Back to home page

openssl Cross Reference

 
 

    


0001 
0002  OpenSSL CHANGES
0003  _______________
0004 
0005  Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
0006 
0007   *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
0008      handshake can force the use of weak keying material in OpenSSL
0009      SSL/TLS clients and servers.
0010 
0011      Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
0012      researching this issue. (CVE-2014-0224)
0013      [KIKUCHI Masashi, Steve Henson]
0014 
0015   *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
0016      OpenSSL DTLS client the code can be made to recurse eventually crashing
0017      in a DoS attack.
0018 
0019      Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
0020      (CVE-2014-0221)
0021      [Imre Rad, Steve Henson]
0022 
0023   *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
0024      be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
0025      client or server. This is potentially exploitable to run arbitrary
0026      code on a vulnerable client or server.
0027 
0028      Thanks to Jüri Aedla for reporting this issue. (CVE-2014-0195)
0029      [Jüri Aedla, Steve Henson]
0030 
0031   *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
0032      are subject to a denial of service attack.
0033 
0034      Thanks to Felix Gröbert and Ivan Fratric at Google for discovering
0035      this issue. (CVE-2014-3470)
0036      [Felix Gröbert, Ivan Fratric, Steve Henson]
0037 
0038   *) Harmonize version and its documentation. -f flag is used to display
0039      compilation flags.
0040      [mancha <mancha1@zoho.com>]
0041 
0042   *) Fix eckey_priv_encode so it immediately returns an error upon a failure
0043      in i2d_ECPrivateKey.
0044      [mancha <mancha1@zoho.com>]
0045 
0046   *) Fix some double frees. These are not thought to be exploitable.
0047      [mancha <mancha1@zoho.com>]
0048 
0049  Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
0050 
0051   *) A missing bounds check in the handling of the TLS heartbeat extension
0052      can be used to reveal up to 64k of memory to a connected client or
0053      server.
0054 
0055      Thanks for Neel Mehta of Google Security for discovering this bug and to
0056      Adam Langley <agl@chromium.org> and Bodo Moeller <bmoeller@acm.org> for
0057      preparing the fix (CVE-2014-0160)
0058      [Adam Langley, Bodo Moeller]
0059 
0060   *) Fix for the attack described in the paper "Recovering OpenSSL
0061      ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
0062      by Yuval Yarom and Naomi Benger. Details can be obtained from:
0063      http://eprint.iacr.org/2014/140
0064 
0065      Thanks to Yuval Yarom and Naomi Benger for discovering this
0066      flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
0067      [Yuval Yarom and Naomi Benger]
0068 
0069   *) TLS pad extension: draft-agl-tls-padding-03
0070 
0071      Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
0072      TLS client Hello record length value would otherwise be > 255 and
0073      less that 512 pad with a dummy extension containing zeroes so it
0074      is at least 512 bytes long.
0075 
0076      [Adam Langley, Steve Henson]
0077 
0078  Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
0079 
0080   *) Fix for TLS record tampering bug. A carefully crafted invalid 
0081      handshake could crash OpenSSL with a NULL pointer exception.
0082      Thanks to Anton Johansson for reporting this issues.
0083      (CVE-2013-4353)
0084 
0085   *) Keep original DTLS digest and encryption contexts in retransmission
0086      structures so we can use the previous session parameters if they need
0087      to be resent. (CVE-2013-6450)
0088      [Steve Henson]
0089 
0090   *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
0091      avoids preferring ECDHE-ECDSA ciphers when the client appears to be
0092      Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
0093      several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
0094      is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
0095      10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
0096      [Rob Stradling, Adam Langley]
0097 
0098  Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
0099 
0100   *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI
0101      supporting platforms or when small records were transferred.
0102      [Andy Polyakov, Steve Henson]
0103 
0104  Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
0105 
0106   *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
0107 
0108      This addresses the flaw in CBC record processing discovered by 
0109      Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
0110      at: http://www.isg.rhul.ac.uk/tls/     
0111 
0112      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
0113      Security Group at Royal Holloway, University of London
0114      (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
0115      Emilia Käsper for the initial patch.
0116      (CVE-2013-0169)
0117      [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
0118 
0119   *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
0120      ciphersuites which can be exploited in a denial of service attack.
0121      Thanks go to and to Adam Langley <agl@chromium.org> for discovering
0122      and detecting this bug and to Wolfgang Ettlinger
0123      <wolfgang.ettlinger@gmail.com> for independently discovering this issue.
0124      (CVE-2012-2686)
0125      [Adam Langley]
0126 
0127   *) Return an error when checking OCSP signatures when key is NULL.
0128      This fixes a DoS attack. (CVE-2013-0166)
0129      [Steve Henson]
0130 
0131   *) Make openssl verify return errors.
0132      [Chris Palmer <palmer@google.com> and Ben Laurie]
0133 
0134   *) Call OCSP Stapling callback after ciphersuite has been chosen, so
0135      the right response is stapled. Also change SSL_get_certificate()
0136      so it returns the certificate actually sent.
0137      See http://rt.openssl.org/Ticket/Display.html?id=2836.
0138      [Rob Stradling <rob.stradling@comodo.com>]
0139 
0140   *) Fix possible deadlock when decoding public keys.
0141      [Steve Henson]
0142 
0143   *) Don't use TLS 1.0 record version number in initial client hello
0144      if renegotiating.
0145      [Steve Henson]
0146 
0147  Changes between 1.0.1b and 1.0.1c [10 May 2012]
0148 
0149   *) Sanity check record length before skipping explicit IV in TLS
0150      1.2, 1.1 and DTLS to fix DoS attack.
0151 
0152      Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
0153      fuzzing as a service testing platform.
0154      (CVE-2012-2333)
0155      [Steve Henson]
0156 
0157   *) Initialise tkeylen properly when encrypting CMS messages.
0158      Thanks to Solar Designer of Openwall for reporting this issue.
0159      [Steve Henson]
0160 
0161   *) In FIPS mode don't try to use composite ciphers as they are not
0162      approved.
0163      [Steve Henson]
0164 
0165  Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
0166 
0167   *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
0168      1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
0169      mean any application compiled against OpenSSL 1.0.0 headers setting
0170      SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng
0171      TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
0172      0x10000000L Any application which was previously compiled against
0173      OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
0174      will need to be recompiled as a result. Letting be results in
0175      inability to disable specifically TLS 1.1 and in client context,
0176      in unlike event, limit maximum offered version to TLS 1.0 [see below].
0177      [Steve Henson]
0178 
0179   *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not
0180      disable just protocol X, but all protocols above X *if* there are
0181      protocols *below* X still enabled. In more practical terms it means
0182      that if application wants to disable TLS1.0 in favor of TLS1.1 and
0183      above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass
0184      SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to
0185      client side.
0186      [Andy Polyakov]
0187 
0188  Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
0189 
0190   *) Check for potentially exploitable overflows in asn1_d2i_read_bio
0191      BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
0192      in CRYPTO_realloc_clean.
0193 
0194      Thanks to Tavis Ormandy, Google Security Team, for discovering this
0195      issue and to Adam Langley <agl@chromium.org> for fixing it.
0196      (CVE-2012-2110)
0197      [Adam Langley (Google), Tavis Ormandy, Google Security Team]
0198 
0199   *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
0200      [Adam Langley]
0201 
0202   *) Workarounds for some broken servers that "hang" if a client hello
0203      record length exceeds 255 bytes.
0204 
0205      1. Do not use record version number > TLS 1.0 in initial client
0206         hello: some (but not all) hanging servers will now work.
0207      2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
0208         the number of ciphers sent in the client hello. This should be
0209         set to an even number, such as 50, for example by passing:
0210         -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
0211         Most broken servers should now work.
0212      3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
0213         TLS 1.2 client support entirely.
0214      [Steve Henson]
0215 
0216   *) Fix SEGV in Vector Permutation AES module observed in OpenSSH.
0217      [Andy Polyakov]
0218 
0219  Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
0220 
0221   *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET
0222      STRING form instead of a DigestInfo.
0223      [Steve Henson]
0224 
0225   *) The format used for MDC2 RSA signatures is inconsistent between EVP
0226      and the RSA_sign/RSA_verify functions. This was made more apparent when
0227      OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
0228      those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 
0229      the correct format in RSA_verify so both forms transparently work.
0230      [Steve Henson]
0231 
0232   *) Some servers which support TLS 1.0 can choke if we initially indicate
0233      support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
0234      encrypted premaster secret. As a workaround use the maximum pemitted
0235      client version in client hello, this should keep such servers happy
0236      and still work with previous versions of OpenSSL.
0237      [Steve Henson]
0238 
0239   *) Add support for TLS/DTLS heartbeats.
0240      [Robin Seggelmann <seggelmann@fh-muenster.de>]
0241 
0242   *) Add support for SCTP.
0243      [Robin Seggelmann <seggelmann@fh-muenster.de>]
0244 
0245   *) Improved PRNG seeding for VOS.
0246      [Paul Green <Paul.Green@stratus.com>]
0247 
0248   *) Extensive assembler packs updates, most notably:
0249 
0250         - x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
0251         - x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
0252         - x86_64:       bit-sliced AES implementation;
0253         - ARM:          NEON support, contemporary platforms optimizations;
0254         - s390x:        z196 support;
0255         - *:            GHASH and GF(2^m) multiplication implementations;
0256 
0257      [Andy Polyakov]
0258 
0259   *) Make TLS-SRP code conformant with RFC 5054 API cleanup
0260      (removal of unnecessary code)
0261      [Peter Sylvester <peter.sylvester@edelweb.fr>]
0262 
0263   *) Add TLS key material exporter from RFC 5705.
0264      [Eric Rescorla]
0265 
0266   *) Add DTLS-SRTP negotiation from RFC 5764.
0267      [Eric Rescorla]
0268 
0269   *) Add Next Protocol Negotiation,
0270      http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
0271      disabled with a no-npn flag to config or Configure. Code donated
0272      by Google.
0273      [Adam Langley <agl@google.com> and Ben Laurie]
0274 
0275   *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
0276      NIST-P256, NIST-P521, with constant-time single point multiplication on
0277      typical inputs. Compiler support for the nonstandard type __uint128_t is
0278      required to use this (present in gcc 4.4 and later, for 64-bit builds).
0279      Code made available under Apache License version 2.0.
0280 
0281      Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
0282      line to include this in your build of OpenSSL, and run "make depend" (or
0283      "make update"). This enables the following EC_METHODs:
0284 
0285          EC_GFp_nistp224_method()
0286          EC_GFp_nistp256_method()
0287          EC_GFp_nistp521_method()
0288 
0289      EC_GROUP_new_by_curve_name() will automatically use these (while
0290      EC_GROUP_new_curve_GFp() currently prefers the more flexible
0291      implementations).
0292      [Emilia Käsper, Adam Langley, Bodo Moeller (Google)]
0293 
0294   *) Use type ossl_ssize_t instad of ssize_t which isn't available on
0295      all platforms. Move ssize_t definition from e_os.h to the public
0296      header file e_os2.h as it now appears in public header file cms.h
0297      [Steve Henson]
0298 
0299   *) New -sigopt option to the ca, req and x509 utilities. Additional
0300      signature parameters can be passed using this option and in
0301      particular PSS. 
0302      [Steve Henson]
0303 
0304   *) Add RSA PSS signing function. This will generate and set the
0305      appropriate AlgorithmIdentifiers for PSS based on those in the
0306      corresponding EVP_MD_CTX structure. No application support yet.
0307      [Steve Henson]
0308 
0309   *) Support for companion algorithm specific ASN1 signing routines.
0310      New function ASN1_item_sign_ctx() signs a pre-initialised
0311      EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
0312      the appropriate parameters.
0313      [Steve Henson]
0314 
0315   *) Add new algorithm specific ASN1 verification initialisation function
0316      to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
0317      handling will be the same no matter what EVP_PKEY_METHOD is used.
0318      Add a PSS handler to support verification of PSS signatures: checked
0319      against a number of sample certificates.
0320      [Steve Henson]
0321 
0322   *) Add signature printing for PSS. Add PSS OIDs.
0323      [Steve Henson, Martin Kaiser <lists@kaiser.cx>]
0324 
0325   *) Add algorithm specific signature printing. An individual ASN1 method
0326      can now print out signatures instead of the standard hex dump. 
0327 
0328      More complex signatures (e.g. PSS) can print out more meaningful
0329      information. Include DSA version that prints out the signature
0330      parameters r, s.
0331      [Steve Henson]
0332 
0333   *) Password based recipient info support for CMS library: implementing
0334      RFC3211.
0335      [Steve Henson]
0336 
0337   *) Split password based encryption into PBES2 and PBKDF2 functions. This
0338      neatly separates the code into cipher and PBE sections and is required
0339      for some algorithms that split PBES2 into separate pieces (such as
0340      password based CMS).
0341      [Steve Henson]
0342 
0343   *) Session-handling fixes:
0344      - Fix handling of connections that are resuming with a session ID,
0345        but also support Session Tickets.
0346      - Fix a bug that suppressed issuing of a new ticket if the client
0347        presented a ticket with an expired session.
0348      - Try to set the ticket lifetime hint to something reasonable.
0349      - Make tickets shorter by excluding irrelevant information.
0350      - On the client side, don't ignore renewed tickets.
0351      [Adam Langley, Bodo Moeller (Google)]
0352 
0353   *) Fix PSK session representation.
0354      [Bodo Moeller]
0355 
0356   *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
0357 
0358      This work was sponsored by Intel.
0359      [Andy Polyakov]
0360 
0361   *) Add GCM support to TLS library. Some custom code is needed to split
0362      the IV between the fixed (from PRF) and explicit (from TLS record)
0363      portions. This adds all GCM ciphersuites supported by RFC5288 and 
0364      RFC5289. Generalise some AES* cipherstrings to inlclude GCM and
0365      add a special AESGCM string for GCM only.
0366      [Steve Henson]
0367 
0368   *) Expand range of ctrls for AES GCM. Permit setting invocation
0369      field on decrypt and retrieval of invocation field only on encrypt.
0370      [Steve Henson]
0371 
0372   *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
0373      As required by RFC5289 these ciphersuites cannot be used if for
0374      versions of TLS earlier than 1.2.
0375      [Steve Henson]
0376 
0377   *) For FIPS capable OpenSSL interpret a NULL default public key method
0378      as unset and return the appopriate default but do *not* set the default.
0379      This means we can return the appopriate method in applications that
0380      swicth between FIPS and non-FIPS modes.
0381      [Steve Henson]
0382 
0383   *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
0384      ENGINE is used then we cannot handle that in the FIPS module so we
0385      keep original code iff non-FIPS operations are allowed.
0386      [Steve Henson]
0387 
0388   *) Add -attime option to openssl utilities.
0389      [Peter Eckersley <pde@eff.org>, Ben Laurie and Steve Henson]
0390 
0391   *) Redirect DSA and DH operations to FIPS module in FIPS mode.
0392      [Steve Henson]
0393 
0394   *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
0395      FIPS EC methods unconditionally for now.
0396      [Steve Henson]
0397 
0398   *) New build option no-ec2m to disable characteristic 2 code.
0399      [Steve Henson]
0400 
0401   *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
0402      all cases can be covered as some introduce binary incompatibilities.
0403      [Steve Henson]
0404 
0405   *) Redirect RSA operations to FIPS module including keygen,
0406      encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
0407      [Steve Henson]
0408 
0409   *) Add similar low level API blocking to ciphers.
0410      [Steve Henson]
0411 
0412   *) Low level digest APIs are not approved in FIPS mode: any attempt
0413      to use these will cause a fatal error. Applications that *really* want
0414      to use them can use the private_* version instead.
0415      [Steve Henson]
0416 
0417   *) Redirect cipher operations to FIPS module for FIPS builds. 
0418      [Steve Henson]
0419 
0420   *) Redirect digest operations to FIPS module for FIPS builds. 
0421      [Steve Henson]
0422 
0423   *) Update build system to add "fips" flag which will link in fipscanister.o
0424      for static and shared library builds embedding a signature if needed.
0425      [Steve Henson]
0426 
0427   *) Output TLS supported curves in preference order instead of numerical
0428      order. This is currently hardcoded for the highest order curves first.
0429      This should be configurable so applications can judge speed vs strength.
0430      [Steve Henson]
0431 
0432   *) Add TLS v1.2 server support for client authentication. 
0433      [Steve Henson]
0434 
0435   *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
0436      and enable MD5.
0437      [Steve Henson]
0438 
0439   *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
0440      FIPS modules versions.
0441      [Steve Henson]
0442 
0443   *) Add TLS v1.2 client side support for client authentication. Keep cache
0444      of handshake records longer as we don't know the hash algorithm to use
0445      until after the certificate request message is received.
0446      [Steve Henson]
0447 
0448   *) Initial TLS v1.2 client support. Add a default signature algorithms
0449      extension including all the algorithms we support. Parse new signature
0450      format in client key exchange. Relax some ECC signing restrictions for
0451      TLS v1.2 as indicated in RFC5246.
0452      [Steve Henson]
0453 
0454   *) Add server support for TLS v1.2 signature algorithms extension. Switch
0455      to new signature format when needed using client digest preference.
0456      All server ciphersuites should now work correctly in TLS v1.2. No client
0457      support yet and no support for client certificates.
0458      [Steve Henson]
0459 
0460   *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
0461      to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
0462      ciphersuites. At present only RSA key exchange ciphersuites work with
0463      TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
0464      SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
0465      and version checking.
0466      [Steve Henson]
0467 
0468   *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
0469      with this defined it will not be affected by any changes to ssl internal
0470      structures. Add several utility functions to allow openssl application
0471      to work with OPENSSL_NO_SSL_INTERN defined.
0472      [Steve Henson]
0473 
0474   *) Add SRP support.
0475      [Tom Wu <tjw@cs.stanford.edu> and Ben Laurie]
0476 
0477   *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
0478      [Steve Henson]
0479 
0480   *) Permit abbreviated handshakes when renegotiating using the function
0481      SSL_renegotiate_abbreviated().
0482      [Robin Seggelmann <seggelmann@fh-muenster.de>]
0483 
0484   *) Add call to ENGINE_register_all_complete() to
0485      ENGINE_load_builtin_engines(), so some implementations get used
0486      automatically instead of needing explicit application support.
0487      [Steve Henson]
0488 
0489   *) Add support for TLS key exporter as described in RFC5705.
0490      [Robin Seggelmann <seggelmann@fh-muenster.de>, Steve Henson]
0491 
0492   *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
0493      a few changes are required:
0494 
0495        Add SSL_OP_NO_TLSv1_1 flag.
0496        Add TLSv1_1 methods.
0497        Update version checking logic to handle version 1.1.
0498        Add explicit IV handling (ported from DTLS code).
0499        Add command line options to s_client/s_server.
0500      [Steve Henson]
0501 
0502  Changes between 1.0.0j and 1.0.0k [5 Feb 2013]
0503 
0504   *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
0505 
0506      This addresses the flaw in CBC record processing discovered by 
0507      Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
0508      at: http://www.isg.rhul.ac.uk/tls/     
0509 
0510      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
0511      Security Group at Royal Holloway, University of London
0512      (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
0513      Emilia Käsper for the initial patch.
0514      (CVE-2013-0169)
0515      [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
0516 
0517   *) Return an error when checking OCSP signatures when key is NULL.
