor callback
 5777      to clean up those corresponding objects before destroying the hash table
 5778      (and losing the object pointers). So some over-zealous constifications in
 5779      LHASH have been relaxed so that lh_insert() does not take (nor store) the
 5780      objects as "const" and the lh_doall[_arg] callback wrappers are not
 5781      prototyped to have "const" restrictions on the object pointers they are
 5782      given (and so aren't required to cast them away any more).
 5783      [Geoff Thorpe]
 5784 
 5785   *) The tmdiff.h API was so ugly and minimal that our own timing utility
 5786      (speed) prefers to use its own implementation. The two implementations
 5787      haven't been consolidated as yet (volunteers?) but the tmdiff API has had
 5788      its object type properly exposed (MS_TM) instead of casting to/from "char
 5789      *". This may still change yet if someone realises MS_TM and "ms_time_***"
 5790      aren't necessarily the greatest nomenclatures - but this is what was used
 5791      internally to the implementation so I've used that for now.
 5792      [Geoff Thorpe]
 5793 
 5794   *) Ensure that deprecated functions do not get compiled when
 5795      OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
 5796      the self-tests were still using deprecated key-generation functions so
 5797      these have been updated also.
 5798      [Geoff Thorpe]
 5799 
 5800   *) Reorganise PKCS#7 code to separate the digest location functionality
 5801      into PKCS7_find_digest(), digest addition into PKCS7_bio_add_digest().
 5802      New function PKCS7_set_digest() to set the digest type for PKCS#7
 5803      digestedData type. Add additional code to correctly generate the
 5804      digestedData type and add support for this type in PKCS7 initialization
 5805      functions.
 5806      [Steve Henson]
 5807 
 5808   *) New function PKCS7_set0_type_other() this initializes a PKCS7
 5809      structure of type "other".
 5810      [Steve Henson]
 5811 
 5812   *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
 5813      sure the loop does correctly stop and breaking ("division by zero")
 5814      modulus operations are not performed. The (pre-generated) prime
 5815      table crypto/bn/bn_prime.h was already correct, but it could not be
 5816      re-generated on some platforms because of the "division by zero"
 5817      situation in the script.
 5818      [Ralf S. Engelschall]
 5819 
 5820   *) Update support for ECC-based TLS ciphersuites according to
 5821      draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
 5822      SHA-1 now is only used for "small" curves (where the
 5823      representation of a field element takes up to 24 bytes); for
 5824      larger curves, the field element resulting from ECDH is directly
 5825      used as premaster secret.
 5826      [Douglas Stebila (Sun Microsystems Laboratories)]
 5827 
 5828   *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
 5829      curve secp160r1 to the tests.
 5830      [Douglas Stebila (Sun Microsystems Laboratories)]
 5831 
 5832   *) Add the possibility to load symbols globally with DSO.
 5833      [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
 5834 
 5835   *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
 5836      control of the error stack.
 5837      [Richard Levitte]
 5838 
 5839   *) Add support for STORE in ENGINE.
 5840      [Richard Levitte]
 5841 
 5842   *) Add the STORE type.  The intention is to provide a common interface
 5843      to certificate and key stores, be they simple file-based stores, or
 5844      HSM-type store, or LDAP stores, or...
 5845      NOTE: The code is currently UNTESTED and isn't really used anywhere.
 5846      [Richard Levitte]
 5847 
 5848   *) Add a generic structure called OPENSSL_ITEM.  This can be used to
 5849      pass a list of arguments to any function as well as provide a way
 5850      for a function to pass data back to the caller.
 5851      [Richard Levitte]
 5852 
 5853   *) Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
 5854      works like BUF_strdup() but can be used to duplicate a portion of
 5855      a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
 5856      a memory area.
 5857      [Richard Levitte]
 5858 
 5859   *) Add the function sk_find_ex() which works like sk_find(), but will
 5860      return an index to an element even if an exact match couldn't be
 5861      found.  The index is guaranteed to point at the element where the
 5862      searched-for key would be inserted to preserve sorting order.
 5863      [Richard Levitte]
 5864 
 5865   *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
 5866      takes an extra flags argument for optional functionality.  Currently,
 5867      the following flags are defined:
 5868 
 5869         OBJ_BSEARCH_VALUE_ON_NOMATCH
 5870         This one gets OBJ_bsearch_ex() to return a pointer to the first
 5871         element where the comparing function returns a negative or zero
 5872         number.
 5873 
 5874         OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
 5875         This one gets OBJ_bsearch_ex() to return a pointer to the first
 5876         element where the comparing function returns zero.  This is useful
 5877         if there are more than one element where the comparing function
 5878         returns zero.
 5879      [Richard Levitte]
 5880 
 5881   *) Make it possible to create self-signed certificates with 'openssl ca'
 5882      in such a way that the self-signed certificate becomes part of the
 5883      CA database and uses the same mechanisms for serial number generation
 5884      as all other certificate signing.  The new flag '-selfsign' enables
 5885      this functionality.  Adapt CA.sh and CA.pl.in.
 5886      [Richard Levitte]
 5887 
 5888   *) Add functionality to check the public key of a certificate request
 5889      against a given private.  This is useful to check that a certificate
 5890      request can be signed by that key (self-signing).
 5891      [Richard Levitte]
 5892 
 5893   *) Make it possible to have multiple active certificates with the same
 5894      subject in the CA index file.  This is done only if the keyword
 5895      'unique_subject' is set to 'no' in the main CA section (default
 5896      if 'CA_default') of the configuration file.  The value is saved
 5897      with the database itself in a separate index attribute file,
 5898      named like the index file with '.attr' appended to the name.
 5899      [Richard Levitte]
 5900 
 5901   *) Generate multi-valued AVAs using '+' notation in config files for
 5902      req and dirName.
 5903      [Steve Henson]
 5904 
 5905   *) Support for nameConstraints certificate extension.
 5906      [Steve Henson]
 5907 
 5908   *) Support for policyConstraints certificate extension.
 5909      [Steve Henson]
 5910 
 5911   *) Support for policyMappings certificate extension.
 5912      [Steve Henson]
 5913 
 5914   *) Make sure the default DSA_METHOD implementation only uses its
 5915      dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
 5916      and change its own handlers to be NULL so as to remove unnecessary
 5917      indirection. This lets alternative implementations fallback to the
 5918      default implementation more easily.
 5919      [Geoff Thorpe]
 5920 
 5921   *) Support for directoryName in GeneralName related extensions
 5922      in config files.
 5923      [Steve Henson]
 5924 
 5925   *) Make it possible to link applications using Makefile.shared.
 5926      Make that possible even when linking against static libraries!
 5927      [Richard Levitte]
 5928 
 5929   *) Support for single pass processing for S/MIME signing. This now
 5930      means that S/MIME signing can be done from a pipe, in addition
 5931      cleartext signing (multipart/signed type) is effectively streaming
 5932      and the signed data does not need to be all held in memory.
 5933 
 5934      This is done with a new flag PKCS7_STREAM. When this flag is set
 5935      PKCS7_sign() only initializes the PKCS7 structure and the actual signing
 5936      is done after the data is output (and digests calculated) in
 5937      SMIME_write_PKCS7().
 5938      [Steve Henson]
 5939 
 5940   *) Add full support for -rpath/-R, both in shared libraries and
 5941      applications, at least on the platforms where it's known how
 5942      to do it.
 5943      [Richard Levitte]
 5944 
 5945   *) In crypto/ec/ec_mult.c, implement fast point multiplication with
 5946      precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
 5947      will now compute a table of multiples of the generator that
 5948      makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
 5949      faster (notably in the case of a single point multiplication,
 5950      scalar * generator).
 5951      [Nils Larsch, Bodo Moeller]
 5952 
 5953   *) IPv6 support for certificate extensions. The various extensions
 5954      which use the IP:a.b.c.d can now take IPv6 addresses using the
 5955      formats of RFC1884 2.2 . IPv6 addresses are now also displayed
 5956      correctly.
 5957      [Steve Henson]
 5958 
 5959   *) Added an ENGINE that implements RSA by performing private key
 5960      exponentiations with the GMP library. The conversions to and from
 5961      GMP's mpz_t format aren't optimised nor are any montgomery forms
 5962      cached, and on x86 it appears OpenSSL's own performance has caught up.
 5963      However there are likely to be other architectures where GMP could
 5964      provide a boost. This ENGINE is not built in by default, but it can be
 5965      specified at Configure time and should be accompanied by the necessary
 5966      linker additions, eg;
 5967          ./config -DOPENSSL_USE_GMP -lgmp
 5968      [Geoff Thorpe]
 5969 
 5970   *) "openssl engine" will not display ENGINE/DSO load failure errors when
 5971      testing availability of engines with "-t" - the old behaviour is
 5972      produced by increasing the feature's verbosity with "-tt".
 5973      [Geoff Thorpe]
 5974 
 5975   *) ECDSA routines: under certain error conditions uninitialized BN objects
 5976      could be freed. Solution: make sure initialization is performed early
 5977      enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
 5978      via PR#459)
 5979      [Lutz Jaenicke]
 5980 
 5981   *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
 5982      and DH_METHOD (eg. by ENGINE implementations) to override the normal
 5983      software implementations. For DSA and DH, parameter generation can
 5984      also be overridden by providing the appropriate method callbacks.
 5985      [Geoff Thorpe]
 5986 
 5987   *) Change the "progress" mechanism used in key-generation and
 5988      primality testing to functions that take a new BN_GENCB pointer in
 5989      place of callback/argument pairs. The new API functions have "_ex"
 5990      postfixes and the older functions are reimplemented as wrappers for
 5991      the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
 5992      declarations of the old functions to help (graceful) attempts to
 5993      migrate to the new functions. Also, the new key-generation API
 5994      functions operate on a caller-supplied key-structure and return
 5995      success/failure rather than returning a key or NULL - this is to
 5996      help make "keygen" another member function of RSA_METHOD etc.
 5997 
 5998      Example for using the new callback interface:
 5999 
 6000           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
 6001           void *my_arg = ...;
 6002           BN_GENCB my_cb;
 6003 
 6004           BN_GENCB_set(&my_cb, my_callback, my_arg);
 6005 
 6006           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
 6007           /* For the meaning of a, b in calls to my_callback(), see the
 6008            * documentation of the function that calls the callback.
 6009            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
 6010            * my_callback should return 1 if it wants BN_is_prime_ex()
 6011            * to continue, or 0 to stop.
 6012            */
 6013 
 6014      [Geoff Thorpe]
 6015 
 6016   *) Change the ZLIB compression method to be stateful, and make it
 6017      available to TLS with the number defined in
 6018      draft-ietf-tls-compression-04.txt.
 6019      [Richard Levitte]
 6020 
 6021   *) Add the ASN.1 structures and functions for CertificatePair, which
 6022      is defined as follows (according to X.509_4thEditionDraftV6.pdf):
 6023 
 6024      CertificatePair ::= SEQUENCE {
 6025         forward         [0]     Certificate OPTIONAL,
 6026         reverse         [1]     Certificate OPTIONAL,
 6027         -- at least one of the pair shall be present -- }
 6028 
 6029      Also implement the PEM functions to read and write certificate
 6030      pairs, and defined the PEM tag as "CERTIFICATE PAIR".
 6031 
 6032      This needed to be defined, mostly for the sake of the LDAP
 6033      attribute crossCertificatePair, but may prove useful elsewhere as
 6034      well.
 6035      [Richard Levitte]
 6036 
 6037   *) Make it possible to inhibit symlinking of shared libraries in
 6038      Makefile.shared, for Cygwin's sake.
 6039      [Richard Levitte]
 6040 
 6041   *) Extend the BIGNUM API by creating a function
 6042           void BN_set_negative(BIGNUM *a, int neg);
 6043      and a macro that behave like
 6044           int  BN_is_negative(const BIGNUM *a);
 6045 
 6046      to avoid the need to access 'a->neg' directly in applications.
 6047      [Nils Larsch]
 6048 
 6049   *) Implement fast modular reduction for pseudo-Mersenne primes
 6050      used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
 6051      EC_GROUP_new_curve_GFp() will now automatically use this
 6052      if applicable.
 6053      [Nils Larsch <nla@trustcenter.de>]
 6054 
 6055   *) Add new lock type (CRYPTO_LOCK_BN).
 6056      [Bodo Moeller]
 6057 
 6058   *) Change the ENGINE framework to automatically load engines
 6059      dynamically from specific directories unless they could be
 6060      found to already be built in or loaded.  Move all the
 6061      current engines except for the cryptodev one to a new
 6062      directory engines/.
 6063      The engines in engines/ are built as shared libraries if
 6064      the "shared" options was given to ./Configure or ./config.
 6065      Otherwise, they are inserted in libcrypto.a.
 6066      /usr/local/ssl/engines is the default directory for dynamic
 6067      engines, but that can be overridden at configure time through
 6068      the usual use of --prefix and/or --openssldir, and at run
 6069      time with the environment variable OPENSSL_ENGINES.
 6070      [Geoff Thorpe and Richard Levitte]
 6071 
 6072   *) Add Makefile.shared, a helper makefile to build shared
 6073      libraries.  Adapt Makefile.org.
 6074      [Richard Levitte]
 6075 
 6076   *) Add version info to Win32 DLLs.
 6077      [Peter 'Luna' Runestig" <peter@runestig.com>]
 6078 
 6079   *) Add new 'medium level' PKCS#12 API. Certificates and keys
 6080      can be added using this API to created arbitrary PKCS#12
 6081      files while avoiding the low level API.
 6082 
 6083      New options to PKCS12_create(), key or cert can be NULL and
 6084      will then be omitted from the output file. The encryption
 6085      algorithm NIDs can be set to -1 for no encryption, the mac
 6086      iteration count can be set to 0 to omit the mac.
 6087 
 6088      Enhance pkcs12 utility by making the -nokeys and -nocerts
 6089      options work when creating a PKCS#12 file. New option -nomac
 6090      to omit the mac, NONE can be set for an encryption algorithm.
 6091      New code is modified to use the enhanced PKCS12_create()
 6092      instead of the low level API.
 6093      [Steve Henson]
 6094 
 6095   *) Extend ASN1 encoder to support indefinite length constructed
 6096      encoding. This can output sequences tags and octet strings in
 6097      this form. Modify pk7_asn1.c to support indefinite length
 6098      encoding. This is experimental and needs additional code to
 6099      be useful, such as an ASN1 bio and some enhanced streaming
 6100      PKCS#7 code.
 6101 
 6102      Extend template encode functionality so that tagging is passed
 6103      down to the template encoder.
 6104      [Steve Henson]
 6105 
 6106   *) Let 'openssl req' fail if an argument to '-newkey' is not
 6107      recognized instead of using RSA as a default.
 6108      [Bodo Moeller]
 6109 
 6110   *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
 6111      As these are not official, they are not included in "ALL";
 6112      the "ECCdraft" ciphersuite group alias can be used to select them.
 6113      [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
 6114 
 6115   *) Add ECDH engine support.
 6116      [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
 6117 
 6118   *) Add ECDH in new directory crypto/ecdh/.
 6119      [Douglas Stebila (Sun Microsystems Laboratories)]
 6120 
 6121   *) Let BN_rand_range() abort with an error after 100 iterations
 6122      without success (which indicates a broken PRNG).
 6123      [Bodo Moeller]
 6124 
 6125   *) Change BN_mod_sqrt() so that it verifies that the input value
 6126      is really the square of the return value.  (Previously,
 6127      BN_mod_sqrt would show GIGO behaviour.)
 6128      [Bodo Moeller]
 6129 
 6130   *) Add named elliptic curves over binary fields from X9.62, SECG,
 6131      and WAP/WTLS; add OIDs that were still missing.
 6132 
 6133      [Sheueling Chang Shantz and Douglas Stebila
 6134      (Sun Microsystems Laboratories)]
 6135 
 6136   *) Extend the EC library for elliptic curves over binary fields
 6137      (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
 6138      New EC_METHOD:
 6139 
 6140           EC_GF2m_simple_method
 6141 
 6142      New API functions:
 6143 
 6144           EC_GROUP_new_curve_GF2m
 6145           EC_GROUP_set_curve_GF2m
 6146           EC_GROUP_get_curve_GF2m
 6147           EC_POINT_set_affine_coordinates_GF2m
 6148           EC_POINT_get_affine_coordinates_GF2m
 6149           EC_POINT_set_compressed_coordinates_GF2m
 6150 
 6151      Point compression for binary fields is disabled by default for
 6152      patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
 6153      enable it).
 6154 
 6155      As binary polynomials are represented as BIGNUMs, various members
 6156      of the EC_GROUP and EC_POINT data structures can be shared
 6157      between the implementations for prime fields and binary fields;
 6158      the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
 6159      are essentially identical to their ..._GFp counterparts.
 6160      (For simplicity, the '..._GFp' prefix has been dropped from
 6161      various internal method names.)
 6162 
 6163      An internal 'field_div' method (similar to 'field_mul' and
 6164      'field_sqr') has been added; this is used only for binary fields.
 6165 
 6166      [Sheueling Chang Shantz and Douglas Stebila
 6167      (Sun Microsystems Laboratories)]
 6168 
 6169   *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
 6170      through methods ('mul', 'precompute_mult').
 6171 
 6172      The generic implementations (now internally called 'ec_wNAF_mul'
 6173      and 'ec_wNAF_precomputed_mult') remain the default if these
 6174      methods are undefined.
 6175 
 6176      [Sheueling Chang Shantz and Douglas Stebila
 6177      (Sun Microsystems Laboratories)]
 6178 
 6179   *) New function EC_GROUP_get_degree, which is defined through
 6180      EC_METHOD.  For curves over prime fields, this returns the bit
 6181      length of the modulus.
 6182 
 6183      [Sheueling Chang Shantz and Douglas Stebila
 6184      (Sun Microsystems Laboratories)]
 6185 
 6186   *) New functions EC_GROUP_dup, EC_POINT_dup.
 6187      (These simply call ..._new  and ..._copy).
 6188 
 6189      [Sheueling Chang Shantz and Douglas Stebila
 6190      (Sun Microsystems Laboratories)]
 6191 
 6192   *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
 6193      Polynomials are represented as BIGNUMs (where the sign bit is not
 6194      used) in the following functions [macros]:
 6195 
 6196           BN_GF2m_add
 6197           BN_GF2m_sub             [= BN_GF2m_add]
 6198           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
 6199           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
 6200           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
 6201           BN_GF2m_mod_inv
 6202           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
 6203           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
 6204           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
 6205           BN_GF2m_cmp             [= BN_ucmp]
 6206 
 6207      (Note that only the 'mod' functions are actually for fields GF(2^m).
 6208      BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
 6209 
 6210      For some functions, an the irreducible polynomial defining a
 6211      field can be given as an 'unsigned int[]' with strictly
 6212      decreasing elements giving the indices of those bits that are set;
 6213      i.e., p[] represents the polynomial
 6214           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
 6215      where
 6216           p[0] > p[1] > ... > p[k] = 0.
 6217      This applies to the following functions:
 6218 
 6219           BN_GF2m_mod_arr
 6220           BN_GF2m_mod_mul_arr
 6221           BN_GF2m_mod_sqr_arr
 6222           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
 6223           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
 6224           BN_GF2m_mod_exp_arr
 6225           BN_GF2m_mod_sqrt_arr
 6226           BN_GF2m_mod_solve_quad_arr
 6227           BN_GF2m_poly2arr
 6228           BN_GF2m_arr2poly
 6229 
 6230      Conversion can be performed by the following functions:
 6231 
 6232           BN_GF2m_poly2arr
 6233           BN_GF2m_arr2poly
 6234 
 6235      bntest.c has additional tests for binary polynomial arithmetic.
 6236 
 6237      Two implementations for BN_GF2m_mod_div() are available.
 6238      The default algorithm simply uses BN_GF2m_mod_inv() and
 6239      BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
 6240      if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
 6241      copyright notice in crypto/bn/bn_gf2m.c before enabling it).
 6242 
 6243      [Sheueling Chang Shantz and Douglas Stebila
 6244      (Sun Microsystems Laboratories)]
 6245 
 6246   *) Add new error code 'ERR_R_DISABLED' that can be used when some
 6247      functionality is disabled at compile-time.
 6248      [Douglas Stebila <douglas.stebila@sun.com>]
 6249 
 6250   *) Change default behaviour of 'openssl asn1parse' so that more
 6251      information is visible when viewing, e.g., a certificate:
 6252 
 6253      Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
 6254      mode the content of non-printable OCTET STRINGs is output in a
 6255      style similar to INTEGERs, but with '[HEX DUMP]' prepended to
 6256      avoid the appearance of a printable string.
 6257      [Nils Larsch <nla@trustcenter.de>]
 6258 
 6259   *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
 6260      functions
 6261           EC_GROUP_set_asn1_flag()
 6262           EC_GROUP_get_asn1_flag()
 6263           EC_GROUP_set_point_conversion_form()
 6264           EC_GROUP_get_point_conversion_form()
 6265      These control ASN1 encoding details:
 6266      - Curves (i.e., groups) are encoded explicitly unless asn1_flag
 6267        has been set to OPENSSL_EC_NAMED_CURVE.
 6268      - Points are encoded in uncompressed form by default; options for
 6269        asn1_for are as for point2oct, namely
 6270           POINT_CONVERSION_COMPRESSED
 6271           POINT_CONVERSION_UNCOMPRESSED
 6272           POINT_CONVERSION_HYBRID
 6273 
 6274      Also add 'seed' and 'seed_len' members to EC_GROUP with access
 6275      functions
 6276           EC_GROUP_set_seed()
 6277           EC_GROUP_get0_seed()
 6278           EC_GROUP_get_seed_len()
 6279      This is used only for ASN1 purposes (so far).
 6280      [Nils Larsch <nla@trustcenter.de>]
 6281 
 6282   *) Add 'field_type' member to EC_METHOD, which holds the NID
 6283      of the appropriate field type OID.  The new function
 6284      EC_METHOD_get_field_type() returns this value.
 6285      [Nils Larsch <nla@trustcenter.de>]
 6286 
 6287   *) Add functions
 6288           EC_POINT_point2bn()
 6289           EC_POINT_bn2point()
 6290           EC_POINT_point2hex()
 6291           EC_POINT_hex2point()
 6292      providing useful interfaces to EC_POINT_point2oct() and
 6293      EC_POINT_oct2point().
 6294      [Nils Larsch <nla@trustcenter.de>]
 6295 
 6296   *) Change internals of the EC library so that the functions
 6297           EC_GROUP_set_generator()
 6298           EC_GROUP_get_generator()
 6299           EC_GROUP_get_order()
 6300           EC_GROUP_get_cofactor()
 6301      are implemented directly in crypto/ec/ec_lib.c and not dispatched
 6302      to methods, which would lead to unnecessary code duplication when
 6303      adding different types of curves.
 6304      [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
 6305 
 6306   *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
 6307      arithmetic, and such that modified wNAFs are generated
 6308      (which avoid length expansion in many cases).
 6309      [Bodo Moeller]
 6310 
 6311   *) Add a function EC_GROUP_check_discriminant() (defined via
 6312      EC_METHOD) that verifies that the curve discriminant is non-zero.
 6313 
 6314      Add a function EC_GROUP_check() that makes some sanity tests
 6315      on a EC_GROUP, its generator and order.  This includes
 6316      EC_GROUP_check_discriminant().
 6317      [Nils Larsch <nla@trustcenter.de>]
 6318 
 6319   *) Add ECDSA in new directory crypto/ecdsa/.
 6320 
 6321      Add applications 'openssl ecparam' and 'openssl ecdsa'
 6322      (these are based on 'openssl dsaparam' and 'openssl dsa').
 6323 
 6324      ECDSA support is also included in various other files across the
 6325      library.  Most notably,
 6326      - 'openssl req' now has a '-newkey ecdsa:file' option;
 6327      - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
 6328      - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
 6329        d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
 6330        them suitable for ECDSA where domain parameters must be
 6331        extracted before the specific public key;
 6332      - ECDSA engine support has been added.
 6333      [Nils Larsch <nla@trustcenter.de>]
 6334 
 6335   *) Include some named elliptic curves, and add OIDs from X9.62,
 6336      SECG, and WAP/WTLS.  Each curve can be obtained from the new
 6337      function
 6338           EC_GROUP_new_by_curve_name(),
 6339      and the list of available named curves can be obtained with
 6340           EC_get_builtin_curves().
 6341      Also add a 'curve_name' member to EC_GROUP objects, which can be
 6342      accessed via
 6343          EC_GROUP_set_curve_name()
 6344          EC_GROUP_get_curve_name()
 6345      [Nils Larsch <larsch@trustcenter.de, Bodo Moeller]
 6346 
 6347   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
 6348      was actually never needed) and in BN_mul().  The removal in BN_mul()
 6349      required a small change in bn_mul_part_recursive() and the addition
 6350      of the functions bn_cmp_part_words(), bn_sub_part_words() and
 6351      bn_add_part_words(), which do the same thing as bn_cmp_words(),
 6352      bn_sub_words() and bn_add_words() except they take arrays with
 6353      differing sizes.
 6354      [Richard Levitte]
 6355 
 6356  Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
 6357 
 6358   *) Cleanse PEM buffers before freeing them since they may contain
 6359      sensitive data.
 6360      [Benjamin Bennett <ben@psc.edu>]
 6361 
 6362   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
 6363      a ciphersuite string such as "DEFAULT:RSA" cannot enable
 6364      authentication-only ciphersuites.
 6365      [Bodo Moeller]
 6366 
 6367   *) Since AES128 and AES256 share a single mask bit in the logic of
 6368      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
 6369      kludge to work properly if AES128 is available and AES256 isn't.
 6370      [Victor Duchovni]
 6371 
 6372   *) Expand security boundary to match 1.1.1 module.
 6373      [Steve Henson]
 6374 
 6375   *) Remove redundant features: hash file source, editing of test vectors
 6376      modify fipsld to use external fips_premain.c signature.
 6377      [Steve Henson]
 6378 
 6379   *) New perl script mkfipsscr.pl to create shell scripts or batch files to
 6380      run algorithm test programs.
 6381      [Steve Henson]
 6382 
 6383   *) Make algorithm test programs more tolerant of whitespace.
 6384      [Steve Henson]
 6385 
 6386   *) Have SSL/TLS server implementation tolerate "mismatched" record
 6387      protocol version while receiving ClientHello even if the
 6388      ClientHello is fragmented.  (The server can't insist on the
 6389      particular protocol version it has chosen before the ServerHello
 6390      message has informed the client about his choice.)
 6391      [Bodo Moeller]
 6392 
 6393   *) Load error codes if they are not already present instead of using a
 6394      static variable. This allows them to be cleanly unloaded and reloaded.
 6395      [Steve Henson]
 6396 
 6397  Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
 6398 
 6399   *) Introduce limits to prevent malicious keys being able to
 6400      cause a denial of service.  (CVE-2006-2940)
 6401      [Steve Henson, Bodo Moeller]
 6402 
 6403   *) Fix ASN.1 parsing of certain invalid structures that can result
 6404      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
 6405 
 6406   *) Fix buffer overflow in SSL_get_shared_ciphers() function.
 6407      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
 6408 
 6409   *) Fix SSL client code which could crash if connecting to a
 6410      malicious SSLv2 server.  (CVE-2006-4343)
 6411      [Tavis Ormandy and Will Drewry, Google Security Team]
 6412 
 6413   *) Change ciphersuite string processing so that an explicit
 6414      ciphersuite selects this one ciphersuite (so that "AES256-SHA"
 6415      will no longer include "AES128-SHA"), and any other similar
 6416      ciphersuite (same bitmap) from *other* protocol versions (so that
 6417      "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
 6418      SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
 6419      changes from 0.9.8b and 0.9.8d.
 6420      [Bodo Moeller]
 6421 
 6422  Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
 6423 
 6424   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
 6425      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
 6426 
 6427   *) Change the Unix randomness entropy gathering to use poll() when
 6428      possible instead of select(), since the latter has some
 6429      undesirable limitations.
 6430      [Darryl Miles via Richard Levitte and Bodo Moeller]
 6431 
 6432   *) Disable rogue ciphersuites:
 6433 
 6434       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
 6435       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
 6436       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
 6437 
 6438      The latter two were purportedly from
 6439      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
 6440      appear there.
 6441 
 6442      Also deactivate the remaining ciphersuites from
 6443      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
 6444      unofficial, and the ID has long expired.
 6445      [Bodo Moeller]
 6446 
 6447   *) Fix RSA blinding Heisenbug (problems sometimes occurred on
 6448      dual-core machines) and other potential thread-safety issues.
 6449      [Bodo Moeller]
 6450 
 6451  Changes between 0.9.7i and 0.9.7j  [04 May 2006]
 6452 
 6453   *) Adapt fipsld and the build system to link against the validated FIPS
 6454      module in FIPS mode.
 6455      [Steve Henson]
 6456 
 6457   *) Fixes for VC++ 2005 build under Windows.
 6458      [Steve Henson]
 6459 
 6460   *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
 6461      from a Windows bash shell such as MSYS. It is autodetected from the
 6462      "config" script when run from a VC++ environment. Modify standard VC++
 6463      build to use fipscanister.o from the GNU make build.