0518      This fixes a DoS attack. (CVE-2013-0166)
0519      [Steve Henson]
0520 
0521   *) Call OCSP Stapling callback after ciphersuite has been chosen, so
0522      the right response is stapled. Also change SSL_get_certificate()
0523      so it returns the certificate actually sent.
0524      See http://rt.openssl.org/Ticket/Display.html?id=2836.
0525      (This is a backport)
0526      [Rob Stradling <rob.stradling@comodo.com>]
0527 
0528   *) Fix possible deadlock when decoding public keys.
0529      [Steve Henson]
0530 
0531  Changes between 1.0.0i and 1.0.0j [10 May 2012]
0532 
0533   [NB: OpenSSL 1.0.0i and later 1.0.0 patch levels were released after
0534   OpenSSL 1.0.1.]
0535 
0536   *) Sanity check record length before skipping explicit IV in DTLS
0537      to fix DoS attack.
0538 
0539      Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
0540      fuzzing as a service testing platform.
0541      (CVE-2012-2333)
0542      [Steve Henson]
0543 
0544   *) Initialise tkeylen properly when encrypting CMS messages.
0545      Thanks to Solar Designer of Openwall for reporting this issue.
0546      [Steve Henson]
0547 
0548  Changes between 1.0.0h and 1.0.0i [19 Apr 2012]
0549 
0550   *) Check for potentially exploitable overflows in asn1_d2i_read_bio
0551      BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
0552      in CRYPTO_realloc_clean.
0553 
0554      Thanks to Tavis Ormandy, Google Security Team, for discovering this
0555      issue and to Adam Langley <agl@chromium.org> for fixing it.
0556      (CVE-2012-2110)
0557      [Adam Langley (Google), Tavis Ormandy, Google Security Team]
0558 
0559  Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
0560 
0561   *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
0562      in CMS and PKCS7 code. When RSA decryption fails use a random key for
0563      content decryption and always return the same error. Note: this attack
0564      needs on average 2^20 messages so it only affects automated senders. The
0565      old behaviour can be reenabled in the CMS code by setting the
0566      CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
0567      an MMA defence is not necessary.
0568      Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
0569      this issue. (CVE-2012-0884)
0570      [Steve Henson]
0571 
0572   *) Fix CVE-2011-4619: make sure we really are receiving a 
0573      client hello before rejecting multiple SGC restarts. Thanks to
0574      Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
0575      [Steve Henson]
0576 
0577  Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
0578 
0579   *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
0580      Thanks to Antonio Martin, Enterprise Secure Access Research and
0581      Development, Cisco Systems, Inc. for discovering this bug and
0582      preparing a fix. (CVE-2012-0050)
0583      [Antonio Martin]
0584 
0585  Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
0586 
0587   *) Nadhem Alfardan and Kenny Paterson have discovered an extension
0588      of the Vaudenay padding oracle attack on CBC mode encryption
0589      which enables an efficient plaintext recovery attack against
0590      the OpenSSL implementation of DTLS. Their attack exploits timing
0591      differences arising during decryption processing. A research
0592      paper describing this attack can be found at:
0593                   http://www.isg.rhul.ac.uk/~kp/dtls.pdf
0594      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
0595      Security Group at Royal Holloway, University of London
0596      (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
0597      <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
0598      for preparing the fix. (CVE-2011-4108)
0599      [Robin Seggelmann, Michael Tuexen]
0600 
0601   *) Clear bytes used for block padding of SSL 3.0 records.
0602      (CVE-2011-4576)
0603      [Adam Langley (Google)]
0604 
0605   *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
0606      Kadianakis <desnacked@gmail.com> for discovering this issue and
0607      Adam Langley for preparing the fix. (CVE-2011-4619)
0608      [Adam Langley (Google)]
0609 
0610   *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027)
0611      [Andrey Kulikov <amdeich@gmail.com>]
0612 
0613   *) Prevent malformed RFC3779 data triggering an assertion failure.
0614      Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
0615      and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577)
0616      [Rob Austein <sra@hactrn.net>]
0617 
0618   *) Improved PRNG seeding for VOS.
0619      [Paul Green <Paul.Green@stratus.com>]
0620 
0621   *) Fix ssl_ciph.c set-up race.
0622      [Adam Langley (Google)]
0623 
0624   *) Fix spurious failures in ecdsatest.c.
0625      [Emilia Käsper (Google)]
0626 
0627   *) Fix the BIO_f_buffer() implementation (which was mixing different
0628      interpretations of the '..._len' fields).
0629      [Adam Langley (Google)]
0630 
0631   *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
0632      BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
0633      threads won't reuse the same blinding coefficients.
0634 
0635      This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
0636      lock to call BN_BLINDING_invert_ex, and avoids one use of
0637      BN_BLINDING_update for each BN_BLINDING structure (previously,
0638      the last update always remained unused).
0639      [Emilia Käsper (Google)]
0640 
0641   *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
0642      [Bob Buckholz (Google)]
0643 
0644  Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
0645 
0646   *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
0647      by initialising X509_STORE_CTX properly. (CVE-2011-3207)
0648      [Kaspar Brand <ossl@velox.ch>]
0649 
0650   *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
0651      for multi-threaded use of ECDH. (CVE-2011-3210)
0652      [Adam Langley (Google)]
0653 
0654   *) Fix x509_name_ex_d2i memory leak on bad inputs.
0655      [Bodo Moeller]
0656 
0657   *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
0658      signature public key algorithm by using OID xref utilities instead.
0659      Before this you could only use some ECC ciphersuites with SHA1 only.
0660      [Steve Henson]
0661 
0662   *) Add protection against ECDSA timing attacks as mentioned in the paper
0663      by Billy Bob Brumley and Nicola Tuveri, see:
0664 
0665         http://eprint.iacr.org/2011/232.pdf
0666 
0667      [Billy Bob Brumley and Nicola Tuveri]
0668 
0669  Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
0670 
0671   *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
0672      [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
0673 
0674   *) Fix bug in string printing code: if *any* escaping is enabled we must
0675      escape the escape character (backslash) or the resulting string is
0676      ambiguous.
0677      [Steve Henson]
0678 
0679  Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
0680 
0681   *) Disable code workaround for ancient and obsolete Netscape browsers
0682      and servers: an attacker can use it in a ciphersuite downgrade attack.
0683      Thanks to Martin Rex for discovering this bug. CVE-2010-4180
0684      [Steve Henson]
0685 
0686   *) Fixed J-PAKE implementation error, originally discovered by
0687      Sebastien Martini, further info and confirmation from Stefan
0688      Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
0689      [Ben Laurie]
0690 
0691  Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
0692 
0693   *) Fix extension code to avoid race conditions which can result in a buffer
0694      overrun vulnerability: resumed sessions must not be modified as they can
0695      be shared by multiple threads. CVE-2010-3864
0696      [Steve Henson]
0697 
0698   *) Fix WIN32 build system to correctly link an ENGINE directory into
0699      a DLL. 
0700      [Steve Henson]
0701 
0702  Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
0703 
0704   *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 
0705      (CVE-2010-1633)
0706      [Steve Henson, Peter-Michael Hager <hager@dortmund.net>]
0707 
0708  Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
0709 
0710   *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
0711      context. The operation can be customised via the ctrl mechanism in
0712      case ENGINEs want to include additional functionality.
0713      [Steve Henson]
0714 
0715   *) Tolerate yet another broken PKCS#8 key format: private key value negative.
0716      [Steve Henson]
0717 
0718   *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
0719      output hashes compatible with older versions of OpenSSL.
0720      [Willy Weisz <weisz@vcpc.univie.ac.at>]
0721 
0722   *) Fix compression algorithm handling: if resuming a session use the
0723      compression algorithm of the resumed session instead of determining
0724      it from client hello again. Don't allow server to change algorithm.
0725      [Steve Henson]
0726 
0727   *) Add load_crls() function to apps tidying load_certs() too. Add option
0728      to verify utility to allow additional CRLs to be included.
0729      [Steve Henson]
0730 
0731   *) Update OCSP request code to permit adding custom headers to the request:
0732      some responders need this.
0733      [Steve Henson]
0734 
0735   *) The function EVP_PKEY_sign() returns <=0 on error: check return code
0736      correctly.
0737      [Julia Lawall <julia@diku.dk>]
0738 
0739   *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
0740      needlessly dereferenced structures, used obsolete functions and
0741      didn't handle all updated verify codes correctly.
0742      [Steve Henson]
0743 
0744   *) Disable MD2 in the default configuration.
0745      [Steve Henson]
0746 
0747   *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
0748      indicate the initial BIO being pushed or popped. This makes it possible
0749      to determine whether the BIO is the one explicitly called or as a result
0750      of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
0751      it handles reference counts correctly and doesn't zero out the I/O bio
0752      when it is not being explicitly popped. WARNING: applications which
0753      included workarounds for the old buggy behaviour will need to be modified
0754      or they could free up already freed BIOs.
0755      [Steve Henson]
0756 
0757   *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
0758      renaming to all platforms (within the 0.9.8 branch, this was
0759      done conditionally on Netware platforms to avoid a name clash).
0760      [Guenter <lists@gknw.net>]
0761 
0762   *) Add ECDHE and PSK support to DTLS.
0763      [Michael Tuexen <tuexen@fh-muenster.de>]
0764 
0765   *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
0766      be used on C++.
0767      [Steve Henson]
0768 
0769   *) Add "missing" function EVP_MD_flags() (without this the only way to
0770      retrieve a digest flags is by accessing the structure directly. Update
0771      EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
0772      or cipher is registered as in the "from" argument. Print out all
0773      registered digests in the dgst usage message instead of manually 
0774      attempting to work them out.
0775      [Steve Henson]
0776 
0777   *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
0778      this allows the use of compression and extensions. Change default cipher
0779      string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
0780      by default unless an application cipher string requests it.
0781      [Steve Henson]
0782 
0783   *) Alter match criteria in PKCS12_parse(). It used to try to use local
0784      key ids to find matching certificates and keys but some PKCS#12 files
0785      don't follow the (somewhat unwritten) rules and this strategy fails.
0786      Now just gather all certificates together and the first private key
0787      then look for the first certificate that matches the key.
0788      [Steve Henson]
0789 
0790   *) Support use of registered digest and cipher names for dgst and cipher
0791      commands instead of having to add each one as a special case. So now
0792      you can do:
0793 
0794         openssl sha256 foo
0795 
0796      as well as:
0797 
0798         openssl dgst -sha256 foo
0799 
0800      and this works for ENGINE based algorithms too.
0801 
0802      [Steve Henson]
0803 
0804   *) Update Gost ENGINE to support parameter files.
0805      [Victor B. Wagner <vitus@cryptocom.ru>]
0806 
0807   *) Support GeneralizedTime in ca utility. 
0808      [Oliver Martin <oliver@volatilevoid.net>, Steve Henson]
0809 
0810   *) Enhance the hash format used for certificate directory links. The new
0811      form uses the canonical encoding (meaning equivalent names will work
0812      even if they aren't identical) and uses SHA1 instead of MD5. This form
0813      is incompatible with the older format and as a result c_rehash should
0814      be used to rebuild symbolic links.
0815      [Steve Henson]
0816 
0817   *) Make PKCS#8 the default write format for private keys, replacing the
0818      traditional format. This form is standardised, more secure and doesn't
0819      include an implicit MD5 dependency.
0820      [Steve Henson]
0821 
0822   *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
0823      committed to OpenSSL should pass this lot as a minimum.
0824      [Steve Henson]
0825 
0826   *) Add session ticket override functionality for use by EAP-FAST.
0827      [Jouni Malinen <j@w1.fi>]
0828 
0829   *) Modify HMAC functions to return a value. Since these can be implemented
0830      in an ENGINE errors can occur.
0831      [Steve Henson]
0832 
0833   *) Type-checked OBJ_bsearch_ex.
0834      [Ben Laurie]
0835 
0836   *) Type-checked OBJ_bsearch. Also some constification necessitated
0837      by type-checking.  Still to come: TXT_DB, bsearch(?),
0838      OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
0839      CONF_VALUE.
0840      [Ben Laurie]
0841 
0842   *) New function OPENSSL_gmtime_adj() to add a specific number of days and
0843      seconds to a tm structure directly, instead of going through OS
0844      specific date routines. This avoids any issues with OS routines such
0845      as the year 2038 bug. New *_adj() functions for ASN1 time structures
0846      and X509_time_adj_ex() to cover the extended range. The existing
0847      X509_time_adj() is still usable and will no longer have any date issues.
0848      [Steve Henson]
0849 
0850   *) Delta CRL support. New use deltas option which will attempt to locate
0851      and search any appropriate delta CRLs available.
0852 
0853      This work was sponsored by Google.
0854      [Steve Henson]
0855 
0856   *) Support for CRLs partitioned by reason code. Reorganise CRL processing
0857      code and add additional score elements. Validate alternate CRL paths
0858      as part of the CRL checking and indicate a new error "CRL path validation
0859      error" in this case. Applications wanting additional details can use
0860      the verify callback and check the new "parent" field. If this is not
0861      NULL CRL path validation is taking place. Existing applications wont
0862      see this because it requires extended CRL support which is off by
0863      default.
0864 
0865      This work was sponsored by Google.
0866      [Steve Henson]
0867 
0868   *) Support for freshest CRL extension.
0869 
0870      This work was sponsored by Google.
0871      [Steve Henson]
0872 
0873   *) Initial indirect CRL support. Currently only supported in the CRLs
0874      passed directly and not via lookup. Process certificate issuer
0875      CRL entry extension and lookup CRL entries by bother issuer name
0876      and serial number. Check and process CRL issuer entry in IDP extension.
0877 
0878      This work was sponsored by Google.
0879      [Steve Henson]
0880 
0881   *) Add support for distinct certificate and CRL paths. The CRL issuer
0882      certificate is validated separately in this case. Only enabled if
0883      an extended CRL support flag is set: this flag will enable additional
0884      CRL functionality in future.
0885 
0886      This work was sponsored by Google.
0887      [Steve Henson]
0888 
0889   *) Add support for policy mappings extension.
0890 
0891      This work was sponsored by Google.
0892      [Steve Henson]
0893 
0894   *) Fixes to pathlength constraint, self issued certificate handling,
0895      policy processing to align with RFC3280 and PKITS tests.
0896 
0897      This work was sponsored by Google.
0898      [Steve Henson]
0899 
0900   *) Support for name constraints certificate extension. DN, email, DNS
0901      and URI types are currently supported.
0902 
0903      This work was sponsored by Google.
0904      [Steve Henson]
0905 
0906   *) To cater for systems that provide a pointer-based thread ID rather
0907      than numeric, deprecate the current numeric thread ID mechanism and
0908      replace it with a structure and associated callback type. This
0909      mechanism allows a numeric "hash" to be extracted from a thread ID in
0910      either case, and on platforms where pointers are larger than 'long',
0911      mixing is done to help ensure the numeric 'hash' is usable even if it
0912      can't be guaranteed unique. The default mechanism is to use "&errno"
0913      as a pointer-based thread ID to distinguish between threads.
0914 
0915      Applications that want to provide their own thread IDs should now use
0916      CRYPTO_THREADID_set_callback() to register a callback that will call
0917      either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
0918 
0919      Note that ERR_remove_state() is now deprecated, because it is tied
0920      to the assumption that thread IDs are numeric.  ERR_remove_state(0)
0921      to free the current thread's error state should be replaced by
0922      ERR_remove_thread_state(NULL).
0923 
0924      (This new approach replaces the functions CRYPTO_set_idptr_callback(),
0925      CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
0926      OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
0927      application was previously providing a numeric thread callback that
0928      was inappropriate for distinguishing threads, then uniqueness might
0929      have been obtained with &errno that happened immediately in the
0930      intermediate development versions of OpenSSL; this is no longer the
0931      case, the numeric thread callback will now override the automatic use
0932      of &errno.)
0933      [Geoff Thorpe, with help from Bodo Moeller]
0934 
0935   *) Initial support for different CRL issuing certificates. This covers a
0936      simple case where the self issued certificates in the chain exist and
0937      the real CRL issuer is higher in the existing chain.
0938 
0939      This work was sponsored by Google.
0940      [Steve Henson]
0941 
0942   *) Removed effectively defunct crypto/store from the build.
0943      [Ben Laurie]
0944 
0945   *) Revamp of STACK to provide stronger type-checking. Still to come:
0946      TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
0947      ASN1_STRING, CONF_VALUE.
0948      [Ben Laurie]
0949 
0950   *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
0951      RAM on SSL connections.  This option can save about 34k per idle SSL.
0952      [Nick Mathewson]
0953 
0954   *) Revamp of LHASH to provide stronger type-checking. Still to come:
0955      STACK, TXT_DB, bsearch, qsort.
0956      [Ben Laurie]
0957 
0958   *) Initial support for Cryptographic Message Syntax (aka CMS) based
0959      on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
0960      support for data, signedData, compressedData, digestedData and
0961      encryptedData, envelopedData types included. Scripts to check against
0962      RFC4134 examples draft and interop and consistency checks of many
0963      content types and variants.
0964      [Steve Henson]
0965 
0966   *) Add options to enc utility to support use of zlib compression BIO.
0967      [Steve Henson]
0968 
0969   *) Extend mk1mf to support importing of options and assembly language
0970      files from Configure script, currently only included in VC-WIN32.
0971      The assembly language rules can now optionally generate the source
0972      files from the associated perl scripts.
0973      [Steve Henson]
0974 
0975   *) Implement remaining functionality needed to support GOST ciphersuites.
0976      Interop testing has been performed using CryptoPro implementations.
0977      [Victor B. Wagner <vitus@cryptocom.ru>]
0978 
0979   *) s390x assembler pack.
0980      [Andy Polyakov]
0981 
0982   *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
0983      "family."
0984      [Andy Polyakov]
0985 
0986   *) Implement Opaque PRF Input TLS extension as specified in
0987      draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
0988      official specification yet and no extension type assignment by
0989      IANA exists, this extension (for now) will have to be explicitly
0990      enabled when building OpenSSL by providing the extension number
0991      to use.  For example, specify an option
0992 
0993          -DTLSEXT_TYPE_opaque_prf_input=0x9527
0994 
0995      to the "config" or "Configure" script to enable the extension,
0996      assuming extension number 0x9527 (which is a completely arbitrary
0997      and unofficial assignment based on the MD5 hash of the Internet
0998      Draft).  Note that by doing so, you potentially lose
0999      interoperability with other TLS implementations since these might
1000      be using the same extension number for other purposes.
1001 
1002      SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
1003      opaque PRF input value to use in the handshake.  This will create
1004      an interal copy of the length-'len' string at 'src', and will
1005      return non-zero for success.
1006 
1007      To get more control and flexibility, provide a callback function
1008      by using
1009 
1010           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
1011           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
1012 
1013      where
1014 
1015           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
1016           void *arg;
1017 
1018      Callback function 'cb' will be called in handshakes, and is
1019      expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
1020      Argument 'arg' is for application purposes (the value as given to
1021      SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
1022      be provided to the callback function).  The callback function
1023      has to return non-zero to report success: usually 1 to use opaque
1024      PRF input just if possible, or 2 to enforce use of the opaque PRF
1025      input.  In the latter case, the library will abort the handshake
1026      if opaque PRF input is not successfully negotiated.
1027 
1028      Arguments 'peerinput' and 'len' given to the callback function
1029      will always be NULL and 0 in the case of a client.  A server will
1030      see the client's opaque PRF input through these variables if
1031      available (NULL and 0 otherwise).  Note that if the server
1032      provides an opaque PRF input, the length must be the same as the
1033      length of the client's opaque PRF input.
1034 
1035      Note that the callback function will only be called when creating
1036      a new session (session resumption can resume whatever was
1037      previously negotiated), and will not be called in SSL 2.0
1038      handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
1039      SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
1040      for applications that need to enforce opaque PRF input.
1041 
1042      [Bodo Moeller]
1043 
1044   *) Update ssl code to support digests other than SHA1+MD5 for handshake
1045      MAC. 
1046 
1047      [Victor B. Wagner <vitus@cryptocom.ru>]
1048 
1049   *) Add RFC4507 support to OpenSSL. This includes the corrections in
1050      RFC4507bis. The encrypted ticket format is an encrypted encoded
1051      SSL_SESSION structure, that way new session features are automatically
1052      supported.
1053 
1054      If a client application caches session in an SSL_SESSION structure
1055      support is transparent because tickets are now stored in the encoded
1056      SSL_SESSION.
1057      
1058      The SSL_CTX structure automatically generates keys for ticket
1059      protection in servers so again support should be possible
1060      with no application modification.
1061 
1062      If a client or server wishes to disable RFC4507 support then the option
1063      SSL_OP_NO_TICKET can be set.
1064 
1065      Add a TLS extension debugging callback to allow the contents of any client
1066      or server extensions to be examined.
1067 
1068      This work was sponsored by Google.
1069      [Steve Henson]
1070 
1071   *) Final changes to avoid use of pointer pointer casts in OpenSSL.
1072      OpenSSL should now compile cleanly on gcc 4.2
1073      [Peter Hartley <pdh@utter.chaos.org.uk>, Steve Henson]
1074 
1075   *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
1076      support including streaming MAC support: this is required for GOST
1077      ciphersuite support.
1078      [Victor B. Wagner <vitus@cryptocom.ru>, Steve Henson]
1079 
1080   *) Add option -stream to use PKCS#7 streaming in smime utility. New
1081      function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
1082      to output in BER and PEM format.
1083      [Steve Henson]
1084 
1085   *) Experimental support for use of HMAC via EVP_PKEY interface. This
1086      allows HMAC to be handled via the EVP_DigestSign*() interface. The
1087      EVP_PKEY "key" in this case is the HMAC key, potentially allowing
1088      ENGINE support for HMAC keys which are unextractable. New -mac and
1089      -macopt options to dgst utility.
1090      [Steve Henson]
1091 
1092   *) New option -sigopt to dgst utility. Update dgst to use
1093      EVP_Digest{Sign,Verify}*. These two changes make it possible to use
1094      alternative signing paramaters such as X9.31 or PSS in the dgst 
1095      utility.
1096      [Steve Henson]
1097 
1098   *) Change ssl_cipher_apply_rule(), the internal function that does
1099      the work each time a ciphersuite string requests enabling
1100      ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
1101      removing ("!foo+bar") a class of ciphersuites: Now it maintains
1102      the order of disabled ciphersuites such that those ciphersuites
1103      that most recently went from enabled to disabled not only stay
1104      in order with respect to each other, but also have higher priority
1105      than other disabled ciphersuites the next time ciphersuites are
1106      enabled again.
1107 
1108      This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
1109      the same ciphersuites as with "HIGH" alone, but in a specific
1110      order where the PSK ciphersuites come first (since they are the
1111      most recently disabled ciphersuites when "HIGH" is parsed).
1112 
1113      Also, change ssl_create_cipher_list() (using this new
1114      funcionality) such that between otherwise identical
1115      cihpersuites, ephemeral ECDH is preferred over ephemeral DH in
1116      the default order.
1117      [Bodo Moeller]
1118 
1119   *) Change ssl_create_cipher_list() so that it automatically
1120      arranges the ciphersuites in reasonable order before starting
1121      to process the rule string.  Thus, the definition for "DEFAULT"
1122      (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
1123      remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH".
1124      This makes it much easier to arrive at a reasonable default order
1125      in applications for which anonymous ciphers are OK (meaning
1126      that you can't actually use DEFAULT).
1127      [Bodo Moeller; suggested by Victor Duchovni]
1128 
1129   *) Split the SSL/TLS algorithm mask (as used for ciphersuite string
1130      processing) into multiple integers instead of setting
1131      "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
1132      "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
1133      (These masks as well as the individual bit definitions are hidden
1134      away into the non-exported interface ssl/ssl_locl.h, so this
1135      change to the definition of the SSL_CIPHER structure shouldn't
1136      affect applications.)  This give us more bits for each of these
1137      categories, so there is no longer a need to coagulate AES128 and
1138      AES256 into a single algorithm bit, and to coagulate Camellia128
1139      and Camellia256 into a single algorithm bit, which has led to all
1140      kinds of kludges.
1141 
1142      Thus, among other things, the kludge introduced in 0.9.7m and
1143      0.9.8e for masking out AES256 independently of AES128 or masking
1144      out Camellia256 independently of AES256 is not needed here in 0.9.9.
1145 
1146      With the change, we also introduce new ciphersuite aliases that
1147      so far were missing: "AES128", "AES256", "CAMELLIA128", and
1148      "CAMELLIA256".
1149      [Bodo Moeller]
1150 
1151   *) Add support for dsa-with-SHA224 and dsa-with-SHA256.
1152      Use the leftmost N bytes of the signature input if the input is
1153      larger than the prime q (with N being the size in bytes of q).
1154      [Nils Larsch]
1155 
1156   *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
1157      it yet and it is largely untested.
1158      [Steve Henson]
1159 
1160   *) Add support for the ecdsa-with-SHA224/256/384/512 signature types.
1161      [Nils Larsch]
1162 
1163   *) Initial incomplete changes to avoid need for function casts in OpenSSL
1164      some compilers (gcc 4.2 and later) reject their use. Safestack is
1165      reimplemented.  Update ASN1 to avoid use of legacy functions. 
1166      [Steve Henson]
1167 
1168   *) Win32/64 targets are linked with Winsock2.
1169      [Andy Polyakov]
1170 
1171   *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
1172      to external functions. This can be used to increase CRL handling 
1173      efficiency especially when CRLs are very large by (for example) storing
1174      the CRL revoked certificates in a database.
1175      [Steve Henson]
1176 
1177   *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so
1178      new CRLs added to a directory can be used. New command line option
1179      -verify_return_error to s_client and s_server. This causes real errors
1180      to be returned by the verify callback instead of carrying on no matter
1181      what. This reflects the way a "real world" verify callback would behave.
1182      [Steve Henson]
1183 
1184   *) GOST engine, supporting several GOST algorithms and public key formats.
1185      Kindly donated by Cryptocom.
1186      [Cryptocom]
1187 
1188   *) Partial support for Issuing Distribution Point CRL extension. CRLs
1189      partitioned by DP are handled but no indirect CRL or reason partitioning
1190      (yet). Complete overhaul of CRL handling: now the most suitable CRL is
1191      selected via a scoring technique which handles IDP and AKID in CRLs.
1192      [Steve Henson]
1193 
1194   *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
1195      will ultimately be used for all verify operations: this will remove the
1196      X509_STORE dependency on certificate verification and allow alternative
1197      lookup methods.  X509_STORE based implementations of these two callbacks.
1198      [Steve Henson]
1199 
1200   *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
1201      Modify get_crl() to find a valid (unexpired) CRL if possible.
1202      [Steve Henson]
1203 
1204   *) New function X509_CRL_match() to check if two CRLs are identical. Normally
1205      this would be called X509_CRL_cmp() but that name is already used by
1206      a function that just compares CRL issuer names. Cache several CRL 
1207      extensions in X509_CRL structure and cache CRLDP in X509.
1208      [Steve Henson]
1209 
1210   *) Store a "canonical" representation of X509_NAME structure (ASN1 Name)
1211      this maps equivalent X509_NAME structures into a consistent structure.
1212      Name comparison can then be performed rapidly using memcmp().
1213      [Steve Henson]
1214 
1215   *) Non-blocking OCSP request processing. Add -timeout option to ocsp 
1216      utility.
1217      [Steve Henson]
1218 
1219   *) Allow digests to supply their own micalg string for S/MIME type using
1220      the ctrl EVP_MD_CTRL_MICALG.
1221      [Steve Henson]
1222 
1223   *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the
1224      EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
1225      ctrl. It can then customise the structure before and/or after signing
1226      if necessary.
1227      [Steve Henson]
1228 
1229   *) New function OBJ_add_sigid() to allow application defined signature OIDs
1230      to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
1231      to free up any added signature OIDs.
1232      [Steve Henson]
1233 
1234   *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
1235      EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
1236      digest and cipher tables. New options added to openssl utility:
1237      list-message-digest-algorithms and list-cipher-algorithms.
1238      [Steve Henson]
1239 
1240   *) Change the array representation of binary polynomials: the list
1241      of degrees of non-zero coefficients is now terminated with -1.
1242      Previously it was terminated with 0, which was also part of the
1243      value; thus, the array representation was not applicable to
1244      polynomials where t^0 has coefficient zero.  This change makes
1245      the array representation useful in a more general context.
1246      [Douglas Stebila]
1247 
1248   *) Various modifications and fixes to SSL/TLS cipher string
1249      handling.  For ECC, the code now distinguishes between fixed ECDH
1250      with RSA certificates on the one hand and with ECDSA certificates
1251      on the other hand, since these are separate ciphersuites.  The
1252      unused code for Fortezza ciphersuites has been removed.
1253 
1254      For consistency with EDH, ephemeral ECDH is now called "EECDH"
1255      (not "ECDHE").  For consistency with the code for DH
1256      certificates, use of ECDH certificates is now considered ECDH
1257      authentication, not RSA or ECDSA authentication (the latter is
1258      merely the CA's signing algorithm and not actively used in the
1259      protocol).
1260 
1261      The temporary ciphersuite alias "ECCdraft" is no longer
1262      available, and ECC ciphersuites are no longer excluded from "ALL"
1263      and "DEFAULT".  The following aliases now exist for RFC 4492
1264      ciphersuites, most of these by analogy with the DH case:
1265 
1266          kECDHr   - ECDH cert, signed with RSA
1267          kECDHe   - ECDH cert, signed with ECDSA
1268          kECDH    - ECDH cert (signed with either RSA or ECDSA)
1269          kEECDH   - ephemeral ECDH
1270          ECDH     - ECDH cert or ephemeral ECDH
1271 
1272          aECDH    - ECDH cert
1273          aECDSA   - ECDSA cert
1274          ECDSA    - ECDSA cert
1275 
1276          AECDH    - anonymous ECDH
1277          EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
1278 
1279      [Bodo Moeller]
1280 
1281   *) Add additional S/MIME capabilities for AES and GOST ciphers if supported.
1282      Use correct micalg parameters depending on digest(s) in signed message.
1283      [Steve Henson]
1284 
1285   *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
1286      an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
1287      [Steve Henson]
1288 
1289   *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
1290      an engine to register a method. Add ENGINE lookups for methods and
1291      functional reference processing.
1292      [Steve Henson]
1293 
1294   *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of
1295      EVP_{Sign,Verify}* which allow an application to customise the signature
1296      process.
1297      [Steve Henson]
1298 
1299   *) New -resign option to smime utility. This adds one or more signers
1300      to an existing PKCS#7 signedData structure. Also -md option to use an
1301      alternative message digest algorithm for signing.
1302      [Steve Henson]
1303 
1304   *) Tidy up PKCS#7 routines and add new functions to make it easier to
1305      create PKCS7 structures containing multiple signers. Update smime
1306      application to support multiple signers.
1307      [Steve Henson]
1308 
1309   *) New -macalg option to pkcs12 utility to allow setting of an alternative
1310      digest MAC.
1311      [Steve Henson]
1312 
1313   *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
1314      Reorganize PBE internals to lookup from a static table using NIDs,
1315      add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
1316      EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
1317      PRF which will be automatically used with PBES2.
1318      [Steve Henson]
1319 
1320   *) Replace the algorithm specific calls to generate keys in "req" with the
1321      new API.
1322      [Steve Henson]
1323 
1324   *) Update PKCS#7 enveloped data routines to use new API. This is now
1325      supported by any public key method supporting the encrypt operation. A
1326      ctrl is added to allow the public key algorithm to examine or modify
1327      the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
1328      a no op.
1329      [Steve Henson]
1330 
1331   *) Add a ctrl to asn1 method to allow a public key algorithm to express
1332      a default digest type to use. In most cases this will be SHA1 but some
1333      algorithms (such as GOST) need to specify an alternative digest. The
1334      return value indicates how strong the prefernce is 1 means optional and
1335      2 is mandatory (that is it is the only supported type). Modify
1336      ASN1_item_sign() to accept a NULL digest argument to indicate it should
1337      use the default md. Update openssl utilities to use the default digest
1338      type for signing if it is not explicitly indicated.
1339      [Steve Henson]
1340 
1341   *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 
1342      EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
1343      signing method from the key type. This effectively removes the link
1344      between digests and public key types.
1345      [Steve Henson]
1346 
1347   *) Add an OID cross reference table and utility functions. Its purpose is to
1348      translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
1349      rsaEncryption. This will allow some of the algorithm specific hackery
1350      needed to use the correct OID to be removed. 
1351      [Steve Henson]
1352 
1353   *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
1354      structures for PKCS7_sign(). They are now set up by the relevant public
1355      key ASN1 method.
1356      [Steve Henson]
1357 
1358   *) Add provisional EC pkey method with support for ECDSA and ECDH.
1359      [Steve Henson]
1360 
1361   *) Add support for key derivation (agreement) in the API, DH method and
1362      pkeyutl.
1363      [Steve Henson]
1364 
1365   *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
1366      public and private key formats. As a side effect these add additional 
1367      command line functionality not previously available: DSA signatures can be
1368      generated and verified using pkeyutl and DH key support and generation in
1369      pkey, genpkey.
1370      [Steve Henson]
1371 
1372   *) BeOS support.
1373      [Oliver Tappe <zooey@hirschkaefer.de>]
1374 
1375   *) New make target "install_html_docs" installs HTML renditions of the
1376      manual pages.
1377      [Oliver Tappe <zooey@hirschkaefer.de>]
1378 
1379   *) New utility "genpkey" this is analagous to "genrsa" etc except it can
1380      generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
1381      support key and parameter generation and add initial key generation
1382      functionality for RSA.
1383      [Steve Henson]
1384 
1385   *) Add functions for main EVP_PKEY_method operations. The undocumented
1386      functions EVP_PKEY_{encrypt,decrypt} have been renamed to
1387      EVP_PKEY_{encrypt,decrypt}_old. 
1388      [Steve Henson]
1389 
1390   *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public
1391      key API, doesn't do much yet.
1392      [Steve Henson]
1393 
1394   *) New function EVP_PKEY_asn1_get0_info() to retrieve information about
1395      public key algorithms. New option to openssl utility:
1396      "list-public-key-algorithms" to print out info.
1397      [Steve Henson]
1398 
1399   *) Implement the Supported Elliptic Curves Extension for
1400      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
1401      [Douglas Stebila]
1402 
1403   *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
1404      EVP_CIPHER structures to avoid later problems in EVP_cleanup().
1405      [Steve Henson]
1406 
1407   *) New utilities pkey and pkeyparam. These are similar to algorithm specific
1408      utilities such as rsa, dsa, dsaparam etc except they process any key
1409      type.
1410      [Steve Henson]
1411 
1412   *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 
1413      functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
1414      EVP_PKEY_print_param() to print public key data from an EVP_PKEY
1415      structure.
1416      [Steve Henson]
1417 
1418   *) Initial support for pluggable public key ASN1.
1419      De-spaghettify the public key ASN1 handling. Move public and private
1420      key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
1421      algorithm specific handling to a single module within the relevant
1422      algorithm directory. Add functions to allow (near) opaque processing
1423      of public and private key structures.
1424      [Steve Henson]
1425 
1426   *) Implement the Supported Point Formats Extension for
1427      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
1428      [Douglas Stebila]
1429 
1430   *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
1431      for the psk identity [hint] and the psk callback functions to the
1432      SSL_SESSION, SSL and SSL_CTX structure.
1433      
1434      New ciphersuites:
1435          PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
1436          PSK-AES256-CBC-SHA
1437  
1438      New functions:
1439          SSL_CTX_use_psk_identity_hint
1440          SSL_get_psk_identity_hint
1441          SSL_get_psk_identity
1442          SSL_use_psk_identity_hint
1443 
1444      [Mika Kousa and Pasi Eronen of Nokia Corporation]
1445 
1446   *) Add RFC 3161 compliant time stamp request creation, response generation
1447      and response verification functionality.
1448      [Zoltán Glózik <zglozik@opentsa.org>, The OpenTSA Project]
1449 
1450   *) Add initial support for TLS extensions, specifically for the server_name
1451      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
1452      have new members for a host name.  The SSL data structure has an
1453      additional member SSL_CTX *initial_ctx so that new sessions can be
1454      stored in that context to allow for session resumption, even after the
1455      SSL has been switched to a new SSL_CTX in reaction to a client's
1456      server_name extension.
1457 
1458      New functions (subject to change):
1459 
1460          SSL_get_servername()
1461          SSL_get_servername_type()
1462          SSL_set_SSL_CTX()
1463 
1464      New CTRL codes and macros (subject to change):
1465 
1466          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
1467                                  - SSL_CTX_set_tlsext_servername_callback()
1468          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
1469                                       - SSL_CTX_set_tlsext_servername_arg()
1470          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
1471 
1472      openssl s_client has a new '-servername ...' option.
1473 
1474      openssl s_server has new options '-servername_host ...', '-cert2 ...',
1475      '-key2 ...', '-servername_fatal' (subject to change).  This allows
1476      testing the HostName extension for a specific single host name ('-cert'
1477      and '-key' remain fallbacks for handshakes without HostName
1478      negotiation).  If the unrecogninzed_name alert has to be sent, this by
1479      default is a warning; it becomes fatal with the '-servername_fatal'
1480      option.