 6464      [Steve Henson]
 6465 
 6466  Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
 6467 
 6468   *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
 6469      The value now differs depending on if you build for FIPS or not.
 6470      BEWARE!  A program linked with a shared FIPSed libcrypto can't be
 6471      safely run with a non-FIPSed libcrypto, as it may crash because of
 6472      the difference induced by this change.
 6473      [Andy Polyakov]
 6474 
 6475  Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
 6476 
 6477   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
 6478      (part of SSL_OP_ALL).  This option used to disable the
 6479      countermeasure against man-in-the-middle protocol-version
 6480      rollback in the SSL 2.0 server implementation, which is a bad
 6481      idea.  (CVE-2005-2969)
 6482 
 6483      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
 6484      for Information Security, National Institute of Advanced Industrial
 6485      Science and Technology [AIST], Japan)]
 6486 
 6487   *) Minimal support for X9.31 signatures and PSS padding modes. This is
 6488      mainly for FIPS compliance and not fully integrated at this stage.
 6489      [Steve Henson]
 6490 
 6491   *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
 6492      the exponentiation using a fixed-length exponent.  (Otherwise,
 6493      the information leaked through timing could expose the secret key
 6494      after many signatures; cf. Bleichenbacher's attack on DSA with
 6495      biased k.)
 6496      [Bodo Moeller]
 6497 
 6498   *) Make a new fixed-window mod_exp implementation the default for
 6499      RSA, DSA, and DH private-key operations so that the sequence of
 6500      squares and multiplies and the memory access pattern are
 6501      independent of the particular secret key.  This will mitigate
 6502      cache-timing and potential related attacks.
 6503 
 6504      BN_mod_exp_mont_consttime() is the new exponentiation implementation,
 6505      and this is automatically used by BN_mod_exp_mont() if the new flag
 6506      BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
 6507      will use this BN flag for private exponents unless the flag
 6508      RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
 6509      DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
 6510 
 6511      [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
 6512 
 6513   *) Change the client implementation for SSLv23_method() and
 6514      SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
 6515      Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
 6516      (Previously, the SSL 2.0 backwards compatible Client Hello
 6517      message format would be used even with SSL_OP_NO_SSLv2.)
 6518      [Bodo Moeller]
 6519 
 6520   *) Add support for smime-type MIME parameter in S/MIME messages which some
 6521      clients need.
 6522      [Steve Henson]
 6523 
 6524   *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
 6525      a threadsafe manner. Modify rsa code to use new function and add calls
 6526      to dsa and dh code (which had race conditions before).
 6527      [Steve Henson]
 6528 
 6529   *) Include the fixed error library code in the C error file definitions
 6530      instead of fixing them up at runtime. This keeps the error code
 6531      structures constant.
 6532      [Steve Henson]
 6533 
 6534  Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
 6535 
 6536   [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
 6537   OpenSSL 0.9.8.]
 6538 
 6539   *) Fixes for newer kerberos headers. NB: the casts are needed because
 6540      the 'length' field is signed on one version and unsigned on another
 6541      with no (?) obvious way to tell the difference, without these VC++
 6542      complains. Also the "definition" of FAR (blank) is no longer included
 6543      nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
 6544      some needed definitions.
 6545      [Steve Henson]
 6546 
 6547   *) Undo Cygwin change.
 6548      [Ulf Möller]
 6549 
 6550   *) Added support for proxy certificates according to RFC 3820.
 6551      Because they may be a security thread to unaware applications,
 6552      they must be explicitly allowed in run-time.  See
 6553      docs/HOWTO/proxy_certificates.txt for further information.
 6554      [Richard Levitte]
 6555 
 6556  Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
 6557 
 6558   *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
 6559      server and client random values. Previously
 6560      (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
 6561      less random data when sizeof(time_t) > 4 (some 64 bit platforms).
 6562 
 6563      This change has negligible security impact because:
 6564 
 6565      1. Server and client random values still have 24 bytes of pseudo random
 6566         data.
 6567 
 6568      2. Server and client random values are sent in the clear in the initial
 6569         handshake.
 6570 
 6571      3. The master secret is derived using the premaster secret (48 bytes in
 6572         size for static RSA ciphersuites) as well as client server and random
 6573         values.
 6574 
 6575      The OpenSSL team would like to thank the UK NISCC for bringing this issue
 6576      to our attention.
 6577 
 6578      [Stephen Henson, reported by UK NISCC]
 6579 
 6580   *) Use Windows randomness collection on Cygwin.
 6581      [Ulf Möller]
 6582 
 6583   *) Fix hang in EGD/PRNGD query when communication socket is closed
 6584      prematurely by EGD/PRNGD.
 6585      [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
 6586 
 6587   *) Prompt for pass phrases when appropriate for PKCS12 input format.
 6588      [Steve Henson]
 6589 
 6590   *) Back-port of selected performance improvements from development
 6591      branch, as well as improved support for PowerPC platforms.
 6592      [Andy Polyakov]
 6593 
 6594   *) Add lots of checks for memory allocation failure, error codes to indicate
 6595      failure and freeing up memory if a failure occurs.
 6596      [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
 6597 
 6598   *) Add new -passin argument to dgst.
 6599      [Steve Henson]
 6600 
 6601   *) Perform some character comparisons of different types in X509_NAME_cmp:
 6602      this is needed for some certificates that re-encode DNs into UTF8Strings
 6603      (in violation of RFC3280) and can't or won't issue name rollover
 6604      certificates.
 6605      [Steve Henson]
 6606 
 6607   *) Make an explicit check during certificate validation to see that
 6608      the CA setting in each certificate on the chain is correct.  As a
 6609      side effect always do the following basic checks on extensions,
 6610      not just when there's an associated purpose to the check:
 6611 
 6612       - if there is an unhandled critical extension (unless the user
 6613         has chosen to ignore this fault)
 6614       - if the path length has been exceeded (if one is set at all)
 6615       - that certain extensions fit the associated purpose (if one has
 6616         been given)
 6617      [Richard Levitte]
 6618 
 6619  Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
 6620 
 6621   *) Avoid a race condition when CRLs are checked in a multi threaded
 6622      environment. This would happen due to the reordering of the revoked
 6623      entries during signature checking and serial number lookup. Now the
 6624      encoding is cached and the serial number sort performed under a lock.
 6625      Add new STACK function sk_is_sorted().
 6626      [Steve Henson]
 6627 
 6628   *) Add Delta CRL to the extension code.
 6629      [Steve Henson]
 6630 
 6631   *) Various fixes to s3_pkt.c so alerts are sent properly.
 6632      [David Holmes <d.holmes@f5.com>]
 6633 
 6634   *) Reduce the chances of duplicate issuer name and serial numbers (in
 6635      violation of RFC3280) using the OpenSSL certificate creation utilities.
 6636      This is done by creating a random 64 bit value for the initial serial
 6637      number when a serial number file is created or when a self signed
 6638      certificate is created using 'openssl req -x509'. The initial serial
 6639      number file is created using 'openssl x509 -next_serial' in CA.pl
 6640      rather than being initialized to 1.
 6641      [Steve Henson]
 6642 
 6643  Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
 6644 
 6645   *) Fix null-pointer assignment in do_change_cipher_spec() revealed
 6646      by using the Codenomicon TLS Test Tool (CVE-2004-0079)
 6647      [Joe Orton, Steve Henson]
 6648 
 6649   *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
 6650      (CVE-2004-0112)
 6651      [Joe Orton, Steve Henson]
 6652 
 6653   *) Make it possible to have multiple active certificates with the same
 6654      subject in the CA index file.  This is done only if the keyword
 6655      'unique_subject' is set to 'no' in the main CA section (default
 6656      if 'CA_default') of the configuration file.  The value is saved
 6657      with the database itself in a separate index attribute file,
 6658      named like the index file with '.attr' appended to the name.
 6659      [Richard Levitte]
 6660 
 6661   *) X509 verify fixes. Disable broken certificate workarounds when
 6662      X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
 6663      keyUsage extension present. Don't accept CRLs with unhandled critical
 6664      extensions: since verify currently doesn't process CRL extensions this
 6665      rejects a CRL with *any* critical extensions. Add new verify error codes
 6666      for these cases.
 6667      [Steve Henson]
 6668 
 6669   *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
 6670      A clarification of RFC2560 will require the use of OCTET STRINGs and
 6671      some implementations cannot handle the current raw format. Since OpenSSL
 6672      copies and compares OCSP nonces as opaque blobs without any attempt at
 6673      parsing them this should not create any compatibility issues.
 6674      [Steve Henson]
 6675 
 6676   *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
 6677      calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
 6678      this HMAC (and other) operations are several times slower than OpenSSL
 6679      < 0.9.7.
 6680      [Steve Henson]
 6681 
 6682   *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
 6683      [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
 6684 
 6685   *) Use the correct content when signing type "other".
 6686      [Steve Henson]
 6687 
 6688  Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
 6689 
 6690   *) Fix various bugs revealed by running the NISCC test suite:
 6691 
 6692      Stop out of bounds reads in the ASN1 code when presented with
 6693      invalid tags (CVE-2003-0543 and CVE-2003-0544).
 6694 
 6695      Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
 6696 
 6697      If verify callback ignores invalid public key errors don't try to check
 6698      certificate signature with the NULL public key.
 6699 
 6700      [Steve Henson]
 6701 
 6702   *) New -ignore_err option in ocsp application to stop the server
 6703      exiting on the first error in a request.
 6704      [Steve Henson]
 6705 
 6706   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
 6707      if the server requested one: as stated in TLS 1.0 and SSL 3.0
 6708      specifications.
 6709      [Steve Henson]
 6710 
 6711   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
 6712      extra data after the compression methods not only for TLS 1.0
 6713      but also for SSL 3.0 (as required by the specification).
 6714      [Bodo Moeller; problem pointed out by Matthias Loepfe]
 6715 
 6716   *) Change X509_certificate_type() to mark the key as exported/exportable
 6717      when it's 512 *bits* long, not 512 bytes.
 6718      [Richard Levitte]
 6719 
 6720   *) Change AES_cbc_encrypt() so it outputs exact multiple of
 6721      blocks during encryption.
 6722      [Richard Levitte]
 6723 
 6724   *) Various fixes to base64 BIO and non blocking I/O. On write
 6725      flushes were not handled properly if the BIO retried. On read
 6726      data was not being buffered properly and had various logic bugs.
 6727      This also affects blocking I/O when the data being decoded is a
 6728      certain size.
 6729      [Steve Henson]
 6730 
 6731   *) Various S/MIME bugfixes and compatibility changes:
 6732      output correct application/pkcs7 MIME type if
 6733      PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
 6734      Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
 6735      of files as .eml work). Correctly handle very long lines in MIME
 6736      parser.
 6737      [Steve Henson]
 6738 
 6739  Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
 6740 
 6741   *) Countermeasure against the Klima-Pokorny-Rosa extension of
 6742      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
 6743      a protocol version number mismatch like a decryption error
 6744      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
 6745      [Bodo Moeller]
 6746 
 6747   *) Turn on RSA blinding by default in the default implementation
 6748      to avoid a timing attack. Applications that don't want it can call
 6749      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
 6750      They would be ill-advised to do so in most cases.
 6751      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
 6752 
 6753   *) Change RSA blinding code so that it works when the PRNG is not
 6754      seeded (in this case, the secret RSA exponent is abused as
 6755      an unpredictable seed -- if it is not unpredictable, there
 6756      is no point in blinding anyway).  Make RSA blinding thread-safe
 6757      by remembering the creator's thread ID in rsa->blinding and
 6758      having all other threads use local one-time blinding factors
 6759      (this requires more computation than sharing rsa->blinding, but
 6760      avoids excessive locking; and if an RSA object is not shared
 6761      between threads, blinding will still be very fast).
 6762      [Bodo Moeller]
 6763 
 6764   *) Fixed a typo bug that would cause ENGINE_set_default() to set an
 6765      ENGINE as defaults for all supported algorithms irrespective of
 6766      the 'flags' parameter. 'flags' is now honoured, so applications
 6767      should make sure they are passing it correctly.
 6768      [Geoff Thorpe]
 6769 
 6770   *) Target "mingw" now allows native Windows code to be generated in
 6771      the Cygwin environment as well as with the MinGW compiler.
 6772      [Ulf Moeller]
 6773 
 6774  Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
 6775 
 6776   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
 6777      via timing by performing a MAC computation even if incorrect
 6778      block cipher padding has been found.  This is a countermeasure
 6779      against active attacks where the attacker has to distinguish
 6780      between bad padding and a MAC verification error. (CVE-2003-0078)
 6781 
 6782      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
 6783      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
 6784      Martin Vuagnoux (EPFL, Ilion)]
 6785 
 6786   *) Make the no-err option work as intended.  The intention with no-err
 6787      is not to have the whole error stack handling routines removed from
 6788      libcrypto, it's only intended to remove all the function name and
 6789      reason texts, thereby removing some of the footprint that may not
 6790      be interesting if those errors aren't displayed anyway.
 6791 
 6792      NOTE: it's still possible for any application or module to have it's
 6793      own set of error texts inserted.  The routines are there, just not
 6794      used by default when no-err is given.
 6795      [Richard Levitte]
 6796 
 6797   *) Add support for FreeBSD on IA64.
 6798      [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
 6799 
 6800   *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
 6801      Kerberos function mit_des_cbc_cksum().  Before this change,
 6802      the value returned by DES_cbc_cksum() was like the one from
 6803      mit_des_cbc_cksum(), except the bytes were swapped.
 6804      [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
 6805 
 6806   *) Allow an application to disable the automatic SSL chain building.
 6807      Before this a rather primitive chain build was always performed in
 6808      ssl3_output_cert_chain(): an application had no way to send the
 6809      correct chain if the automatic operation produced an incorrect result.
 6810 
 6811      Now the chain builder is disabled if either:
 6812 
 6813      1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
 6814 
 6815      2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
 6816 
 6817      The reasoning behind this is that an application would not want the
 6818      auto chain building to take place if extra chain certificates are
 6819      present and it might also want a means of sending no additional
 6820      certificates (for example the chain has two certificates and the
 6821      root is omitted).
 6822      [Steve Henson]
 6823 
 6824   *) Add the possibility to build without the ENGINE framework.
 6825      [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
 6826 
 6827   *) Under Win32 gmtime() can return NULL: check return value in
 6828      OPENSSL_gmtime(). Add error code for case where gmtime() fails.
 6829      [Steve Henson]
 6830 
 6831   *) DSA routines: under certain error conditions uninitialized BN objects
 6832      could be freed. Solution: make sure initialization is performed early
 6833      enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
 6834      Nils Larsch <nla@trustcenter.de> via PR#459)
 6835      [Lutz Jaenicke]
 6836 
 6837   *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
 6838      checked on reconnect on the client side, therefore session resumption
 6839      could still fail with a "ssl session id is different" error. This
 6840      behaviour is masked when SSL_OP_ALL is used due to
 6841      SSL_OP_MICROSOFT_SESS_ID_BUG being set.
 6842      Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
 6843      followup to PR #377.
 6844      [Lutz Jaenicke]
 6845 
 6846   *) IA-32 assembler support enhancements: unified ELF targets, support
 6847      for SCO/Caldera platforms, fix for Cygwin shared build.
 6848      [Andy Polyakov]
 6849 
 6850   *) Add support for FreeBSD on sparc64.  As a consequence, support for
 6851      FreeBSD on non-x86 processors is separate from x86 processors on
 6852      the config script, much like the NetBSD support.
 6853      [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
 6854 
 6855  Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
 6856 
 6857   [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
 6858   OpenSSL 0.9.7.]
 6859 
 6860   *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
 6861      code (06) was taken as the first octet of the session ID and the last
 6862      octet was ignored consequently. As a result SSLv2 client side session
 6863      caching could not have worked due to the session ID mismatch between
 6864      client and server.
 6865      Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
 6866      PR #377.
 6867      [Lutz Jaenicke]
 6868 
 6869   *) Change the declaration of needed Kerberos libraries to use EX_LIBS
 6870      instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
 6871      removed entirely.
 6872      [Richard Levitte]
 6873 
 6874   *) The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
 6875      seems that in spite of existing for more than a year, many application
 6876      author have done nothing to provide the necessary callbacks, which
 6877      means that this particular engine will not work properly anywhere.
 6878      This is a very unfortunate situation which forces us, in the name
 6879      of usability, to give the hw_ncipher.c a static lock, which is part
 6880      of libcrypto.
 6881      NOTE: This is for the 0.9.7 series ONLY.  This hack will never
 6882      appear in 0.9.8 or later.  We EXPECT application authors to have
 6883      dealt properly with this when 0.9.8 is released (unless we actually
 6884      make such changes in the libcrypto locking code that changes will
 6885      have to be made anyway).
 6886      [Richard Levitte]
 6887 
 6888   *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
 6889      octets have been read, EOF or an error occurs. Without this change
 6890      some truncated ASN1 structures will not produce an error.
 6891      [Steve Henson]
 6892 
 6893   *) Disable Heimdal support, since it hasn't been fully implemented.
 6894      Still give the possibility to force the use of Heimdal, but with
 6895      warnings and a request that patches get sent to openssl-dev.
 6896      [Richard Levitte]
 6897 
 6898   *) Add the VC-CE target, introduce the WINCE sysname, and add
 6899      INSTALL.WCE and appropriate conditionals to make it build.
 6900      [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
 6901 
 6902   *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
 6903      cygssl-x.y.z.dll, where x, y and z are the major, minor and
 6904      edit numbers of the version.
 6905      [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
 6906 
 6907   *) Introduce safe string copy and catenation functions
 6908      (BUF_strlcpy() and BUF_strlcat()).
 6909      [Ben Laurie (CHATS) and Richard Levitte]
 6910 
 6911   *) Avoid using fixed-size buffers for one-line DNs.
 6912      [Ben Laurie (CHATS)]
 6913 
 6914   *) Add BUF_MEM_grow_clean() to avoid information leakage when
 6915      resizing buffers containing secrets, and use where appropriate.
 6916      [Ben Laurie (CHATS)]
 6917 
 6918   *) Avoid using fixed size buffers for configuration file location.
 6919      [Ben Laurie (CHATS)]
 6920 
 6921   *) Avoid filename truncation for various CA files.
 6922      [Ben Laurie (CHATS)]
 6923 
 6924   *) Use sizeof in preference to magic numbers.
 6925      [Ben Laurie (CHATS)]
 6926 
 6927   *) Avoid filename truncation in cert requests.
 6928      [Ben Laurie (CHATS)]
 6929 
 6930   *) Add assertions to check for (supposedly impossible) buffer
 6931      overflows.
 6932      [Ben Laurie (CHATS)]
 6933 
 6934   *) Don't cache truncated DNS entries in the local cache (this could
 6935      potentially lead to a spoofing attack).
 6936      [Ben Laurie (CHATS)]
 6937 
 6938   *) Fix various buffers to be large enough for hex/decimal
 6939      representations in a platform independent manner.
 6940      [Ben Laurie (CHATS)]
 6941 
 6942   *) Add CRYPTO_realloc_clean() to avoid information leakage when
 6943      resizing buffers containing secrets, and use where appropriate.
 6944      [Ben Laurie (CHATS)]
 6945 
 6946   *) Add BIO_indent() to avoid much slightly worrying code to do
 6947      indents.
 6948      [Ben Laurie (CHATS)]
 6949 
 6950   *) Convert sprintf()/BIO_puts() to BIO_printf().
 6951      [Ben Laurie (CHATS)]
 6952 
 6953   *) buffer_gets() could terminate with the buffer only half
 6954      full. Fixed.
 6955      [Ben Laurie (CHATS)]
 6956 
 6957   *) Add assertions to prevent user-supplied crypto functions from
 6958      overflowing internal buffers by having large block sizes, etc.
 6959      [Ben Laurie (CHATS)]
 6960 
 6961   *) New OPENSSL_assert() macro (similar to assert(), but enabled
 6962      unconditionally).
 6963      [Ben Laurie (CHATS)]
 6964 
 6965   *) Eliminate unused copy of key in RC4.
 6966      [Ben Laurie (CHATS)]
 6967 
 6968   *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
 6969      [Ben Laurie (CHATS)]
 6970 
 6971   *) Fix off-by-one error in EGD path.
 6972      [Ben Laurie (CHATS)]
 6973 
 6974   *) If RANDFILE path is too long, ignore instead of truncating.
 6975      [Ben Laurie (CHATS)]
 6976 
 6977   *) Eliminate unused and incorrectly sized X.509 structure
 6978      CBCParameter.
 6979      [Ben Laurie (CHATS)]
 6980 
 6981   *) Eliminate unused and dangerous function knumber().
 6982      [Ben Laurie (CHATS)]
 6983 
 6984   *) Eliminate unused and dangerous structure, KSSL_ERR.
 6985      [Ben Laurie (CHATS)]
 6986 
 6987   *) Protect against overlong session ID context length in an encoded
 6988      session object. Since these are local, this does not appear to be
 6989      exploitable.
 6990      [Ben Laurie (CHATS)]
 6991 
 6992   *) Change from security patch (see 0.9.6e below) that did not affect
 6993      the 0.9.6 release series:
 6994 
 6995      Remote buffer overflow in SSL3 protocol - an attacker could
 6996      supply an oversized master key in Kerberos-enabled versions.
 6997      (CVE-2002-0657)
 6998      [Ben Laurie (CHATS)]
 6999 
 7000   *) Change the SSL kerb5 codes to match RFC 2712.
 7001      [Richard Levitte]
 7002 
 7003   *) Make -nameopt work fully for req and add -reqopt switch.
 7004      [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
 7005 
 7006   *) The "block size" for block ciphers in CFB and OFB mode should be 1.
 7007      [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
 7008 
 7009   *) Make sure tests can be performed even if the corresponding algorithms
 7010      have been removed entirely.  This was also the last step to make
 7011      OpenSSL compilable with DJGPP under all reasonable conditions.
 7012      [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
 7013 
 7014   *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
 7015      to allow version independent disabling of normally unselected ciphers,
 7016      which may be activated as a side-effect of selecting a single cipher.
 7017 
 7018      (E.g., cipher list string "RSA" enables ciphersuites that are left
 7019      out of "ALL" because they do not provide symmetric encryption.
 7020      "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
 7021      [Lutz Jaenicke, Bodo Moeller]
 7022 
 7023   *) Add appropriate support for separate platform-dependent build
 7024      directories.  The recommended way to make a platform-dependent
 7025      build directory is the following (tested on Linux), maybe with
 7026      some local tweaks:
 7027 
 7028         # Place yourself outside of the OpenSSL source tree.  In
 7029         # this example, the environment variable OPENSSL_SOURCE
 7030         # is assumed to contain the absolute OpenSSL source directory.
 7031         mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
 7032         cd objtree/"`uname -s`-`uname -r`-`uname -m`"
 7033         (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
 7034                 mkdir -p `dirname $F`
 7035                 ln -s $OPENSSL_SOURCE/$F $F
 7036         done
 7037 
 7038      To be absolutely sure not to disturb the source tree, a "make clean"
 7039      is a good thing.  If it isn't successful, don't worry about it,
 7040      it probably means the source directory is very clean.
 7041      [Richard Levitte]
 7042 
 7043   *) Make sure any ENGINE control commands make local copies of string
 7044      pointers passed to them whenever necessary. Otherwise it is possible
 7045      the caller may have overwritten (or deallocated) the original string
 7046      data when a later ENGINE operation tries to use the stored values.
 7047      [Götz Babin-Ebell <babinebell@trustcenter.de>]
 7048 
 7049   *) Improve diagnostics in file reading and command-line digests.
 7050      [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
 7051 
 7052   *) Add AES modes CFB and OFB to the object database.  Correct an
 7053      error in AES-CFB decryption.
 7054      [Richard Levitte]
 7055 
 7056   *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
 7057      allows existing EVP_CIPHER_CTX structures to be reused after
 7058      calling EVP_*Final(). This behaviour is used by encryption
 7059      BIOs and some applications. This has the side effect that
 7060      applications must explicitly clean up cipher contexts with
 7061      EVP_CIPHER_CTX_cleanup() or they will leak memory.
 7062      [Steve Henson]
 7063 
 7064   *) Check the values of dna and dnb in bn_mul_recursive before calling
 7065      bn_mul_comba (a non zero value means the a or b arrays do not contain
 7066      n2 elements) and fallback to bn_mul_normal if either is not zero.
 7067      [Steve Henson]
 7068 
 7069   *) Fix escaping of non-ASCII characters when using the -subj option
 7070      of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
 7071      [Lutz Jaenicke]
 7072 
 7073   *) Make object definitions compliant to LDAP (RFC2256): SN is the short
 7074      form for "surname", serialNumber has no short form.
 7075      Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
 7076      therefore remove "mail" short name for "internet 7".
 7077      The OID for unique identifiers in X509 certificates is
 7078      x500UniqueIdentifier, not uniqueIdentifier.
 7079      Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
 7080      [Lutz Jaenicke]
 7081 
 7082   *) Add an "init" command to the ENGINE config module and auto initialize
 7083      ENGINEs. Without any "init" command the ENGINE will be initialized
 7084      after all ctrl commands have been executed on it. If init=1 the
 7085      ENGINE is initialized at that point (ctrls before that point are run
 7086      on the uninitialized ENGINE and after on the initialized one). If
 7087      init=0 then the ENGINE will not be initialized at all.
 7088      [Steve Henson]
 7089 
 7090   *) Fix the 'app_verify_callback' interface so that the user-defined
 7091      argument is actually passed to the callback: In the
 7092      SSL_CTX_set_cert_verify_callback() prototype, the callback
 7093      declaration has been changed from
 7094           int (*cb)()
 7095      into
 7096           int (*cb)(X509_STORE_CTX *,void *);
 7097      in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
 7098           i=s->ctx->app_verify_callback(&ctx)
 7099      has been changed into
 7100           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
 7101 
 7102      To update applications using SSL_CTX_set_cert_verify_callback(),
 7103      a dummy argument can be added to their callback functions.
 7104      [D. K. Smetters <smetters@parc.xerox.com>]
 7105 
 7106   *) Added the '4758cca' ENGINE to support IBM 4758 cards.
 7107      [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
 7108 
 7109   *) Add and OPENSSL_LOAD_CONF define which will cause
 7110      OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
 7111      This allows older applications to transparently support certain
 7112      OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
 7113      Two new functions OPENSSL_add_all_algorithms_noconf() which will never
 7114      load the config file and OPENSSL_add_all_algorithms_conf() which will
 7115      always load it have also been added.
 7116      [Steve Henson]
 7117 
 7118   *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
 7119      Adjust NIDs and EVP layer.
 7120      [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
 7121 
 7122   *) Config modules support in openssl utility.
 7123 
 7124      Most commands now load modules from the config file,
 7125      though in a few (such as version) this isn't done
 7126      because it couldn't be used for anything.
 7127 
 7128      In the case of ca and req the config file used is
 7129      the same as the utility itself: that is the -config
 7130      command line option can be used to specify an
 7131      alternative file.
 7132      [Steve Henson]
 7133 
 7134   *) Move default behaviour from OPENSSL_config(). If appname is NULL
 7135      use "openssl_conf" if filename is NULL use default openssl config file.
 7136      [Steve Henson]
 7137 
 7138   *) Add an argument to OPENSSL_config() to allow the use of an alternative
 7139      config section name. Add a new flag to tolerate a missing config file
 7140      and move code to CONF_modules_load_file().
 7141      [Steve Henson]
 7142 
 7143   *) Support for crypto accelerator cards from Accelerated Encryption
 7144      Processing, www.aep.ie.  (Use engine 'aep')
 7145      The support was copied from 0.9.6c [engine] and adapted/corrected
 7146      to work with the new engine framework.
 7147      [AEP Inc. and Richard Levitte]
 7148 
 7149   *) Support for SureWare crypto accelerator cards from Baltimore
 7150      Technologies.  (Use engine 'sureware')
 7151      The support was copied from 0.9.6c [engine] and adapted
 7152      to work with the new engine framework.
 7153      [Richard Levitte]
 7154 
 7155   *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
 7156      make the newer ENGINE framework commands for the CHIL engine work.
 7157      [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
 7158 
 7159   *) Make it possible to produce shared libraries on ReliantUNIX.
 7160      [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
 7161 
 7162   *) Add the configuration target debug-linux-ppro.
 7163      Make 'openssl rsa' use the general key loading routines
 7164      implemented in apps.c, and make those routines able to
 7165      handle the key format FORMAT_NETSCAPE and the variant
 7166      FORMAT_IISSGC.
 7167      [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
 7168 
 7169  *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
 7170      [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
 7171 
 7172   *) Add -keyform to rsautl, and document -engine.
 7173      [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
 7174 
 7175   *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
 7176      BIO_R_NO_SUCH_FILE error code rather than the generic
 7177      ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
 7178      [Ben Laurie]
 7179 
 7180   *) Add new functions
 7181           ERR_peek_last_error
 7182           ERR_peek_last_error_line
 7183           ERR_peek_last_error_line_data.
 7184      These are similar to
 7185           ERR_peek_error
 7186           ERR_peek_error_line
 7187           ERR_peek_error_line_data,
 7188      but report on the latest error recorded rather than the first one
 7189      still in the error queue.
 7190      [Ben Laurie, Bodo Moeller]
 7191 
 7192   *) default_algorithms option in ENGINE config module. This allows things
 7193      like:
 7194      default_algorithms = ALL
 7195      default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
 7196      [Steve Henson]
 7197 
 7198   *) Preliminary ENGINE config module.