1481 
1482      [Peter Sylvester,  Remy Allais, Christophe Renou]
1483 
1484   *) Whirlpool hash implementation is added.
1485      [Andy Polyakov]
1486 
1487   *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
1488      bn(64,32). Because of instruction set limitations it doesn't have
1489      any negative impact on performance. This was done mostly in order
1490      to make it possible to share assembler modules, such as bn_mul_mont
1491      implementations, between 32- and 64-bit builds without hassle.
1492      [Andy Polyakov]
1493 
1494   *) Move code previously exiled into file crypto/ec/ec2_smpt.c
1495      to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
1496      macro.
1497      [Bodo Moeller]
1498 
1499   *) New candidate for BIGNUM assembler implementation, bn_mul_mont,
1500      dedicated Montgomery multiplication procedure, is introduced.
1501      BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
1502      "64-bit" performance on certain 32-bit targets.
1503      [Andy Polyakov]
1504 
1505   *) New option SSL_OP_NO_COMP to disable use of compression selectively
1506      in SSL structures. New SSL ctrl to set maximum send fragment size. 
1507      Save memory by seeting the I/O buffer sizes dynamically instead of
1508      using the maximum available value.
1509      [Steve Henson]
1510 
1511   *) New option -V for 'openssl ciphers'. This prints the ciphersuite code
1512      in addition to the text details.
1513      [Bodo Moeller]
1514 
1515   *) Very, very preliminary EXPERIMENTAL support for printing of general
1516      ASN1 structures. This currently produces rather ugly output and doesn't
1517      handle several customised structures at all.
1518      [Steve Henson]
1519 
1520   *) Integrated support for PVK file format and some related formats such
1521      as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
1522      these in the 'rsa' and 'dsa' utilities.
1523      [Steve Henson]
1524 
1525   *) Support for PKCS#1 RSAPublicKey format on rsa utility command line.
1526      [Steve Henson]
1527 
1528   *) Remove the ancient ASN1_METHOD code. This was only ever used in one
1529      place for the (very old) "NETSCAPE" format certificates which are now
1530      handled using new ASN1 code equivalents.
1531      [Steve Henson]
1532 
1533   *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
1534      pointer and make the SSL_METHOD parameter in SSL_CTX_new,
1535      SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
1536      [Nils Larsch]
1537 
1538   *) Modify CRL distribution points extension code to print out previously
1539      unsupported fields. Enhance extension setting code to allow setting of
1540      all fields.
1541      [Steve Henson]
1542 
1543   *) Add print and set support for Issuing Distribution Point CRL extension.
1544      [Steve Henson]
1545 
1546   *) Change 'Configure' script to enable Camellia by default.
1547      [NTT]
1548 
1549  Changes between 0.9.8x and 0.9.8y [5 Feb 2013]
1550 
1551   *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
1552 
1553      This addresses the flaw in CBC record processing discovered by 
1554      Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
1555      at: http://www.isg.rhul.ac.uk/tls/     
1556 
1557      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
1558      Security Group at Royal Holloway, University of London
1559      (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
1560      Emilia Käsper for the initial patch.
1561      (CVE-2013-0169)
1562      [Emilia Käsper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
1563 
1564   *) Return an error when checking OCSP signatures when key is NULL.
1565      This fixes a DoS attack. (CVE-2013-0166)
1566      [Steve Henson]
1567 
1568   *) Call OCSP Stapling callback after ciphersuite has been chosen, so
1569      the right response is stapled. Also change SSL_get_certificate()
1570      so it returns the certificate actually sent.
1571      See http://rt.openssl.org/Ticket/Display.html?id=2836.
1572      (This is a backport)
1573      [Rob Stradling <rob.stradling@comodo.com>]
1574 
1575   *) Fix possible deadlock when decoding public keys.
1576      [Steve Henson]
1577 
1578  Changes between 0.9.8w and 0.9.8x [10 May 2012]
1579 
1580   *) Sanity check record length before skipping explicit IV in DTLS
1581      to fix DoS attack.
1582 
1583      Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
1584      fuzzing as a service testing platform.
1585      (CVE-2012-2333)
1586      [Steve Henson]
1587 
1588   *) Initialise tkeylen properly when encrypting CMS messages.
1589      Thanks to Solar Designer of Openwall for reporting this issue.
1590      [Steve Henson]
1591 
1592  Changes between 0.9.8v and 0.9.8w [23 Apr 2012]
1593 
1594   *) The fix for CVE-2012-2110 did not take into account that the 
1595      'len' argument to BUF_MEM_grow and BUF_MEM_grow_clean is an
1596      int in OpenSSL 0.9.8, making it still vulnerable. Fix by 
1597      rejecting negative len parameter. (CVE-2012-2131)
1598      [Tomas Hoger <thoger@redhat.com>]
1599 
1600  Changes between 0.9.8u and 0.9.8v [19 Apr 2012]
1601 
1602   *) Check for potentially exploitable overflows in asn1_d2i_read_bio
1603      BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
1604      in CRYPTO_realloc_clean.
1605 
1606      Thanks to Tavis Ormandy, Google Security Team, for discovering this
1607      issue and to Adam Langley <agl@chromium.org> for fixing it.
1608      (CVE-2012-2110)
1609      [Adam Langley (Google), Tavis Ormandy, Google Security Team]
1610 
1611  Changes between 0.9.8t and 0.9.8u [12 Mar 2012]
1612 
1613   *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
1614      in CMS and PKCS7 code. When RSA decryption fails use a random key for
1615      content decryption and always return the same error. Note: this attack
1616      needs on average 2^20 messages so it only affects automated senders. The
1617      old behaviour can be reenabled in the CMS code by setting the
1618      CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
1619      an MMA defence is not necessary.
1620      Thanks to Ivan Nestlerode <inestlerode@us.ibm.com> for discovering
1621      this issue. (CVE-2012-0884)
1622      [Steve Henson]
1623 
1624   *) Fix CVE-2011-4619: make sure we really are receiving a 
1625      client hello before rejecting multiple SGC restarts. Thanks to
1626      Ivan Nestlerode <inestlerode@us.ibm.com> for discovering this bug.
1627      [Steve Henson]
1628 
1629  Changes between 0.9.8s and 0.9.8t [18 Jan 2012]
1630 
1631   *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
1632      Thanks to Antonio Martin, Enterprise Secure Access Research and
1633      Development, Cisco Systems, Inc. for discovering this bug and
1634      preparing a fix. (CVE-2012-0050)
1635      [Antonio Martin]
1636 
1637  Changes between 0.9.8r and 0.9.8s [4 Jan 2012]
1638 
1639   *) Nadhem Alfardan and Kenny Paterson have discovered an extension
1640      of the Vaudenay padding oracle attack on CBC mode encryption
1641      which enables an efficient plaintext recovery attack against
1642      the OpenSSL implementation of DTLS. Their attack exploits timing
1643      differences arising during decryption processing. A research
1644      paper describing this attack can be found at:
1645                   http://www.isg.rhul.ac.uk/~kp/dtls.pdf
1646      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
1647      Security Group at Royal Holloway, University of London
1648      (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
1649      <seggelmann@fh-muenster.de> and Michael Tuexen <tuexen@fh-muenster.de>
1650      for preparing the fix. (CVE-2011-4108)
1651      [Robin Seggelmann, Michael Tuexen]
1652 
1653   *) Stop policy check failure freeing same buffer twice. (CVE-2011-4109)
1654      [Ben Laurie, Kasper <ekasper@google.com>]
1655 
1656   *) Clear bytes used for block padding of SSL 3.0 records.
1657      (CVE-2011-4576)
1658      [Adam Langley (Google)]
1659 
1660   *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
1661      Kadianakis <desnacked@gmail.com> for discovering this issue and
1662      Adam Langley for preparing the fix. (CVE-2011-4619)
1663      [Adam Langley (Google)]
1664  
1665   *) Prevent malformed RFC3779 data triggering an assertion failure.
1666      Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
1667      and Rob Austein <sra@hactrn.net> for fixing it. (CVE-2011-4577)
1668      [Rob Austein <sra@hactrn.net>]
1669 
1670   *) Fix ssl_ciph.c set-up race.
1671      [Adam Langley (Google)]
1672 
1673   *) Fix spurious failures in ecdsatest.c.
1674      [Emilia Käsper (Google)]
1675 
1676   *) Fix the BIO_f_buffer() implementation (which was mixing different
1677      interpretations of the '..._len' fields).
1678      [Adam Langley (Google)]
1679 
1680   *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
1681      BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
1682      threads won't reuse the same blinding coefficients.
1683 
1684      This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
1685      lock to call BN_BLINDING_invert_ex, and avoids one use of
1686      BN_BLINDING_update for each BN_BLINDING structure (previously,
1687      the last update always remained unused).
1688      [Emilia Käsper (Google)]
1689 
1690   *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
1691      for multi-threaded use of ECDH.
1692      [Adam Langley (Google)]
1693 
1694   *) Fix x509_name_ex_d2i memory leak on bad inputs.
1695      [Bodo Moeller]
1696 
1697   *) Add protection against ECDSA timing attacks as mentioned in the paper
1698      by Billy Bob Brumley and Nicola Tuveri, see:
1699 
1700         http://eprint.iacr.org/2011/232.pdf
1701 
1702      [Billy Bob Brumley and Nicola Tuveri]
1703 
1704  Changes between 0.9.8q and 0.9.8r [8 Feb 2011]
1705 
1706   *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
1707      [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
1708 
1709   *) Fix bug in string printing code: if *any* escaping is enabled we must
1710      escape the escape character (backslash) or the resulting string is
1711      ambiguous.
1712      [Steve Henson]
1713 
1714  Changes between 0.9.8p and 0.9.8q [2 Dec 2010]
1715 
1716   *) Disable code workaround for ancient and obsolete Netscape browsers
1717      and servers: an attacker can use it in a ciphersuite downgrade attack.
1718      Thanks to Martin Rex for discovering this bug. CVE-2010-4180
1719      [Steve Henson]
1720 
1721   *) Fixed J-PAKE implementation error, originally discovered by
1722      Sebastien Martini, further info and confirmation from Stefan
1723      Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
1724      [Ben Laurie]
1725 
1726  Changes between 0.9.8o and 0.9.8p [16 Nov 2010]
1727 
1728   *) Fix extension code to avoid race conditions which can result in a buffer
1729      overrun vulnerability: resumed sessions must not be modified as they can
1730      be shared by multiple threads. CVE-2010-3864
1731      [Steve Henson]
1732 
1733   *) Fix for double free bug in ssl/s3_clnt.c CVE-2010-2939
1734      [Steve Henson]
1735 
1736   *) Don't reencode certificate when calculating signature: cache and use
1737      the original encoding instead. This makes signature verification of
1738      some broken encodings work correctly.
1739      [Steve Henson]
1740 
1741   *) ec2_GF2m_simple_mul bugfix: compute correct result if the output EC_POINT
1742      is also one of the inputs.
1743      [Emilia Käsper <emilia.kasper@esat.kuleuven.be> (Google)]
1744 
1745   *) Don't repeatedly append PBE algorithms to table if they already exist.
1746      Sort table on each new add. This effectively makes the table read only
1747      after all algorithms are added and subsequent calls to PKCS12_pbe_add
1748      etc are non-op.
1749      [Steve Henson]
1750 
1751  Changes between 0.9.8n and 0.9.8o [01 Jun 2010]
1752 
1753   [NB: OpenSSL 0.9.8o and later 0.9.8 patch levels were released after
1754   OpenSSL 1.0.0.]
1755 
1756   *) Correct a typo in the CMS ASN1 module which can result in invalid memory
1757      access or freeing data twice (CVE-2010-0742)
1758      [Steve Henson, Ronald Moesbergen <intercommit@gmail.com>]
1759 
1760   *) Add SHA2 algorithms to SSL_library_init(). SHA2 is becoming far more
1761      common in certificates and some applications which only call
1762      SSL_library_init and not OpenSSL_add_all_algorithms() will fail.
1763      [Steve Henson]
1764 
1765   *) VMS fixes: 
1766      Reduce copying into .apps and .test in makevms.com
1767      Don't try to use blank CA certificate in CA.com
1768      Allow use of C files from original directories in maketests.com
1769      [Steven M. Schweda" <sms@antinode.info>]
1770 
1771  Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
1772 
1773   *) When rejecting SSL/TLS records due to an incorrect version number, never
1774      update s->server with a new major version number.  As of
1775      - OpenSSL 0.9.8m if 'short' is a 16-bit type,
1776      - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
1777      the previous behavior could result in a read attempt at NULL when
1778      receiving specific incorrect SSL/TLS records once record payload
1779      protection is active.  (CVE-2010-0740)
1780      [Bodo Moeller, Adam Langley <agl@chromium.org>]
1781 
1782   *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 
1783      could be crashed if the relevant tables were not present (e.g. chrooted).
1784      [Tomas Hoger <thoger@redhat.com>]
1785 
1786  Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
1787 
1788   *) Always check bn_wexpend() return values for failure.  (CVE-2009-3245)
1789      [Martin Olsson, Neel Mehta]
1790 
1791   *) Fix X509_STORE locking: Every 'objs' access requires a lock (to
1792      accommodate for stack sorting, always a write lock!).
1793      [Bodo Moeller]
1794 
1795   *) On some versions of WIN32 Heap32Next is very slow. This can cause
1796      excessive delays in the RAND_poll(): over a minute. As a workaround
1797      include a time check in the inner Heap32Next loop too.
1798      [Steve Henson]
1799 
1800   *) The code that handled flushing of data in SSL/TLS originally used the
1801      BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
1802      the problem outlined in PR#1949. The fix suggested there however can
1803      trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
1804      of Apache). So instead simplify the code to flush unconditionally.
1805      This should be fine since flushing with no data to flush is a no op.
1806      [Steve Henson]
1807 
1808   *) Handle TLS versions 2.0 and later properly and correctly use the
1809      highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
1810      off ancient servers have a habit of sticking around for a while...
1811      [Steve Henson]
1812 
1813   *) Modify compression code so it frees up structures without using the
1814      ex_data callbacks. This works around a problem where some applications
1815      call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
1816      restarting) then use compression (e.g. SSL with compression) later.
1817      This results in significant per-connection memory leaks and
1818      has caused some security issues including CVE-2008-1678 and
1819      CVE-2009-4355.
1820      [Steve Henson]
1821 
1822   *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
1823      change when encrypting or decrypting.
1824      [Bodo Moeller]
1825 
1826   *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
1827      connect and renegotiate with servers which do not support RI.
1828      Until RI is more widely deployed this option is enabled by default.
1829      [Steve Henson]
1830 
1831   *) Add "missing" ssl ctrls to clear options and mode.
1832      [Steve Henson]
1833 
1834   *) If client attempts to renegotiate and doesn't support RI respond with
1835      a no_renegotiation alert as required by RFC5746.  Some renegotiating
1836      TLS clients will continue a connection gracefully when they receive
1837      the alert. Unfortunately OpenSSL mishandled this alert and would hang
1838      waiting for a server hello which it will never receive. Now we treat a
1839      received no_renegotiation alert as a fatal error. This is because
1840      applications requesting a renegotiation might well expect it to succeed
1841      and would have no code in place to handle the server denying it so the
1842      only safe thing to do is to terminate the connection.
1843      [Steve Henson]
1844 
1845   *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
1846      peer supports secure renegotiation and 0 otherwise. Print out peer
1847      renegotiation support in s_client/s_server.
1848      [Steve Henson]
1849 
1850   *) Replace the highly broken and deprecated SPKAC certification method with
1851      the updated NID creation version. This should correctly handle UTF8.
1852      [Steve Henson]
1853 
1854   *) Implement RFC5746. Re-enable renegotiation but require the extension
1855      as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
1856      turns out to be a bad idea. It has been replaced by
1857      SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
1858      SSL_CTX_set_options(). This is really not recommended unless you
1859      know what you are doing.
1860      [Eric Rescorla <ekr@networkresonance.com>, Ben Laurie, Steve Henson]
1861 
1862   *) Fixes to stateless session resumption handling. Use initial_ctx when
1863      issuing and attempting to decrypt tickets in case it has changed during
1864      servername handling. Use a non-zero length session ID when attempting
1865      stateless session resumption: this makes it possible to determine if
1866      a resumption has occurred immediately after receiving server hello
1867      (several places in OpenSSL subtly assume this) instead of later in
1868      the handshake.
1869      [Steve Henson]
1870 
1871   *) The functions ENGINE_ctrl(), OPENSSL_isservice(),
1872      CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
1873      fixes for a few places where the return code is not checked
1874      correctly.
1875      [Julia Lawall <julia@diku.dk>]
1876 
1877   *) Add --strict-warnings option to Configure script to include devteam
1878      warnings in other configurations.
1879      [Steve Henson]
1880 
1881   *) Add support for --libdir option and LIBDIR variable in makefiles. This
1882      makes it possible to install openssl libraries in locations which
1883      have names other than "lib", for example "/usr/lib64" which some
1884      systems need.
1885      [Steve Henson, based on patch from Jeremy Utley]
1886 
1887   *) Don't allow the use of leading 0x80 in OIDs. This is a violation of
1888      X690 8.9.12 and can produce some misleading textual output of OIDs.
1889      [Steve Henson, reported by Dan Kaminsky]
1890 
1891   *) Delete MD2 from algorithm tables. This follows the recommendation in
1892      several standards that it is not used in new applications due to
1893      several cryptographic weaknesses. For binary compatibility reasons
1894      the MD2 API is still compiled in by default.
1895      [Steve Henson]
1896 
1897   *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
1898      and restored.
1899      [Steve Henson]
1900 
1901   *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
1902      OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
1903      clash.
1904      [Guenter <lists@gknw.net>]
1905 
1906   *) Fix the server certificate chain building code to use X509_verify_cert(),
1907      it used to have an ad-hoc builder which was unable to cope with anything
1908      other than a simple chain.
1909      [David Woodhouse <dwmw2@infradead.org>, Steve Henson]
1910 
1911   *) Don't check self signed certificate signatures in X509_verify_cert()
1912      by default (a flag can override this): it just wastes time without
1913      adding any security. As a useful side effect self signed root CAs
1914      with non-FIPS digests are now usable in FIPS mode.
1915      [Steve Henson]
1916 
1917   *) In dtls1_process_out_of_seq_message() the check if the current message
1918      is already buffered was missing. For every new message was memory
1919      allocated, allowing an attacker to perform an denial of service attack
1920      with sending out of seq handshake messages until there is no memory
1921      left. Additionally every future messege was buffered, even if the
1922      sequence number made no sense and would be part of another handshake.