 7199      [Steve Henson]
 7200 
 7201   *) New experimental application configuration code.
 7202      [Steve Henson]
 7203 
 7204   *) Change the AES code to follow the same name structure as all other
 7205      symmetric ciphers, and behave the same way.  Move everything to
 7206      the directory crypto/aes, thereby obsoleting crypto/rijndael.
 7207      [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
 7208 
 7209   *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
 7210      [Ben Laurie and Theo de Raadt]
 7211 
 7212   *) Add option to output public keys in req command.
 7213      [Massimiliano Pala madwolf@openca.org]
 7214 
 7215   *) Use wNAFs in EC_POINTs_mul() for improved efficiency
 7216      (up to about 10% better than before for P-192 and P-224).
 7217      [Bodo Moeller]
 7218 
 7219   *) New functions/macros
 7220 
 7221           SSL_CTX_set_msg_callback(ctx, cb)
 7222           SSL_CTX_set_msg_callback_arg(ctx, arg)
 7223           SSL_set_msg_callback(ssl, cb)
 7224           SSL_set_msg_callback_arg(ssl, arg)
 7225 
 7226      to request calling a callback function
 7227 
 7228           void cb(int write_p, int version, int content_type,
 7229                   const void *buf, size_t len, SSL *ssl, void *arg)
 7230 
 7231      whenever a protocol message has been completely received
 7232      (write_p == 0) or sent (write_p == 1).  Here 'version' is the
 7233      protocol version  according to which the SSL library interprets
 7234      the current protocol message (SSL2_VERSION, SSL3_VERSION, or
 7235      TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
 7236      the content type as defined in the SSL 3.0/TLS 1.0 protocol
 7237      specification (change_cipher_spec(20), alert(21), handshake(22)).
 7238      'buf' and 'len' point to the actual message, 'ssl' to the
 7239      SSL object, and 'arg' is the application-defined value set by
 7240      SSL[_CTX]_set_msg_callback_arg().
 7241 
 7242      'openssl s_client' and 'openssl s_server' have new '-msg' options
 7243      to enable a callback that displays all protocol messages.
 7244      [Bodo Moeller]
 7245 
 7246   *) Change the shared library support so shared libraries are built as
 7247      soon as the corresponding static library is finished, and thereby get
 7248      openssl and the test programs linked against the shared library.
 7249      This still only happens when the keyword "shard" has been given to
 7250      the configuration scripts.
 7251 
 7252      NOTE: shared library support is still an experimental thing, and
 7253      backward binary compatibility is still not guaranteed.
 7254      ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
 7255 
 7256   *) Add support for Subject Information Access extension.
 7257      [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
 7258 
 7259   *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
 7260      additional bytes when new memory had to be allocated, not just
 7261      when reusing an existing buffer.
 7262      [Bodo Moeller]
 7263 
 7264   *) New command line and configuration option 'utf8' for the req command.
 7265      This allows field values to be specified as UTF8 strings.
 7266      [Steve Henson]
 7267 
 7268   *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
 7269      runs for the former and machine-readable output for the latter.
 7270      [Ben Laurie]
 7271 
 7272   *) Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
 7273      of the e-mail address in the DN (i.e., it will go into a certificate
 7274      extension only).  The new configuration file option 'email_in_dn = no'
 7275      has the same effect.
 7276      [Massimiliano Pala madwolf@openca.org]
 7277 
 7278   *) Change all functions with names starting with des_ to be starting
 7279      with DES_ instead.  Add wrappers that are compatible with libdes,
 7280      but are named _ossl_old_des_*.  Finally, add macros that map the
 7281      des_* symbols to the corresponding _ossl_old_des_* if libdes
 7282      compatibility is desired.  If OpenSSL 0.9.6c compatibility is
 7283      desired, the des_* symbols will be mapped to DES_*, with one
 7284      exception.
 7285 
 7286      Since we provide two compatibility mappings, the user needs to
 7287      define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
 7288      compatibility is desired.  The default (i.e., when that macro
 7289      isn't defined) is OpenSSL 0.9.6c compatibility.
 7290 
 7291      There are also macros that enable and disable the support of old
 7292      des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
 7293      and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
 7294      are defined, the default will apply: to support the old des routines.
 7295 
 7296      In either case, one must include openssl/des.h to get the correct
 7297      definitions.  Do not try to just include openssl/des_old.h, that
 7298      won't work.
 7299 
 7300      NOTE: This is a major break of an old API into a new one.  Software
 7301      authors are encouraged to switch to the DES_ style functions.  Some
 7302      time in the future, des_old.h and the libdes compatibility functions
 7303      will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
 7304      default), and then completely removed.
 7305      [Richard Levitte]
 7306 
 7307   *) Test for certificates which contain unsupported critical extensions.
 7308      If such a certificate is found during a verify operation it is
 7309      rejected by default: this behaviour can be overridden by either
 7310      handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
 7311      by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
 7312      X509_supported_extension() has also been added which returns 1 if a
 7313      particular extension is supported.
 7314      [Steve Henson]
 7315 
 7316   *) Modify the behaviour of EVP cipher functions in similar way to digests
 7317      to retain compatibility with existing code.
 7318      [Steve Henson]
 7319 
 7320   *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
 7321      compatibility with existing code. In particular the 'ctx' parameter does
 7322      not have to be to be initialized before the call to EVP_DigestInit() and
 7323      it is tidied up after a call to EVP_DigestFinal(). New function
 7324      EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
 7325      EVP_MD_CTX_copy() changed to not require the destination to be
 7326      initialized valid and new function EVP_MD_CTX_copy_ex() added which
 7327      requires the destination to be valid.
 7328 
 7329      Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
 7330      EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
 7331      [Steve Henson]
 7332 
 7333   *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
 7334      so that complete 'Handshake' protocol structures are kept in memory
 7335      instead of overwriting 'msg_type' and 'length' with 'body' data.
 7336      [Bodo Moeller]
 7337 
 7338   *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
 7339      [Massimo Santin via Richard Levitte]
 7340 
 7341   *) Major restructuring to the underlying ENGINE code. This includes
 7342      reduction of linker bloat, separation of pure "ENGINE" manipulation
 7343      (initialisation, etc) from functionality dealing with implementations
 7344      of specific crypto interfaces. This change also introduces integrated
 7345      support for symmetric ciphers and digest implementations - so ENGINEs
 7346      can now accelerate these by providing EVP_CIPHER and EVP_MD
 7347      implementations of their own. This is detailed in crypto/engine/README
 7348      as it couldn't be adequately described here. However, there are a few
 7349      API changes worth noting - some RSA, DSA, DH, and RAND functions that
 7350      were changed in the original introduction of ENGINE code have now
 7351      reverted back - the hooking from this code to ENGINE is now a good
 7352      deal more passive and at run-time, operations deal directly with
 7353      RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
 7354      dereferencing through an ENGINE pointer any more. Also, the ENGINE
 7355      functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
 7356      they were not being used by the framework as there is no concept of a
 7357      BIGNUM_METHOD and they could not be generalised to the new
 7358      'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
 7359      ENGINE_cpy() has been removed as it cannot be consistently defined in
 7360      the new code.
 7361      [Geoff Thorpe]
 7362 
 7363   *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
 7364      [Steve Henson]
 7365 
 7366   *) Change mkdef.pl to sort symbols that get the same entry number,
 7367      and make sure the automatically generated functions ERR_load_*
 7368      become part of libeay.num as well.
 7369      [Richard Levitte]
 7370 
 7371   *) New function SSL_renegotiate_pending().  This returns true once
 7372      renegotiation has been requested (either SSL_renegotiate() call
 7373      or HelloRequest/ClientHello received from the peer) and becomes
 7374      false once a handshake has been completed.
 7375      (For servers, SSL_renegotiate() followed by SSL_do_handshake()
 7376      sends a HelloRequest, but does not ensure that a handshake takes
 7377      place.  SSL_renegotiate_pending() is useful for checking if the
 7378      client has followed the request.)
 7379      [Bodo Moeller]
 7380 
 7381   *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
 7382      By default, clients may request session resumption even during
 7383      renegotiation (if session ID contexts permit); with this option,
 7384      session resumption is possible only in the first handshake.
 7385 
 7386      SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
 7387      more bits available for options that should not be part of
 7388      SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
 7389      [Bodo Moeller]
 7390 
 7391   *) Add some demos for certificate and certificate request creation.
 7392      [Steve Henson]
 7393 
 7394   *) Make maximum certificate chain size accepted from the peer application
 7395      settable (SSL*_get/set_max_cert_list()), as proposed by
 7396      "Douglas E. Engert" <deengert@anl.gov>.
 7397      [Lutz Jaenicke]
 7398 
 7399   *) Add support for shared libraries for Unixware-7
 7400      (Boyd Lynn Gerber <gerberb@zenez.com>).
 7401      [Lutz Jaenicke]
 7402 
 7403   *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
 7404      be done prior to destruction. Use this to unload error strings from
 7405      ENGINEs that load their own error strings. NB: This adds two new API
 7406      functions to "get" and "set" this destroy handler in an ENGINE.
 7407      [Geoff Thorpe]
 7408 
 7409   *) Alter all existing ENGINE implementations (except "openssl" and
 7410      "openbsd") to dynamically instantiate their own error strings. This
 7411      makes them more flexible to be built both as statically-linked ENGINEs
 7412      and self-contained shared-libraries loadable via the "dynamic" ENGINE.
 7413      Also, add stub code to each that makes building them as self-contained
 7414      shared-libraries easier (see README.ENGINE).
 7415      [Geoff Thorpe]
 7416 
 7417   *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
 7418      implementations into applications that are completely implemented in
 7419      self-contained shared-libraries. The "dynamic" ENGINE exposes control
 7420      commands that can be used to configure what shared-library to load and
 7421      to control aspects of the way it is handled. Also, made an update to
 7422      the README.ENGINE file that brings its information up-to-date and
 7423      provides some information and instructions on the "dynamic" ENGINE
 7424      (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
 7425      [Geoff Thorpe]
 7426 
 7427   *) Make it possible to unload ranges of ERR strings with a new
 7428      "ERR_unload_strings" function.
 7429      [Geoff Thorpe]
 7430 
 7431   *) Add a copy() function to EVP_MD.
 7432      [Ben Laurie]
 7433 
 7434   *) Make EVP_MD routines take a context pointer instead of just the
 7435      md_data void pointer.
 7436      [Ben Laurie]
 7437 
 7438   *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
 7439      that the digest can only process a single chunk of data
 7440      (typically because it is provided by a piece of
 7441      hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
 7442      is only going to provide a single chunk of data, and hence the
 7443      framework needn't accumulate the data for oneshot drivers.
 7444      [Ben Laurie]
 7445 
 7446   *) As with "ERR", make it possible to replace the underlying "ex_data"
 7447      functions. This change also alters the storage and management of global
 7448      ex_data state - it's now all inside ex_data.c and all "class" code (eg.
 7449      RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
 7450      index counters. The API functions that use this state have been changed
 7451      to take a "class_index" rather than pointers to the class's local STACK
 7452      and counter, and there is now an API function to dynamically create new
 7453      classes. This centralisation allows us to (a) plug a lot of the
 7454      thread-safety problems that existed, and (b) makes it possible to clean
 7455      up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
 7456      such data would previously have always leaked in application code and
 7457      workarounds were in place to make the memory debugging turn a blind eye
 7458      to it. Application code that doesn't use this new function will still
 7459      leak as before, but their memory debugging output will announce it now
 7460      rather than letting it slide.
 7461 
 7462      Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
 7463      induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
 7464      has a return value to indicate success or failure.
 7465      [Geoff Thorpe]
 7466 
 7467   *) Make it possible to replace the underlying "ERR" functions such that the
 7468      global state (2 LHASH tables and 2 locks) is only used by the "default"
 7469      implementation. This change also adds two functions to "get" and "set"
 7470      the implementation prior to it being automatically set the first time
 7471      any other ERR function takes place. Ie. an application can call "get",
 7472      pass the return value to a module it has just loaded, and that module
 7473      can call its own "set" function using that value. This means the
 7474      module's "ERR" operations will use (and modify) the error state in the
 7475      application and not in its own statically linked copy of OpenSSL code.
 7476      [Geoff Thorpe]
 7477 
 7478   *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
 7479      reference counts. This performs normal REF_PRINT/REF_CHECK macros on
 7480      the operation, and provides a more encapsulated way for external code
 7481      (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
 7482      to use these functions rather than manually incrementing the counts.
 7483 
 7484      Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
 7485      [Geoff Thorpe]
 7486 
 7487   *) Add EVP test program.
 7488      [Ben Laurie]
 7489 
 7490   *) Add symmetric cipher support to ENGINE. Expect the API to change!
 7491      [Ben Laurie]
 7492 
 7493   *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
 7494      X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
 7495      X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
 7496      These allow a CRL to be built without having to access X509_CRL fields
 7497      directly. Modify 'ca' application to use new functions.
 7498      [Steve Henson]
 7499 
 7500   *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
 7501      bug workarounds. Rollback attack detection is a security feature.
 7502      The problem will only arise on OpenSSL servers when TLSv1 is not
 7503      available (sslv3_server_method() or SSL_OP_NO_TLSv1).
 7504      Software authors not wanting to support TLSv1 will have special reasons
 7505      for their choice and can explicitly enable this option.
 7506      [Bodo Moeller, Lutz Jaenicke]
 7507 
 7508   *) Rationalise EVP so it can be extended: don't include a union of
 7509      cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
 7510      (similar to those existing for EVP_CIPHER_CTX).
 7511      Usage example:
 7512 
 7513          EVP_MD_CTX md;
 7514 
 7515          EVP_MD_CTX_init(&md);             /* new function call */
 7516          EVP_DigestInit(&md, EVP_sha1());
 7517          EVP_DigestUpdate(&md, in, len);
 7518          EVP_DigestFinal(&md, out, NULL);
 7519          EVP_MD_CTX_cleanup(&md);          /* new function call */
 7520 
 7521      [Ben Laurie]
 7522 
 7523   *) Make DES key schedule conform to the usual scheme, as well as
 7524      correcting its structure. This means that calls to DES functions
 7525      now have to pass a pointer to a des_key_schedule instead of a
 7526      plain des_key_schedule (which was actually always a pointer
 7527      anyway): E.g.,
 7528 
 7529          des_key_schedule ks;
 7530 
 7531          des_set_key_checked(..., &ks);
 7532          des_ncbc_encrypt(..., &ks, ...);
 7533 
 7534      (Note that a later change renames 'des_...' into 'DES_...'.)
 7535      [Ben Laurie]
 7536 
 7537   *) Initial reduction of linker bloat: the use of some functions, such as
 7538      PEM causes large amounts of unused functions to be linked in due to
 7539      poor organisation. For example pem_all.c contains every PEM function
 7540      which has a knock on effect of linking in large amounts of (unused)
 7541      ASN1 code. Grouping together similar functions and splitting unrelated
 7542      functions prevents this.
 7543      [Steve Henson]
 7544 
 7545   *) Cleanup of EVP macros.
 7546      [Ben Laurie]
 7547 
 7548   *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
 7549      correct _ecb suffix.
 7550      [Ben Laurie]
 7551 
 7552   *) Add initial OCSP responder support to ocsp application. The
 7553      revocation information is handled using the text based index
 7554      use by the ca application. The responder can either handle
 7555      requests generated internally, supplied in files (for example
 7556      via a CGI script) or using an internal minimal server.
 7557      [Steve Henson]
 7558 
 7559   *) Add configuration choices to get zlib compression for TLS.
 7560      [Richard Levitte]
 7561 
 7562   *) Changes to Kerberos SSL for RFC 2712 compliance:
 7563      1.  Implemented real KerberosWrapper, instead of just using
 7564          KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
 7565      2.  Implemented optional authenticator field of KerberosWrapper.
 7566 
 7567      Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
 7568      and authenticator structs; see crypto/krb5/.
 7569 
 7570      Generalized Kerberos calls to support multiple Kerberos libraries.
 7571      [Vern Staats <staatsvr@asc.hpc.mil>,
 7572       Jeffrey Altman <jaltman@columbia.edu>
 7573       via Richard Levitte]
 7574 
 7575   *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
 7576      already does with RSA. testdsa.h now has 'priv_key/pub_key'
 7577      values for each of the key sizes rather than having just
 7578      parameters (and 'speed' generating keys each time).
 7579      [Geoff Thorpe]
 7580 
 7581   *) Speed up EVP routines.
 7582      Before:
 7583 encrypt
 7584 type              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
 7585 des-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
 7586 des-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
 7587 des-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
 7588 decrypt
 7589 des-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
 7590 des-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
 7591 des-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
 7592      After:
 7593 encrypt
 7594 des-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
 7595 decrypt
 7596 des-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
 7597      [Ben Laurie]
 7598 
 7599   *) Added the OS2-EMX target.
 7600      ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
 7601 
 7602   *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
 7603      to support NCONF routines in extension code. New function CONF_set_nconf()
 7604      to allow functions which take an NCONF to also handle the old LHASH
 7605      structure: this means that the old CONF compatible routines can be
 7606      retained (in particular wrt extensions) without having to duplicate the
 7607      code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
 7608      [Steve Henson]
 7609 
 7610   *) Enhance the general user interface with mechanisms for inner control
 7611      and with possibilities to have yes/no kind of prompts.
 7612      [Richard Levitte]
 7613 
 7614   *) Change all calls to low level digest routines in the library and
 7615      applications to use EVP. Add missing calls to HMAC_cleanup() and
 7616      don't assume HMAC_CTX can be copied using memcpy().
 7617      [Verdon Walker <VWalker@novell.com>, Steve Henson]
 7618 
 7619   *) Add the possibility to control engines through control names but with
 7620      arbitrary arguments instead of just a string.
 7621      Change the key loaders to take a UI_METHOD instead of a callback
 7622      function pointer.  NOTE: this breaks binary compatibility with earlier
 7623      versions of OpenSSL [engine].
 7624      Adapt the nCipher code for these new conditions and add a card insertion
 7625      callback.
 7626      [Richard Levitte]
 7627 
 7628   *) Enhance the general user interface with mechanisms to better support
 7629      dialog box interfaces, application-defined prompts, the possibility
 7630      to use defaults (for example default passwords from somewhere else)
 7631      and interrupts/cancellations.
 7632      [Richard Levitte]
 7633 
 7634   *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
 7635      attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
 7636      [Steve Henson]
 7637 
 7638   *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
 7639      tidy up some unnecessarily weird code in 'sk_new()').
 7640      [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
 7641 
 7642   *) Change the key loading routines for ENGINEs to use the same kind
 7643      callback (pem_password_cb) as all other routines that need this
 7644      kind of callback.
 7645      [Richard Levitte]
 7646 
 7647   *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
 7648      256 bit (=32 byte) keys. Of course seeding with more entropy bytes
 7649      than this minimum value is recommended.
 7650      [Lutz Jaenicke]
 7651 
 7652   *) New random seeder for OpenVMS, using the system process statistics
 7653      that are easily reachable.
 7654      [Richard Levitte]
 7655 
 7656   *) Windows apparently can't transparently handle global
 7657      variables defined in DLLs. Initialisations such as:
 7658 
 7659         const ASN1_ITEM *it = &ASN1_INTEGER_it;
 7660 
 7661      won't compile. This is used by the any applications that need to
 7662      declare their own ASN1 modules. This was fixed by adding the option
 7663      EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
 7664      needed for static libraries under Win32.
 7665      [Steve Henson]
 7666 
 7667   *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
 7668      setting of purpose and trust fields. New X509_STORE trust and
 7669      purpose functions and tidy up setting in other SSL functions.
 7670      [Steve Henson]
 7671 
 7672   *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
 7673      structure. These are inherited by X509_STORE_CTX when it is
 7674      initialised. This allows various defaults to be set in the
 7675      X509_STORE structure (such as flags for CRL checking and custom
 7676      purpose or trust settings) for functions which only use X509_STORE_CTX
 7677      internally such as S/MIME.
 7678 
 7679      Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
 7680      trust settings if they are not set in X509_STORE. This allows X509_STORE
 7681      purposes and trust (in S/MIME for example) to override any set by default.
 7682 
 7683      Add command line options for CRL checking to smime, s_client and s_server
 7684      applications.
 7685      [Steve Henson]
 7686 
 7687   *) Initial CRL based revocation checking. If the CRL checking flag(s)
 7688      are set then the CRL is looked up in the X509_STORE structure and
 7689      its validity and signature checked, then if the certificate is found
 7690      in the CRL the verify fails with a revoked error.
 7691 
 7692      Various new CRL related callbacks added to X509_STORE_CTX structure.
 7693 
 7694      Command line options added to 'verify' application to support this.
 7695 
 7696      This needs some additional work, such as being able to handle multiple
 7697      CRLs with different times, extension based lookup (rather than just
 7698      by subject name) and ultimately more complete V2 CRL extension
 7699      handling.
 7700      [Steve Henson]
 7701 
 7702   *) Add a general user interface API (crypto/ui/).  This is designed
 7703      to replace things like des_read_password and friends (backward
 7704      compatibility functions using this new API are provided).
 7705      The purpose is to remove prompting functions from the DES code
 7706      section as well as provide for prompting through dialog boxes in
 7707      a window system and the like.
 7708      [Richard Levitte]
 7709 
 7710   *) Add "ex_data" support to ENGINE so implementations can add state at a
 7711      per-structure level rather than having to store it globally.
 7712      [Geoff]
 7713 
 7714   *) Make it possible for ENGINE structures to be copied when retrieved by
 7715      ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
 7716      This causes the "original" ENGINE structure to act like a template,
 7717      analogous to the RSA vs. RSA_METHOD type of separation. Because of this
 7718      operational state can be localised to each ENGINE structure, despite the
 7719      fact they all share the same "methods". New ENGINE structures returned in
 7720      this case have no functional references and the return value is the single
 7721      structural reference. This matches the single structural reference returned
 7722      by ENGINE_by_id() normally, when it is incremented on the pre-existing
 7723      ENGINE structure.
 7724      [Geoff]
 7725 
 7726   *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
 7727      needs to match any other type at all we need to manually clear the
 7728      tag cache.
 7729      [Steve Henson]
 7730 
 7731   *) Changes to the "openssl engine" utility to include;
 7732      - verbosity levels ('-v', '-vv', and '-vvv') that provide information
 7733        about an ENGINE's available control commands.
 7734      - executing control commands from command line arguments using the
 7735        '-pre' and '-post' switches. '-post' is only used if '-t' is
 7736        specified and the ENGINE is successfully initialised. The syntax for
 7737        the individual commands are colon-separated, for example;
 7738          openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
 7739      [Geoff]
 7740 
 7741   *) New dynamic control command support for ENGINEs. ENGINEs can now
 7742      declare their own commands (numbers), names (strings), descriptions,
 7743      and input types for run-time discovery by calling applications. A
 7744      subset of these commands are implicitly classed as "executable"
 7745      depending on their input type, and only these can be invoked through
 7746      the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
 7747      can be based on user input, config files, etc). The distinction is
 7748      that "executable" commands cannot return anything other than a boolean
 7749      result and can only support numeric or string input, whereas some
 7750      discoverable commands may only be for direct use through
 7751      ENGINE_ctrl(), eg. supporting the exchange of binary data, function
 7752      pointers, or other custom uses. The "executable" commands are to
 7753      support parameterisations of ENGINE behaviour that can be
 7754      unambiguously defined by ENGINEs and used consistently across any
 7755      OpenSSL-based application. Commands have been added to all the
 7756      existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
 7757      control over shared-library paths without source code alterations.
 7758      [Geoff]
 7759 
 7760   *) Changed all ENGINE implementations to dynamically allocate their
 7761      ENGINEs rather than declaring them statically. Apart from this being
 7762      necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
 7763      this also allows the implementations to compile without using the
 7764      internal engine_int.h header.
 7765      [Geoff]
 7766 
 7767   *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
 7768      'const' value. Any code that should be able to modify a RAND_METHOD
 7769      should already have non-const pointers to it (ie. they should only
 7770      modify their own ones).
 7771      [Geoff]
 7772 
 7773   *) Made a variety of little tweaks to the ENGINE code.
 7774      - "atalla" and "ubsec" string definitions were moved from header files
 7775        to C code. "nuron" string definitions were placed in variables
 7776        rather than hard-coded - allowing parameterisation of these values
 7777        later on via ctrl() commands.
 7778      - Removed unused "#if 0"'d code.
 7779      - Fixed engine list iteration code so it uses ENGINE_free() to release
 7780        structural references.
 7781      - Constified the RAND_METHOD element of ENGINE structures.
 7782      - Constified various get/set functions as appropriate and added
 7783        missing functions (including a catch-all ENGINE_cpy that duplicates
 7784        all ENGINE values onto a new ENGINE except reference counts/state).
 7785      - Removed NULL parameter checks in get/set functions. Setting a method
 7786        or function to NULL is a way of cancelling out a previously set
 7787        value.  Passing a NULL ENGINE parameter is just plain stupid anyway
 7788        and doesn't justify the extra error symbols and code.
 7789      - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
 7790        flags from engine_int.h to engine.h.
 7791      - Changed prototypes for ENGINE handler functions (init(), finish(),
 7792        ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
 7793      [Geoff]
 7794 
 7795   *) Implement binary inversion algorithm for BN_mod_inverse in addition
 7796      to the algorithm using long division.  The binary algorithm can be
 7797      used only if the modulus is odd.  On 32-bit systems, it is faster
 7798      only for relatively small moduli (roughly 20-30% for 128-bit moduli,
 7799      roughly 5-15% for 256-bit moduli), so we use it only for moduli
 7800      up to 450 bits.  In 64-bit environments, the binary algorithm
 7801      appears to be advantageous for much longer moduli; here we use it
 7802      for moduli up to 2048 bits.
 7803      [Bodo Moeller]
 7804 
 7805   *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
 7806      could not support the combine flag in choice fields.
 7807      [Steve Henson]
 7808 
 7809   *) Add a 'copy_extensions' option to the 'ca' utility. This copies
 7810      extensions from a certificate request to the certificate.
 7811      [Steve Henson]
 7812 
 7813   *) Allow multiple 'certopt' and 'nameopt' options to be separated
 7814      by commas. Add 'namopt' and 'certopt' options to the 'ca' config
 7815      file: this allows the display of the certificate about to be
 7816      signed to be customised, to allow certain fields to be included
 7817      or excluded and extension details. The old system didn't display
 7818      multicharacter strings properly, omitted fields not in the policy
 7819      and couldn't display additional details such as extensions.
 7820      [Steve Henson]
 7821 
 7822   *) Function EC_POINTs_mul for multiple scalar multiplication
 7823      of an arbitrary number of elliptic curve points
 7824           \sum scalars[i]*points[i],
 7825      optionally including the generator defined for the EC_GROUP:
 7826           scalar*generator +  \sum scalars[i]*points[i].
 7827 
 7828      EC_POINT_mul is a simple wrapper function for the typical case
 7829      that the point list has just one item (besides the optional
 7830      generator).
 7831      [Bodo Moeller]
 7832 
 7833   *) First EC_METHODs for curves over GF(p):
 7834 
 7835      EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
 7836      operations and provides various method functions that can also
 7837      operate with faster implementations of modular arithmetic.
 7838 
 7839      EC_GFp_mont_method() reuses most functions that are part of
 7840      EC_GFp_simple_method, but uses Montgomery arithmetic.
 7841 
 7842      [Bodo Moeller; point addition and point doubling
 7843      implementation directly derived from source code provided by
 7844      Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
 7845 
 7846   *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
 7847      crypto/ec/ec_lib.c):
 7848 
 7849      Curves are EC_GROUP objects (with an optional group generator)
 7850      based on EC_METHODs that are built into the library.
 7851 
 7852      Points are EC_POINT objects based on EC_GROUP objects.
 7853 
 7854      Most of the framework would be able to handle curves over arbitrary
 7855      finite fields, but as there are no obvious types for fields other
 7856      than GF(p), some functions are limited to that for now.
 7857      [Bodo Moeller]
 7858 
 7859   *) Add the -HTTP option to s_server.  It is similar to -WWW, but requires
 7860      that the file contains a complete HTTP response.
 7861      [Richard Levitte]
 7862 
 7863   *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
 7864      change the def and num file printf format specifier from "%-40sXXX"
 7865      to "%-39s XXX". The latter will always guarantee a space after the
 7866      field while the former will cause them to run together if the field
 7867      is 40 of more characters long.
 7868      [Steve Henson]
 7869 
 7870   *) Constify the cipher and digest 'method' functions and structures
 7871      and modify related functions to take constant EVP_MD and EVP_CIPHER
 7872      pointers.
 7873      [Steve Henson]
 7874 
 7875   *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
 7876      in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
 7877      [Bodo Moeller]
 7878 
 7879   *) Modify EVP_Digest*() routines so they now return values. Although the
 7880      internal software routines can never fail additional hardware versions
 7881      might.
 7882      [Steve Henson]
 7883 
 7884   *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
 7885 
 7886      Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
 7887      (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
 7888 
 7889      ASN1 error codes
 7890           ERR_R_NESTED_ASN1_ERROR
 7891           ...
 7892           ERR_R_MISSING_ASN1_EOS
 7893      were 4 .. 9, conflicting with
 7894           ERR_LIB_RSA (= ERR_R_RSA_LIB)
 7895           ...
 7896           ERR_LIB_PEM (= ERR_R_PEM_LIB).
 7897      They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
 7898 
 7899      Add new error code 'ERR_R_INTERNAL_ERROR'.
 7900      [Bodo Moeller]
 7901 
 7902   *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
 7903      suffices.
 7904      [Bodo Moeller]
 7905 
 7906   *) New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
 7907      sets the subject name for a new request or supersedes the
 7908      subject name in a given request. Formats that can be parsed are
 7909           'CN=Some Name, OU=myOU, C=IT'
 7910      and
 7911           'CN=Some Name/OU=myOU/C=IT'.
 7912 
 7913      Add options '-batch' and '-verbose' to 'openssl req'.
 7914      [Massimiliano Pala <madwolf@hackmasters.net>]
 7915 
 7916   *) Introduce the possibility to access global variables through
 7917      functions on platform were that's the best way to handle exporting
 7918      global variables in shared libraries.  To enable this functionality,
 7919      one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
 7920      "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
 7921      is normally done by Configure or something similar).
 7922 
 7923      To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
 7924      in the source file (foo.c) like this:
 7925 
 7926         OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
 7927         OPENSSL_IMPLEMENT_GLOBAL(double,bar);
 7928 
 7929      To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
 7930      and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
 7931 
 7932         OPENSSL_DECLARE_GLOBAL(int,foo);
 7933         #define foo OPENSSL_GLOBAL_REF(foo)
 7934         OPENSSL_DECLARE_GLOBAL(double,bar);
 7935         #define bar OPENSSL_GLOBAL_REF(bar)
 7936 
 7937      The #defines are very important, and therefore so is including the
 7938      header file everywhere where the defined globals are used.
 7939 
 7940      The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
 7941      of ASN.1 items, but that structure is a bit different.
 7942 
 7943      The largest change is in util/mkdef.pl which has been enhanced with
 7944      better and easier to understand logic to choose which symbols should
 7945      go into the Windows .def files as well as a number of fixes and code
 7946      cleanup (among others, algorithm keywords are now sorted
 7947      lexicographically to avoid constant rewrites).
 7948      [Richard Levitte]
 7949 
 7950   *) In BN_div() keep a copy of the sign of 'num' before writing the
 7951      result to 'rm' because if rm==num the value will be overwritten
 7952      and produce the wrong result if 'num' is negative: this caused
 7953      problems with BN_mod() and BN_nnmod().
 7954      [Steve Henson]
 7955 
 7956   *) Function OCSP_request_verify(). This checks the signature on an
 7957      OCSP request and verifies the signer certificate. The signer
 7958      certificate is just checked for a generic purpose and OCSP request
 7959      trust settings.
 7960      [Steve Henson]
 7961 
 7962   *) Add OCSP_check_validity() function to check the validity of OCSP
 7963      responses. OCSP responses are prepared in real time and may only
 7964      be a few seconds old. Simply checking that the current time lies
 7965      between thisUpdate and nextUpdate max reject otherwise valid responses
 7966      caused by either OCSP responder or client clock inaccuracy. Instead
 7967      we allow thisUpdate and nextUpdate to fall within a certain period of
 7968      the current time. The age of the response can also optionally be
 7969      checked. Two new options -validity_period and -status_age added to
 7970      ocsp utility.
 7971      [Steve Henson]
 7972 
 7973   *) If signature or public key algorithm is unrecognized print out its
 7974      OID rather that just UNKNOWN.
 7975      [Steve Henson]
 7976 
 7977   *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
 7978      OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
 7979      ID to be generated from the issuer certificate alone which can then be
 7980      passed to OCSP_id_issuer_cmp().
 7981      [Steve Henson]
 7982 
 7983   *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
 7984      ASN1 modules to export functions returning ASN1_ITEM pointers
 7985      instead of the ASN1_ITEM structures themselves. This adds several
 7986      new macros which allow the underlying ASN1 function/structure to
 7987      be accessed transparently. As a result code should not use ASN1_ITEM
 7988      references directly (such as &X509_it) but instead use the relevant
 7989      macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
 7990      use of the new ASN1 code on platforms where exporting structures
 7991      is problematical (for example in shared libraries) but exporting
 7992      functions returning pointers to structures is not.
 7993      [Steve Henson]
 7994 
 7995   *) Add support for overriding the generation of SSL/TLS session IDs.
 7996      These callbacks can be registered either in an SSL_CTX or per SSL.
 7997      The purpose of this is to allow applications to control, if they wish,
 7998      the arbitrary values chosen for use as session IDs, particularly as it
 7999      can be useful for session caching in multiple-server environments. A
 8000      command-line switch for testing this (and any client code that wishes
 8001      to use such a feature) has been added to "s_server".
 8002      [Geoff Thorpe, Lutz Jaenicke]
 8003 
 8004   *) Modify mkdef.pl to recognise and parse preprocessor conditionals
 8005      of the form '#if defined(...) || defined(...) || ...' and
 8006      '#if !defined(...) && !defined(...) && ...'.  This also avoids
 8007      the growing number of special cases it was previously handling.
 8008      [Richard Levitte]
 8009 
 8010   *) Make all configuration macros available for application by making
 8011      sure they are available in opensslconf.h, by giving them names starting
 8012      with "OPENSSL_" to avoid conflicts with other packages and by making
 8013      sure e_os2.h will cover all platform-specific cases together with
 8014      opensslconf.h.
 8015      Additionally, it is now possible to define configuration/platform-
 8016      specific names (called "system identities").  In the C code, these
 8017      are prefixed with "OPENSSL_SYSNAME_".  e_os2.h will create another
 8018      macro with the name beginning with "OPENSSL_SYS_", which is determined
 8019      from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
 8020      what is available.
 8021      [Richard Levitte]
 8022 
 8023   *) New option -set_serial to 'req' and 'x509' this allows the serial
 8024      number to use to be specified on the command line. Previously self
 8025      signed certificates were hard coded with serial number 0 and the
 8026      CA options of 'x509' had to use a serial number in a file which was
 8027      auto incremented.
 8028      [Steve Henson]
 8029 
 8030   *) New options to 'ca' utility to support V2 CRL entry extensions.
 8031      Currently CRL reason, invalidity date and hold instruction are
 8032      supported. Add new CRL extensions to V3 code and some new objects.
 8033      [Steve Henson]
 8034 
 8035   *) New function EVP_CIPHER_CTX_set_padding() this is used to
 8036      disable standard block padding (aka PKCS#5 padding) in the EVP
 8037      API, which was previously mandatory. This means that the data is
 8038      not padded in any way and so the total length much be a multiple
 8039      of the block size, otherwise an error occurs.
 8040      [Steve Henson]
 8041 
 8042   *) Initial (incomplete) OCSP SSL support.
 8043      [Steve Henson]
 8044 
 8045   *) New function OCSP_parse_url(). This splits up a URL into its host,
 8046      port and path components: primarily to parse OCSP URLs. New -url
 8047      option to ocsp utility.
 8048      [Steve Henson]
 8049 
 8050   *) New nonce behavior. The return value of OCSP_check_nonce() now
 8051      reflects the various checks performed. Applications can decide
 8052      whether to tolerate certain situations such as an absent nonce
 8053      in a response when one was present in a request: the ocsp application
 8054      just prints out a warning. New function OCSP_add1_basic_nonce()
 8055      this is to allow responders to include a nonce in a response even if
 8056      the request is nonce-less.
 8057      [Steve Henson]
 8058 
 8059   *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
 8060      skipped when using openssl x509 multiple times on a single input file,
 8061      e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
 8062      [Bodo Moeller]
 8063 
 8064   *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
 8065      set string type: to handle setting ASN1_TIME structures. Fix ca
 8066      utility to correctly initialize revocation date of CRLs.
 8067      [Steve Henson]
 8068 
 8069   *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
 8070      the clients preferred ciphersuites and rather use its own preferences.
 8071      Should help to work around M$ SGC (Server Gated Cryptography) bug in
 8072      Internet Explorer by ensuring unchanged hash method during stepup.
 8073      (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
 8074      [Lutz Jaenicke]
 8075 
 8076   *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
 8077      to aes and add a new 'exist' option to print out symbols that don't
 8078      appear to exist.
 8079      [Steve Henson]
 8080 
 8081   *) Additional options to ocsp utility to allow flags to be set and
 8082      additional certificates supplied.
 8083      [Steve Henson]
 8084 
 8085   *) Add the option -VAfile to 'openssl ocsp', so the user can give the
 8086      OCSP client a number of certificate to only verify the response
 8087      signature against.
 8088      [Richard Levitte]
 8089 
 8090   *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
 8091      handle the new API. Currently only ECB, CBC modes supported. Add new
 8092      AES OIDs.
 8093 
 8094      Add TLS AES ciphersuites as described in RFC3268, "Advanced
 8095      Encryption Standard (AES) Ciphersuites for Transport Layer
 8096      Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
 8097      not enabled by default and were not part of the "ALL" ciphersuite
 8098      alias because they were not yet official; they could be
 8099      explicitly requested by specifying the "AESdraft" ciphersuite
 8100      group alias.  In the final release of OpenSSL 0.9.7, the group
 8101      alias is called "AES" and is part of "ALL".)
 8102      [Ben Laurie, Steve  Henson, Bodo Moeller]
 8103 
 8104   *) New function OCSP_copy_nonce() to copy nonce value (if present) from
 8105      request to response.
 8106      [Steve Henson]
 8107 
 8108   *) Functions for OCSP responders. OCSP_request_onereq_count(),
 8109      OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
 8110      extract information from a certificate request. OCSP_response_create()
 8111      creates a response and optionally adds a basic response structure.
 8112      OCSP_basic_add1_status() adds a complete single response to a basic
 8113      response and returns the OCSP_SINGLERESP structure just added (to allow
 8114      extensions to be included for example). OCSP_basic_add1_cert() adds a
 8115      certificate to a basic response and OCSP_basic_sign() signs a basic
 8116      response with various flags. New helper functions ASN1_TIME_check()
 8117      (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
 8118      (converts ASN1_TIME to GeneralizedTime).
 8119      [Steve Henson]
 8120 
 8121   *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
 8122      in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
 8123      structure from a certificate. X509_pubkey_digest() digests the public_key
 8124      contents: this is used in various key identifiers.
 8125      [Steve Henson]
 8126 
 8127   *) Make sk_sort() tolerate a NULL argument.
 8128      [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
 8129 
 8130   *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
 8131      passed by the function are trusted implicitly. If any of them signed the
 8132      response then it is assumed to be valid and is not verified.
 8133      [Steve Henson]
 8134 
 8135   *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
 8136      to data. This was previously part of the PKCS7 ASN1 code. This
 8137      was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
 8138      [Steve Henson, reported by Kenneth R. Robinette
 8139                                 <support@securenetterm.com>]
 8140 
 8141   *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
 8142      routines: without these tracing memory leaks is very painful.
 8143      Fix leaks in PKCS12 and PKCS7 routines.
 8144      [Steve Henson]
 8145 
 8146   *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
 8147      Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
 8148      effectively meant GeneralizedTime would never be used. Now it
 8149      is initialised to -1 but X509_time_adj() now has to check the value
 8150      and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
 8151      V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
 8152      [Steve Henson, reported by Kenneth R. Robinette
 8153                                 <support@securenetterm.com>]
 8154 
 8155   *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
 8156      result in a zero length in the ASN1_INTEGER structure which was
 8157      not consistent with the structure when d2i_ASN1_INTEGER() was used
 8158      and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
 8159      to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
 8160      where it did not print out a minus for negative ASN1_INTEGER.
 8161      [Steve Henson]
 8162 
 8163   *) Add summary printout to ocsp utility. The various functions which
 8164      convert status values to strings have been renamed to:
 8165      OCSP_response_status_str(), OCSP_cert_status_str() and
 8166      OCSP_crl_reason_str() and are no longer static. New options
 8167      to verify nonce values and to disable verification. OCSP response
 8168      printout format cleaned up.
 8169      [Steve Henson]
 8170 
 8171   *) Add additional OCSP certificate checks. These are those specified
 8172      in RFC2560. This consists of two separate checks: the CA of the
 8173      certificate being checked must either be the OCSP signer certificate
 8174      or the issuer of the OCSP signer certificate. In the latter case the
 8175      OCSP signer certificate must contain the OCSP signing extended key
 8176      usage. This check is performed by attempting to match the OCSP
 8177      signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
 8178      in the OCSP_CERTID structures of the response.
 8179      [Steve Henson]
 8180 
 8181   *) Initial OCSP certificate verification added to OCSP_basic_verify()
 8182      and related routines. This uses the standard OpenSSL certificate
 8183      verify routines to perform initial checks (just CA validity) and
 8184      to obtain the certificate chain. Then additional checks will be
 8185      performed on the chain. Currently the root CA is checked to see
 8186      if it is explicitly trusted for OCSP signing. This is used to set
 8187      a root CA as a global signing root: that is any certificate that
 8188      chains to that CA is an acceptable OCSP signing certificate.
 8189      [Steve Henson]
 8190 
 8191   *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
 8192      extensions from a separate configuration file.
 8193      As when reading extensions from the main configuration file,
 8194      the '-extensions ...' option may be used for specifying the
 8195      section to use.
 8196      [Massimiliano Pala <madwolf@comune.modena.it>]
 8197 
 8198   *) New OCSP utility. Allows OCSP requests to be generated or
 8199      read. The request can be sent to a responder and the output
 8200      parsed, outputed or printed in text form. Not complete yet:
 8201      still needs to check the OCSP response validity.
 8202      [Steve Henson]
 8203 
 8204   *) New subcommands for 'openssl ca':
 8205      'openssl ca -status <serial>' prints the status of the cert with
 8206      the given serial number (according to the index file).
 8207      'openssl ca -updatedb' updates the expiry status of certificates
 8208      in the index file.
 8209      [Massimiliano Pala <madwolf@comune.modena.it>]
 8210 
 8211   *) New '-newreq-nodes' command option to CA.pl.  This is like
 8212      '-newreq', but calls 'openssl req' with the '-nodes' option
 8213      so that the resulting key is not encrypted.
 8214      [Damien Miller <djm@mindrot.org>]
 8215 
 8216   *) New configuration for the GNU Hurd.
 8217      [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
 8218 
 8219   *) Initial code to implement OCSP basic response verify. This
 8220      is currently incomplete. Currently just finds the signer's
 8221      certificate and verifies the signature on the response.
 8222      [Steve Henson]
 8223 
 8224   *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
 8225      value of OPENSSLDIR.  This is available via the new '-d' option
 8226      to 'openssl version', and is also included in 'openssl version -a'.
 8227      [Bodo Moeller]
 8228 
 8229   *) Allowing defining memory allocation callbacks that will be given
 8230      file name and line number information in additional arguments
 8231      (a const char* and an int).  The basic functionality remains, as
 8232      well as the original possibility to just replace malloc(),
 8233      realloc() and free() by functions that do not know about these
 8234      additional arguments.  To register and find out the current
 8235      settings for extended allocation functions, the following
 8236      functions are provided:
 8237 
 8238         CRYPTO_set_mem_ex_functions
 8239         CRYPTO_set_locked_mem_ex_functions
 8240         CRYPTO_get_mem_ex_functions
 8241         CRYPTO_get_locked_mem_ex_functions
 8242 
 8243      These work the same way as CRYPTO_set_mem_functions and friends.
 8244      CRYPTO_get_[locked_]mem_functions now writes 0 where such an
 8245      extended allocation function is enabled.
 8246      Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
 8247      a conventional allocation function is enabled.
 8248      [Richard Levitte, Bodo Moeller]
 8249 
 8250   *) Finish off removing the remaining LHASH function pointer casts.
 8251      There should no longer be any prototype-casting required when using
 8252      the LHASH abstraction, and any casts that remain are "bugs". See
 8253      the callback types and macros at the head of lhash.h for details
 8254      (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
 8255      [Geoff Thorpe]
 8256 
 8257   *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
 8258      If /dev/[u]random devices are not available or do not return enough
 8259      entropy, EGD style sockets (served by EGD or PRNGD) will automatically
 8260      be queried.
 8261      The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
 8262      /etc/entropy will be queried once each in this sequence, querying stops
 8263      when enough entropy was collected without querying more sockets.
 8264      [Lutz Jaenicke]
 8265 
 8266   *) Change the Unix RAND_poll() variant to be able to poll several
 8267      random devices, as specified by DEVRANDOM, until a sufficient amount
 8268      of data has been collected.   We spend at most 10 ms on each file
 8269      (select timeout) and read in non-blocking mode.  DEVRANDOM now
 8270      defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
 8271      (previously it was just the string "/dev/urandom"), so on typical
 8272      platforms the 10 ms delay will never occur.
 8273      Also separate out the Unix variant to its own file, rand_unix.c.
 8274      For VMS, there's a currently-empty rand_vms.c.
 8275      [Richard Levitte]
 8276 
 8277   *) Move OCSP client related routines to ocsp_cl.c. These
 8278      provide utility functions which an application needing
 8279      to issue a request to an OCSP responder and analyse the
 8280      response will typically need: as opposed to those which an
 8281      OCSP responder itself would need which will be added later.
 8282 
 8283      OCSP_request_sign() signs an OCSP request with an API similar
 8284      to PKCS7_sign(). OCSP_response_status() returns status of OCSP
 8285      response. OCSP_response_get1_basic() extracts basic response
 8286      from response. OCSP_resp_find_status(): finds and extracts status
 8287      information from an OCSP_CERTID structure (which will be created
 8288      when the request structure is built). These are built from lower
 8289      level functions which work on OCSP_SINGLERESP structures but
 8290      won't normally be used unless the application wishes to examine
 8291      extensions in the OCSP response for example.
 8292 
 8293      Replace nonce routines with a pair of functions.
 8294      OCSP_request_add1_nonce() adds a nonce value and optionally
 8295      generates a random value. OCSP_check_nonce() checks the
 8296      validity of the nonce in an OCSP response.
 8297      [Steve Henson]
 8298 
 8299   *) Change function OCSP_request_add() to OCSP_request_add0_id().
 8300      This doesn't copy the supplied OCSP_CERTID and avoids the
 8301      need to free up the newly created id. Change return type
 8302      to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
 8303      This can then be used to add extensions to the request.
 8304      Deleted OCSP_request_new(), since most of its functionality
 8305      is now in OCSP_REQUEST_new() (and the case insensitive name
 8306      clash) apart from the ability to set the request name which
 8307      will be added elsewhere.
 8308      [Steve Henson]
 8309 
 8310   *) Update OCSP API. Remove obsolete extensions argument from
 8311      various functions. Extensions are now handled using the new
 8312      OCSP extension code. New simple OCSP HTTP function which
 8313      can be used to send requests and parse the response.
 8314      [Steve Henson]
 8315 
 8316   *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
 8317      ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
 8318      uses the special reorder version of SET OF to sort the attributes
 8319      and reorder them to match the encoded order. This resolves a long
 8320      standing problem: a verify on a PKCS7 structure just after signing
 8321      it used to fail because the attribute order did not match the
 8322      encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
 8323      it uses the received order. This is necessary to tolerate some broken
 8324      software that does not order SET OF. This is handled by encoding
 8325      as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
 8326      to produce the required SET OF.
 8327      [Steve Henson]
 8328 
 8329   *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
 8330      OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
 8331      files to get correct declarations of the ASN.1 item variables.
 8332      [Richard Levitte]
 8333 
 8334   *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
 8335      PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
 8336      asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
 8337      NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
 8338      New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
 8339      ASN1_ITEM and no wrapper functions.
 8340      [Steve Henson]
 8341 
 8342   *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
 8343      replace the old function pointer based I/O routines. Change most of
 8344      the *_d2i_bio() and *_d2i_fp() functions to use these.
 8345      [Steve Henson]
 8346 
 8347   *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
 8348      lines, recognize more "algorithms" that can be deselected, and make
 8349      it complain about algorithm deselection that isn't recognised.
 8350      [Richard Levitte]
 8351 
 8352   *) New ASN1 functions to handle dup, sign, verify, digest, pack and
 8353      unpack operations in terms of ASN1_ITEM. Modify existing wrappers
 8354      to use new functions. Add NO_ASN1_OLD which can be set to remove
 8355      some old style ASN1 functions: this can be used to determine if old
 8356      code will still work when these eventually go away.
 8357      [Steve Henson]
 8358 
 8359   *) New extension functions for OCSP structures, these follow the
 8360      same conventions as certificates and CRLs.
 8361      [Steve Henson]
 8362 
 8363   *) New function X509V3_add1_i2d(). This automatically encodes and
 8364      adds an extension. Its behaviour can be customised with various
 8365      flags to append, replace or delete. Various wrappers added for
 8366      certificates and CRLs.
 8367      [Steve Henson]
 8368 
 8369   *) Fix to avoid calling the underlying ASN1 print routine when
 8370      an extension cannot be parsed. Correct a typo in the
 8371      OCSP_SERVICELOC extension. Tidy up print OCSP format.
 8372      [Steve Henson]
 8373 
 8374   *) Make mkdef.pl parse some of the ASN1 macros and add appropriate
 8375      entries for variables.
 8376      [Steve Henson]
 8377 
 8378   *) Add functionality to apps/openssl.c for detecting locking
 8379      problems: As the program is single-threaded, all we have
 8380      to do is register a locking callback using an array for
 8381      storing which locks are currently held by the program.
 8382      [Bodo Moeller]
 8383 
 8384   *) Use a lock around the call to CRYPTO_get_ex_new_index() in
 8385      SSL_get_ex_data_X509_STORE_idx(), which is used in
 8386      ssl_verify_cert_chain() and thus can be called at any time
 8387      during TLS/SSL handshakes so that thread-safety is essential.
 8388      Unfortunately, the ex_data design is not at all suited
 8389      for multi-threaded use, so it probably should be abolished.
 8390      [Bodo Moeller]
 8391 
 8392   *) Added Broadcom "ubsec" ENGINE to OpenSSL.
 8393      [Broadcom, tweaked and integrated by Geoff Thorpe]
 8394 
 8395   *) Move common extension printing code to new function
 8396      X509V3_print_extensions(). Reorganise OCSP print routines and
 8397      implement some needed OCSP ASN1 functions. Add OCSP extensions.
 8398      [Steve Henson]
 8399 
 8400   *) New function X509_signature_print() to remove duplication in some
 8401      print routines.
 8402      [Steve Henson]
 8403 
 8404   *) Add a special meaning when SET OF and SEQUENCE OF flags are both
 8405      set (this was treated exactly the same as SET OF previously). This
 8406      is used to reorder the STACK representing the structure to match the
 8407      encoding. This will be used to get round a problem where a PKCS7
 8408      structure which was signed could not be verified because the STACK
 8409      order did not reflect the encoded order.
 8410      [Steve Henson]
 8411 
 8412   *) Reimplement the OCSP ASN1 module using the new code.
 8413      [Steve Henson]
 8414 
 8415   *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
 8416      for its ASN1 operations. The old style function pointers still exist
 8417      for now but they will eventually go away.
 8418      [Steve Henson]
 8419 
 8420   *) Merge in replacement ASN1 code from the ASN1 branch. This almost
 8421      completely replaces the old ASN1 functionality with a table driven
 8422      encoder and decoder which interprets an ASN1_ITEM structure describing
 8423      the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
 8424      largely maintained. Almost all of the old asn1_mac.h macro based ASN1
 8425      has also been converted to the new form.
 8426      [Steve Henson]
 8427 
 8428   *) Change BN_mod_exp_recp so that negative moduli are tolerated
 8429      (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
 8430      so that BN_mod_exp_mont and BN_mod_exp_mont_word work
 8431      for negative moduli.
 8432      [Bodo Moeller]
 8433 
 8434   *) Fix BN_uadd and BN_usub: Always return non-negative results instead
 8435      of not touching the result's sign bit.
 8436      [Bodo Moeller]
 8437 
 8438   *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
 8439      set.
 8440      [Bodo Moeller]
 8441 
 8442   *) Changed the LHASH code to use prototypes for callbacks, and created
 8443      macros to declare and implement thin (optionally static) functions
 8444      that provide type-safety and avoid function pointer casting for the
 8445      type-specific callbacks.
 8446      [Geoff Thorpe]
 8447 
 8448   *) Added Kerberos Cipher Suites to be used with TLS, as written in
 8449      RFC 2712.
 8450      [Veers Staats <staatsvr@asc.hpc.mil>,
 8451       Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
 8452 
 8453   *) Reformat the FAQ so the different questions and answers can be divided
 8454      in sections depending on the subject.
 8455      [Richard Levitte]
 8456 
 8457   *) Have the zlib compression code load ZLIB.DLL dynamically under
 8458      Windows.
 8459      [Richard Levitte]
 8460 
 8461   *) New function BN_mod_sqrt for computing square roots modulo a prime
 8462      (using the probabilistic Tonelli-Shanks algorithm unless
 8463      p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
 8464      be handled deterministically).
 8465      [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
 8466 
 8467   *) Make BN_mod_inverse faster by explicitly handling small quotients
 8468      in the Euclid loop. (Speed gain about 20% for small moduli [256 or
 8469      512 bits], about 30% for larger ones [1024 or 2048 bits].)
 8470      [Bodo Moeller]
 8471 
 8472   *) New function BN_kronecker.
 8473      [Bodo Moeller]
 8474 
 8475   *) Fix BN_gcd so that it works on negative inputs; the result is
 8476      positive unless both parameters are zero.
 8477      Previously something reasonably close to an infinite loop was
 8478      possible because numbers could be growing instead of shrinking
 8479      in the implementation of Euclid's algorithm.
 8480      [Bodo Moeller]
 8481 
 8482   *) Fix BN_is_word() and BN_is_one() macros to take into account the
 8483      sign of the number in question.
 8484 
 8485      Fix BN_is_word(a,w) to work correctly for w == 0.
 8486 
 8487      The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
 8488      because its test if the absolute value of 'a' equals 'w'.
 8489      Note that BN_abs_is_word does *not* handle w == 0 reliably;
 8490      it exists mostly for use in the implementations of BN_is_zero(),
 8491      BN_is_one(), and BN_is_word().
 8492      [Bodo Moeller]
 8493 
 8494   *) New function BN_swap.
 8495      [Bodo Moeller]
 8496 
 8497   *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
 8498      the exponentiation functions are more likely to produce reasonable
 8499      results on negative inputs.
 8500      [Bodo Moeller]
 8501 
 8502   *) Change BN_mod_mul so that the result is always non-negative.
 8503      Previously, it could be negative if one of the factors was negative;
 8504      I don't think anyone really wanted that behaviour.
 8505      [Bodo Moeller]
 8506 
 8507   *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
 8508      (except for exponentiation, which stays in crypto/bn/bn_exp.c,
 8509      and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
 8510      and add new functions:
 8511 
 8512           BN_nnmod
 8513           BN_mod_sqr
 8514           BN_mod_add
 8515           BN_mod_add_quick
 8516           BN_mod_sub
 8517           BN_mod_sub_quick
 8518           BN_mod_lshift1
 8519           BN_mod_lshift1_quick
 8520           BN_mod_lshift
 8521           BN_mod_lshift_quick
 8522 
 8523      These functions always generate non-negative results.
 8524 
 8525      BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder  r
 8526      such that  |m| < r < 0,  BN_nnmod will output  rem + |m|  instead).
 8527 
 8528      BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
 8529      BN_mod_XXX(r, a, [b,] m, ctx), but requires that  a  [and  b]
 8530      be reduced modulo  m.
 8531      [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
 8532 
 8533 #if 0
 8534      The following entry accidentally appeared in the CHANGES file
 8535      distributed with OpenSSL 0.9.7.  The modifications described in
 8536      it do *not* apply to OpenSSL 0.9.7.
 8537 
 8538   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
 8539      was actually never needed) and in BN_mul().  The removal in BN_mul()
 8540      required a small change in bn_mul_part_recursive() and the addition
 8541      of the functions bn_cmp_part_words(), bn_sub_part_words() and
 8542      bn_add_part_words(), which do the same thing as bn_cmp_words(),
 8543      bn_sub_words() and bn_add_words() except they take arrays with
 8544      differing sizes.
 8545      [Richard Levitte]
 8546 #endif
 8547 
 8548   *) In 'openssl passwd', verify passwords read from the terminal
 8549      unless the '-salt' option is used (which usually means that
 8550      verification would just waste user's time since the resulting
 8551      hash is going to be compared with some given password hash)
 8552      or the new '-noverify' option is used.
 8553 
 8554      This is an incompatible change, but it does not affect
 8555      non-interactive use of 'openssl passwd' (passwords on the command
 8556      line, '-stdin' option, '-in ...' option) and thus should not
 8557      cause any problems.
 8558      [Bodo Moeller]
 8559 
 8560   *) Remove all references to RSAref, since there's no more need for it.
 8561      [Richard Levitte]
 8562 
 8563   *) Make DSO load along a path given through an environment variable
 8564      (SHLIB_PATH) with shl_load().
 8565      [Richard Levitte]
 8566 
 8567   *) Constify the ENGINE code as a result of BIGNUM constification.
 8568      Also constify the RSA code and most things related to it.  In a
 8569      few places, most notable in the depth of the ASN.1 code, ugly
 8570      casts back to non-const were required (to be solved at a later
 8571      time)
 8572      [Richard Levitte]
 8573 
 8574   *) Make it so the openssl application has all engines loaded by default.