1923      So only messages with sequence numbers less than 10 in advance will be
1924      buffered.  (CVE-2009-1378)
1925      [Robin Seggelmann, discovered by Daniel Mentz]     
1926 
1927   *) Records are buffered if they arrive with a future epoch to be
1928      processed after finishing the corresponding handshake. There is
1929      currently no limitation to this buffer allowing an attacker to perform
1930      a DOS attack with sending records with future epochs until there is no
1931      memory left. This patch adds the pqueue_size() function to detemine
1932      the size of a buffer and limits the record buffer to 100 entries.
1933      (CVE-2009-1377)
1934      [Robin Seggelmann, discovered by Daniel Mentz]     
1935 
1936   *) Keep a copy of frag->msg_header.frag_len so it can be used after the
1937      parent structure is freed.  (CVE-2009-1379)
1938      [Daniel Mentz]     
1939 
1940   *) Handle non-blocking I/O properly in SSL_shutdown() call.
1941      [Darryl Miles <darryl-mailinglists@netbauds.net>]
1942 
1943   *) Add 2.5.4.* OIDs
1944      [Ilya O. <vrghost@gmail.com>]
1945 
1946  Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
1947 
1948   *) Disable renegotiation completely - this fixes a severe security
1949      problem (CVE-2009-3555) at the cost of breaking all
1950      renegotiation. Renegotiation can be re-enabled by setting
1951      SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
1952      run-time. This is really not recommended unless you know what
1953      you're doing.
1954      [Ben Laurie]
1955 
1956  Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
1957 
1958   *) Don't set val to NULL when freeing up structures, it is freed up by
1959      underlying code. If sizeof(void *) > sizeof(long) this can result in
1960      zeroing past the valid field. (CVE-2009-0789)
1961      [Paolo Ganci <Paolo.Ganci@AdNovum.CH>]
1962 
1963   *) Fix bug where return value of CMS_SignerInfo_verify_content() was not
1964      checked correctly. This would allow some invalid signed attributes to
1965      appear to verify correctly. (CVE-2009-0591)
1966      [Ivan Nestlerode <inestlerode@us.ibm.com>]
1967 
1968   *) Reject UniversalString and BMPString types with invalid lengths. This
1969      prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
1970      a legal length. (CVE-2009-0590)
1971      [Steve Henson]
1972 
1973   *) Set S/MIME signing as the default purpose rather than setting it 
1974      unconditionally. This allows applications to override it at the store
1975      level.
1976      [Steve Henson]
1977 
1978   *) Permit restricted recursion of ASN1 strings. This is needed in practice
1979      to handle some structures.
1980      [Steve Henson]
1981 
1982   *) Improve efficiency of mem_gets: don't search whole buffer each time
1983      for a '\n'
1984      [Jeremy Shapiro <jnshapir@us.ibm.com>]
1985 
1986   *) New -hex option for openssl rand.
1987      [Matthieu Herrb]
1988 
1989   *) Print out UTF8String and NumericString when parsing ASN1.
1990      [Steve Henson]
1991 
1992   *) Support NumericString type for name components.
1993      [Steve Henson]
1994 
1995   *) Allow CC in the environment to override the automatically chosen
1996      compiler. Note that nothing is done to ensure flags work with the
1997      chosen compiler.
1998      [Ben Laurie]
1999 
2000  Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
2001 
2002   *) Properly check EVP_VerifyFinal() and similar return values
2003      (CVE-2008-5077).
2004      [Ben Laurie, Bodo Moeller, Google Security Team]
2005 
2006   *) Enable TLS extensions by default.
2007      [Ben Laurie]
2008 
2009   *) Allow the CHIL engine to be loaded, whether the application is
2010      multithreaded or not. (This does not release the developer from the
2011      obligation to set up the dynamic locking callbacks.)
2012      [Sander Temme <sander@temme.net>]
2013 
2014   *) Use correct exit code if there is an error in dgst command.
2015      [Steve Henson; problem pointed out by Roland Dirlewanger]
2016 
2017   *) Tweak Configure so that you need to say "experimental-jpake" to enable
2018      JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
2019      [Bodo Moeller]
2020 
2021   *) Add experimental JPAKE support, including demo authentication in
2022      s_client and s_server.
2023      [Ben Laurie]
2024 
2025   *) Set the comparison function in v3_addr_canonize().
2026      [Rob Austein <sra@hactrn.net>]
2027 
2028   *) Add support for XMPP STARTTLS in s_client.
2029      [Philip Paeps <philip@freebsd.org>]
2030 
2031   *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
2032      to ensure that even with this option, only ciphersuites in the
2033      server's preference list will be accepted.  (Note that the option
2034      applies only when resuming a session, so the earlier behavior was
2035      just about the algorithm choice for symmetric cryptography.)
2036      [Bodo Moeller]
2037 
2038  Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
2039 
2040   *) Fix NULL pointer dereference if a DTLS server received
2041      ChangeCipherSpec as first record (CVE-2009-1386).
2042      [PR #1679]
2043 
2044   *) Fix a state transitition in s3_srvr.c and d1_srvr.c
2045      (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
2046      [Nagendra Modadugu]
2047 
2048   *) The fix in 0.9.8c that supposedly got rid of unsafe
2049      double-checked locking was incomplete for RSA blinding,
2050      addressing just one layer of what turns out to have been
2051      doubly unsafe triple-checked locking.
2052 
2053      So now fix this for real by retiring the MONT_HELPER macro
2054      in crypto/rsa/rsa_eay.c.
2055 
2056      [Bodo Moeller; problem pointed out by Marius Schilder]
2057 
2058   *) Various precautionary measures:
2059 
2060      - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
2061 
2062      - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
2063        (NB: This would require knowledge of the secret session ticket key
2064        to exploit, in which case you'd be SOL either way.)
2065 
2066      - Change bn_nist.c so that it will properly handle input BIGNUMs
2067        outside the expected range.
2068 
2069      - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
2070        builds.
2071 
2072      [Neel Mehta, Bodo Moeller]
2073 
2074   *) Allow engines to be "soft loaded" - i.e. optionally don't die if
2075      the load fails. Useful for distros.
2076      [Ben Laurie and the FreeBSD team]
2077 
2078   *) Add support for Local Machine Keyset attribute in PKCS#12 files.
2079      [Steve Henson]
2080 
2081   *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
2082      [Huang Ying]
2083 
2084   *) Expand ENGINE to support engine supplied SSL client certificate functions.
2085 
2086      This work was sponsored by Logica.
2087      [Steve Henson]
2088 
2089   *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
2090      keystores. Support for SSL/TLS client authentication too.
2091      Not compiled unless enable-capieng specified to Configure.
2092 
2093      This work was sponsored by Logica.
2094      [Steve Henson]
2095 
2096   *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using
2097      ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
2098      attribute creation routines such as certifcate requests and PKCS#12
2099      files.
2100      [Steve Henson]
2101 
2102  Changes between 0.9.8g and 0.9.8h  [28 May 2008]
2103 
2104   *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
2105      handshake which could lead to a cilent crash as found using the
2106      Codenomicon TLS test suite (CVE-2008-1672) 
2107      [Steve Henson, Mark Cox]
2108 
2109   *) Fix double free in TLS server name extensions which could lead to
2110      a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 
2111      [Joe Orton]
2112 
2113   *) Clear error queue in SSL_CTX_use_certificate_chain_file()
2114 
2115      Clear the error queue to ensure that error entries left from
2116      older function calls do not interfere with the correct operation.
2117      [Lutz Jaenicke, Erik de Castro Lopo]
2118 
2119   *) Remove root CA certificates of commercial CAs:
2120 
2121      The OpenSSL project does not recommend any specific CA and does not
2122      have any policy with respect to including or excluding any CA.
2123      Therefore it does not make any sense to ship an arbitrary selection
2124      of root CA certificates with the OpenSSL software.
2125      [Lutz Jaenicke]
2126 
2127   *) RSA OAEP patches to fix two separate invalid memory reads.
2128      The first one involves inputs when 'lzero' is greater than
2129      'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
2130      before the beginning of from). The second one involves inputs where
2131      the 'db' section contains nothing but zeroes (there is a one-byte
2132      invalid read after the end of 'db').
2133      [Ivan Nestlerode <inestlerode@us.ibm.com>]
2134 
2135   *) Partial backport from 0.9.9-dev:
2136 
2137      Introduce bn_mul_mont (dedicated Montgomery multiplication
2138      procedure) as a candidate for BIGNUM assembler implementation.
2139      While 0.9.9-dev uses assembler for various architectures, only
2140      x86_64 is available by default here in the 0.9.8 branch, and
2141      32-bit x86 is available through a compile-time setting.
2142 
2143      To try the 32-bit x86 assembler implementation, use Configure
2144      option "enable-montasm" (which exists only for this backport).
2145 
2146      As "enable-montasm" for 32-bit x86 disclaims code stability
2147      anyway, in this constellation we activate additional code
2148      backported from 0.9.9-dev for further performance improvements,
2149      namely BN_from_montgomery_word.  (To enable this otherwise,
2150      e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
2151 
2152      [Andy Polyakov (backport partially by Bodo Moeller)]
2153 
2154   *) Add TLS session ticket callback. This allows an application to set
2155      TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
2156      values. This is useful for key rollover for example where several key
2157      sets may exist with different names.
2158      [Steve Henson]
2159 
2160   *) Reverse ENGINE-internal logic for caching default ENGINE handles.
2161      This was broken until now in 0.9.8 releases, such that the only way
2162      a registered ENGINE could be used (assuming it initialises
2163      successfully on the host) was to explicitly set it as the default
2164      for the relevant algorithms. This is in contradiction with 0.9.7
2165      behaviour and the documentation. With this fix, when an ENGINE is
2166      registered into a given algorithm's table of implementations, the
2167      'uptodate' flag is reset so that auto-discovery will be used next
2168      time a new context for that algorithm attempts to select an
2169      implementation.
2170      [Ian Lister (tweaked by Geoff Thorpe)]
2171 
2172   *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
2173      implemention in the following ways:
2174 
2175      Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
2176      hard coded.
2177 
2178      Lack of BER streaming support means one pass streaming processing is
2179      only supported if data is detached: setting the streaming flag is
2180      ignored for embedded content.
2181 
2182      CMS support is disabled by default and must be explicitly enabled
2183      with the enable-cms configuration option.
2184      [Steve Henson]
2185 
2186   *) Update the GMP engine glue to do direct copies between BIGNUM and
2187      mpz_t when openssl and GMP use the same limb size. Otherwise the
2188      existing "conversion via a text string export" trick is still used.
2189      [Paul Sheer <paulsheer@gmail.com>]
2190 
2191   *) Zlib compression BIO. This is a filter BIO which compressed and
2192      uncompresses any data passed through it.
2193      [Steve Henson]
2194 
2195   *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
2196      RFC3394 compatible AES key wrapping.
2197      [Steve Henson]
2198 
2199   *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
2200      sets string data without copying. X509_ALGOR_set0() and
2201      X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
2202      data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
2203      from an X509_ATTRIBUTE structure optionally checking it occurs only
2204      once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
2205      data.
2206      [Steve Henson]
2207 
2208   *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
2209      to get the expected BN_FLG_CONSTTIME behavior.
2210      [Bodo Moeller (Google)]
2211   
2212   *) Netware support:
2213 
2214      - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
2215      - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
2216      - added some more tests to do_tests.pl
2217      - fixed RunningProcess usage so that it works with newer LIBC NDKs too
2218      - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
2219      - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
2220        netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
2221      - various changes to netware.pl to enable gcc-cross builds on Win32
2222        platform
2223      - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
2224      - various changes to fix missing prototype warnings
2225      - fixed x86nasm.pl to create correct asm files for NASM COFF output
2226      - added AES, WHIRLPOOL and CPUID assembler code to build files
2227      - added missing AES assembler make rules to mk1mf.pl
2228      - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
2229      [Guenter Knauf <eflash@gmx.net>]
2230 
2231   *) Implement certificate status request TLS extension defined in RFC3546.
2232      A client can set the appropriate parameters and receive the encoded
2233      OCSP response via a callback. A server can query the supplied parameters
2234      and set the encoded OCSP response in the callback. Add simplified examples
2235      to s_client and s_server.
2236      [Steve Henson]
2237 
2238  Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
2239 
2240   *) Fix various bugs:
2241      + Binary incompatibility of ssl_ctx_st structure
2242      + DTLS interoperation with non-compliant servers
2243      + Don't call get_session_cb() without proposed session
2244      + Fix ia64 assembler code
2245      [Andy Polyakov, Steve Henson]
2246 
2247  Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
2248 
2249   *) DTLS Handshake overhaul. There were longstanding issues with
2250      OpenSSL DTLS implementation, which were making it impossible for
2251      RFC 4347 compliant client to communicate with OpenSSL server.
2252      Unfortunately just fixing these incompatibilities would "cut off"
2253      pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
2254      server keeps tolerating non RFC compliant syntax. The opposite is
2255      not true, 0.9.8f client can not communicate with earlier server.
2256      This update even addresses CVE-2007-4995.
2257      [Andy Polyakov]
2258 
2259   *) Changes to avoid need for function casts in OpenSSL: some compilers
2260      (gcc 4.2 and later) reject their use.
2261      [Kurt Roeckx <kurt@roeckx.be>, Peter Hartley <pdh@utter.chaos.org.uk>,
2262       Steve Henson]
2263   
2264   *) Add RFC4507 support to OpenSSL. This includes the corrections in
2265      RFC4507bis. The encrypted ticket format is an encrypted encoded
2266      SSL_SESSION structure, that way new session features are automatically
2267      supported.
2268 
2269      If a client application caches session in an SSL_SESSION structure
2270      support is transparent because tickets are now stored in the encoded
2271      SSL_SESSION.
2272      
2273      The SSL_CTX structure automatically generates keys for ticket
2274      protection in servers so again support should be possible
2275      with no application modification.
2276 
2277      If a client or server wishes to disable RFC4507 support then the option
2278      SSL_OP_NO_TICKET can be set.
2279 
2280      Add a TLS extension debugging callback to allow the contents of any client
2281      or server extensions to be examined.
2282 
2283      This work was sponsored by Google.
2284      [Steve Henson]
2285 
2286   *) Add initial support for TLS extensions, specifically for the server_name
2287      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
2288      have new members for a host name.  The SSL data structure has an
2289      additional member SSL_CTX *initial_ctx so that new sessions can be
2290      stored in that context to allow for session resumption, even after the
2291      SSL has been switched to a new SSL_CTX in reaction to a client's
2292      server_name extension.
2293 
2294      New functions (subject to change):
2295 
2296          SSL_get_servername()
2297          SSL_get_servername_type()
2298          SSL_set_SSL_CTX()
2299 
2300      New CTRL codes and macros (subject to change):
2301 
2302          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
2303                                  - SSL_CTX_set_tlsext_servername_callback()
2304          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
2305                                       - SSL_CTX_set_tlsext_servername_arg()
2306          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
2307 
2308      openssl s_client has a new '-servername ...' option.
2309 
2310      openssl s_server has new options '-servername_host ...', '-cert2 ...',
2311      '-key2 ...', '-servername_fatal' (subject to change).  This allows
2312      testing the HostName extension for a specific single host name ('-cert'
2313      and '-key' remain fallbacks for handshakes without HostName
2314      negotiation).  If the unrecogninzed_name alert has to be sent, this by
2315      default is a warning; it becomes fatal with the '-servername_fatal'
2316      option.
2317 
2318      [Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson]
2319 
2320   *) Add AES and SSE2 assembly language support to VC++ build.
2321      [Steve Henson]
2322 
2323   *) Mitigate attack on final subtraction in Montgomery reduction.
2324      [Andy Polyakov]
2325 
2326   *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
2327      (which previously caused an internal error).
2328      [Bodo Moeller]
2329 
2330   *) Squeeze another 10% out of IGE mode when in != out.
2331      [Ben Laurie]
2332 
2333   *) AES IGE mode speedup.
2334      [Dean Gaudet (Google)]
2335 
2336   *) Add the Korean symmetric 128-bit cipher SEED (see
2337      http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
2338      add SEED ciphersuites from RFC 4162:
2339 
2340         TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
2341         TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
2342         TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
2343         TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
2344 
2345      To minimize changes between patchlevels in the OpenSSL 0.9.8
2346      series, SEED remains excluded from compilation unless OpenSSL
2347      is configured with 'enable-seed'.
2348      [KISA, Bodo Moeller]
2349 
2350   *) Mitigate branch prediction attacks, which can be practical if a
2351      single processor is shared, allowing a spy process to extract
2352      information.  For detailed background information, see
2353      http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
2354      J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
2355      and Necessary Software Countermeasures").  The core of the change
2356      are new versions BN_div_no_branch() and
2357      BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
2358      respectively, which are slower, but avoid the security-relevant
2359      conditional branches.  These are automatically called by BN_div()
2360      and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
2361      of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
2362      remove a conditional branch.
2363 
2364      BN_FLG_CONSTTIME is the new name for the previous
2365      BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
2366      modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
2367      in the exponent causes BN_mod_exp_mont() to use the alternative
2368      implementation in BN_mod_exp_mont_consttime().)  The old name
2369      remains as a deprecated alias.
2370 
2371      Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
2372      RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
2373      constant-time implementations for more than just exponentiation.
2374      Here too the old name is kept as a deprecated alias.
2375 
2376      BN_BLINDING_new() will now use BN_dup() for the modulus so that
2377      the BN_BLINDING structure gets an independent copy of the
2378      modulus.  This means that the previous "BIGNUM *m" argument to
2379      BN_BLINDING_new() and to BN_BLINDING_create_param() now
2380      essentially becomes "const BIGNUM *m", although we can't actually
2381      change this in the header file before 0.9.9.  It allows
2382      RSA_setup_blinding() to use BN_with_flags() on the modulus to
2383      enable BN_FLG_CONSTTIME.
2384 
2385      [Matthew D Wood (Intel Corp)]
2386 
2387   *) In the SSL/TLS server implementation, be strict about session ID
2388      context matching (which matters if an application uses a single
2389      external cache for different purposes).  Previously,
2390      out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
2391      set.  This did ensure strict client verification, but meant that,
2392      with applications using a single external cache for quite
2393      different requirements, clients could circumvent ciphersuite
2394      restrictions for a given session ID context by starting a session
2395      in a different context.
2396      [Bodo Moeller]
2397 
2398   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
2399      a ciphersuite string such as "DEFAULT:RSA" cannot enable
2400      authentication-only ciphersuites.
2401      [Bodo Moeller]
2402 
2403   *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
2404      not complete and could lead to a possible single byte overflow
2405      (CVE-2007-5135) [Ben Laurie]
2406 
2407  Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
2408 
2409   *) Since AES128 and AES256 (and similarly Camellia128 and
2410      Camellia256) share a single mask bit in the logic of
2411      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
2412      kludge to work properly if AES128 is available and AES256 isn't
2413      (or if Camellia128 is available and Camellia256 isn't).