 8575      [Richard Levitte]
 8576 
 8577   *) Constify the BIGNUM routines a little more.
 8578      [Richard Levitte]
 8579 
 8580   *) Add the following functions:
 8581 
 8582         ENGINE_load_cswift()
 8583         ENGINE_load_chil()
 8584         ENGINE_load_atalla()
 8585         ENGINE_load_nuron()
 8586         ENGINE_load_builtin_engines()
 8587 
 8588      That way, an application can itself choose if external engines that
 8589      are built-in in OpenSSL shall ever be used or not.  The benefit is
 8590      that applications won't have to be linked with libdl or other dso
 8591      libraries unless it's really needed.
 8592 
 8593      Changed 'openssl engine' to load all engines on demand.
 8594      Changed the engine header files to avoid the duplication of some
 8595      declarations (they differed!).
 8596      [Richard Levitte]
 8597 
 8598   *) 'openssl engine' can now list capabilities.
 8599      [Richard Levitte]
 8600 
 8601   *) Better error reporting in 'openssl engine'.
 8602      [Richard Levitte]
 8603 
 8604   *) Never call load_dh_param(NULL) in s_server.
 8605      [Bodo Moeller]
 8606 
 8607   *) Add engine application.  It can currently list engines by name and
 8608      identity, and test if they are actually available.
 8609      [Richard Levitte]
 8610 
 8611   *) Improve RPM specification file by forcing symbolic linking and making
 8612      sure the installed documentation is also owned by root.root.
 8613      [Damien Miller <djm@mindrot.org>]
 8614 
 8615   *) Give the OpenSSL applications more possibilities to make use of
 8616      keys (public as well as private) handled by engines.
 8617      [Richard Levitte]
 8618 
 8619   *) Add OCSP code that comes from CertCo.
 8620      [Richard Levitte]
 8621 
 8622   *) Add VMS support for the Rijndael code.
 8623      [Richard Levitte]
 8624 
 8625   *) Added untested support for Nuron crypto accelerator.
 8626      [Ben Laurie]
 8627 
 8628   *) Add support for external cryptographic devices.  This code was
 8629      previously distributed separately as the "engine" branch.
 8630      [Geoff Thorpe, Richard Levitte]
 8631 
 8632   *) Rework the filename-translation in the DSO code. It is now possible to
 8633      have far greater control over how a "name" is turned into a filename
 8634      depending on the operating environment and any oddities about the
 8635      different shared library filenames on each system.
 8636      [Geoff Thorpe]
 8637 
 8638   *) Support threads on FreeBSD-elf in Configure.
 8639      [Richard Levitte]
 8640 
 8641   *) Fix for SHA1 assembly problem with MASM: it produces
 8642      warnings about corrupt line number information when assembling
 8643      with debugging information. This is caused by the overlapping
 8644      of two sections.
 8645      [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
 8646 
 8647   *) NCONF changes.
 8648      NCONF_get_number() has no error checking at all.  As a replacement,
 8649      NCONF_get_number_e() is defined (_e for "error checking") and is
 8650      promoted strongly.  The old NCONF_get_number is kept around for
 8651      binary backward compatibility.
 8652      Make it possible for methods to load from something other than a BIO,
 8653      by providing a function pointer that is given a name instead of a BIO.
 8654      For example, this could be used to load configuration data from an
 8655      LDAP server.
 8656      [Richard Levitte]
 8657 
 8658   *) Fix for non blocking accept BIOs. Added new I/O special reason
 8659      BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
 8660      with non blocking I/O was not possible because no retry code was
 8661      implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
 8662      this case.
 8663      [Steve Henson]
 8664 
 8665   *) Added the beginnings of Rijndael support.
 8666      [Ben Laurie]
 8667 
 8668   *) Fix for bug in DirectoryString mask setting. Add support for
 8669      X509_NAME_print_ex() in 'req' and X509_print_ex() function
 8670      to allow certificate printing to more controllable, additional
 8671      'certopt' option to 'x509' to allow new printing options to be
 8672      set.
 8673      [Steve Henson]
 8674 
 8675   *) Clean old EAY MD5 hack from e_os.h.
 8676      [Richard Levitte]
 8677 
 8678  Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
 8679 
 8680   *) Fix null-pointer assignment in do_change_cipher_spec() revealed
 8681      by using the Codenomicon TLS Test Tool (CVE-2004-0079)
 8682      [Joe Orton, Steve Henson]
 8683 
 8684  Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
 8685 
 8686   *) Fix additional bug revealed by the NISCC test suite:
 8687 
 8688      Stop bug triggering large recursion when presented with
 8689      certain ASN.1 tags (CVE-2003-0851)
 8690      [Steve Henson]
 8691 
 8692  Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
 8693 
 8694   *) Fix various bugs revealed by running the NISCC test suite:
 8695 
 8696      Stop out of bounds reads in the ASN1 code when presented with
 8697      invalid tags (CVE-2003-0543 and CVE-2003-0544).
 8698 
 8699      If verify callback ignores invalid public key errors don't try to check
 8700      certificate signature with the NULL public key.
 8701 
 8702      [Steve Henson]
 8703 
 8704   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
 8705      if the server requested one: as stated in TLS 1.0 and SSL 3.0
 8706      specifications.
 8707      [Steve Henson]
 8708 
 8709   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
 8710      extra data after the compression methods not only for TLS 1.0
 8711      but also for SSL 3.0 (as required by the specification).
 8712      [Bodo Moeller; problem pointed out by Matthias Loepfe]
 8713 
 8714   *) Change X509_certificate_type() to mark the key as exported/exportable
 8715      when it's 512 *bits* long, not 512 bytes.
 8716      [Richard Levitte]
 8717 
 8718  Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
 8719 
 8720   *) Countermeasure against the Klima-Pokorny-Rosa extension of
 8721      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
 8722      a protocol version number mismatch like a decryption error
 8723      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
 8724      [Bodo Moeller]
 8725 
 8726   *) Turn on RSA blinding by default in the default implementation
 8727      to avoid a timing attack. Applications that don't want it can call
 8728      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
 8729      They would be ill-advised to do so in most cases.
 8730      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
 8731 
 8732   *) Change RSA blinding code so that it works when the PRNG is not
 8733      seeded (in this case, the secret RSA exponent is abused as
 8734      an unpredictable seed -- if it is not unpredictable, there
 8735      is no point in blinding anyway).  Make RSA blinding thread-safe
 8736      by remembering the creator's thread ID in rsa->blinding and
 8737      having all other threads use local one-time blinding factors
 8738      (this requires more computation than sharing rsa->blinding, but
 8739      avoids excessive locking; and if an RSA object is not shared
 8740      between threads, blinding will still be very fast).
 8741      [Bodo Moeller]
 8742 
 8743  Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
 8744 
 8745   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
 8746      via timing by performing a MAC computation even if incorrect
 8747      block cipher padding has been found.  This is a countermeasure
 8748      against active attacks where the attacker has to distinguish
 8749      between bad padding and a MAC verification error. (CVE-2003-0078)
 8750 
 8751      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
 8752      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
 8753      Martin Vuagnoux (EPFL, Ilion)]
 8754 
 8755  Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
 8756 
 8757   *) New function OPENSSL_cleanse(), which is used to cleanse a section of
 8758      memory from it's contents.  This is done with a counter that will
 8759      place alternating values in each byte.  This can be used to solve
 8760      two issues: 1) the removal of calls to memset() by highly optimizing
 8761      compilers, and 2) cleansing with other values than 0, since those can
 8762      be read through on certain media, for example a swap space on disk.
 8763      [Geoff Thorpe]
 8764 
 8765   *) Bugfix: client side session caching did not work with external caching,
 8766      because the session->cipher setting was not restored when reloading
 8767      from the external cache. This problem was masked, when
 8768      SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
 8769      (Found by Steve Haslam <steve@araqnid.ddts.net>.)
 8770      [Lutz Jaenicke]
 8771 
 8772   *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
 8773      length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
 8774      [Zeev Lieber <zeev-l@yahoo.com>]
 8775 
 8776   *) Undo an undocumented change introduced in 0.9.6e which caused
 8777      repeated calls to OpenSSL_add_all_ciphers() and
 8778      OpenSSL_add_all_digests() to be ignored, even after calling
 8779      EVP_cleanup().
 8780      [Richard Levitte]
 8781 
 8782   *) Change the default configuration reader to deal with last line not
 8783      being properly terminated.
 8784      [Richard Levitte]
 8785 
 8786   *) Change X509_NAME_cmp() so it applies the special rules on handling
 8787      DN values that are of type PrintableString, as well as RDNs of type
 8788      emailAddress where the value has the type ia5String.
 8789      [stefank@valicert.com via Richard Levitte]
 8790 
 8791   *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
 8792      the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
 8793      doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
 8794      the bitwise-OR of the two for use by the majority of applications
 8795      wanting this behaviour, and update the docs. The documented
 8796      behaviour and actual behaviour were inconsistent and had been
 8797      changing anyway, so this is more a bug-fix than a behavioural
 8798      change.
 8799      [Geoff Thorpe, diagnosed by Nadav Har'El]
 8800 
 8801   *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
 8802      (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
 8803      [Bodo Moeller]
 8804 
 8805   *) Fix initialization code race conditions in
 8806         SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
 8807         SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
 8808         SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
 8809         TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
 8810         ssl2_get_cipher_by_char(),
 8811         ssl3_get_cipher_by_char().
 8812      [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
 8813 
 8814   *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
 8815      the cached sessions are flushed, as the remove_cb() might use ex_data
 8816      contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
 8817      (see [openssl.org #212]).
 8818      [Geoff Thorpe, Lutz Jaenicke]
 8819 
 8820   *) Fix typo in OBJ_txt2obj which incorrectly passed the content
 8821      length, instead of the encoding length to d2i_ASN1_OBJECT.
 8822      [Steve Henson]
 8823 
 8824  Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
 8825 
 8826   *) [In 0.9.6g-engine release:]
 8827      Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
 8828      [Lynn Gazis <lgazis@rainbow.com>]
 8829 
 8830  Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
 8831 
 8832   *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
 8833      and get fix the header length calculation.
 8834      [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
 8835         Alon Kantor <alonk@checkpoint.com> (and others),
 8836         Steve Henson]
 8837 
 8838   *) Use proper error handling instead of 'assertions' in buffer
 8839      overflow checks added in 0.9.6e.  This prevents DoS (the
 8840      assertions could call abort()).
 8841      [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
 8842 
 8843  Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
 8844 
 8845   *) Add various sanity checks to asn1_get_length() to reject
 8846      the ASN1 length bytes if they exceed sizeof(long), will appear
 8847      negative or the content length exceeds the length of the
 8848      supplied buffer.
 8849      [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
 8850 
 8851   *) Fix cipher selection routines: ciphers without encryption had no flags
 8852      for the cipher strength set and where therefore not handled correctly
 8853      by the selection routines (PR #130).
 8854      [Lutz Jaenicke]
 8855 
 8856   *) Fix EVP_dsa_sha macro.
 8857      [Nils Larsch]
 8858 
 8859   *) New option
 8860           SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
 8861      for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
 8862      that was added in OpenSSL 0.9.6d.
 8863 
 8864      As the countermeasure turned out to be incompatible with some
 8865      broken SSL implementations, the new option is part of SSL_OP_ALL.
 8866      SSL_OP_ALL is usually employed when compatibility with weird SSL
 8867      implementations is desired (e.g. '-bugs' option to 's_client' and
 8868      's_server'), so the new option is automatically set in many
 8869      applications.
 8870      [Bodo Moeller]
 8871 
 8872   *) Changes in security patch:
 8873 
 8874      Changes marked "(CHATS)" were sponsored by the Defense Advanced
 8875      Research Projects Agency (DARPA) and Air Force Research Laboratory,
 8876      Air Force Materiel Command, USAF, under agreement number
 8877      F30602-01-2-0537.
 8878 
 8879   *) Add various sanity checks to asn1_get_length() to reject
 8880      the ASN1 length bytes if they exceed sizeof(long), will appear
 8881      negative or the content length exceeds the length of the
 8882      supplied buffer. (CVE-2002-0659)
 8883      [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
 8884 
 8885   *) Assertions for various potential buffer overflows, not known to
 8886      happen in practice.
 8887      [Ben Laurie (CHATS)]
 8888 
 8889   *) Various temporary buffers to hold ASCII versions of integers were
 8890      too small for 64 bit platforms. (CVE-2002-0655)
 8891      [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
 8892 
 8893   *) Remote buffer overflow in SSL3 protocol - an attacker could
 8894      supply an oversized session ID to a client. (CVE-2002-0656)
 8895      [Ben Laurie (CHATS)]
 8896 
 8897   *) Remote buffer overflow in SSL2 protocol - an attacker could
 8898      supply an oversized client master key. (CVE-2002-0656)
 8899      [Ben Laurie (CHATS)]
 8900 
 8901  Changes between 0.9.6c and 0.9.6d  [9 May 2002]
 8902 
 8903   *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
 8904      encoded as NULL) with id-dsa-with-sha1.
 8905      [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
 8906 
 8907   *) Check various X509_...() return values in apps/req.c.
 8908      [Nils Larsch <nla@trustcenter.de>]
 8909 
 8910   *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
 8911      an end-of-file condition would erroneously be flagged, when the CRLF
 8912      was just at the end of a processed block. The bug was discovered when
 8913      processing data through a buffering memory BIO handing the data to a
 8914      BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
 8915      <ptsekov@syntrex.com> and Nedelcho Stanev.
 8916      [Lutz Jaenicke]
 8917 
 8918   *) Implement a countermeasure against a vulnerability recently found
 8919      in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
 8920      before application data chunks to avoid the use of known IVs
 8921      with data potentially chosen by the attacker.
 8922      [Bodo Moeller]
 8923 
 8924   *) Fix length checks in ssl3_get_client_hello().
 8925      [Bodo Moeller]
 8926 
 8927   *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
 8928      to prevent ssl3_read_internal() from incorrectly assuming that
 8929      ssl3_read_bytes() found application data while handshake
 8930      processing was enabled when in fact s->s3->in_read_app_data was
 8931      merely automatically cleared during the initial handshake.
 8932      [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
 8933 
 8934   *) Fix object definitions for Private and Enterprise: they were not
 8935      recognized in their shortname (=lowercase) representation. Extend
 8936      obj_dat.pl to issue an error when using undefined keywords instead
 8937      of silently ignoring the problem (Svenning Sorensen
 8938      <sss@sss.dnsalias.net>).
 8939      [Lutz Jaenicke]
 8940 
 8941   *) Fix DH_generate_parameters() so that it works for 'non-standard'
 8942      generators, i.e. generators other than 2 and 5.  (Previously, the
 8943      code did not properly initialise the 'add' and 'rem' values to
 8944      BN_generate_prime().)
 8945 
 8946      In the new general case, we do not insist that 'generator' is
 8947      actually a primitive root: This requirement is rather pointless;
 8948      a generator of the order-q subgroup is just as good, if not
 8949      better.
 8950      [Bodo Moeller]
 8951 
 8952   *) Map new X509 verification errors to alerts. Discovered and submitted by
 8953      Tom Wu <tom@arcot.com>.
 8954      [Lutz Jaenicke]
 8955 
 8956   *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
 8957      returning non-zero before the data has been completely received
 8958      when using non-blocking I/O.
 8959      [Bodo Moeller; problem pointed out by John Hughes]
 8960 
 8961   *) Some of the ciphers missed the strength entry (SSL_LOW etc).
 8962      [Ben Laurie, Lutz Jaenicke]
 8963 
 8964   *) Fix bug in SSL_clear(): bad sessions were not removed (found by
 8965      Yoram Zahavi <YoramZ@gilian.com>).
 8966      [Lutz Jaenicke]
 8967 
 8968   *) Add information about CygWin 1.3 and on, and preserve proper
 8969      configuration for the versions before that.
 8970      [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
 8971 
 8972   *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
 8973      check whether we deal with a copy of a session and do not delete from
 8974      the cache in this case. Problem reported by "Izhar Shoshani Levi"
 8975      <izhar@checkpoint.com>.
 8976      [Lutz Jaenicke]
 8977 
 8978   *) Do not store session data into the internal session cache, if it
 8979      is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
 8980      flag is set). Proposed by Aslam <aslam@funk.com>.
 8981      [Lutz Jaenicke]
 8982 
 8983   *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
 8984      value is 0.
 8985      [Richard Levitte]
 8986 
 8987   *) [In 0.9.6d-engine release:]
 8988      Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
 8989      [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
 8990 
 8991   *) Add the configuration target linux-s390x.
 8992      [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
 8993 
 8994   *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
 8995      ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
 8996      variable as an indication that a ClientHello message has been
 8997      received.  As the flag value will be lost between multiple
 8998      invocations of ssl3_accept when using non-blocking I/O, the
 8999      function may not be aware that a handshake has actually taken
 9000      place, thus preventing a new session from being added to the
 9001      session cache.
 9002 
 9003      To avoid this problem, we now set s->new_session to 2 instead of
 9004      using a local variable.
 9005      [Lutz Jaenicke, Bodo Moeller]
 9006 
 9007   *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
 9008      if the SSL_R_LENGTH_MISMATCH error is detected.
 9009      [Geoff Thorpe, Bodo Moeller]
 9010 
 9011   *) New 'shared_ldflag' column in Configure platform table.
 9012      [Richard Levitte]
 9013 
 9014   *) Fix EVP_CIPHER_mode macro.
 9015      ["Dan S. Camper" <dan@bti.net>]
 9016 
 9017   *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
 9018      type, we must throw them away by setting rr->length to 0.
 9019      [D P Chang <dpc@qualys.com>]
 9020 
 9021  Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
 9022 
 9023   *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
 9024      <Dominikus.Scherkl@biodata.com>.  (The previous implementation
 9025      worked incorrectly for those cases where  range = 10..._2  and
 9026      3*range  is two bits longer than  range.)
 9027      [Bodo Moeller]
 9028 
 9029   *) Only add signing time to PKCS7 structures if it is not already
 9030      present.
 9031      [Steve Henson]
 9032 
 9033   *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
 9034      OBJ_ld_ce should be OBJ_id_ce.
 9035      Also some ip-pda OIDs in crypto/objects/objects.txt were
 9036      incorrect (cf. RFC 3039).
 9037      [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
 9038 
 9039   *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
 9040      returns early because it has nothing to do.
 9041      [Andy Schneider <andy.schneider@bjss.co.uk>]
 9042 
 9043   *) [In 0.9.6c-engine release:]
 9044      Fix mutex callback return values in crypto/engine/hw_ncipher.c.
 9045      [Andy Schneider <andy.schneider@bjss.co.uk>]
 9046 
 9047   *) [In 0.9.6c-engine release:]
 9048      Add support for Cryptographic Appliance's keyserver technology.
 9049      (Use engine 'keyclient')
 9050      [Cryptographic Appliances and Geoff Thorpe]
 9051 
 9052   *) Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
 9053      is called via tools/c89.sh because arguments have to be
 9054      rearranged (all '-L' options must appear before the first object
 9055      modules).
 9056      [Richard Shapiro <rshapiro@abinitio.com>]
 9057 
 9058   *) [In 0.9.6c-engine release:]
 9059      Add support for Broadcom crypto accelerator cards, backported
 9060      from 0.9.7.
 9061      [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
 9062 
 9063   *) [In 0.9.6c-engine release:]
 9064      Add support for SureWare crypto accelerator cards from
 9065      Baltimore Technologies.  (Use engine 'sureware')
 9066      [Baltimore Technologies and Mark Cox]
 9067 
 9068   *) [In 0.9.6c-engine release:]
 9069      Add support for crypto accelerator cards from Accelerated
 9070      Encryption Processing, www.aep.ie.  (Use engine 'aep')
 9071      [AEP Inc. and Mark Cox]
 9072 
 9073   *) Add a configuration entry for gcc on UnixWare.
 9074      [Gary Benson <gbenson@redhat.com>]
 9075 
 9076   *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
 9077      messages are stored in a single piece (fixed-length part and
 9078      variable-length part combined) and fix various bugs found on the way.
 9079      [Bodo Moeller]
 9080 
 9081   *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
 9082      instead.  BIO_gethostbyname() does not know what timeouts are
 9083      appropriate, so entries would stay in cache even when they have
 9084      become invalid.
 9085      [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
 9086 
 9087   *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
 9088      faced with a pathologically small ClientHello fragment that does
 9089      not contain client_version: Instead of aborting with an error,
 9090      simply choose the highest available protocol version (i.e.,
 9091      TLS 1.0 unless it is disabled).  In practice, ClientHello
 9092      messages are never sent like this, but this change gives us
 9093      strictly correct behaviour at least for TLS.
 9094      [Bodo Moeller]
 9095 
 9096   *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
 9097      never resets s->method to s->ctx->method when called from within
 9098      one of the SSL handshake functions.
 9099      [Bodo Moeller; problem pointed out by Niko Baric]
 9100 
 9101   *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
 9102      (sent using the client's version number) if client_version is
 9103      smaller than the protocol version in use.  Also change
 9104      ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
 9105      the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
 9106      the client will at least see that alert.
 9107      [Bodo Moeller]
 9108 
 9109   *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
 9110      correctly.
 9111      [Bodo Moeller]
 9112 
 9113   *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
 9114      client receives HelloRequest while in a handshake.
 9115      [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
 9116 
 9117   *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
 9118      should end in 'break', not 'goto end' which circumvents various
 9119      cleanups done in state SSL_ST_OK.   But session related stuff
 9120      must be disabled for SSL_ST_OK in the case that we just sent a
 9121      HelloRequest.
 9122 
 9123      Also avoid some overhead by not calling ssl_init_wbio_buffer()
 9124      before just sending a HelloRequest.
 9125      [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
 9126 
 9127   *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
 9128      reveal whether illegal block cipher padding was found or a MAC
 9129      verification error occurred.  (Neither SSLerr() codes nor alerts
 9130      are directly visible to potential attackers, but the information
 9131      may leak via logfiles.)
 9132 
 9133      Similar changes are not required for the SSL 2.0 implementation
 9134      because the number of padding bytes is sent in clear for SSL 2.0,
 9135      and the extra bytes are just ignored.  However ssl/s2_pkt.c
 9136      failed to verify that the purported number of padding bytes is in
 9137      the legal range.
 9138      [Bodo Moeller]
 9139 
 9140   *) Add OpenUNIX-8 support including shared libraries
 9141      (Boyd Lynn Gerber <gerberb@zenez.com>).
 9142      [Lutz Jaenicke]
 9143 
 9144   *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
 9145      'wristwatch attack' using huge encoding parameters (cf.
 9146      James H. Manger's CRYPTO 2001 paper).  Note that the
 9147      RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
 9148      encoding parameters and hence was not vulnerable.
 9149      [Bodo Moeller]
 9150 
 9151   *) BN_sqr() bug fix.
 9152      [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
 9153 
 9154   *) Rabin-Miller test analyses assume uniformly distributed witnesses,
 9155      so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
 9156      followed by modular reduction.
 9157      [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
 9158 
 9159   *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
 9160      equivalent based on BN_pseudo_rand() instead of BN_rand().
 9161      [Bodo Moeller]
 9162 
 9163   *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
 9164      This function was broken, as the check for a new client hello message
 9165      to handle SGC did not allow these large messages.
 9166      (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
 9167      [Lutz Jaenicke]
 9168 
 9169   *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
 9170      [Lutz Jaenicke]
 9171 
 9172   *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
 9173      for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
 9174      [Lutz Jaenicke]
 9175 
 9176   *) Rework the configuration and shared library support for Tru64 Unix.
 9177      The configuration part makes use of modern compiler features and
 9178      still retains old compiler behavior for those that run older versions
 9179      of the OS.  The shared library support part includes a variant that
 9180      uses the RPATH feature, and is available through the special
 9181      configuration target "alpha-cc-rpath", which will never be selected
 9182      automatically.
 9183      [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
 9184 
 9185   *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
 9186      with the same message size as in ssl3_get_certificate_request().
 9187      Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
 9188      messages might inadvertently be reject as too long.
 9189      [Petr Lampa <lampa@fee.vutbr.cz>]
 9190 
 9191   *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
 9192      [Andy Polyakov]
 9193 
 9194   *) Modified SSL library such that the verify_callback that has been set
 9195      specificly for an SSL object with SSL_set_verify() is actually being
 9196      used. Before the change, a verify_callback set with this function was
 9197      ignored and the verify_callback() set in the SSL_CTX at the time of
 9198      the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
 9199      to allow the necessary settings.
 9200      [Lutz Jaenicke]
 9201 
 9202   *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
 9203      explicitly to NULL, as at least on Solaris 8 this seems not always to be
 9204      done automatically (in contradiction to the requirements of the C
 9205      standard). This made problems when used from OpenSSH.
 9206      [Lutz Jaenicke]
 9207 
 9208   *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
 9209      dh->length and always used
 9210 
 9211           BN_rand_range(priv_key, dh->p).
 9212 
 9213      BN_rand_range() is not necessary for Diffie-Hellman, and this
 9214      specific range makes Diffie-Hellman unnecessarily inefficient if
 9215      dh->length (recommended exponent length) is much smaller than the
 9216      length of dh->p.  We could use BN_rand_range() if the order of
 9217      the subgroup was stored in the DH structure, but we only have
 9218      dh->length.
 9219 
 9220      So switch back to
 9221 
 9222           BN_rand(priv_key, l, ...)
 9223 
 9224      where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
 9225      otherwise.
 9226      [Bodo Moeller]
 9227 
 9228   *) In
 9229 
 9230           RSA_eay_public_encrypt
 9231           RSA_eay_private_decrypt
 9232           RSA_eay_private_encrypt (signing)
 9233           RSA_eay_public_decrypt (signature verification)
 9234 
 9235      (default implementations for RSA_public_encrypt,
 9236      RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
 9237      always reject numbers >= n.
 9238      [Bodo Moeller]
 9239 
 9240   *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
 9241      to synchronize access to 'locking_thread'.  This is necessary on
 9242      systems where access to 'locking_thread' (an 'unsigned long'
 9243      variable) is not atomic.
 9244      [Bodo Moeller]
 9245 
 9246   *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
 9247      *before* setting the 'crypto_lock_rand' flag.  The previous code had
 9248      a race condition if 0 is a valid thread ID.
 9249      [Travis Vitek <vitek@roguewave.com>]
 9250 
 9251   *) Add support for shared libraries under Irix.
 9252      [Albert Chin-A-Young <china@thewrittenword.com>]
 9253 
 9254   *) Add configuration option to build on Linux on both big-endian and
 9255      little-endian MIPS.
 9256      [Ralf Baechle <ralf@uni-koblenz.de>]
 9257 
 9258   *) Add the possibility to create shared libraries on HP-UX.
 9259      [Richard Levitte]
 9260 
 9261  Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
 9262 
 9263   *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
 9264      to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
 9265      Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
 9266      PRNG state recovery was possible based on the output of
 9267      one PRNG request appropriately sized to gain knowledge on
 9268      'md' followed by enough consecutive 1-byte PRNG requests
 9269      to traverse all of 'state'.
 9270 
 9271      1. When updating 'md_local' (the current thread's copy of 'md')
 9272         during PRNG output generation, hash all of the previous
 9273         'md_local' value, not just the half used for PRNG output.
 9274 
 9275      2. Make the number of bytes from 'state' included into the hash
 9276         independent from the number of PRNG bytes requested.
 9277 
 9278      The first measure alone would be sufficient to avoid
 9279      Markku-Juhani's attack.  (Actually it had never occurred
 9280      to me that the half of 'md_local' used for chaining was the
 9281      half from which PRNG output bytes were taken -- I had always
 9282      assumed that the secret half would be used.)  The second
 9283      measure makes sure that additional data from 'state' is never
 9284      mixed into 'md_local' in small portions; this heuristically
 9285      further strengthens the PRNG.
 9286      [Bodo Moeller]
 9287 
 9288   *) Fix crypto/bn/asm/mips3.s.
 9289      [Andy Polyakov]
 9290 
 9291   *) When only the key is given to "enc", the IV is undefined. Print out
 9292      an error message in this case.
 9293      [Lutz Jaenicke]
 9294 
 9295   *) Handle special case when X509_NAME is empty in X509 printing routines.
 9296      [Steve Henson]
 9297 
 9298   *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
 9299      positive and less than q.
 9300      [Bodo Moeller]
 9301 
 9302   *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
 9303      used: it isn't thread safe and the add_lock_callback should handle
 9304      that itself.
 9305      [Paul Rose <Paul.Rose@bridge.com>]
 9306 
 9307   *) Verify that incoming data obeys the block size in
 9308      ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
 9309      [Bodo Moeller]
 9310 
 9311   *) Fix OAEP check.
 9312      [Ulf Möller, Bodo Möller]
 9313 
 9314   *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
 9315      RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
 9316      when fixing the server behaviour for backwards-compatible 'client
 9317      hello' messages.  (Note that the attack is impractical against
 9318      SSL 3.0 and TLS 1.0 anyway because length and version checking
 9319      means that the probability of guessing a valid ciphertext is
 9320      around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
 9321      paper.)
 9322 
 9323      Before 0.9.5, the countermeasure (hide the error by generating a
 9324      random 'decryption result') did not work properly because
 9325      ERR_clear_error() was missing, meaning that SSL_get_error() would
 9326      detect the supposedly ignored error.