2414      [Victor Duchovni]
2415 
2416   *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
2417      (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
2418      When a point or a seed is encoded in a BIT STRING, we need to
2419      prevent the removal of trailing zero bits to get the proper DER
2420      encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
2421      of a NamedBitList, for which trailing 0 bits need to be removed.)
2422      [Bodo Moeller]
2423 
2424   *) Have SSL/TLS server implementation tolerate "mismatched" record
2425      protocol version while receiving ClientHello even if the
2426      ClientHello is fragmented.  (The server can't insist on the
2427      particular protocol version it has chosen before the ServerHello
2428      message has informed the client about his choice.)
2429      [Bodo Moeller]
2430 
2431   *) Add RFC 3779 support.
2432      [Rob Austein for ARIN, Ben Laurie]
2433 
2434   *) Load error codes if they are not already present instead of using a
2435      static variable. This allows them to be cleanly unloaded and reloaded.
2436      Improve header file function name parsing.
2437      [Steve Henson]
2438 
2439   *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
2440      or CAPABILITY handshake as required by RFCs.
2441      [Goetz Babin-Ebell]
2442 
2443  Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
2444 
2445   *) Introduce limits to prevent malicious keys being able to
2446      cause a denial of service.  (CVE-2006-2940)
2447      [Steve Henson, Bodo Moeller]
2448 
2449   *) Fix ASN.1 parsing of certain invalid structures that can result
2450      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
2451 
2452   *) Fix buffer overflow in SSL_get_shared_ciphers() function. 
2453      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
2454 
2455   *) Fix SSL client code which could crash if connecting to a
2456      malicious SSLv2 server.  (CVE-2006-4343)
2457      [Tavis Ormandy and Will Drewry, Google Security Team]
2458 
2459   *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
2460      match only those.  Before that, "AES256-SHA" would be interpreted
2461      as a pattern and match "AES128-SHA" too (since AES128-SHA got
2462      the same strength classification in 0.9.7h) as we currently only
2463      have a single AES bit in the ciphersuite description bitmap.
2464      That change, however, also applied to ciphersuite strings such as
2465      "RC4-MD5" that intentionally matched multiple ciphersuites --
2466      namely, SSL 2.0 ciphersuites in addition to the more common ones
2467      from SSL 3.0/TLS 1.0.
2468 
2469      So we change the selection algorithm again: Naming an explicit
2470      ciphersuite selects this one ciphersuite, and any other similar
2471      ciphersuite (same bitmap) from *other* protocol versions.
2472      Thus, "RC4-MD5" again will properly select both the SSL 2.0
2473      ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
2474 
2475      Since SSL 2.0 does not have any ciphersuites for which the
2476      128/256 bit distinction would be relevant, this works for now.
2477      The proper fix will be to use different bits for AES128 and
2478      AES256, which would have avoided the problems from the beginning;
2479      however, bits are scarce, so we can only do this in a new release
2480      (not just a patchlevel) when we can change the SSL_CIPHER
2481      definition to split the single 'unsigned long mask' bitmap into
2482      multiple values to extend the available space.
2483 
2484      [Bodo Moeller]
2485 
2486  Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
2487 
2488   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
2489      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
2490 
2491   *) Add AES IGE and biIGE modes.
2492      [Ben Laurie]
2493 
2494   *) Change the Unix randomness entropy gathering to use poll() when
2495      possible instead of select(), since the latter has some
2496      undesirable limitations.
2497      [Darryl Miles via Richard Levitte and Bodo Moeller]
2498 
2499   *) Disable "ECCdraft" ciphersuites more thoroughly.  Now special
2500      treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
2501      cannot be implicitly activated as part of, e.g., the "AES" alias.
2502      However, please upgrade to OpenSSL 0.9.9[-dev] for
2503      non-experimental use of the ECC ciphersuites to get TLS extension
2504      support, which is required for curve and point format negotiation
2505      to avoid potential handshake problems.
2506      [Bodo Moeller]
2507 
2508   *) Disable rogue ciphersuites:
2509 
2510       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
2511       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
2512       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
2513 
2514      The latter two were purportedly from
2515      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
2516      appear there.
2517 
2518      Also deactivate the remaining ciphersuites from
2519      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
2520      unofficial, and the ID has long expired.
2521      [Bodo Moeller]
2522 
2523   *) Fix RSA blinding Heisenbug (problems sometimes occured on
2524      dual-core machines) and other potential thread-safety issues.
2525      [Bodo Moeller]
2526 
2527   *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
2528      versions), which is now available for royalty-free use
2529      (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
2530      Also, add Camellia TLS ciphersuites from RFC 4132.
2531 
2532      To minimize changes between patchlevels in the OpenSSL 0.9.8
2533      series, Camellia remains excluded from compilation unless OpenSSL
2534      is configured with 'enable-camellia'.
2535      [NTT]
2536 
2537   *) Disable the padding bug check when compression is in use. The padding
2538      bug check assumes the first packet is of even length, this is not
2539      necessarily true if compresssion is enabled and can result in false
2540      positives causing handshake failure. The actual bug test is ancient
2541      code so it is hoped that implementations will either have fixed it by
2542      now or any which still have the bug do not support compression.
2543      [Steve Henson]
2544 
2545  Changes between 0.9.8a and 0.9.8b  [04 May 2006]
2546 
2547   *) When applying a cipher rule check to see if string match is an explicit
2548      cipher suite and only match that one cipher suite if it is.
2549      [Steve Henson]
2550 
2551   *) Link in manifests for VC++ if needed.
2552      [Austin Ziegler <halostatue@gmail.com>]
2553 
2554   *) Update support for ECC-based TLS ciphersuites according to
2555      draft-ietf-tls-ecc-12.txt with proposed changes (but without
2556      TLS extensions, which are supported starting with the 0.9.9
2557      branch, not in the OpenSSL 0.9.8 branch).
2558      [Douglas Stebila]
2559 
2560   *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
2561      opaque EVP_CIPHER_CTX handling.
2562      [Steve Henson]
2563 
2564   *) Fixes and enhancements to zlib compression code. We now only use
2565      "zlib1.dll" and use the default __cdecl calling convention on Win32
2566      to conform with the standards mentioned here:
2567            http://www.zlib.net/DLL_FAQ.txt
2568      Static zlib linking now works on Windows and the new --with-zlib-include
2569      --with-zlib-lib options to Configure can be used to supply the location
2570      of the headers and library. Gracefully handle case where zlib library
2571      can't be loaded.
2572      [Steve Henson]
2573 
2574   *) Several fixes and enhancements to the OID generation code. The old code
2575      sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
2576      handle numbers larger than ULONG_MAX, truncated printing and had a
2577      non standard OBJ_obj2txt() behaviour.
2578      [Steve Henson]
2579 
2580   *) Add support for building of engines under engine/ as shared libraries
2581      under VC++ build system.
2582      [Steve Henson]
2583 
2584   *) Corrected the numerous bugs in the Win32 path splitter in DSO.
2585      Hopefully, we will not see any false combination of paths any more.
2586      [Richard Levitte]
2587 
2588  Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
2589 
2590   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
2591      (part of SSL_OP_ALL).  This option used to disable the
2592      countermeasure against man-in-the-middle protocol-version
2593      rollback in the SSL 2.0 server implementation, which is a bad
2594      idea.  (CVE-2005-2969)
2595 
2596      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
2597      for Information Security, National Institute of Advanced Industrial
2598      Science and Technology [AIST], Japan)]
2599 
2600   *) Add two function to clear and return the verify parameter flags.
2601      [Steve Henson]
2602 
2603   *) Keep cipherlists sorted in the source instead of sorting them at
2604      runtime, thus removing the need for a lock.
2605      [Nils Larsch]
2606 
2607   *) Avoid some small subgroup attacks in Diffie-Hellman.
2608      [Nick Mathewson and Ben Laurie]
2609 
2610   *) Add functions for well-known primes.
2611      [Nick Mathewson]
2612 
2613   *) Extended Windows CE support.
2614      [Satoshi Nakamura and Andy Polyakov]
2615 
2616   *) Initialize SSL_METHOD structures at compile time instead of during
2617      runtime, thus removing the need for a lock.
2618      [Steve Henson]
2619 
2620   *) Make PKCS7_decrypt() work even if no certificate is supplied by
2621      attempting to decrypt each encrypted key in turn. Add support to
2622      smime utility.
2623      [Steve Henson]
2624 
2625  Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
2626 
2627   [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
2628   OpenSSL 0.9.8.]
2629 
2630   *) Add libcrypto.pc and libssl.pc for those who feel they need them.
2631      [Richard Levitte]
2632 
2633   *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
2634      key into the same file any more.
2635      [Richard Levitte]
2636 
2637   *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
2638      [Andy Polyakov]
2639 
2640   *) Add -utf8 command line and config file option to 'ca'.
2641      [Stefan <stf@udoma.org]
2642 
2643   *) Removed the macro des_crypt(), as it seems to conflict with some
2644      libraries.  Use DES_crypt().
2645      [Richard Levitte]
2646 
2647   *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
2648      involves renaming the source and generated shared-libs for
2649      both. The engines will accept the corrected or legacy ids
2650      ('ncipher' and '4758_cca' respectively) when binding. NB,
2651      this only applies when building 'shared'.
2652      [Corinna Vinschen <vinschen@redhat.com> and Geoff Thorpe]
2653 
2654   *) Add attribute functions to EVP_PKEY structure. Modify
2655      PKCS12_create() to recognize a CSP name attribute and
2656      use it. Make -CSP option work again in pkcs12 utility.
2657      [Steve Henson]
2658 
2659   *) Add new functionality to the bn blinding code:
2660      - automatic re-creation of the BN_BLINDING parameters after
2661        a fixed number of uses (currently 32)
2662      - add new function for parameter creation
2663      - introduce flags to control the update behaviour of the
2664        BN_BLINDING parameters
2665      - hide BN_BLINDING structure
2666      Add a second BN_BLINDING slot to the RSA structure to improve
2667      performance when a single RSA object is shared among several
2668      threads.
2669      [Nils Larsch]
2670 
2671   *) Add support for DTLS.
2672      [Nagendra Modadugu <nagendra@cs.stanford.edu> and Ben Laurie]
2673 
2674   *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
2675      to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
2676      [Walter Goulet]
2677 
2678   *) Remove buggy and incompletet DH cert support from
2679      ssl/ssl_rsa.c and ssl/s3_both.c
2680      [Nils Larsch]
2681 
2682   *) Use SHA-1 instead of MD5 as the default digest algorithm for
2683      the apps/openssl applications.
2684      [Nils Larsch]
2685 
2686   *) Compile clean with "-Wall -Wmissing-prototypes
2687      -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
2688      DEBUG_SAFESTACK must also be set.
2689      [Ben Laurie]
2690 
2691   *) Change ./Configure so that certain algorithms can be disabled by default.
2692      The new counterpiece to "no-xxx" is "enable-xxx".
2693 
2694      The patented RC5 and MDC2 algorithms will now be disabled unless
2695      "enable-rc5" and "enable-mdc2", respectively, are specified.
2696 
2697      (IDEA remains enabled despite being patented.  This is because IDEA
2698      is frequently required for interoperability, and there is no license
2699      fee for non-commercial use.  As before, "no-idea" can be used to
2700      avoid this algorithm.)
2701 
2702      [Bodo Moeller]
2703 
2704   *) Add processing of proxy certificates (see RFC 3820).  This work was
2705      sponsored by KTH (The Royal Institute of Technology in Stockholm) and
2706      EGEE (Enabling Grids for E-science in Europe).
2707      [Richard Levitte]
2708 
2709   *) RC4 performance overhaul on modern architectures/implementations, such
2710      as Intel P4, IA-64 and AMD64.
2711      [Andy Polyakov]
2712 
2713   *) New utility extract-section.pl. This can be used specify an alternative
2714      section number in a pod file instead of having to treat each file as
2715      a separate case in Makefile. This can be done by adding two lines to the
2716      pod file:
2717 
2718      =for comment openssl_section:XXX
2719 
2720      The blank line is mandatory.
2721 
2722      [Steve Henson]
2723 
2724   *) New arguments -certform, -keyform and -pass for s_client and s_server
2725      to allow alternative format key and certificate files and passphrase
2726      sources.
2727      [Steve Henson]
2728 
2729   *) New structure X509_VERIFY_PARAM which combines current verify parameters,
2730      update associated structures and add various utility functions.
2731 
2732      Add new policy related verify parameters, include policy checking in 
2733      standard verify code. Enhance 'smime' application with extra parameters
2734      to support policy checking and print out.
2735      [Steve Henson]
2736 
2737   *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
2738      Nehemiah processors. These extensions support AES encryption in hardware
2739      as well as RNG (though RNG support is currently disabled).
2740      [Michal Ludvig <michal@logix.cz>, with help from Andy Polyakov]
2741 
2742   *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
2743      [Geoff Thorpe]
2744 
2745   *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
2746      [Andy Polyakov and a number of other people]
2747 
2748   *) Improved PowerPC platform support. Most notably BIGNUM assembler
2749      implementation contributed by IBM.
2750      [Suresh Chari, Peter Waltenberg, Andy Polyakov]
2751 
2752   *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
2753      exponent rather than 'unsigned long'. There is a corresponding change to
2754      the new 'rsa_keygen' element of the RSA_METHOD structure.
2755      [Jelte Jansen, Geoff Thorpe]
2756 
2757   *) Functionality for creating the initial serial number file is now
2758      moved from CA.pl to the 'ca' utility with a new option -create_serial.
2759 
2760      (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
2761      number file to 1, which is bound to cause problems.  To avoid
2762      the problems while respecting compatibility between different 0.9.7
2763      patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
2764      CA.pl for serial number initialization.  With the new release 0.9.8,
2765      we can fix the problem directly in the 'ca' utility.)
2766      [Steve Henson]
2767 
2768   *) Reduced header interdepencies by declaring more opaque objects in
2769      ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
2770      give fewer recursive includes, which could break lazy source code - so
2771      this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
2772      developers should define this symbol when building and using openssl to
2773      ensure they track the recommended behaviour, interfaces, [etc], but
2774      backwards-compatible behaviour prevails when this isn't defined.
2775      [Geoff Thorpe]
2776 
2777   *) New function X509_POLICY_NODE_print() which prints out policy nodes.
2778      [Steve Henson]
2779 
2780   *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
2781      This will generate a random key of the appropriate length based on the 
2782      cipher context. The EVP_CIPHER can provide its own random key generation
2783      routine to support keys of a specific form. This is used in the des and 
2784      3des routines to generate a key of the correct parity. Update S/MIME
2785      code to use new functions and hence generate correct parity DES keys.
2786      Add EVP_CHECK_DES_KEY #define to return an error if the key is not 
2787      valid (weak or incorrect parity).
2788      [Steve Henson]
2789 
2790   *) Add a local set of CRLs that can be used by X509_verify_cert() as well
2791      as looking them up. This is useful when the verified structure may contain
2792      CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
2793      present unless the new PKCS7_NO_CRL flag is asserted.
2794      [Steve Henson]
2795 
2796   *) Extend ASN1 oid configuration module. It now additionally accepts the
2797      syntax:
2798 
2799      shortName = some long name, 1.2.3.4
2800      [Steve Henson]
2801 
2802   *) Reimplemented the BN_CTX implementation. There is now no more static
2803      limitation on the number of variables it can handle nor the depth of the
2804      "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
2805      information can now expand as required, and rather than having a single
2806      static array of bignums, BN_CTX now uses a linked-list of such arrays
2807      allowing it to expand on demand whilst maintaining the usefulness of
2808      BN_CTX's "bundling".
2809      [Geoff Thorpe]
2810 
2811   *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
2812      to allow all RSA operations to function using a single BN_CTX.
2813      [Geoff Thorpe]
2814 
2815   *) Preliminary support for certificate policy evaluation and checking. This
2816      is initially intended to pass the tests outlined in "Conformance Testing
2817      of Relying Party Client Certificate Path Processing Logic" v1.07.
2818      [Steve Henson]
2819 
2820   *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
2821      remained unused and not that useful. A variety of other little bignum
2822      tweaks and fixes have also been made continuing on from the audit (see
2823      below).
2824      [Geoff Thorpe]
2825 
2826   *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
2827      associated ASN1, EVP and SSL functions and old ASN1 macros.
2828      [Richard Levitte]
2829 
2830   *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
2831      and this should never fail. So the return value from the use of
2832      BN_set_word() (which can fail due to needless expansion) is now deprecated;
2833      if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
2834      [Geoff Thorpe]
2835 
2836   *) BN_CTX_get() should return zero-valued bignums, providing the same
2837      initialised value as BN_new().
2838      [Geoff Thorpe, suggested by Ulf Möller]
2839 
2840   *) Support for inhibitAnyPolicy certificate extension.
2841      [Steve Henson]
2842 
2843   *) An audit of the BIGNUM code is underway, for which debugging code is
2844      enabled when BN_DEBUG is defined. This makes stricter enforcements on what
2845      is considered valid when processing BIGNUMs, and causes execution to
2846      assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
2847      further steps are taken to deliberately pollute unused data in BIGNUM
2848      structures to try and expose faulty code further on. For now, openssl will
2849      (in its default mode of operation) continue to tolerate the inconsistent
2850      forms that it has tolerated in the past, but authors and packagers should
2851      consider trying openssl and their own applications when compiled with
2852      these debugging symbols defined. It will help highlight potential bugs in
2853      their own code, and will improve the test coverage for OpenSSL itself. At
2854      some point, these tighter rules will become openssl's default to improve
2855      maintainability, though the assert()s and other overheads will remain only
2856      in debugging configurations. See bn.h for more details.
2857      [Geoff Thorpe, Nils Larsch, Ulf Möller]
2858 
2859   *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
2860      that can only be obtained through BN_CTX_new() (which implicitly
2861      initialises it). The presence of this function only made it possible
2862      to overwrite an existing structure (and cause memory leaks).
2863      [Geoff Thorpe]
2864 
2865   *) Because of the callback-based approach for implementing LHASH as a
2866      template type, lh_insert() adds opaque objects to hash-tables and
2867      lh_doall() or lh_doall_arg() are typically used with a destructor callback
2868      to clean up those corresponding objects before destroying the hash table
2869      (and losing the object pointers). So some over-zealous constifications in
2870      LHASH have been relaxed so that lh_insert() does not take (nor store) the
2871      objects as "const" and the lh_doall[_arg] callback wrappers are not
2872      prototyped to have "const" restrictions on the object pointers they are
2873      given (and so aren't required to cast them away any more).
2874      [Geoff Thorpe]
2875 
2876   *) The tmdiff.h API was so ugly and minimal that our own timing utility
2877      (speed) prefers to use its own implementation. The two implementations
2878      haven't been consolidated as yet (volunteers?) but the tmdiff API has had
2879      its object type properly exposed (MS_TM) instead of casting to/from "char
2880      *". This may still change yet if someone realises MS_TM and "ms_time_***"
2881      aren't necessarily the greatest nomenclatures - but this is what was used
2882      internally to the implementation so I've used that for now.