 9327 
 9328      Both problems are now fixed.
 9329      [Bodo Moeller]
 9330 
 9331   *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
 9332      (previously it was 1024).
 9333      [Bodo Moeller]
 9334 
 9335   *) Fix for compatibility mode trust settings: ignore trust settings
 9336      unless some valid trust or reject settings are present.
 9337      [Steve Henson]
 9338 
 9339   *) Fix for blowfish EVP: its a variable length cipher.
 9340      [Steve Henson]
 9341 
 9342   *) Fix various bugs related to DSA S/MIME verification. Handle missing
 9343      parameters in DSA public key structures and return an error in the
 9344      DSA routines if parameters are absent.
 9345      [Steve Henson]
 9346 
 9347   *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
 9348      in the current directory if neither $RANDFILE nor $HOME was set.
 9349      RAND_file_name() in 0.9.6a returned NULL in this case.  This has
 9350      caused some confusion to Windows users who haven't defined $HOME.
 9351      Thus RAND_file_name() is changed again: e_os.h can define a
 9352      DEFAULT_HOME, which will be used if $HOME is not set.
 9353      For Windows, we use "C:"; on other platforms, we still require
 9354      environment variables.
 9355 
 9356   *) Move 'if (!initialized) RAND_poll()' into regions protected by
 9357      CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
 9358      having multiple threads call RAND_poll() concurrently.
 9359      [Bodo Moeller]
 9360 
 9361   *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
 9362      combination of a flag and a thread ID variable.
 9363      Otherwise while one thread is in ssleay_rand_bytes (which sets the
 9364      flag), *other* threads can enter ssleay_add_bytes without obeying
 9365      the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
 9366      that they do not hold after the first thread unsets add_do_not_lock).
 9367      [Bodo Moeller]
 9368 
 9369   *) Change bctest again: '-x' expressions are not available in all
 9370      versions of 'test'.
 9371      [Bodo Moeller]
 9372 
 9373  Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
 9374 
 9375   *) Fix a couple of memory leaks in PKCS7_dataDecode()
 9376      [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
 9377 
 9378   *) Change Configure and Makefiles to provide EXE_EXT, which will contain
 9379      the default extension for executables, if any.  Also, make the perl
 9380      scripts that use symlink() to test if it really exists and use "cp"
 9381      if it doesn't.  All this made OpenSSL compilable and installable in
 9382      CygWin.
 9383      [Richard Levitte]
 9384 
 9385   *) Fix for asn1_GetSequence() for indefinite length constructed data.
 9386      If SEQUENCE is length is indefinite just set c->slen to the total
 9387      amount of data available.
 9388      [Steve Henson, reported by shige@FreeBSD.org]
 9389      [This change does not apply to 0.9.7.]
 9390 
 9391   *) Change bctest to avoid here-documents inside command substitution
 9392      (workaround for FreeBSD /bin/sh bug).
 9393      For compatibility with Ultrix, avoid shell functions (introduced
 9394      in the bctest version that searches along $PATH).
 9395      [Bodo Moeller]
 9396 
 9397   *) Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
 9398      with des_encrypt() defined on some operating systems, like Solaris
 9399      and UnixWare.
 9400      [Richard Levitte]
 9401 
 9402   *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
 9403      On the Importance of Eliminating Errors in Cryptographic
 9404      Computations, J. Cryptology 14 (2001) 2, 101-119,
 9405      http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
 9406      [Ulf Moeller]
 9407 
 9408   *) MIPS assembler BIGNUM division bug fix.
 9409      [Andy Polyakov]
 9410 
 9411   *) Disabled incorrect Alpha assembler code.
 9412      [Richard Levitte]
 9413 
 9414   *) Fix PKCS#7 decode routines so they correctly update the length
 9415      after reading an EOC for the EXPLICIT tag.
 9416      [Steve Henson]
 9417      [This change does not apply to 0.9.7.]
 9418 
 9419   *) Fix bug in PKCS#12 key generation routines. This was triggered
 9420      if a 3DES key was generated with a 0 initial byte. Include
 9421      PKCS12_BROKEN_KEYGEN compilation option to retain the old
 9422      (but broken) behaviour.
 9423      [Steve Henson]
 9424 
 9425   *) Enhance bctest to search for a working bc along $PATH and print
 9426      it when found.
 9427      [Tim Rice <tim@multitalents.net> via Richard Levitte]
 9428 
 9429   *) Fix memory leaks in err.c: free err_data string if necessary;
 9430      don't write to the wrong index in ERR_set_error_data.
 9431      [Bodo Moeller]
 9432 
 9433   *) Implement ssl23_peek (analogous to ssl23_read), which previously
 9434      did not exist.
 9435      [Bodo Moeller]
 9436 
 9437   *) Replace rdtsc with _emit statements for VC++ version 5.
 9438      [Jeremy Cooper <jeremy@baymoo.org>]
 9439 
 9440   *) Make it possible to reuse SSLv2 sessions.
 9441      [Richard Levitte]
 9442 
 9443   *) In copy_email() check for >= 0 as a return value for
 9444      X509_NAME_get_index_by_NID() since 0 is a valid index.
 9445      [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
 9446 
 9447   *) Avoid coredump with unsupported or invalid public keys by checking if
 9448      X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
 9449      PKCS7_verify() fails with non detached data.
 9450      [Steve Henson]
 9451 
 9452   *) Don't use getenv in library functions when run as setuid/setgid.
 9453      New function OPENSSL_issetugid().
 9454      [Ulf Moeller]
 9455 
 9456   *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
 9457      due to incorrect handling of multi-threading:
 9458 
 9459      1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
 9460 
 9461      2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
 9462 
 9463      3. Count how many times MemCheck_off() has been called so that
 9464         nested use can be treated correctly.  This also avoids
 9465         inband-signalling in the previous code (which relied on the
 9466         assumption that thread ID 0 is impossible).
 9467      [Bodo Moeller]
 9468 
 9469   *) Add "-rand" option also to s_client and s_server.
 9470      [Lutz Jaenicke]
 9471 
 9472   *) Fix CPU detection on Irix 6.x.
 9473      [Kurt Hockenbury <khockenb@stevens-tech.edu> and
 9474       "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
 9475 
 9476   *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
 9477      was empty.
 9478      [Steve Henson]
 9479      [This change does not apply to 0.9.7.]
 9480 
 9481   *) Use the cached encoding of an X509_NAME structure rather than
 9482      copying it. This is apparently the reason for the libsafe "errors"
 9483      but the code is actually correct.
 9484      [Steve Henson]
 9485 
 9486   *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
 9487      Bleichenbacher's DSA attack.
 9488      Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
 9489      to be set and top=0 forces the highest bit to be set; top=-1 is new
 9490      and leaves the highest bit random.
 9491      [Ulf Moeller, Bodo Moeller]
 9492 
 9493   *) In the NCONF_...-based implementations for CONF_... queries
 9494      (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
 9495      a temporary CONF structure with the data component set to NULL
 9496      (which gives segmentation faults in lh_retrieve).
 9497      Instead, use NULL for the CONF pointer in CONF_get_string and
 9498      CONF_get_number (which may use environment variables) and directly
 9499      return NULL from CONF_get_section.
 9500      [Bodo Moeller]
 9501 
 9502   *) Fix potential buffer overrun for EBCDIC.
 9503      [Ulf Moeller]
 9504 
 9505   *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
 9506      keyUsage if basicConstraints absent for a CA.
 9507      [Steve Henson]
 9508 
 9509   *) Make SMIME_write_PKCS7() write mail header values with a format that
 9510      is more generally accepted (no spaces before the semicolon), since
 9511      some programs can't parse those values properly otherwise.  Also make
 9512      sure BIO's that break lines after each write do not create invalid
 9513      headers.
 9514      [Richard Levitte]
 9515 
 9516   *) Make the CRL encoding routines work with empty SEQUENCE OF. The
 9517      macros previously used would not encode an empty SEQUENCE OF
 9518      and break the signature.
 9519      [Steve Henson]
 9520      [This change does not apply to 0.9.7.]
 9521 
 9522   *) Zero the premaster secret after deriving the master secret in
 9523      DH ciphersuites.
 9524      [Steve Henson]
 9525 
 9526   *) Add some EVP_add_digest_alias registrations (as found in
 9527      OpenSSL_add_all_digests()) to SSL_library_init()
 9528      aka OpenSSL_add_ssl_algorithms().  This provides improved
 9529      compatibility with peers using X.509 certificates
 9530      with unconventional AlgorithmIdentifier OIDs.
 9531      [Bodo Moeller]
 9532 
 9533   *) Fix for Irix with NO_ASM.
 9534      ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
 9535 
 9536   *) ./config script fixes.
 9537      [Ulf Moeller, Richard Levitte]
 9538 
 9539   *) Fix 'openssl passwd -1'.
 9540      [Bodo Moeller]
 9541 
 9542   *) Change PKCS12_key_gen_asc() so it can cope with non null
 9543      terminated strings whose length is passed in the passlen
 9544      parameter, for example from PEM callbacks. This was done
 9545      by adding an extra length parameter to asc2uni().
 9546      [Steve Henson, reported by <oddissey@samsung.co.kr>]
 9547 
 9548   *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
 9549      call failed, free the DSA structure.
 9550      [Bodo Moeller]
 9551 
 9552   *) Fix to uni2asc() to cope with zero length Unicode strings.
 9553      These are present in some PKCS#12 files.
 9554      [Steve Henson]
 9555 
 9556   *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
 9557      Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
 9558      when writing a 32767 byte record.
 9559      [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
 9560 
 9561   *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
 9562      obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
 9563 
 9564      (RSA objects have a reference count access to which is protected
 9565      by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
 9566      so they are meant to be shared between threads.)
 9567      [Bodo Moeller, Geoff Thorpe; original patch submitted by
 9568      "Reddie, Steven" <Steven.Reddie@ca.com>]
 9569 
 9570   *) Fix a deadlock in CRYPTO_mem_leaks().
 9571      [Bodo Moeller]
 9572 
 9573   *) Use better test patterns in bntest.
 9574      [Ulf Möller]
 9575 
 9576   *) rand_win.c fix for Borland C.
 9577      [Ulf Möller]
 9578 
 9579   *) BN_rshift bugfix for n == 0.
 9580      [Bodo Moeller]
 9581 
 9582   *) Add a 'bctest' script that checks for some known 'bc' bugs
 9583      so that 'make test' does not abort just because 'bc' is broken.
 9584      [Bodo Moeller]
 9585 
 9586   *) Store verify_result within SSL_SESSION also for client side to
 9587      avoid potential security hole. (Re-used sessions on the client side
 9588      always resulted in verify_result==X509_V_OK, not using the original
 9589      result of the server certificate verification.)
 9590      [Lutz Jaenicke]
 9591 
 9592   *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
 9593      SSL3_RT_APPLICATION_DATA, return 0.
 9594      Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
 9595      [Bodo Moeller]
 9596 
 9597   *) Fix SSL_peek:
 9598      Both ssl2_peek and ssl3_peek, which were totally broken in earlier
 9599      releases, have been re-implemented by renaming the previous
 9600      implementations of ssl2_read and ssl3_read to ssl2_read_internal
 9601      and ssl3_read_internal, respectively, and adding 'peek' parameters
 9602      to them.  The new ssl[23]_{read,peek} functions are calls to
 9603      ssl[23]_read_internal with the 'peek' flag set appropriately.
 9604      A 'peek' parameter has also been added to ssl3_read_bytes, which
 9605      does the actual work for ssl3_read_internal.
 9606      [Bodo Moeller]
 9607 
 9608   *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
 9609      the method-specific "init()" handler. Also clean up ex_data after
 9610      calling the method-specific "finish()" handler. Previously, this was
 9611      happening the other way round.
 9612      [Geoff Thorpe]
 9613 
 9614   *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
 9615      The previous value, 12, was not always sufficient for BN_mod_exp().
 9616      [Bodo Moeller]
 9617 
 9618   *) Make sure that shared libraries get the internal name engine with
 9619      the full version number and not just 0.  This should mark the
 9620      shared libraries as not backward compatible.  Of course, this should
 9621      be changed again when we can guarantee backward binary compatibility.
 9622      [Richard Levitte]
 9623 
 9624   *) Fix typo in get_cert_by_subject() in by_dir.c
 9625      [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
 9626 
 9627   *) Rework the system to generate shared libraries:
 9628 
 9629      - Make note of the expected extension for the shared libraries and
 9630        if there is a need for symbolic links from for example libcrypto.so.0
 9631        to libcrypto.so.0.9.7.  There is extended info in Configure for
 9632        that.
 9633 
 9634      - Make as few rebuilds of the shared libraries as possible.
 9635 
 9636      - Still avoid linking the OpenSSL programs with the shared libraries.
 9637 
 9638      - When installing, install the shared libraries separately from the
 9639        static ones.
 9640      [Richard Levitte]
 9641 
 9642   *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
 9643 
 9644      Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
 9645      and not in SSL_clear because the latter is also used by the
 9646      accept/connect functions; previously, the settings made by
 9647      SSL_set_read_ahead would be lost during the handshake.
 9648      [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
 9649 
 9650   *) Correct util/mkdef.pl to be selective about disabled algorithms.
 9651      Previously, it would create entries for disabled algorithms no
 9652      matter what.
 9653      [Richard Levitte]
 9654 
 9655   *) Added several new manual pages for SSL_* function.
 9656      [Lutz Jaenicke]
 9657 
 9658  Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
 9659 
 9660   *) In ssl23_get_client_hello, generate an error message when faced
 9661      with an initial SSL 3.0/TLS record that is too small to contain the
 9662      first two bytes of the ClientHello message, i.e. client_version.
 9663      (Note that this is a pathologic case that probably has never happened
 9664      in real life.)  The previous approach was to use the version number
 9665      from the record header as a substitute; but our protocol choice
 9666      should not depend on that one because it is not authenticated
 9667      by the Finished messages.
 9668      [Bodo Moeller]
 9669 
 9670   *) More robust randomness gathering functions for Windows.
 9671      [Jeffrey Altman <jaltman@columbia.edu>]
 9672 
 9673   *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
 9674      not set then we don't setup the error code for issuer check errors
 9675      to avoid possibly overwriting other errors which the callback does
 9676      handle. If an application does set the flag then we assume it knows
 9677      what it is doing and can handle the new informational codes
 9678      appropriately.
 9679      [Steve Henson]
 9680 
 9681   *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
 9682      a general "ANY" type, as such it should be able to decode anything
 9683      including tagged types. However it didn't check the class so it would
 9684      wrongly interpret tagged types in the same way as their universal
 9685      counterpart and unknown types were just rejected. Changed so that the
 9686      tagged and unknown types are handled in the same way as a SEQUENCE:
 9687      that is the encoding is stored intact. There is also a new type
 9688      "V_ASN1_OTHER" which is used when the class is not universal, in this
 9689      case we have no idea what the actual type is so we just lump them all
 9690      together.
 9691      [Steve Henson]
 9692 
 9693   *) On VMS, stdout may very well lead to a file that is written to
 9694      in a record-oriented fashion.  That means that every write() will
 9695      write a separate record, which will be read separately by the
 9696      programs trying to read from it.  This can be very confusing.
 9697 
 9698      The solution is to put a BIO filter in the way that will buffer
 9699      text until a linefeed is reached, and then write everything a
 9700      line at a time, so every record written will be an actual line,
 9701      not chunks of lines and not (usually doesn't happen, but I've
 9702      seen it once) several lines in one record.  BIO_f_linebuffer() is
 9703      the answer.
 9704 
 9705      Currently, it's a VMS-only method, because that's where it has
 9706      been tested well enough.
 9707      [Richard Levitte]
 9708 
 9709   *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
 9710      it can return incorrect results.
 9711      (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
 9712      but it was in 0.9.6-beta[12].)
 9713      [Bodo Moeller]
 9714 
 9715   *) Disable the check for content being present when verifying detached
 9716      signatures in pk7_smime.c. Some versions of Netscape (wrongly)
 9717      include zero length content when signing messages.
 9718      [Steve Henson]
 9719 
 9720   *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
 9721      BIO_ctrl (for BIO pairs).
 9722      [Bodo Möller]
 9723 
 9724   *) Add DSO method for VMS.
 9725      [Richard Levitte]
 9726 
 9727   *) Bug fix: Montgomery multiplication could produce results with the
 9728      wrong sign.
 9729      [Ulf Möller]
 9730 
 9731   *) Add RPM specification openssl.spec and modify it to build three
 9732      packages.  The default package contains applications, application
 9733      documentation and run-time libraries.  The devel package contains
 9734      include files, static libraries and function documentation.  The
 9735      doc package contains the contents of the doc directory.  The original
 9736      openssl.spec was provided by Damien Miller <djm@mindrot.org>.
 9737      [Richard Levitte]
 9738 
 9739   *) Add a large number of documentation files for many SSL routines.
 9740      [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
 9741 
 9742   *) Add a configuration entry for Sony News 4.
 9743      [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
 9744 
 9745   *) Don't set the two most significant bits to one when generating a
 9746      random number < q in the DSA library.
 9747      [Ulf Möller]
 9748 
 9749   *) New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
 9750      behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
 9751      the underlying transport is blocking) if a handshake took place.
 9752      (The default behaviour is needed by applications such as s_client
 9753      and s_server that use select() to determine when to use SSL_read;
 9754      but for applications that know in advance when to expect data, it
 9755      just makes things more complicated.)
 9756      [Bodo Moeller]
 9757 
 9758   *) Add RAND_egd_bytes(), which gives control over the number of bytes read
 9759      from EGD.
 9760      [Ben Laurie]
 9761 
 9762   *) Add a few more EBCDIC conditionals that make `req' and `x509'
 9763      work better on such systems.
 9764      [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
 9765 
 9766   *) Add two demo programs for PKCS12_parse() and PKCS12_create().
 9767      Update PKCS12_parse() so it copies the friendlyName and the
 9768      keyid to the certificates aux info.
 9769      [Steve Henson]
 9770 
 9771   *) Fix bug in PKCS7_verify() which caused an infinite loop
 9772      if there was more than one signature.
 9773      [Sven Uszpelkat <su@celocom.de>]
 9774 
 9775   *) Major change in util/mkdef.pl to include extra information
 9776      about each symbol, as well as presenting variables as well
 9777      as functions.  This change means that there's n more need
 9778      to rebuild the .num files when some algorithms are excluded.
 9779      [Richard Levitte]
 9780 
 9781   *) Allow the verify time to be set by an application,
 9782      rather than always using the current time.
 9783      [Steve Henson]
 9784 
 9785   *) Phase 2 verify code reorganisation. The certificate
 9786      verify code now looks up an issuer certificate by a
 9787      number of criteria: subject name, authority key id
 9788      and key usage. It also verifies self signed certificates
 9789      by the same criteria. The main comparison function is
 9790      X509_check_issued() which performs these checks.
 9791 
 9792      Lot of changes were necessary in order to support this
 9793      without completely rewriting the lookup code.
 9794 
 9795      Authority and subject key identifier are now cached.
 9796 
 9797      The LHASH 'certs' is X509_STORE has now been replaced
 9798      by a STACK_OF(X509_OBJECT). This is mainly because an
 9799      LHASH can't store or retrieve multiple objects with
 9800      the same hash value.
 9801 
 9802      As a result various functions (which were all internal
 9803      use only) have changed to handle the new X509_STORE
 9804      structure. This will break anything that messed round
 9805      with X509_STORE internally.
 9806 
 9807      The functions X509_STORE_add_cert() now checks for an
 9808      exact match, rather than just subject name.
 9809 
 9810      The X509_STORE API doesn't directly support the retrieval
 9811      of multiple certificates matching a given criteria, however
 9812      this can be worked round by performing a lookup first
 9813      (which will fill the cache with candidate certificates)
 9814      and then examining the cache for matches. This is probably
 9815      the best we can do without throwing out X509_LOOKUP
 9816      entirely (maybe later...).
 9817 
 9818      The X509_VERIFY_CTX structure has been enhanced considerably.
 9819 
 9820      All certificate lookup operations now go via a get_issuer()
 9821      callback. Although this currently uses an X509_STORE it
 9822      can be replaced by custom lookups. This is a simple way
 9823      to bypass the X509_STORE hackery necessary to make this
 9824      work and makes it possible to use more efficient techniques
 9825      in future. A very simple version which uses a simple
 9826      STACK for its trusted certificate store is also provided
 9827      using X509_STORE_CTX_trusted_stack().
 9828 
 9829      The verify_cb() and verify() callbacks now have equivalents
 9830      in the X509_STORE_CTX structure.
 9831 
 9832      X509_STORE_CTX also has a 'flags' field which can be used
 9833      to customise the verify behaviour.
 9834      [Steve Henson]
 9835 
 9836   *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
 9837      excludes S/MIME capabilities.
 9838      [Steve Henson]
 9839 
 9840   *) When a certificate request is read in keep a copy of the
 9841      original encoding of the signed data and use it when outputting
 9842      again. Signatures then use the original encoding rather than
 9843      a decoded, encoded version which may cause problems if the
 9844      request is improperly encoded.
 9845      [Steve Henson]
 9846 
 9847   *) For consistency with other BIO_puts implementations, call
 9848      buffer_write(b, ...) directly in buffer_puts instead of calling
 9849      BIO_write(b, ...).
 9850 
 9851      In BIO_puts, increment b->num_write as in BIO_write.
 9852      [Peter.Sylvester@EdelWeb.fr]
 9853 
 9854   *) Fix BN_mul_word for the case where the word is 0. (We have to use
 9855      BN_zero, we may not return a BIGNUM with an array consisting of
 9856      words set to zero.)
 9857      [Bodo Moeller]
 9858 
 9859   *) Avoid calling abort() from within the library when problems are
 9860      detected, except if preprocessor symbols have been defined
 9861      (such as REF_CHECK, BN_DEBUG etc.).
 9862      [Bodo Moeller]
 9863 
 9864   *) New openssl application 'rsautl'. This utility can be
 9865      used for low level RSA operations. DER public key
 9866      BIO/fp routines also added.
 9867      [Steve Henson]
 9868 
 9869   *) New Configure entry and patches for compiling on QNX 4.
 9870      [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
 9871 
 9872   *) A demo state-machine implementation was sponsored by
 9873      Nuron (http://www.nuron.com/) and is now available in
 9874      demos/state_machine.
 9875      [Ben Laurie]
 9876 
 9877   *) New options added to the 'dgst' utility for signature
 9878      generation and verification.
 9879      [Steve Henson]
 9880 
 9881   *) Unrecognized PKCS#7 content types are now handled via a
 9882      catch all ASN1_TYPE structure. This allows unsupported
 9883      types to be stored as a "blob" and an application can
 9884      encode and decode it manually.
 9885      [Steve Henson]
 9886 
 9887   *) Fix various signed/unsigned issues to make a_strex.c
 9888      compile under VC++.
 9889      [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
 9890 
 9891   *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
 9892      length if passed a buffer. ASN1_INTEGER_to_BN failed
 9893      if passed a NULL BN and its argument was negative.
 9894      [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
 9895 
 9896   *) Modification to PKCS#7 encoding routines to output definite
 9897      length encoding. Since currently the whole structures are in
 9898      memory there's not real point in using indefinite length
 9899      constructed encoding. However if OpenSSL is compiled with
 9900      the flag PKCS7_INDEFINITE_ENCODING the old form is used.
 9901      [Steve Henson]
 9902 
 9903   *) Added BIO_vprintf() and BIO_vsnprintf().
 9904      [Richard Levitte]
 9905 
 9906   *) Added more prefixes to parse for in the strings written
 9907      through a logging bio, to cover all the levels that are available
 9908      through syslog.  The prefixes are now:
 9909 
 9910         PANIC, EMERG, EMR       =>      LOG_EMERG
 9911         ALERT, ALR              =>      LOG_ALERT
 9912         CRIT, CRI               =>      LOG_CRIT
 9913         ERROR, ERR              =>      LOG_ERR
 9914         WARNING, WARN, WAR      =>      LOG_WARNING
 9915         NOTICE, NOTE, NOT       =>      LOG_NOTICE
 9916         INFO, INF               =>      LOG_INFO
 9917         DEBUG, DBG              =>      LOG_DEBUG
 9918 
 9919      and as before, if none of those prefixes are present at the
 9920      beginning of the string, LOG_ERR is chosen.
 9921 
 9922      On Win32, the LOG_* levels are mapped according to this:
 9923 
 9924         LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
 9925         LOG_WARNING                             => EVENTLOG_WARNING_TYPE
 9926         LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE
 9927 
 9928      [Richard Levitte]
 9929 
 9930   *) Made it possible to reconfigure with just the configuration
 9931      argument "reconf" or "reconfigure".  The command line arguments
 9932      are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
 9933      and are retrieved from there when reconfiguring.
 9934      [Richard Levitte]
 9935 
 9936   *) MD4 implemented.
 9937      [Assar Westerlund <assar@sics.se>, Richard Levitte]
 9938 
 9939   *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
 9940      [Richard Levitte]
 9941 
 9942   *) The obj_dat.pl script was messing up the sorting of object
 9943      names. The reason was that it compared the quoted version
 9944      of strings as a result "OCSP" > "OCSP Signing" because
 9945      " > SPACE. Changed script to store unquoted versions of
 9946      names and add quotes on output. It was also omitting some
 9947      names from the lookup table if they were given a default
 9948      value (that is if SN is missing it is given the same
 9949      value as LN and vice versa), these are now added on the
 9950      grounds that if an object has a name we should be able to
 9951      look it up. Finally added warning output when duplicate
 9952      short or long names are found.
 9953      [Steve Henson]
 9954 
 9955   *) Changes needed for Tandem NSK.
 9956      [Scott Uroff <scott@xypro.com>]
 9957 
 9958   *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
 9959      RSA_padding_check_SSLv23(), special padding was never detected
 9960      and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
 9961      version rollback attacks was not effective.
 9962 
 9963      In s23_clnt.c, don't use special rollback-attack detection padding
 9964      (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
 9965      client; similarly, in s23_srvr.c, don't do the rollback check if
 9966      SSL 2.0 is the only protocol enabled in the server.
 9967      [Bodo Moeller]
 9968 
 9969   *) Make it possible to get hexdumps of unprintable data with 'openssl
 9970      asn1parse'.  By implication, the functions ASN1_parse_dump() and
 9971      BIO_dump_indent() are added.
 9972      [Richard Levitte]
 9973 
 9974   *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
 9975      these print out strings and name structures based on various
 9976      flags including RFC2253 support and proper handling of
 9977      multibyte characters. Added options to the 'x509' utility
 9978      to allow the various flags to be set.
 9979      [Steve Henson]
 9980 
 9981   *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
 9982      Also change the functions X509_cmp_current_time() and
 9983      X509_gmtime_adj() work with an ASN1_TIME structure,
 9984      this will enable certificates using GeneralizedTime in validity
 9985      dates to be checked.
 9986      [Steve Henson]
 9987 
 9988   *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
 9989      negative public key encodings) on by default,
 9990      NO_NEG_PUBKEY_BUG can be set to disable it.
 9991      [Steve Henson]
 9992 
 9993   *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
 9994      content octets. An i2c_ASN1_OBJECT is unnecessary because
 9995      the encoding can be trivially obtained from the structure.
 9996      [Steve Henson]
 9997 
 9998   *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
 9999      not read locks (CRYPTO_r_[un]lock).
10000      [Bodo Moeller]
10001 
10002   *) A first attempt at creating official support for shared
10003      libraries through configuration.  I've kept it so the
10004      default is static libraries only, and the OpenSSL programs
10005      are always statically linked for now, but there are
10006      preparations for dynamic linking in place.
10007      This has been tested on Linux and Tru64.
10008      [Richard Levitte]
10009 
10010   *) Randomness polling function for Win9x, as described in:
10011      Peter Gutmann, Software Generation of Practically Strong
10012      Random Numbers.
10013      [Ulf Möller]
10014 
10015   *) Fix so PRNG is seeded in req if using an already existing
10016      DSA key.
10017      [Steve Henson]
10018 
10019   *) New options to smime application. -inform and -outform
10020      allow alternative formats for the S/MIME message including
10021      PEM and DER. The -content option allows the content to be
10022      specified separately. This should allow things like Netscape
10023      form signing output easier to verify.
10024      [Steve Henson]
10025 
10026   *) Fix the ASN1 encoding of tags using the 'long form'.
10027      [Steve Henson]
10028 
10029   *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
10030      STRING types. These convert content octets to and from the
10031      underlying type. The actual tag and length octets are
10032      already assumed to have been read in and checked. These
10033      are needed because all other string types have virtually
10034      identical handling apart from the tag. By having versions
10035      of the ASN1 functions that just operate on content octets
10036      IMPLICIT tagging can be handled properly. It also allows
10037      the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
10038      and ASN1_INTEGER are identical apart from the tag.
10039      [Steve Henson]
10040 
10041   *) Change the handling of OID objects as follows:
10042 
10043      - New object identifiers are inserted in objects.txt, following
10044        the syntax given in objects.README.
10045      - objects.pl is used to process obj_mac.num and create a new
10046        obj_mac.h.
10047      - obj_dat.pl is used to create a new obj_dat.h, using the data in
10048        obj_mac.h.