2883      [Geoff Thorpe]
2884 
2885   *) Ensure that deprecated functions do not get compiled when
2886      OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
2887      the self-tests were still using deprecated key-generation functions so
2888      these have been updated also.
2889      [Geoff Thorpe]
2890 
2891   *) Reorganise PKCS#7 code to separate the digest location functionality
2892      into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
2893      New function PKCS7_set_digest() to set the digest type for PKCS#7
2894      digestedData type. Add additional code to correctly generate the
2895      digestedData type and add support for this type in PKCS7 initialization
2896      functions.
2897      [Steve Henson]
2898 
2899   *) New function PKCS7_set0_type_other() this initializes a PKCS7 
2900      structure of type "other".
2901      [Steve Henson]
2902 
2903   *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
2904      sure the loop does correctly stop and breaking ("division by zero")
2905      modulus operations are not performed. The (pre-generated) prime
2906      table crypto/bn/bn_prime.h was already correct, but it could not be
2907      re-generated on some platforms because of the "division by zero"
2908      situation in the script.
2909      [Ralf S. Engelschall]
2910 
2911   *) Update support for ECC-based TLS ciphersuites according to
2912      draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
2913      SHA-1 now is only used for "small" curves (where the
2914      representation of a field element takes up to 24 bytes); for
2915      larger curves, the field element resulting from ECDH is directly
2916      used as premaster secret.
2917      [Douglas Stebila (Sun Microsystems Laboratories)]
2918 
2919   *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
2920      curve secp160r1 to the tests.
2921      [Douglas Stebila (Sun Microsystems Laboratories)]
2922 
2923   *) Add the possibility to load symbols globally with DSO.
2924      [Götz Babin-Ebell <babin-ebell@trustcenter.de> via Richard Levitte]
2925 
2926   *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
2927      control of the error stack.
2928      [Richard Levitte]
2929 
2930   *) Add support for STORE in ENGINE.
2931      [Richard Levitte]
2932 
2933   *) Add the STORE type.  The intention is to provide a common interface
2934      to certificate and key stores, be they simple file-based stores, or
2935      HSM-type store, or LDAP stores, or...
2936      NOTE: The code is currently UNTESTED and isn't really used anywhere.
2937      [Richard Levitte]
2938 
2939   *) Add a generic structure called OPENSSL_ITEM.  This can be used to
2940      pass a list of arguments to any function as well as provide a way
2941      for a function to pass data back to the caller.
2942      [Richard Levitte]
2943 
2944   *) Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
2945      works like BUF_strdup() but can be used to duplicate a portion of
2946      a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
2947      a memory area.
2948      [Richard Levitte]
2949 
2950   *) Add the function sk_find_ex() which works like sk_find(), but will
2951      return an index to an element even if an exact match couldn't be
2952      found.  The index is guaranteed to point at the element where the
2953      searched-for key would be inserted to preserve sorting order.
2954      [Richard Levitte]
2955 
2956   *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
2957      takes an extra flags argument for optional functionality.  Currently,
2958      the following flags are defined:
2959 
2960         OBJ_BSEARCH_VALUE_ON_NOMATCH
2961         This one gets OBJ_bsearch_ex() to return a pointer to the first
2962         element where the comparing function returns a negative or zero
2963         number.
2964 
2965         OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
2966         This one gets OBJ_bsearch_ex() to return a pointer to the first
2967         element where the comparing function returns zero.  This is useful
2968         if there are more than one element where the comparing function
2969         returns zero.
2970      [Richard Levitte]
2971 
2972   *) Make it possible to create self-signed certificates with 'openssl ca'
2973      in such a way that the self-signed certificate becomes part of the
2974      CA database and uses the same mechanisms for serial number generation
2975      as all other certificate signing.  The new flag '-selfsign' enables
2976      this functionality.  Adapt CA.sh and CA.pl.in.
2977      [Richard Levitte]
2978 
2979   *) Add functionality to check the public key of a certificate request
2980      against a given private.  This is useful to check that a certificate
2981      request can be signed by that key (self-signing).
2982      [Richard Levitte]
2983 
2984   *) Make it possible to have multiple active certificates with the same
2985      subject in the CA index file.  This is done only if the keyword
2986      'unique_subject' is set to 'no' in the main CA section (default
2987      if 'CA_default') of the configuration file.  The value is saved
2988      with the database itself in a separate index attribute file,
2989      named like the index file with '.attr' appended to the name.
2990      [Richard Levitte]
2991 
2992   *) Generate muti valued AVAs using '+' notation in config files for
2993      req and dirName.
2994      [Steve Henson]
2995 
2996   *) Support for nameConstraints certificate extension.
2997      [Steve Henson]
2998 
2999   *) Support for policyConstraints certificate extension.
3000      [Steve Henson]
3001 
3002   *) Support for policyMappings certificate extension.
3003      [Steve Henson]
3004 
3005   *) Make sure the default DSA_METHOD implementation only uses its
3006      dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
3007      and change its own handlers to be NULL so as to remove unnecessary
3008      indirection. This lets alternative implementations fallback to the
3009      default implementation more easily.
3010      [Geoff Thorpe]
3011 
3012   *) Support for directoryName in GeneralName related extensions
3013      in config files.
3014      [Steve Henson]
3015 
3016   *) Make it possible to link applications using Makefile.shared.
3017      Make that possible even when linking against static libraries!
3018      [Richard Levitte]
3019 
3020   *) Support for single pass processing for S/MIME signing. This now
3021      means that S/MIME signing can be done from a pipe, in addition
3022      cleartext signing (multipart/signed type) is effectively streaming
3023      and the signed data does not need to be all held in memory.
3024 
3025      This is done with a new flag PKCS7_STREAM. When this flag is set
3026      PKCS7_sign() only initializes the PKCS7 structure and the actual signing
3027      is done after the data is output (and digests calculated) in
3028      SMIME_write_PKCS7().
3029      [Steve Henson]
3030 
3031   *) Add full support for -rpath/-R, both in shared libraries and
3032      applications, at least on the platforms where it's known how
3033      to do it.
3034      [Richard Levitte]
3035 
3036   *) In crypto/ec/ec_mult.c, implement fast point multiplication with
3037      precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
3038      will now compute a table of multiples of the generator that
3039      makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
3040      faster (notably in the case of a single point multiplication,
3041      scalar * generator).
3042      [Nils Larsch, Bodo Moeller]
3043 
3044   *) IPv6 support for certificate extensions. The various extensions
3045      which use the IP:a.b.c.d can now take IPv6 addresses using the
3046      formats of RFC1884 2.2 . IPv6 addresses are now also displayed
3047      correctly.
3048      [Steve Henson]
3049 
3050   *) Added an ENGINE that implements RSA by performing private key
3051      exponentiations with the GMP library. The conversions to and from
3052      GMP's mpz_t format aren't optimised nor are any montgomery forms
3053      cached, and on x86 it appears OpenSSL's own performance has caught up.
3054      However there are likely to be other architectures where GMP could
3055      provide a boost. This ENGINE is not built in by default, but it can be
3056      specified at Configure time and should be accompanied by the necessary
3057      linker additions, eg;
3058          ./config -DOPENSSL_USE_GMP -lgmp
3059      [Geoff Thorpe]
3060 
3061   *) "openssl engine" will not display ENGINE/DSO load failure errors when
3062      testing availability of engines with "-t" - the old behaviour is
3063      produced by increasing the feature's verbosity with "-tt".
3064      [Geoff Thorpe]
3065 
3066   *) ECDSA routines: under certain error conditions uninitialized BN objects
3067      could be freed. Solution: make sure initialization is performed early
3068      enough. (Reported and fix supplied by Nils Larsch <nla@trustcenter.de>
3069      via PR#459)
3070      [Lutz Jaenicke]
3071 
3072   *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
3073      and DH_METHOD (eg. by ENGINE implementations) to override the normal
3074      software implementations. For DSA and DH, parameter generation can
3075      also be overriden by providing the appropriate method callbacks.
3076      [Geoff Thorpe]
3077 
3078   *) Change the "progress" mechanism used in key-generation and
3079      primality testing to functions that take a new BN_GENCB pointer in
3080      place of callback/argument pairs. The new API functions have "_ex"
3081      postfixes and the older functions are reimplemented as wrappers for
3082      the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
3083      declarations of the old functions to help (graceful) attempts to
3084      migrate to the new functions. Also, the new key-generation API
3085      functions operate on a caller-supplied key-structure and return
3086      success/failure rather than returning a key or NULL - this is to
3087      help make "keygen" another member function of RSA_METHOD etc.
3088 
3089      Example for using the new callback interface:
3090 
3091           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
3092           void *my_arg = ...;
3093           BN_GENCB my_cb;
3094 
3095           BN_GENCB_set(&my_cb, my_callback, my_arg);
3096 
3097           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
3098           /* For the meaning of a, b in calls to my_callback(), see the
3099            * documentation of the function that calls the callback.
3100            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
3101            * my_callback should return 1 if it wants BN_is_prime_ex()
3102            * to continue, or 0 to stop.
3103            */
3104 
3105      [Geoff Thorpe]
3106 
3107   *) Change the ZLIB compression method to be stateful, and make it
3108      available to TLS with the number defined in 
3109      draft-ietf-tls-compression-04.txt.
3110      [Richard Levitte]
3111 
3112   *) Add the ASN.1 structures and functions for CertificatePair, which
3113      is defined as follows (according to X.509_4thEditionDraftV6.pdf):
3114 
3115      CertificatePair ::= SEQUENCE {
3116         forward         [0]     Certificate OPTIONAL,
3117         reverse         [1]     Certificate OPTIONAL,
3118         -- at least one of the pair shall be present -- }
3119 
3120      Also implement the PEM functions to read and write certificate
3121      pairs, and defined the PEM tag as "CERTIFICATE PAIR".
3122 
3123      This needed to be defined, mostly for the sake of the LDAP
3124      attribute crossCertificatePair, but may prove useful elsewhere as
3125      well.
3126      [Richard Levitte]
3127 
3128   *) Make it possible to inhibit symlinking of shared libraries in
3129      Makefile.shared, for Cygwin's sake.
3130      [Richard Levitte]
3131 
3132   *) Extend the BIGNUM API by creating a function 
3133           void BN_set_negative(BIGNUM *a, int neg);
3134      and a macro that behave like
3135           int  BN_is_negative(const BIGNUM *a);
3136 
3137      to avoid the need to access 'a->neg' directly in applications.
3138      [Nils Larsch]
3139 
3140   *) Implement fast modular reduction for pseudo-Mersenne primes
3141      used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
3142      EC_GROUP_new_curve_GFp() will now automatically use this
3143      if applicable.
3144      [Nils Larsch <nla@trustcenter.de>]
3145 
3146   *) Add new lock type (CRYPTO_LOCK_BN).
3147      [Bodo Moeller]
3148 
3149   *) Change the ENGINE framework to automatically load engines
3150      dynamically from specific directories unless they could be
3151      found to already be built in or loaded.  Move all the
3152      current engines except for the cryptodev one to a new
3153      directory engines/.
3154      The engines in engines/ are built as shared libraries if
3155      the "shared" options was given to ./Configure or ./config.
3156      Otherwise, they are inserted in libcrypto.a.
3157      /usr/local/ssl/engines is the default directory for dynamic
3158      engines, but that can be overriden at configure time through
3159      the usual use of --prefix and/or --openssldir, and at run
3160      time with the environment variable OPENSSL_ENGINES.
3161      [Geoff Thorpe and Richard Levitte]
3162 
3163   *) Add Makefile.shared, a helper makefile to build shared
3164      libraries.  Addapt Makefile.org.
3165      [Richard Levitte]
3166 
3167   *) Add version info to Win32 DLLs.
3168      [Peter 'Luna' Runestig" <peter@runestig.com>]
3169 
3170   *) Add new 'medium level' PKCS#12 API. Certificates and keys
3171      can be added using this API to created arbitrary PKCS#12
3172      files while avoiding the low level API.
3173 
3174      New options to PKCS12_create(), key or cert can be NULL and
3175      will then be omitted from the output file. The encryption
3176      algorithm NIDs can be set to -1 for no encryption, the mac
3177      iteration count can be set to 0 to omit the mac.
3178 
3179      Enhance pkcs12 utility by making the -nokeys and -nocerts
3180      options work when creating a PKCS#12 file. New option -nomac
3181      to omit the mac, NONE can be set for an encryption algorithm.
3182      New code is modified to use the enhanced PKCS12_create()
3183      instead of the low level API.
3184      [Steve Henson]
3185 
3186   *) Extend ASN1 encoder to support indefinite length constructed
3187      encoding. This can output sequences tags and octet strings in
3188      this form. Modify pk7_asn1.c to support indefinite length
3189      encoding. This is experimental and needs additional code to
3190      be useful, such as an ASN1 bio and some enhanced streaming
3191      PKCS#7 code.
3192 
3193      Extend template encode functionality so that tagging is passed
3194      down to the template encoder.
3195      [Steve Henson]
3196 
3197   *) Let 'openssl req' fail if an argument to '-newkey' is not
3198      recognized instead of using RSA as a default.
3199      [Bodo Moeller]
3200 
3201   *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
3202      As these are not official, they are not included in "ALL";
3203      the "ECCdraft" ciphersuite group alias can be used to select them.
3204      [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
3205 
3206   *) Add ECDH engine support.
3207      [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
3208 
3209   *) Add ECDH in new directory crypto/ecdh/.
3210      [Douglas Stebila (Sun Microsystems Laboratories)]
3211 
3212   *) Let BN_rand_range() abort with an error after 100 iterations
3213      without success (which indicates a broken PRNG).
3214      [Bodo Moeller]
3215 
3216   *) Change BN_mod_sqrt() so that it verifies that the input value
3217      is really the square of the return value.  (Previously,
3218      BN_mod_sqrt would show GIGO behaviour.)
3219      [Bodo Moeller]
3220 
3221   *) Add named elliptic curves over binary fields from X9.62, SECG,
3222      and WAP/WTLS; add OIDs that were still missing.
3223 
3224      [Sheueling Chang Shantz and Douglas Stebila
3225      (Sun Microsystems Laboratories)]
3226 
3227   *) Extend the EC library for elliptic curves over binary fields
3228      (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
3229      New EC_METHOD:
3230 
3231           EC_GF2m_simple_method
3232 
3233      New API functions:
3234 
3235           EC_GROUP_new_curve_GF2m
3236           EC_GROUP_set_curve_GF2m
3237           EC_GROUP_get_curve_GF2m
3238           EC_POINT_set_affine_coordinates_GF2m
3239           EC_POINT_get_affine_coordinates_GF2m
3240           EC_POINT_set_compressed_coordinates_GF2m
3241 
3242      Point compression for binary fields is disabled by default for
3243      patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
3244      enable it).
3245 
3246      As binary polynomials are represented as BIGNUMs, various members
3247      of the EC_GROUP and EC_POINT data structures can be shared
3248      between the implementations for prime fields and binary fields;
3249      the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
3250      are essentially identical to their ..._GFp counterparts.
3251      (For simplicity, the '..._GFp' prefix has been dropped from
3252      various internal method names.)
3253 
3254      An internal 'field_div' method (similar to 'field_mul' and
3255      'field_sqr') has been added; this is used only for binary fields.
3256 
3257      [Sheueling Chang Shantz and Douglas Stebila
3258      (Sun Microsystems Laboratories)]
3259 
3260   *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
3261      through methods ('mul', 'precompute_mult').
3262 
3263      The generic implementations (now internally called 'ec_wNAF_mul'
3264      and 'ec_wNAF_precomputed_mult') remain the default if these
3265      methods are undefined.
3266 
3267      [Sheueling Chang Shantz and Douglas Stebila
3268      (Sun Microsystems Laboratories)]
3269 
3270   *) New function EC_GROUP_get_degree, which is defined through
3271      EC_METHOD.  For curves over prime fields, this returns the bit
3272      length of the modulus.
3273 
3274      [Sheueling Chang Shantz and Douglas Stebila
3275      (Sun Microsystems Laboratories)]
3276 
3277   *) New functions EC_GROUP_dup, EC_POINT_dup.
3278      (These simply call ..._new  and ..._copy).
3279 
3280      [Sheueling Chang Shantz and Douglas Stebila
3281      (Sun Microsystems Laboratories)]
3282 
3283   *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
3284      Polynomials are represented as BIGNUMs (where the sign bit is not
3285      used) in the following functions [macros]:  
3286 
3287           BN_GF2m_add
3288           BN_GF2m_sub             [= BN_GF2m_add]
3289           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
3290           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
3291           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
3292           BN_GF2m_mod_inv
3293           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
3294           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
3295           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
3296           BN_GF2m_cmp             [= BN_ucmp]
3297 
3298      (Note that only the 'mod' functions are actually for fields GF(2^m).
3299      BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
3300 
3301      For some functions, an the irreducible polynomial defining a
3302      field can be given as an 'unsigned int[]' with strictly
3303      decreasing elements giving the indices of those bits that are set;
3304      i.e., p[] represents the polynomial
3305           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
3306      where
3307           p[0] > p[1] > ... > p[k] = 0.
3308      This applies to the following functions:
3309 
3310           BN_GF2m_mod_arr
3311           BN_GF2m_mod_mul_arr
3312           BN_GF2m_mod_sqr_arr
3313           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
3314           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
3315           BN_GF2m_mod_exp_arr
3316           BN_GF2m_mod_sqrt_arr
3317           BN_GF2m_mod_solve_quad_arr
3318           BN_GF2m_poly2arr
3319           BN_GF2m_arr2poly
3320 
3321      Conversion can be performed by the following functions:
3322 
3323           BN_GF2m_poly2arr
3324           BN_GF2m_arr2poly
3325 
3326      bntest.c has additional tests for binary polynomial arithmetic.
3327 
3328      Two implementations for BN_GF2m_mod_div() are available.
3329      The default algorithm simply uses BN_GF2m_mod_inv() and
3330      BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
3331      if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
3332      copyright notice in crypto/bn/bn_gf2m.c before enabling it).
3333 
3334      [Sheueling Chang Shantz and Douglas Stebila
3335      (Sun Microsystems Laboratories)]
3336 
3337   *) Add new error code 'ERR_R_DISABLED' that can be used when some
3338      functionality is disabled at compile-time.
3339      [Douglas Stebila <douglas.stebila@sun.com>]
3340 
3341   *) Change default behaviour of 'openssl asn1parse' so that more
3342      information is visible when viewing, e.g., a certificate:
3343 
3344      Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
3345      mode the content of non-printable OCTET STRINGs is output in a
3346      style similar to INTEGERs, but with '[HEX DUMP]' prepended to
3347      avoid the appearance of a printable string.