10049 
10050      This is currently kind of a hack, and the perl code in objects.pl
10051      isn't very elegant, but it works as I intended.  The simplest way
10052      to check that it worked correctly is to look in obj_dat.h and
10053      check the array nid_objs and make sure the objects haven't moved
10054      around (this is important!).  Additions are OK, as well as
10055      consistent name changes.
10056      [Richard Levitte]
10057 
10058   *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
10059      [Bodo Moeller]
10060 
10061   *) Addition of the command line parameter '-rand file' to 'openssl req'.
10062      The given file adds to whatever has already been seeded into the
10063      random pool through the RANDFILE configuration file option or
10064      environment variable, or the default random state file.
10065      [Richard Levitte]
10066 
10067   *) mkstack.pl now sorts each macro group into lexical order.
10068      Previously the output order depended on the order the files
10069      appeared in the directory, resulting in needless rewriting
10070      of safestack.h .
10071      [Steve Henson]
10072 
10073   *) Patches to make OpenSSL compile under Win32 again. Mostly
10074      work arounds for the VC++ problem that it treats func() as
10075      func(void). Also stripped out the parts of mkdef.pl that
10076      added extra typesafe functions: these no longer exist.
10077      [Steve Henson]
10078 
10079   *) Reorganisation of the stack code. The macros are now all
10080      collected in safestack.h . Each macro is defined in terms of
10081      a "stack macro" of the form SKM_<name>(type, a, b). The
10082      DEBUG_SAFESTACK is now handled in terms of function casts,
10083      this has the advantage of retaining type safety without the
10084      use of additional functions. If DEBUG_SAFESTACK is not defined
10085      then the non typesafe macros are used instead. Also modified the
10086      mkstack.pl script to handle the new form. Needs testing to see
10087      if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
10088      the default if no major problems. Similar behaviour for ASN1_SET_OF
10089      and PKCS12_STACK_OF.
10090      [Steve Henson]
10091 
10092   *) When some versions of IIS use the 'NET' form of private key the
10093      key derivation algorithm is different. Normally MD5(password) is
10094      used as a 128 bit RC4 key. In the modified case
10095      MD5(MD5(password) + "SGCKEYSALT")  is used instead. Added some
10096      new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
10097      as the old Netscape_RSA functions except they have an additional
10098      'sgckey' parameter which uses the modified algorithm. Also added
10099      an -sgckey command line option to the rsa utility. Thanks to
10100      Adrian Peck <bertie@ncipher.com> for posting details of the modified
10101      algorithm to openssl-dev.
10102      [Steve Henson]
10103 
10104   *) The evp_local.h macros were using 'c.##kname' which resulted in
10105      invalid expansion on some systems (SCO 5.0.5 for example).
10106      Corrected to 'c.kname'.
10107      [Phillip Porch <root@theporch.com>]
10108 
10109   *) New X509_get1_email() and X509_REQ_get1_email() functions that return
10110      a STACK of email addresses from a certificate or request, these look
10111      in the subject name and the subject alternative name extensions and
10112      omit any duplicate addresses.
10113      [Steve Henson]
10114 
10115   *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
10116      This makes DSA verification about 2 % faster.
10117      [Bodo Moeller]
10118 
10119   *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
10120      (meaning that now 2^5 values will be precomputed, which is only 4 KB
10121      plus overhead for 1024 bit moduli).
10122      This makes exponentiations about 0.5 % faster for 1024 bit
10123      exponents (as measured by "openssl speed rsa2048").
10124      [Bodo Moeller]
10125 
10126   *) Rename memory handling macros to avoid conflicts with other
10127      software:
10128           Malloc         =>  OPENSSL_malloc
10129           Malloc_locked  =>  OPENSSL_malloc_locked
10130           Realloc        =>  OPENSSL_realloc
10131           Free           =>  OPENSSL_free
10132      [Richard Levitte]
10133 
10134   *) New function BN_mod_exp_mont_word for small bases (roughly 15%
10135      faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
10136      [Bodo Moeller]
10137 
10138   *) CygWin32 support.
10139      [John Jarvie <jjarvie@newsguy.com>]
10140 
10141   *) The type-safe stack code has been rejigged. It is now only compiled
10142      in when OpenSSL is configured with the DEBUG_SAFESTACK option and
10143      by default all type-specific stack functions are "#define"d back to
10144      standard stack functions. This results in more streamlined output
10145      but retains the type-safety checking possibilities of the original
10146      approach.
10147      [Geoff Thorpe]
10148 
10149   *) The STACK code has been cleaned up, and certain type declarations
10150      that didn't make a lot of sense have been brought in line. This has
10151      also involved a cleanup of sorts in safestack.h to more correctly
10152      map type-safe stack functions onto their plain stack counterparts.
10153      This work has also resulted in a variety of "const"ifications of
10154      lots of the code, especially "_cmp" operations which should normally
10155      be prototyped with "const" parameters anyway.
10156      [Geoff Thorpe]
10157 
10158   *) When generating bytes for the first time in md_rand.c, 'stir the pool'
10159      by seeding with STATE_SIZE dummy bytes (with zero entropy count).
10160      (The PRNG state consists of two parts, the large pool 'state' and 'md',
10161      where all of 'md' is used each time the PRNG is used, but 'state'
10162      is used only indexed by a cyclic counter. As entropy may not be
10163      well distributed from the beginning, 'md' is important as a
10164      chaining variable. However, the output function chains only half
10165      of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
10166      all of 'md', and seeding with STATE_SIZE dummy bytes will result
10167      in all of 'state' being rewritten, with the new values depending
10168      on virtually all of 'md'.  This overcomes the 80 bit limitation.)
10169      [Bodo Moeller]
10170 
10171   *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
10172      the handshake is continued after ssl_verify_cert_chain();
10173      otherwise, if SSL_VERIFY_NONE is set, remaining error codes
10174      can lead to 'unexplainable' connection aborts later.
10175      [Bodo Moeller; problem tracked down by Lutz Jaenicke]
10176 
10177   *) Major EVP API cipher revision.
10178      Add hooks for extra EVP features. This allows various cipher
10179      parameters to be set in the EVP interface. Support added for variable
10180      key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
10181      setting of RC2 and RC5 parameters.
10182 
10183      Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
10184      ciphers.
10185 
10186      Remove lots of duplicated code from the EVP library. For example *every*
10187      cipher init() function handles the 'iv' in the same way according to the
10188      cipher mode. They also all do nothing if the 'key' parameter is NULL and
10189      for CFB and OFB modes they zero ctx->num.
10190 
10191      New functionality allows removal of S/MIME code RC2 hack.
10192 
10193      Most of the routines have the same form and so can be declared in terms
10194      of macros.
10195 
10196      By shifting this to the top level EVP_CipherInit() it can be removed from
10197      all individual ciphers. If the cipher wants to handle IVs or keys
10198      differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
10199      flags.
10200 
10201      Change lots of functions like EVP_EncryptUpdate() to now return a
10202      value: although software versions of the algorithms cannot fail
10203      any installed hardware versions can.
10204      [Steve Henson]
10205 
10206   *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
10207      this option is set, tolerate broken clients that send the negotiated
10208      protocol version number instead of the requested protocol version
10209      number.
10210      [Bodo Moeller]
10211 
10212   *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
10213      i.e. non-zero for export ciphersuites, zero otherwise.
10214      Previous versions had this flag inverted, inconsistent with
10215      rsa_tmp_cb (..._TMP_RSA_CB).
10216      [Bodo Moeller; problem reported by Amit Chopra]
10217 
10218   *) Add missing DSA library text string. Work around for some IIS
10219      key files with invalid SEQUENCE encoding.
10220      [Steve Henson]
10221 
10222   *) Add a document (doc/standards.txt) that list all kinds of standards
10223      and so on that are implemented in OpenSSL.
10224      [Richard Levitte]
10225 
10226   *) Enhance c_rehash script. Old version would mishandle certificates
10227      with the same subject name hash and wouldn't handle CRLs at all.
10228      Added -fingerprint option to crl utility, to support new c_rehash
10229      features.
10230      [Steve Henson]
10231 
10232   *) Eliminate non-ANSI declarations in crypto.h and stack.h.
10233      [Ulf Möller]
10234 
10235   *) Fix for SSL server purpose checking. Server checking was
10236      rejecting certificates which had extended key usage present
10237      but no ssl client purpose.
10238      [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
10239 
10240   *) Make PKCS#12 code work with no password. The PKCS#12 spec
10241      is a little unclear about how a blank password is handled.
10242      Since the password in encoded as a BMPString with terminating
10243      double NULL a zero length password would end up as just the
10244      double NULL. However no password at all is different and is
10245      handled differently in the PKCS#12 key generation code. NS
10246      treats a blank password as zero length. MSIE treats it as no
10247      password on export: but it will try both on import. We now do
10248      the same: PKCS12_parse() tries zero length and no password if
10249      the password is set to "" or NULL (NULL is now a valid password:
10250      it wasn't before) as does the pkcs12 application.
10251      [Steve Henson]
10252 
10253   *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
10254      perror when PEM_read_bio_X509_REQ fails, the error message must
10255      be obtained from the error queue.
10256      [Bodo Moeller]
10257 
10258   *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
10259      it in ERR_remove_state if appropriate, and change ERR_get_state
10260      accordingly to avoid race conditions (this is necessary because
10261      thread_hash is no longer constant once set).
10262      [Bodo Moeller]
10263 
10264   *) Bugfix for linux-elf makefile.one.
10265      [Ulf Möller]
10266 
10267   *) RSA_get_default_method() will now cause a default
10268      RSA_METHOD to be chosen if one doesn't exist already.
10269      Previously this was only set during a call to RSA_new()
10270      or RSA_new_method(NULL) meaning it was possible for
10271      RSA_get_default_method() to return NULL.
10272      [Geoff Thorpe]
10273 
10274   *) Added native name translation to the existing DSO code
10275      that will convert (if the flag to do so is set) filenames
10276      that are sufficiently small and have no path information
10277      into a canonical native form. Eg. "blah" converted to
10278      "libblah.so" or "blah.dll" etc.
10279      [Geoff Thorpe]
10280 
10281   *) New function ERR_error_string_n(e, buf, len) which is like
10282      ERR_error_string(e, buf), but writes at most 'len' bytes
10283      including the 0 terminator.  For ERR_error_string_n, 'buf'
10284      may not be NULL.
10285      [Damien Miller <djm@mindrot.org>, Bodo Moeller]
10286 
10287   *) CONF library reworked to become more general.  A new CONF
10288      configuration file reader "class" is implemented as well as a
10289      new functions (NCONF_*, for "New CONF") to handle it.  The now
10290      old CONF_* functions are still there, but are reimplemented to
10291      work in terms of the new functions.  Also, a set of functions
10292      to handle the internal storage of the configuration data is
10293      provided to make it easier to write new configuration file
10294      reader "classes" (I can definitely see something reading a
10295      configuration file in XML format, for example), called _CONF_*,
10296      or "the configuration storage API"...
10297 
10298      The new configuration file reading functions are:
10299 
10300         NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
10301         NCONF_get_section, NCONF_get_string, NCONF_get_numbre
10302 
10303         NCONF_default, NCONF_WIN32
10304 
10305         NCONF_dump_fp, NCONF_dump_bio
10306 
10307      NCONF_default and NCONF_WIN32 are method (or "class") choosers,
10308      NCONF_new creates a new CONF object.  This works in the same way
10309      as other interfaces in OpenSSL, like the BIO interface.
10310      NCONF_dump_* dump the internal storage of the configuration file,
10311      which is useful for debugging.  All other functions take the same
10312      arguments as the old CONF_* functions wth the exception of the
10313      first that must be a `CONF *' instead of a `LHASH *'.
10314 
10315      To make it easer to use the new classes with the old CONF_* functions,
10316      the function CONF_set_default_method is provided.
10317      [Richard Levitte]
10318 
10319   *) Add '-tls1' option to 'openssl ciphers', which was already
10320      mentioned in the documentation but had not been implemented.
10321      (This option is not yet really useful because even the additional
10322      experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
10323      [Bodo Moeller]
10324 
10325   *) Initial DSO code added into libcrypto for letting OpenSSL (and
10326      OpenSSL-based applications) load shared libraries and bind to
10327      them in a portable way.
10328      [Geoff Thorpe, with contributions from Richard Levitte]
10329 
10330  Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
10331 
10332   *) Make sure _lrotl and _lrotr are only used with MSVC.
10333 
10334   *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
10335      (the default implementation of RAND_status).
10336 
10337   *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
10338      to '-clrext' (= clear extensions), as intended and documented.
10339      [Bodo Moeller; inconsistency pointed out by Michael Attili
10340      <attili@amaxo.com>]
10341 
10342   *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
10343      was larger than the MD block size.
10344      [Steve Henson, pointed out by Yost William <YostW@tce.com>]
10345 
10346   *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
10347      fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
10348      using the passed key: if the passed key was a private key the result
10349      of X509_print(), for example, would be to print out all the private key
10350      components.
10351      [Steve Henson]
10352 
10353   *) des_quad_cksum() byte order bug fix.
10354      [Ulf Möller, using the problem description in krb4-0.9.7, where
10355       the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
10356 
10357   *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
10358      discouraged.
10359      [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
10360 
10361   *) For easily testing in shell scripts whether some command
10362      'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
10363      returns with exit code 0 iff no command of the given name is available.
10364      'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
10365      the output goes to stdout and nothing is printed to stderr.
10366      Additional arguments are always ignored.
10367 
10368      Since for each cipher there is a command of the same name,
10369      the 'no-cipher' compilation switches can be tested this way.
10370 
10371      ('openssl no-XXX' is not able to detect pseudo-commands such
10372      as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
10373      [Bodo Moeller]
10374 
10375   *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
10376      [Bodo Moeller]
10377 
10378   *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
10379      is set; it will be thrown away anyway because each handshake creates
10380      its own key.
10381      ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
10382      to parameters -- in previous versions (since OpenSSL 0.9.3) the
10383      'default key' from SSL_CTX_set_tmp_dh would always be lost, meaning
10384      you effectively got SSL_OP_SINGLE_DH_USE when using this macro.
10385      [Bodo Moeller]
10386 
10387   *) New s_client option -ign_eof: EOF at stdin is ignored, and
10388      'Q' and 'R' lose their special meanings (quit/renegotiate).
10389      This is part of what -quiet does; unlike -quiet, -ign_eof
10390      does not suppress any output.
10391      [Richard Levitte]
10392 
10393   *) Add compatibility options to the purpose and trust code. The
10394      purpose X509_PURPOSE_ANY is "any purpose" which automatically
10395      accepts a certificate or CA, this was the previous behaviour,
10396      with all the associated security issues.
10397 
10398      X509_TRUST_COMPAT is the old trust behaviour: only and
10399      automatically trust self signed roots in certificate store. A
10400      new trust setting X509_TRUST_DEFAULT is used to specify that
10401      a purpose has no associated trust setting and it should instead
10402      use the value in the default purpose.
10403      [Steve Henson]
10404 
10405   *) Fix the PKCS#8 DSA private key code so it decodes keys again
10406      and fix a memory leak.
10407      [Steve Henson]
10408 
10409   *) In util/mkerr.pl (which implements 'make errors'), preserve
10410      reason strings from the previous version of the .c file, as
10411      the default to have only downcase letters (and digits) in
10412      automatically generated reasons codes is not always appropriate.
10413      [Bodo Moeller]
10414 
10415   *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
10416      using strerror.  Previously, ERR_reason_error_string() returned
10417      library names as reason strings for SYSerr; but SYSerr is a special
10418      case where small numbers are errno values, not library numbers.
10419      [Bodo Moeller]
10420 
10421   *) Add '-dsaparam' option to 'openssl dhparam' application.  This
10422      converts DSA parameters into DH parameters. (When creating parameters,
10423      DSA_generate_parameters is used.)
10424      [Bodo Moeller]
10425 
10426   *) Include 'length' (recommended exponent length) in C code generated
10427      by 'openssl dhparam -C'.
10428      [Bodo Moeller]
10429 
10430   *) The second argument to set_label in perlasm was already being used
10431      so couldn't be used as a "file scope" flag. Moved to third argument
10432      which was free.
10433      [Steve Henson]
10434 
10435   *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
10436      instead of RAND_bytes for encryption IVs and salts.
10437      [Bodo Moeller]
10438 
10439   *) Include RAND_status() into RAND_METHOD instead of implementing
10440      it only for md_rand.c  Otherwise replacing the PRNG by calling
10441      RAND_set_rand_method would be impossible.
10442      [Bodo Moeller]
10443 
10444   *) Don't let DSA_generate_key() enter an infinite loop if the random
10445      number generation fails.
10446      [Bodo Moeller]
10447 
10448   *) New 'rand' application for creating pseudo-random output.
10449      [Bodo Moeller]
10450 
10451   *) Added configuration support for Linux/IA64
10452      [Rolf Haberrecker <rolf@suse.de>]
10453 
10454   *) Assembler module support for Mingw32.
10455      [Ulf Möller]
10456 
10457   *) Shared library support for HPUX (in shlib/).
10458      [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
10459 
10460   *) Shared library support for Solaris gcc.
10461      [Lutz Behnke <behnke@trustcenter.de>]
10462 
10463  Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
10464 
10465   *) PKCS7_encrypt() was adding text MIME headers twice because they
10466      were added manually and by SMIME_crlf_copy().
10467      [Steve Henson]
10468 
10469   *) In bntest.c don't call BN_rand with zero bits argument.
10470      [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
10471 
10472   *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
10473      case was implemented. This caused BN_div_recp() to fail occasionally.
10474      [Ulf Möller]
10475 
10476   *) Add an optional second argument to the set_label() in the perl
10477      assembly language builder. If this argument exists and is set
10478      to 1 it signals that the assembler should use a symbol whose
10479      scope is the entire file, not just the current function. This
10480      is needed with MASM which uses the format label:: for this scope.
10481      [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
10482 
10483   *) Change the ASN1 types so they are typedefs by default. Before
10484      almost all types were #define'd to ASN1_STRING which was causing
10485      STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
10486      for example.
10487      [Steve Henson]
10488 
10489   *) Change names of new functions to the new get1/get0 naming
10490      convention: After 'get1', the caller owns a reference count
10491      and has to call ..._free; 'get0' returns a pointer to some
10492      data structure without incrementing reference counters.
10493      (Some of the existing 'get' functions increment a reference
10494      counter, some don't.)
10495      Similarly, 'set1' and 'add1' functions increase reference
10496      counters or duplicate objects.
10497      [Steve Henson]
10498 
10499   *) Allow for the possibility of temp RSA key generation failure:
10500      the code used to assume it always worked and crashed on failure.
10501      [Steve Henson]
10502 
10503   *) Fix potential buffer overrun problem in BIO_printf().
10504      [Ulf Möller, using public domain code by Patrick Powell; problem
10505       pointed out by David Sacerdote <das33@cornell.edu>]
10506 
10507   *) Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
10508      RAND_egd() and RAND_status().  In the command line application,
10509      the EGD socket can be specified like a seed file using RANDFILE
10510      or -rand.
10511      [Ulf Möller]
10512 
10513   *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
10514      Some CAs (e.g. Verisign) distribute certificates in this form.
10515      [Steve Henson]
10516 
10517   *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
10518      list to exclude them. This means that no special compilation option
10519      is needed to use anonymous DH: it just needs to be included in the
10520      cipher list.
10521      [Steve Henson]
10522 
10523   *) Change the EVP_MD_CTX_type macro so its meaning consistent with
10524      EVP_MD_type. The old functionality is available in a new macro called
10525      EVP_MD_md(). Change code that uses it and update docs.
10526      [Steve Henson]
10527 
10528   *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
10529      where the 'void *' argument is replaced by a function pointer argument.
10530      Previously 'void *' was abused to point to functions, which works on
10531      many platforms, but is not correct.  As these functions are usually
10532      called by macros defined in OpenSSL header files, most source code
10533      should work without changes.
10534      [Richard Levitte]
10535 
10536   *) <openssl/opensslconf.h> (which is created by Configure) now contains
10537      sections with information on -D... compiler switches used for
10538      compiling the library so that applications can see them.  To enable
10539      one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
10540      must be defined.  E.g.,
10541         #define OPENSSL_ALGORITHM_DEFINES
10542         #include <openssl/opensslconf.h>
10543      defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
10544      [Richard Levitte, Ulf and Bodo Möller]
10545 
10546   *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
10547      record layer.
10548      [Bodo Moeller]
10549 
10550   *) Change the 'other' type in certificate aux info to a STACK_OF
10551      X509_ALGOR. Although not an AlgorithmIdentifier as such it has
10552      the required ASN1 format: arbitrary types determined by an OID.
10553      [Steve Henson]
10554 
10555   *) Add some PEM_write_X509_REQ_NEW() functions and a command line
10556      argument to 'req'. This is not because the function is newer or
10557      better than others it just uses the work 'NEW' in the certificate
10558      request header lines. Some software needs this.
10559      [Steve Henson]
10560 
10561   *) Reorganise password command line arguments: now passwords can be
10562      obtained from various sources. Delete the PEM_cb function and make
10563      it the default behaviour: i.e. if the callback is NULL and the
10564      usrdata argument is not NULL interpret it as a null terminated pass
10565      phrase. If usrdata and the callback are NULL then the pass phrase
10566      is prompted for as usual.
10567      [Steve Henson]
10568 
10569   *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
10570      the support is automatically enabled. The resulting binaries will
10571      autodetect the card and use it if present.
10572      [Ben Laurie and Compaq Inc.]
10573 
10574   *) Work around for Netscape hang bug. This sends certificate request
10575      and server done in one record. Since this is perfectly legal in the
10576      SSL/TLS protocol it isn't a "bug" option and is on by default. See
10577      the bugs/SSLv3 entry for more info.
10578      [Steve Henson]
10579 
10580   *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
10581      [Andy Polyakov]
10582 
10583   *) Add -rand argument to smime and pkcs12 applications and read/write
10584      of seed file.
10585      [Steve Henson]
10586 
10587   *) New 'passwd' tool for crypt(3) and apr1 password hashes.
10588      [Bodo Moeller]
10589 
10590   *) Add command line password options to the remaining applications.
10591      [Steve Henson]
10592 
10593   *) Bug fix for BN_div_recp() for numerators with an even number of
10594      bits.
10595      [Ulf Möller]
10596 
10597   *) More tests in bntest.c, and changed test_bn output.
10598      [Ulf Möller]
10599 
10600   *) ./config recognizes MacOS X now.
10601      [Andy Polyakov]
10602 
10603   *) Bug fix for BN_div() when the first words of num and divisor are
10604      equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
10605      [Ulf Möller]
10606 
10607   *) Add support for various broken PKCS#8 formats, and command line
10608      options to produce them.
10609      [Steve Henson]
10610 
10611   *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
10612      get temporary BIGNUMs from a BN_CTX.
10613      [Ulf Möller]
10614 
10615   *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
10616      for p == 0.
10617      [Ulf Möller]
10618 
10619   *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
10620      include a #define from the old name to the new. The original intent
10621      was that statically linked binaries could for example just call
10622      SSLeay_add_all_ciphers() to just add ciphers to the table and not
10623      link with digests. This never worked because SSLeay_add_all_digests()
10624      and SSLeay_add_all_ciphers() were in the same source file so calling
10625      one would link with the other. They are now in separate source files.
10626      [Steve Henson]
10627 
10628   *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
10629      [Steve Henson]
10630 
10631   *) Use a less unusual form of the Miller-Rabin primality test (it used
10632      a binary algorithm for exponentiation integrated into the Miller-Rabin
10633      loop, our standard modexp algorithms are faster).
10634      [Bodo Moeller]
10635 
10636   *) Support for the EBCDIC character set completed.
10637      [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
10638 
10639   *) Source code cleanups: use const where appropriate, eliminate casts,
10640      use void * instead of char * in lhash.
10641      [Ulf Möller]
10642 
10643   *) Bugfix: ssl3_send_server_key_exchange was not restartable
10644      (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
10645      this the server could overwrite ephemeral keys that the client
10646      has already seen).
10647      [Bodo Moeller]
10648 
10649   *) Turn DSA_is_prime into a macro that calls BN_is_prime,
10650      using 50 iterations of the Rabin-Miller test.
10651 
10652      DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
10653      iterations of the Rabin-Miller test as required by the appendix
10654      to FIPS PUB 186[-1]) instead of DSA_is_prime.
10655      As BN_is_prime_fasttest includes trial division, DSA parameter
10656      generation becomes much faster.
10657 
10658      This implies a change for the callback functions in DSA_is_prime
10659      and DSA_generate_parameters: The callback function is called once
10660      for each positive witness in the Rabin-Miller test, not just
10661      occasionally in the inner loop; and the parameters to the
10662      callback function now provide an iteration count for the outer
10663      loop rather than for the current invocation of the inner loop.
10664      DSA_generate_parameters additionally can call the callback
10665      function with an 'iteration count' of -1, meaning that a
10666      candidate has passed the trial division test (when q is generated
10667      from an application-provided seed, trial division is skipped).
10668      [Bodo Moeller]
10669 
10670   *) New function BN_is_prime_fasttest that optionally does trial
10671      division before starting the Rabin-Miller test and has
10672      an additional BN_CTX * argument (whereas BN_is_prime always
10673      has to allocate at least one BN_CTX).
10674      'callback(1, -1, cb_arg)' is called when a number has passed the
10675      trial division stage.
10676      [Bodo Moeller]
10677 
10678   *) Fix for bug in CRL encoding. The validity dates weren't being handled
10679      as ASN1_TIME.
10680      [Steve Henson]
10681 
10682   *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
10683      [Steve Henson]
10684 
10685   *) New function BN_pseudo_rand().
10686      [Ulf Möller]
10687 
10688   *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
10689      bignum version of BN_from_montgomery() with the working code from
10690      SSLeay 0.9.0 (the word based version is faster anyway), and clean up
10691      the comments.
10692      [Ulf Möller]
10693 
10694   *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
10695      made it impossible to use the same SSL_SESSION data structure in
10696      SSL2 clients in multiple threads.
10697      [Bodo Moeller]
10698 
10699   *) The return value of RAND_load_file() no longer counts bytes obtained
10700      by stat().  RAND_load_file(..., -1) is new and uses the complete file
10701      to seed the PRNG (previously an explicit byte count was required).
10702      [Ulf Möller, Bodo Möller]
10703 
10704   *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
10705      used (char *) instead of (void *) and had casts all over the place.
10706      [Steve Henson]
10707 
10708   *) Make BN_generate_prime() return NULL on error if ret!=NULL.
10709      [Ulf Möller]
10710 
10711   *) Retain source code compatibility for BN_prime_checks macro:
10712      BN_is_prime(..., BN_prime_checks, ...) now uses
10713      BN_prime_checks_for_size to determine the appropriate number of
10714      Rabin-Miller iterations.
10715      [Ulf Möller]
10716 
10717   *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
10718      DH_CHECK_P_NOT_SAFE_PRIME.
10719      (Check if this is true? OpenPGP calls them "strong".)
10720      [Ulf Möller]
10721 
10722   *) Merge the functionality of "dh" and "gendh" programs into a new program
10723      "dhparam". The old programs are retained for now but will handle DH keys
10724      (instead of parameters) in future.
10725      [Steve Henson]
10726 
10727   *) Make the ciphers, s_server and s_client programs check the return values
10728      when a new cipher list is set.
10729      [Steve Henson]
10730 
10731   *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
10732      ciphers. Before when the 56bit ciphers were enabled the sorting was
10733      wrong.
10734 
10735      The syntax for the cipher sorting has been extended to support sorting by
10736      cipher-strength (using the strength_bits hard coded in the tables).
10737      The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
10738 
10739      Fix a bug in the cipher-command parser: when supplying a cipher command
10740      string with an "undefined" symbol (neither command nor alphanumeric
10741      [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
10742      an error is flagged.
10743 
10744      Due to the strength-sorting extension, the code of the
10745      ssl_create_cipher_list() function was completely rearranged. I hope that
10746      the readability was also increased :-)
10747      [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
10748 
10749   *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
10750      for the first serial number and places 2 in the serial number file. This
10751      avoids problems when the root CA is created with serial number zero and
10752      the first user certificate has the same issuer name and serial number
10753      as the root CA.
10754      [Steve Henson]
10755 
10756   *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
10757      the new code. Add documentation for this stuff.
10758      [Steve Henson]
10759 
10760   *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
10761      X509_*() to X509at_*() on the grounds that they don't handle X509
10762      structures and behave in an analogous way to the X509v3 functions:
10763      they shouldn't be called directly but wrapper functions should be used
10764      instead.
10765 
10766      So we also now have some wrapper functions that call the X509at functions
10767      when passed certificate requests. (TO DO: similar things can be done with
10768      PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
10769      things. Some of these need some d2i or i2d and print functionality
10770      because they handle more complex structures.)
10771      [Steve Henson]
10772 
10773   *) Add missing #ifndefs that caused missing symbols when building libssl
10774      as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
10775      NO_RSA in ssl/s2*.c.
10776      [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
10777 
10778   *) Precautions against using the PRNG uninitialized: RAND_bytes() now
10779      has a return value which indicates the quality of the random data
10780      (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
10781      error queue. New function RAND_pseudo_bytes() generates output that is
10782      guaranteed to be unique but not unpredictable. RAND_add is like
10783      RAND_seed, but takes an extra argument for an entropy estimate
10784      (RAND_seed always assumes full entropy).
10785      [Ulf Möller]
10786 
10787   *) Do more iterations of Rabin-Miller probable prime test (specifically,
10788      3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
10789      instead of only 2 for all lengths; see BN_prime_checks_for_size definition
10790      in crypto/bn/bn_prime.c for the complete table).  This guarantees a
10791      false-positive rate of at most 2^-80 for random input.
10792      [Bodo Moeller]
10793 
10794   *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
10795      [Bodo Moeller]
10796 
10797   *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
10798      in the 0.9.5 release), this returns the chain
10799      from an X509_CTX structure with a dup of the stack and all
10800      the X509 reference counts upped: so the stack will exist
10801      after X509_CTX_cleanup() has been called. Modify pkcs12.c
10802      to use this.
10803 
10804      Also make SSL_SESSION_print() print out the verify return
10805      code.
10806      [Steve Henson]
10807 
10808   *) Add manpage for the pkcs12 command. Also change the default
10809      behaviour so MAC iteration counts are used unless the new
10810      -nomaciter option is used. This improves file security and
10811      only older versions of MSIE (4.0 for example) need it.
10812      [Steve Henson]
10813 
10814   *) Honor the no-xxx Configure options when creating .DEF files.
10815      [Ulf Möller]
10816 
10817   *) Add PKCS#10 attributes to field table: challengePassword,
10818      unstructuredName and unstructuredAddress. These are taken from
10819      draft PKCS#9 v2.0 but are compatible with v1.2 provided no
10820      international characters are used.
10821 
10822      More changes to X509_ATTRIBUTE code: allow the setting of types
10823      based on strings. Remove the 'loc' parameter when adding
10824      attributes because these will be a SET OF encoding which is sorted
10825      in ASN1 order.
10826      [Steve Henson]
10827 
10828   *) Initial changes to the 'req' utility to allow request generation
10829      automation. This will allow an application to just generate a template
10830      file containing all the field values and have req construct the
10831      request.
10832 
10833      Initial support for X509_ATTRIBUTE handling. Stacks of these are
10834      used all over the place including certificate requests and PKCS#7
10835      structures. They are currently handled manually where necessary with
10836      some primitive wrappers for PKCS#7. The new functions behave in a
10837      manner analogous to the X509 extension functions: they allow
10838      attributes to be looked up by NID and added.
10839 
10840      Later something similar to the X509V3 code would be desirable to
10841      automatically handle the encoding, decoding and printing of the
10842      more complex types. The string types like challengePassword can
10843      be handled by the string table functions.
10844 
10845      Also modified the multi byte string table handling. Now there is
10846      a 'global mask' which masks out certain types. The table itself
10847      can use the flag STABLE_NO_MASK to ignore the mask setting: this
10848      is useful when for example there is only one permissible type
10849      (as in countryName) and using the mask might result in no valid
10850      types at all.
10851      [Steve Henson]
10852 
10853   *) Clean up 'Finished' handling, and add functions SSL_get_finished and
10854      SSL_get_peer_finished to allow applications to obtain the latest
10855      Finished messages sent to the peer or expected from the peer,
10856      respectively.  (SSL_get_peer_finished is usually the Finished message
10857      actually received from the peer, otherwise the protocol will be aborted.)
10858 
10859      As the Finished message are message digests of the complete handshake
10860      (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
10861      be used for external authentication procedures when the authentication
10862      provided by SSL/TLS is not desired or is not enough.
10863      [Bodo Moeller]
10864 
10865   *) Enhanced support for Alpha Linux is added. Now ./config checks if
10866      the host supports BWX extension and if Compaq C is present on the
10867      $PATH. Just exploiting of the BWX extension results in 20-30%
10868      performance kick for some algorithms, e.g. DES and RC4 to mention
10869      a couple. Compaq C in turn generates ~20% faster code for MD5 and
10870      SHA1.
10871      [Andy Polyakov]
10872 
10873   *) Add support for MS "fast SGC". This is arguably a violation of the
10874      SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
10875      weak crypto and after checking the certificate is SGC a second one
10876      with strong crypto. MS SGC stops the first handshake after receiving
10877      the server certificate message and sends a second client hello. Since
10878      a server will typically do all the time consuming operations before
10879      expecting any further messages from the client (server key exchange
10880      is the most expensive) there is little difference between the two.
10881 
10882      To get OpenSSL to support MS SGC we have to permit a second client
10883      hello message after we have sent server done. In addition we have to
10884      reset the MAC if we do get this second client hello.
10885      [Steve Henson]
10886 
10887   *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
10888      if a DER encoded private key is RSA or DSA traditional format. Changed
10889      d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
10890      format DER encoded private key. Newer code should use PKCS#8 format which
10891      has the key type encoded in the ASN1 structure. Added DER private key
10892      support to pkcs8 application.
10893      [Steve Henson]
10894 
10895   *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
10896      ciphersuites has been selected (as required by the SSL 3/TLS 1
10897      specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
10898      is set, we interpret this as a request to violate the specification
10899      (the worst that can happen is a handshake failure, and 'correct'
10900      behaviour would result in a handshake failure anyway).
10901      [Bodo Moeller]
10902 
10903   *) In SSL_CTX_add_session, take into account that there might be multiple
10904      SSL_SESSION structures with the same session ID (e.g. when two threads
10905      concurrently obtain them from an external cache).
10906      The internal cache can handle only one SSL_SESSION with a given ID,
10907      so if there's a conflict, we now throw out the old one to achieve
10908      consistency.
10909      [Bodo Moeller]
10910 
10911   *) Add OIDs for idea and blowfish in CBC mode. This will allow both
10912      to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
10913      some routines that use cipher OIDs: some ciphers do not have OIDs
10914      defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
10915      example.
10916      [Steve Henson]
10917 
10918   *) Simplify the trust setting structure and code. Now we just have
10919      two sequences of OIDs for trusted and rejected settings. These will
10920      typically have values the same as the extended key usage extension
10921      and any application specific purposes.
10922 
10923      The trust checking code now has a default behaviour: it will just
10924      check for an object with the same NID as the passed id. Functions can
10925      be provided to override either the default behaviour or the behaviour
10926      for a given id. SSL client, server and email already have functions
10927      in place for compatibility: they check the NID and also return "trusted"
10928      if the certificate is self signed.
10929      [Steve Henson]
10930 
10931   *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
10932      traditional format into an EVP_PKEY structure.
10933      [Steve Henson]
10934 
10935   *) Add a password callback function PEM_cb() which either prompts for
10936      a password if usr_data is NULL or otherwise assumes it is a null
10937      terminated password. Allow passwords to be passed on command line
10938      environment or config files in a few more utilities.
10939      [Steve Henson]
10940 
10941   *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
10942      keys. Add some short names for PKCS#8 PBE algorithms and allow them
10943      to be specified on the command line for the pkcs8 and pkcs12 utilities.
10944      Update documentation.
10945      [Steve Henson]
10946 
10947   *) Support for ASN1 "NULL" type. This could be handled before by using
10948      ASN1_TYPE but there wasn't any function that would try to read a NULL
10949      and produce an error if it couldn't. For compatibility we also have
10950      ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
10951      don't allocate anything because they don't need to.
10952      [Steve Henson]
10953 
10954   *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
10955      for details.
10956      [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
10957 
10958   *) Rebuild of the memory allocation routines used by OpenSSL code and
10959      possibly others as well.  The purpose is to make an interface that
10960      provide hooks so anyone can build a separate set of allocation and
10961      deallocation routines to be used by OpenSSL, for example memory
10962      pool implementations, or something else, which was previously hard
10963      since Malloc(), Realloc() and Free() were defined as macros having
10964      the values malloc, realloc and free, respectively (except for Win32
10965      compilations).  The same is provided for memory debugging code.
10966      OpenSSL already comes with functionality to find memory leaks, but
10967      this gives people a chance to debug other memory problems.
10968 
10969      With these changes, a new set of functions and macros have appeared:
10970 
10971        CRYPTO_set_mem_debug_functions()         [F]
10972        CRYPTO_get_mem_debug_functions()         [F]
10973        CRYPTO_dbg_set_options()                 [F]
10974        CRYPTO_dbg_get_options()                 [F]
10975        CRYPTO_malloc_debug_init()               [M]
10976 
10977      The memory debug functions are NULL by default, unless the library
10978      is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
10979      wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
10980      gives the standard debugging functions that come with OpenSSL) or
10981      CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
10982      provided by the library user) must be used.  When the standard
10983      debugging functions are used, CRYPTO_dbg_set_options can be used to
10984      request additional information:
10985      CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
10986      the CRYPTO_MDEBUG_xxx macro when compiling the library.
10987 
10988      Also, things like CRYPTO_set_mem_functions will always give the
10989      expected result (the new set of functions is used for allocation
10990      and deallocation) at all times, regardless of platform and compiler
10991      options.
10992 
10993      To finish it up, some functions that were never use in any other
10994      way than through macros have a new API and new semantic:
10995 
10996        CRYPTO_dbg_malloc()
10997        CRYPTO_dbg_realloc()
10998        CRYPTO_dbg_free()
10999 
11000      All macros of value have retained their old syntax.
11001      [Richard Levitte and Bodo Moeller]
11002 
11003   *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
11004      ordering of SMIMECapabilities wasn't in "strength order" and there
11005      was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
11006      algorithm.
11007      [Steve Henson]
11008 
11009   *) Some ASN1 types with illegal zero length encoding (INTEGER,
11010      ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
11011      [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
11012 
11013   *) Merge in my S/MIME library for OpenSSL. This provides a simple
11014      S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
11015      functionality to handle multipart/signed properly) and a utility
11016      called 'smime' to call all this stuff. This is based on code I
11017      originally wrote for Celo who have kindly allowed it to be
11018      included in OpenSSL.
11019      [Steve Henson]
11020 
11021   *) Add variants des_set_key_checked and des_set_key_unchecked of
11022      des_set_key (aka des_key_sched).  Global variable des_check_key
11023      decides which of these is called by des_set_key; this way
11024      des_check_key behaves as it always did, but applications and
11025      the library itself, which was buggy for des_check_key == 1,
11026      have a cleaner way to pick the version they need.
11027      [Bodo Moeller]
11028 
11029   *) New function PKCS12_newpass() which changes the password of a
11030      PKCS12 structure.
11031      [Steve Henson]
11032 
11033   *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
11034      dynamic mix. In both cases the ids can be used as an index into the
11035      table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
11036      functions so they accept a list of the field values and the
11037      application doesn't need to directly manipulate the X509_TRUST
11038      structure.
11039      [Steve Henson]
11040 
11041   *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
11042      need initialising.
11043      [Steve Henson]
11044 
11045   *) Modify the way the V3 extension code looks up extensions. This now
11046      works in a similar way to the object code: we have some "standard"
11047      extensions in a static table which is searched with OBJ_bsearch()
11048      and the application can add dynamic ones if needed. The file
11049      crypto/x509v3/ext_dat.h now has the info: this file needs to be
11050      updated whenever a new extension is added to the core code and kept
11051      in ext_nid order. There is a simple program 'tabtest.c' which checks
11052      this. New extensions are not added too often so this file can readily
11053      be maintained manually.
11054 
11055      There are two big advantages in doing things this way. The extensions
11056      can be looked up immediately and no longer need to be "added" using
11057      X509V3_add_standard_extensions(): this function now does nothing.
11058      [Side note: I get *lots* of email saying the extension code doesn't
11059       work because people forget to call this function]
11060      Also no dynamic allocation is done unless new extensions are added:
11061      so if we don't add custom extensions there is no need to call
11062      X509V3_EXT_cleanup().
11063      [Steve Henson]
11064 
11065   *) Modify enc utility's salting as follows: make salting the default. Add a
11066      magic header, so unsalted files fail gracefully instead of just decrypting
11067      to garbage. This is because not salting is a big security hole, so people
11068      should be discouraged from doing it.
11069      [Ben Laurie]
11070 
11071   *) Fixes and enhancements to the 'x509' utility. It allowed a message
11072      digest to be passed on the command line but it only used this
11073      parameter when signing a certificate. Modified so all relevant
11074      operations are affected by the digest parameter including the
11075      -fingerprint and -x509toreq options. Also -x509toreq choked if a
11076      DSA key was used because it didn't fix the digest.
11077      [Steve Henson]
11078 
11079   *) Initial certificate chain verify code. Currently tests the untrusted
11080      certificates for consistency with the verify purpose (which is set
11081      when the X509_STORE_CTX structure is set up) and checks the pathlength.
11082 
11083      There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
11084      this is because it will reject chains with invalid extensions whereas
11085      every previous version of OpenSSL and SSLeay made no checks at all.
11086 
11087      Trust code: checks the root CA for the relevant trust settings. Trust
11088      settings have an initial value consistent with the verify purpose: e.g.
11089      if the verify purpose is for SSL client use it expects the CA to be
11090      trusted for SSL client use. However the default value can be changed to
11091      permit custom trust settings: one example of this would be to only trust
11092      certificates from a specific "secure" set of CAs.
11093 
11094      Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
11095      which should be used for version portability: especially since the
11096      verify structure is likely to change more often now.
11097 
11098      SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
11099      to set them. If not set then assume SSL clients will verify SSL servers
11100      and vice versa.
11101 
11102      Two new options to the verify program: -untrusted allows a set of
11103      untrusted certificates to be passed in and -purpose which sets the
11104      intended purpose of the certificate. If a purpose is set then the
11105      new chain verify code is used to check extension consistency.
11106      [Steve Henson]
11107 
11108   *) Support for the authority information access extension.
11109      [Steve Henson]
11110 
11111   *) Modify RSA and DSA PEM read routines to transparently handle
11112      PKCS#8 format private keys. New *_PUBKEY_* functions that handle
11113      public keys in a format compatible with certificate
11114      SubjectPublicKeyInfo structures. Unfortunately there were already
11115      functions called *_PublicKey_* which used various odd formats so
11116      these are retained for compatibility: however the DSA variants were
11117      never in a public release so they have been deleted. Changed dsa/rsa
11118      utilities to handle the new format: note no releases ever handled public
11119      keys so we should be OK.
11120 
11121      The primary motivation for this change is to avoid the same fiasco
11122      that dogs private keys: there are several incompatible private key
11123      formats some of which are standard and some OpenSSL specific and
11124      require various evil hacks to allow partial transparent handling and
11125      even then it doesn't work with DER formats. Given the option anything
11126      other than PKCS#8 should be dumped: but the other formats have to
11127      stay in the name of compatibility.
11128 
11129      With public keys and the benefit of hindsight one standard format
11130      is used which works with EVP_PKEY, RSA or DSA structures: though
11131      it clearly returns an error if you try to read the wrong kind of key.
11132 
11133      Added a -pubkey option to the 'x509' utility to output the public key.
11134      Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
11135      (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
11136      EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
11137      that do the same as the EVP_PKEY_assign_*() except they up the
11138      reference count of the added key (they don't "swallow" the
11139      supplied key).
11140      [Steve Henson]
11141 
11142   *) Fixes to crypto/x509/by_file.c the code to read in certificates and
11143      CRLs would fail if the file contained no certificates or no CRLs:
11144      added a new function to read in both types and return the number
11145      read: this means that if none are read it will be an error. The
11146      DER versions of the certificate and CRL reader would always fail
11147      because it isn't possible to mix certificates and CRLs in DER format
11148      without choking one or the other routine. Changed this to just read
11149      a certificate: this is the best we can do. Also modified the code
11150      in apps/verify.c to take notice of return codes: it was previously
11151      attempting to read in certificates from NULL pointers and ignoring
11152      any errors: this is one reason why the cert and CRL reader seemed
11153      to work. It doesn't check return codes from the default certificate
11154      routines: these may well fail if the certificates aren't installed.
11155      [Steve Henson]
11156 
11157   *) Code to support otherName option in GeneralName.
11158      [Steve Henson]
11159 
11160   *) First update to verify code. Change the verify utility
11161      so it warns if it is passed a self signed certificate:
11162      for consistency with the normal behaviour. X509_verify
11163      has been modified to it will now verify a self signed
11164      certificate if *exactly* the same certificate appears
11165      in the store: it was previously impossible to trust a
11166      single self signed certificate. This means that:
11167      openssl verify ss.pem
11168      now gives a warning about a self signed certificate but
11169      openssl verify -CAfile ss.pem ss.pem
11170      is OK.
11171      [Steve Henson]
11172 
11173   *) For servers, store verify_result in SSL_SESSION data structure
11174      (and add it to external session representation).
11175      This is needed when client certificate verifications fails,
11176      but an application-provided verification callback (set by
11177      SSL_CTX_set_cert_verify_callback) allows accepting the session
11178      anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
11179      but returns 1): When the session is reused, we have to set
11180      ssl->verify_result to the appropriate error code to avoid
11181      security holes.
11182      [Bodo Moeller, problem pointed out by Lutz Jaenicke]
11183 
11184   *) Fix a bug in the new PKCS#7 code: it didn't consider the
11185      case in PKCS7_dataInit() where the signed PKCS7 structure
11186      didn't contain any existing data because it was being created.
11187      [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
11188 
11189   *) Add a salt to the key derivation routines in enc.c. This
11190      forms the first 8 bytes of the encrypted file. Also add a
11191      -S option to allow a salt to be input on the command line.
11192      [Steve Henson]
11193 
11194   *) New function X509_cmp(). Oddly enough there wasn't a function
11195      to compare two certificates. We do this by working out the SHA1
11196      hash and comparing that. X509_cmp() will be needed by the trust
11197      code.
11198      [Steve Henson]
11199 
11200   *) SSL_get1_session() is like SSL_get_session(), but increments
11201      the reference count in the SSL_SESSION returned.
11202      [Geoff Thorpe <geoff@eu.c2.net>]
11203 
11204   *) Fix for 'req': it was adding a null to request attributes.
11205      Also change the X509_LOOKUP and X509_INFO code to handle
11206      certificate auxiliary information.
11207      [Steve Henson]
11208 
11209   *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
11210      the 'enc' command.
11211      [Steve Henson]
11212 
11213   *) Add the possibility to add extra information to the memory leak
11214      detecting output, to form tracebacks, showing from where each
11215      allocation was originated: CRYPTO_push_info("constant string") adds
11216      the string plus current file name and line number to a per-thread
11217      stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
11218      is like calling CYRPTO_pop_info() until the stack is empty.
11219      Also updated memory leak detection code to be multi-thread-safe.
11220      [Richard Levitte]
11221 
11222   *) Add options -text and -noout to pkcs7 utility and delete the
11223      encryption options which never did anything. Update docs.
11224      [Steve Henson]
11225 
11226   *) Add options to some of the utilities to allow the pass phrase
11227      to be included on either the command line (not recommended on
11228      OSes like Unix) or read from the environment. Update the
11229      manpages and fix a few bugs.
11230      [Steve Henson]
11231 
11232   *) Add a few manpages for some of the openssl commands.
11233      [Steve Henson]
11234 
11235   *) Fix the -revoke option in ca. It was freeing up memory twice,
11236      leaking and not finding already revoked certificates.
11237      [Steve Henson]
11238 
11239   *) Extensive changes to support certificate auxiliary information.
11240      This involves the use of X509_CERT_AUX structure and X509_AUX
11241      functions. An X509_AUX function such as PEM_read_X509_AUX()
11242      can still read in a certificate file in the usual way but it
11243      will also read in any additional "auxiliary information". By
11244      doing things this way a fair degree of compatibility can be
11245      retained: existing certificates can have this information added
11246      using the new 'x509' options.
11247 
11248      Current auxiliary information includes an "alias" and some trust
11249      settings. The trust settings will ultimately be used in enhanced
11250      certificate chain verification routines: currently a certificate
11251      can only be trusted if it is self signed and then it is trusted
11252      for all purposes.
11253      [Steve Henson]
11254 
11255   *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
11256      The problem was that one of the replacement routines had not been working
11257      since SSLeay releases.  For now the offending routine has been replaced
11258      with non-optimised assembler.  Even so, this now gives around 95%
11259      performance improvement for 1024 bit RSA signs.
11260      [Mark Cox]
11261 
11262   *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
11263      handling. Most clients have the effective key size in bits equal to
11264      the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
11265      A few however don't do this and instead use the size of the decrypted key
11266      to determine the RC2 key length and the AlgorithmIdentifier to determine
11267      the effective key length. In this case the effective key length can still
11268      be 40 bits but the key length can be 168 bits for example. This is fixed
11269      by manually forcing an RC2 key into the EVP_PKEY structure because the
11270      EVP code can't currently handle unusual RC2 key sizes: it always assumes
11271      the key length and effective key length are equal.
11272      [Steve Henson]
11273 
11274   *) Add a bunch of functions that should simplify the creation of
11275      X509_NAME structures. Now you should be able to do:
11276      X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
11277      and have it automatically work out the correct field type and fill in
11278      the structures. The more adventurous can try:
11279      X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
11280      and it will (hopefully) work out the correct multibyte encoding.
11281      [Steve Henson]
11282 
11283   *) Change the 'req' utility to use the new field handling and multibyte
11284      copy routines. Before the DN field creation was handled in an ad hoc
11285      way in req, ca, and x509 which was rather broken and didn't support
11286      BMPStrings or UTF8Strings. Since some software doesn't implement
11287      BMPStrings or UTF8Strings yet, they can be enabled using the config file
11288      using the dirstring_type option. See the new comment in the default
11289      openssl.cnf for more info.
11290      [Steve Henson]
11291 
11292   *) Make crypto/rand/md_rand.c more robust:
11293      - Assure unique random numbers after fork().
11294      - Make sure that concurrent threads access the global counter and
11295        md serializably so that we never lose entropy in them
11296        or use exactly the same state in multiple threads.
11297        Access to the large state is not always serializable because
11298        the additional locking could be a performance killer, and
11299        md should be large enough anyway.
11300      [Bodo Moeller]
11301 
11302   *) New file apps/app_rand.c with commonly needed functionality
11303      for handling the random seed file.
11304 
11305      Use the random seed file in some applications that previously did not:
11306           ca,
11307           dsaparam -genkey (which also ignored its '-rand' option),
11308           s_client,
11309           s_server,
11310           x509 (when signing).
11311      Except on systems with /dev/urandom, it is crucial to have a random
11312      seed file at least for key creation, DSA signing, and for DH exchanges;
11313      for RSA signatures we could do without one.
11314 
11315      gendh and gendsa (unlike genrsa) used to read only the first byte
11316      of each file listed in the '-rand' option.  The function as previously
11317      found in genrsa is now in app_rand.c and is used by all programs
11318      that support '-rand'.
11319      [Bodo Moeller]
11320 
11321   *) In RAND_write_file, use mode 0600 for creating files;
11322      don't just chmod when it may be too late.
11323      [Bodo Moeller]
11324 
11325   *) Report an error from X509_STORE_load_locations
11326      when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
11327      [Bill Perry]
11328 
11329   *) New function ASN1_mbstring_copy() this copies a string in either
11330      ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
11331      into an ASN1_STRING type. A mask of permissible types is passed
11332      and it chooses the "minimal" type to use or an error if not type
11333      is suitable.
11334      [Steve Henson]
11335 
11336   *) Add function equivalents to the various macros in asn1.h. The old
11337      macros are retained with an M_ prefix. Code inside the library can
11338      use the M_ macros. External code (including the openssl utility)
11339      should *NOT* in order to be "shared library friendly".
11340      [Steve Henson]
11341 
11342   *) Add various functions that can check a certificate's extensions
11343      to see if it usable for various purposes such as SSL client,
11344      server or S/MIME and CAs of these types. This is currently
11345      VERY EXPERIMENTAL but will ultimately be used for certificate chain
11346      verification. Also added a -purpose flag to x509 utility to
11347      print out all the purposes.
11348      [Steve Henson]
11349 
11350   *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
11351      functions.
11352      [Steve Henson]
11353 
11354   *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
11355      for, obtain and decode and extension and obtain its critical flag.
11356      This allows all the necessary extension code to be handled in a
11357      single function call.
11358      [Steve Henson]
11359 
11360   *) RC4 tune-up featuring 30-40% performance improvement on most RISC
11361      platforms. See crypto/rc4/rc4_enc.c for further details.
11362      [Andy Polyakov]
11363 
11364   *) New -noout option to asn1parse. This causes no output to be produced
11365      its main use is when combined with -strparse and -out to extract data
11366      from a file (which may not be in ASN.1 format).
11367      [Steve Henson]
11368 
11369   *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
11370      when producing the local key id.
11371      [Richard Levitte <levitte@stacken.kth.se>]
11372 
11373   *) New option -dhparam in s_server. This allows a DH parameter file to be
11374      stated explicitly. If it is not stated then it tries the first server
11375      certificate file. The previous behaviour hard coded the filename
11376      "server.pem".
11377      [Steve Henson]
11378 
11379   *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
11380      a public key to be input or output. For example:
11381      openssl rsa -in key.pem -pubout -out pubkey.pem
11382      Also added necessary DSA public key functions to handle this.
11383      [Steve Henson]
11384 
11385   *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
11386      in the message. This was handled by allowing
11387      X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
11388      [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
11389 
11390   *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
11391      to the end of the strings whereas this didn't. This would cause problems
11392      if strings read with d2i_ASN1_bytes() were later modified.
11393      [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
11394 
11395   *) Fix for base64 decode bug. When a base64 bio reads only one line of
11396      data and it contains EOF it will end up returning an error. This is
11397      caused by input 46 bytes long. The cause is due to the way base64
11398      BIOs find the start of base64 encoded data. They do this by trying a
11399      trial decode on each line until they find one that works. When they
11400      do a flag is set and it starts again knowing it can pass all the
11401      data directly through the decoder. Unfortunately it doesn't reset
11402      the context it uses. This means that if EOF is reached an attempt
11403      is made to pass two EOFs through the context and this causes the
11404      resulting error. This can also cause other problems as well. As is
11405      usual with these problems it takes *ages* to find and the fix is
11406      trivial: move one line.
11407      [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
11408 
11409   *) Ugly workaround to get s_client and s_server working under Windows. The
11410      old code wouldn't work because it needed to select() on sockets and the
11411      tty (for keypresses and to see if data could be written). Win32 only
11412      supports select() on sockets so we select() with a 1s timeout on the
11413      sockets and then see if any characters are waiting to be read, if none
11414      are present then we retry, we also assume we can always write data to
11415      the tty. This isn't nice because the code then blocks until we've
11416      received a complete line of data and it is effectively polling the
11417      keyboard at 1s intervals: however it's quite a bit better than not
11418      working at all :-) A dedicated Windows application might handle this
11419      with an event loop for example.
11420      [Steve Henson]
11421 
11422   *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
11423      and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
11424      will be called when RSA_sign() and RSA_verify() are used. This is useful
11425      if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
11426      For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
11427      should *not* be used: RSA_sign() and RSA_verify() must be used instead.
11428      This necessitated the support of an extra signature type NID_md5_sha1
11429      for SSL signatures and modifications to the SSL library to use it instead
11430      of calling RSA_public_decrypt() and RSA_private_encrypt().
11431      [Steve Henson]
11432 
11433   *) Add new -verify -CAfile and -CApath options to the crl program, these
11434      will lookup a CRL issuers certificate and verify the signature in a
11435      similar way to the verify program. Tidy up the crl program so it
11436      no longer accesses structures directly. Make the ASN1 CRL parsing a bit
11437      less strict. It will now permit CRL extensions even if it is not
11438      a V2 CRL: this will allow it to tolerate some broken CRLs.
11439      [Steve Henson]
11440 
11441   *) Initialize all non-automatic variables each time one of the openssl
11442      sub-programs is started (this is necessary as they may be started
11443      multiple times from the "OpenSSL>" prompt).
11444      [Lennart Bang, Bodo Moeller]
11445 
11446   *) Preliminary compilation option RSA_NULL which disables RSA crypto without
11447      removing all other RSA functionality (this is what NO_RSA does). This
11448      is so (for example) those in the US can disable those operations covered
11449      by the RSA patent while allowing storage and parsing of RSA keys and RSA
11450      key generation.
11451      [Steve Henson]
11452 
11453   *) Non-copying interface to BIO pairs.
11454      (still largely untested)
11455      [Bodo Moeller]
11456 
11457   *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
11458      ASCII string. This was handled independently in various places before.
11459      [Steve Henson]
11460 
11461   *) New functions UTF8_getc() and UTF8_putc() that parse and generate
11462      UTF8 strings a character at a time.
11463      [Steve Henson]
11464 
11465   *) Use client_version from client hello to select the protocol
11466      (s23_srvr.c) and for RSA client key exchange verification
11467      (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
11468      [Bodo Moeller]
11469 
11470   *) Add various utility functions to handle SPKACs, these were previously
11471      handled by poking round in the structure internals. Added new function
11472      NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
11473      print, verify and generate SPKACs. Based on an original idea from
11474      Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
11475      [Steve Henson]
11476 
11477   *) RIPEMD160 is operational on all platforms and is back in 'make test'.
11478      [Andy Polyakov]
11479 
11480   *) Allow the config file extension section to be overwritten on the
11481      command line. Based on an original idea from Massimiliano Pala
11482      <madwolf@comune.modena.it>. The new option is called -extensions
11483      and can be applied to ca, req and x509. Also -reqexts to override
11484      the request extensions in req and -crlexts to override the crl extensions
11485      in ca.
11486      [Steve Henson]
114