3348      [Nils Larsch <nla@trustcenter.de>]
3349 
3350   *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
3351      functions
3352           EC_GROUP_set_asn1_flag()
3353           EC_GROUP_get_asn1_flag()
3354           EC_GROUP_set_point_conversion_form()
3355           EC_GROUP_get_point_conversion_form()
3356      These control ASN1 encoding details:
3357      - Curves (i.e., groups) are encoded explicitly unless asn1_flag
3358        has been set to OPENSSL_EC_NAMED_CURVE.
3359      - Points are encoded in uncompressed form by default; options for
3360        asn1_for are as for point2oct, namely
3361           POINT_CONVERSION_COMPRESSED
3362           POINT_CONVERSION_UNCOMPRESSED
3363           POINT_CONVERSION_HYBRID
3364 
3365      Also add 'seed' and 'seed_len' members to EC_GROUP with access
3366      functions
3367           EC_GROUP_set_seed()
3368           EC_GROUP_get0_seed()
3369           EC_GROUP_get_seed_len()
3370      This is used only for ASN1 purposes (so far).
3371      [Nils Larsch <nla@trustcenter.de>]
3372 
3373   *) Add 'field_type' member to EC_METHOD, which holds the NID
3374      of the appropriate field type OID.  The new function
3375      EC_METHOD_get_field_type() returns this value.
3376      [Nils Larsch <nla@trustcenter.de>]
3377 
3378   *) Add functions 
3379           EC_POINT_point2bn()
3380           EC_POINT_bn2point()
3381           EC_POINT_point2hex()
3382           EC_POINT_hex2point()
3383      providing useful interfaces to EC_POINT_point2oct() and
3384      EC_POINT_oct2point().
3385      [Nils Larsch <nla@trustcenter.de>]
3386 
3387   *) Change internals of the EC library so that the functions
3388           EC_GROUP_set_generator()
3389           EC_GROUP_get_generator()
3390           EC_GROUP_get_order()
3391           EC_GROUP_get_cofactor()
3392      are implemented directly in crypto/ec/ec_lib.c and not dispatched
3393      to methods, which would lead to unnecessary code duplication when
3394      adding different types of curves.
3395      [Nils Larsch <nla@trustcenter.de> with input by Bodo Moeller]
3396 
3397   *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
3398      arithmetic, and such that modified wNAFs are generated
3399      (which avoid length expansion in many cases).
3400      [Bodo Moeller]
3401 
3402   *) Add a function EC_GROUP_check_discriminant() (defined via
3403      EC_METHOD) that verifies that the curve discriminant is non-zero.
3404 
3405      Add a function EC_GROUP_check() that makes some sanity tests
3406      on a EC_GROUP, its generator and order.  This includes
3407      EC_GROUP_check_discriminant().
3408      [Nils Larsch <nla@trustcenter.de>]
3409 
3410   *) Add ECDSA in new directory crypto/ecdsa/.
3411 
3412      Add applications 'openssl ecparam' and 'openssl ecdsa'
3413      (these are based on 'openssl dsaparam' and 'openssl dsa').
3414 
3415      ECDSA support is also included in various other files across the
3416      library.  Most notably,
3417      - 'openssl req' now has a '-newkey ecdsa:file' option;
3418      - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
3419      - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
3420        d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
3421        them suitable for ECDSA where domain parameters must be
3422        extracted before the specific public key;
3423      - ECDSA engine support has been added.
3424      [Nils Larsch <nla@trustcenter.de>]
3425 
3426   *) Include some named elliptic curves, and add OIDs from X9.62,
3427      SECG, and WAP/WTLS.  Each curve can be obtained from the new
3428      function
3429           EC_GROUP_new_by_curve_name(),
3430      and the list of available named curves can be obtained with
3431           EC_get_builtin_curves().
3432      Also add a 'curve_name' member to EC_GROUP objects, which can be
3433      accessed via
3434          EC_GROUP_set_curve_name()
3435          EC_GROUP_get_curve_name()
3436      [Nils Larsch <larsch@trustcenter.de, Bodo Moeller]
3437  
3438   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
3439      was actually never needed) and in BN_mul().  The removal in BN_mul()
3440      required a small change in bn_mul_part_recursive() and the addition
3441      of the functions bn_cmp_part_words(), bn_sub_part_words() and
3442      bn_add_part_words(), which do the same thing as bn_cmp_words(),
3443      bn_sub_words() and bn_add_words() except they take arrays with
3444      differing sizes.
3445      [Richard Levitte]
3446 
3447  Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
3448 
3449   *) Cleanse PEM buffers before freeing them since they may contain 
3450      sensitive data.
3451      [Benjamin Bennett <ben@psc.edu>]
3452 
3453   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
3454      a ciphersuite string such as "DEFAULT:RSA" cannot enable
3455      authentication-only ciphersuites.
3456      [Bodo Moeller]
3457 
3458   *) Since AES128 and AES256 share a single mask bit in the logic of
3459      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
3460      kludge to work properly if AES128 is available and AES256 isn't.
3461      [Victor Duchovni]
3462 
3463   *) Expand security boundary to match 1.1.1 module.
3464      [Steve Henson]
3465 
3466   *) Remove redundant features: hash file source, editing of test vectors
3467      modify fipsld to use external fips_premain.c signature.
3468      [Steve Henson]
3469 
3470   *) New perl script mkfipsscr.pl to create shell scripts or batch files to
3471      run algorithm test programs.
3472      [Steve Henson]
3473 
3474   *) Make algorithm test programs more tolerant of whitespace.
3475      [Steve Henson]
3476 
3477   *) Have SSL/TLS server implementation tolerate "mismatched" record
3478      protocol version while receiving ClientHello even if the
3479      ClientHello is fragmented.  (The server can't insist on the
3480      particular protocol version it has chosen before the ServerHello
3481      message has informed the client about his choice.)
3482      [Bodo Moeller]
3483 
3484   *) Load error codes if they are not already present instead of using a
3485      static variable. This allows them to be cleanly unloaded and reloaded.
3486      [Steve Henson]
3487 
3488  Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
3489 
3490   *) Introduce limits to prevent malicious keys being able to
3491      cause a denial of service.  (CVE-2006-2940)
3492      [Steve Henson, Bodo Moeller]
3493 
3494   *) Fix ASN.1 parsing of certain invalid structures that can result
3495      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
3496 
3497   *) Fix buffer overflow in SSL_get_shared_ciphers() function. 
3498      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
3499 
3500   *) Fix SSL client code which could crash if connecting to a
3501      malicious SSLv2 server.  (CVE-2006-4343)
3502      [Tavis Ormandy and Will Drewry, Google Security Team]
3503 
3504   *) Change ciphersuite string processing so that an explicit
3505      ciphersuite selects this one ciphersuite (so that "AES256-SHA"
3506      will no longer include "AES128-SHA"), and any other similar
3507      ciphersuite (same bitmap) from *other* protocol versions (so that
3508      "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
3509      SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
3510      changes from 0.9.8b and 0.9.8d.
3511      [Bodo Moeller]
3512 
3513  Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
3514 
3515   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
3516      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
3517 
3518   *) Change the Unix randomness entropy gathering to use poll() when
3519      possible instead of select(), since the latter has some
3520      undesirable limitations.
3521      [Darryl Miles via Richard Levitte and Bodo Moeller]
3522 
3523   *) Disable rogue ciphersuites:
3524 
3525       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
3526       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
3527       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
3528 
3529      The latter two were purportedly from
3530      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
3531      appear there.
3532 
3533      Also deactive the remaining ciphersuites from
3534      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
3535      unofficial, and the ID has long expired.
3536      [Bodo Moeller]
3537 
3538   *) Fix RSA blinding Heisenbug (problems sometimes occured on
3539      dual-core machines) and other potential thread-safety issues.
3540      [Bodo Moeller]
3541 
3542  Changes between 0.9.7i and 0.9.7j  [04 May 2006]
3543 
3544   *) Adapt fipsld and the build system to link against the validated FIPS
3545      module in FIPS mode.
3546      [Steve Henson]
3547 
3548   *) Fixes for VC++ 2005 build under Windows.
3549      [Steve Henson]
3550 
3551   *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 
3552      from a Windows bash shell such as MSYS. It is autodetected from the
3553      "config" script when run from a VC++ environment. Modify standard VC++
3554      build to use fipscanister.o from the GNU make build. 
3555      [Steve Henson]
3556 
3557  Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
3558 
3559   *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
3560      The value now differs depending on if you build for FIPS or not.
3561      BEWARE!  A program linked with a shared FIPSed libcrypto can't be
3562      safely run with a non-FIPSed libcrypto, as it may crash because of
3563      the difference induced by this change.
3564      [Andy Polyakov]
3565 
3566  Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
3567 
3568   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
3569      (part of SSL_OP_ALL).  This option used to disable the
3570      countermeasure against man-in-the-middle protocol-version
3571      rollback in the SSL 2.0 server implementation, which is a bad
3572      idea.  (CVE-2005-2969)
3573 
3574      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
3575      for Information Security, National Institute of Advanced Industrial
3576      Science and Technology [AIST], Japan)]
3577 
3578   *) Minimal support for X9.31 signatures and PSS padding modes. This is
3579      mainly for FIPS compliance and not fully integrated at this stage.
3580      [Steve Henson]
3581 
3582   *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
3583      the exponentiation using a fixed-length exponent.  (Otherwise,
3584      the information leaked through timing could expose the secret key
3585      after many signatures; cf. Bleichenbacher's attack on DSA with
3586      biased k.)
3587      [Bodo Moeller]
3588 
3589   *) Make a new fixed-window mod_exp implementation the default for
3590      RSA, DSA, and DH private-key operations so that the sequence of
3591      squares and multiplies and the memory access pattern are
3592      independent of the particular secret key.  This will mitigate
3593      cache-timing and potential related attacks.
3594 
3595      BN_mod_exp_mont_consttime() is the new exponentiation implementation,
3596      and this is automatically used by BN_mod_exp_mont() if the new flag
3597      BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
3598      will use this BN flag for private exponents unless the flag
3599      RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
3600      DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
3601 
3602      [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
3603 
3604   *) Change the client implementation for SSLv23_method() and
3605      SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
3606      Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
3607      (Previously, the SSL 2.0 backwards compatible Client Hello
3608      message format would be used even with SSL_OP_NO_SSLv2.)
3609      [Bodo Moeller]
3610 
3611   *) Add support for smime-type MIME parameter in S/MIME messages which some
3612      clients need.
3613      [Steve Henson]
3614 
3615   *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
3616      a threadsafe manner. Modify rsa code to use new function and add calls
3617      to dsa and dh code (which had race conditions before).
3618      [Steve Henson]
3619 
3620   *) Include the fixed error library code in the C error file definitions
3621      instead of fixing them up at runtime. This keeps the error code
3622      structures constant.
3623      [Steve Henson]
3624 
3625  Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
3626 
3627   [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
3628   OpenSSL 0.9.8.]
3629 
3630   *) Fixes for newer kerberos headers. NB: the casts are needed because
3631      the 'length' field is signed on one version and unsigned on another
3632      with no (?) obvious way to tell the difference, without these VC++
3633      complains. Also the "definition" of FAR (blank) is no longer included
3634      nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
3635      some needed definitions.
3636      [Steve Henson]
3637 
3638   *) Undo Cygwin change.
3639      [Ulf Möller]
3640 
3641   *) Added support for proxy certificates according to RFC 3820.
3642      Because they may be a security thread to unaware applications,
3643      they must be explicitely allowed in run-time.  See
3644      docs/HOWTO/proxy_certificates.txt for further information.
3645      [Richard Levitte]
3646 
3647  Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
3648 
3649   *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
3650      server and client random values. Previously
3651      (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
3652      less random data when sizeof(time_t) > 4 (some 64 bit platforms).
3653 
3654      This change has negligible security impact because:
3655 
3656      1. Server and client random values still have 24 bytes of pseudo random
3657         data.
3658 
3659      2. Server and client random values are sent in the clear in the initial
3660         handshake.
3661 
3662      3. The master secret is derived using the premaster secret (48 bytes in
3663         size for static RSA ciphersuites) as well as client server and random
3664         values.
3665 
3666      The OpenSSL team would like to thank the UK NISCC for bringing this issue
3667      to our attention. 
3668 
3669      [Stephen Henson, reported by UK NISCC]
3670 
3671   *) Use Windows randomness collection on Cygwin.
3672      [Ulf Möller]
3673 
3674   *) Fix hang in EGD/PRNGD query when communication socket is closed
3675      prematurely by EGD/PRNGD.
3676      [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
3677 
3678   *) Prompt for pass phrases when appropriate for PKCS12 input format.
3679      [Steve Henson]
3680 
3681   *) Back-port of selected performance improvements from development
3682      branch, as well as improved support for PowerPC platforms.
3683      [Andy Polyakov]
3684 
3685   *) Add lots of checks for memory allocation failure, error codes to indicate
3686      failure and freeing up memory if a failure occurs.
3687      [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
3688 
3689   *) Add new -passin argument to dgst.
3690      [Steve Henson]
3691 
3692   *) Perform some character comparisons of different types in X509_NAME_cmp:
3693      this is needed for some certificates that reencode DNs into UTF8Strings
3694      (in violation of RFC3280) and can't or wont issue name rollover
3695      certificates.
3696      [Steve Henson]
3697 
3698   *) Make an explicit check during certificate validation to see that
3699      the CA setting in each certificate on the chain is correct.  As a
3700      side effect always do the following basic checks on extensions,
3701      not just when there's an associated purpose to the check:
3702 
3703       - if there is an unhandled critical extension (unless the user
3704         has chosen to ignore this fault)
3705       - if the path length has been exceeded (if one is set at all)
3706       - that certain extensions fit the associated purpose (if one has
3707         been given)
3708      [Richard Levitte]
3709 
3710  Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
3711 
3712   *) Avoid a race condition when CRLs are checked in a multi threaded 
3713      environment. This would happen due to the reordering of the revoked
3714      entries during signature checking and serial number lookup. Now the
3715      encoding is cached and the serial number sort performed under a lock.
3716      Add new STACK function sk_is_sorted().
3717      [Steve Henson]
3718 
3719   *) Add Delta CRL to the extension code.
3720      [Steve Henson]
3721 
3722   *) Various fixes to s3_pkt.c so alerts are sent properly.
3723      [David Holmes <d.holmes@f5.com>]
3724 
3725   *) Reduce the chances of duplicate issuer name and serial numbers (in
3726      violation of RFC3280) using the OpenSSL certificate creation utilities.
3727      This is done by creating a random 64 bit value for the initial serial
3728      number when a serial number file is created or when a self signed
3729      certificate is created using 'openssl req -x509'. The initial serial
3730      number file is created using 'openssl x509 -next_serial' in CA.pl
3731      rather than being initialized to 1.
3732      [Steve Henson]
3733 
3734  Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
3735 
3736   *) Fix null-pointer assignment in do_change_cipher_spec() revealed           
3737      by using the Codenomicon TLS Test Tool (CVE-2004-0079)                    
3738      [Joe Orton, Steve Henson]   
3739 
3740   *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
3741      (CVE-2004-0112)
3742      [Joe Orton, Steve Henson]   
3743 
3744   *) Make it possible to have multiple active certificates with the same
3745      subject in the CA index file.  This is done only if the keyword
3746      'unique_subject' is set to 'no' in the main CA section (default
3747      if 'CA_default') of the configuration file.  The value is saved
3748      with the database itself in a separate index attribute file,
3749      named like the index file with '.attr' appended to the name.
3750      [Richard Levitte]
3751 
3752   *) X509 verify fixes. Disable broken certificate workarounds when 
3753      X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
3754      keyUsage extension present. Don't accept CRLs with unhandled critical
3755      extensions: since verify currently doesn't process CRL extensions this
3756      rejects a CRL with *any* critical extensions. Add new verify error codes
3757      for these cases.
3758      [Steve Henson]
3759 
3760   *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
3761      A clarification of RFC2560 will require the use of OCTET STRINGs and 
3762      some implementations cannot handle the current raw format. Since OpenSSL
3763      copies and compares OCSP nonces as opaque blobs without any attempt at
3764      parsing them this should not create any compatibility issues.
3765      [Steve Henson]
3766 
3767   *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
3768      calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
3769      this HMAC (and other) operations are several times slower than OpenSSL
3770      < 0.9.7.
3771      [Steve Henson]
3772 
3773   *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
3774      [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
3775 
3776   *) Use the correct content when signing type "other".
3777      [Steve Henson]
3778 
3779  Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
3780 
3781   *) Fix various bugs revealed by running the NISCC test suite:
3782 
3783      Stop out of bounds reads in the ASN1 code when presented with
3784      invalid tags (CVE-2003-0543 and CVE-2003-0544).
3785      
3786      Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
3787 
3788      If verify callback ignores invalid public key errors don't try to check
3789      certificate signature with the NULL public key.
3790 
3791      [Steve Henson]
3792 
3793   *) New -ignore_err option in ocsp application to stop the server
3794      exiting on the first error in a request.
3795      [Steve Henson]
3796 
3797   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
3798      if the server requested one: as stated in TLS 1.0 and SSL 3.0
3799      specifications.
3800      [Steve Henson]
3801 
3802   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
3803      extra data after the compression methods not only for TLS 1.0
3804      but also for SSL 3.0 (as required by the specification).
3805      [Bodo Moeller; problem pointed out by Matthias Loepfe]
3806 
3807   *) Change X509_certificate_type() to mark the key as exported/exportable
3808      when it's 512 *bits* long, not 512 bytes.
3809      [Richard Levitte]
3810 
3811   *) Change AES_cbc_encrypt() so it outputs exact multiple of
3812      blocks during encryption.
3813      [Richard Levitte]
3814 
3815   *) Various fixes to base64 BIO and non blocking I/O. On write 
3816      flushes were not handled properly if the BIO retried. On read
3817      data was not being buffered properly and had various logic bugs.
3818      This also affects blocking I/O when the data being decoded is a
3819      certain size.
3820      [Steve Henson]
3821 
3822   *) Various S/MIME bugfixes and compatibility changes:
3823      output correct application/pkcs7 MIME type if
3824      PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
3825      Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
3826      of files as .eml work). Correctly handle very long lines in MIME
3827      parser.
3828      [Steve Henson]
3829 
3830  Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
3831 
3832   *) Countermeasure against the Klima-Pokorny-Rosa extension of
3833      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
3834      a protocol version number mismatch like a decryption error
3835      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
3836      [Bodo Moeller]
3837 
3838   *) Turn on RSA blinding by default in the default implementation
3839      to avoid a timing attack. Applications that don't want it can call
3840      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
3841      They would be ill-advised to do so in most cases.
3842      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
3843 
3844   *) Change RSA blinding code so that it works when the PRNG is not
3845      seeded (in this case, the secret RSA exponent is abused as
3846      an unpredictable seed -- if it is not unpredictable, there
3847      is no point in blinding anyway).  Make RSA blinding thread-safe
3848      by remembering the creator's thread ID in rsa->blinding and
3849      having all other threads use local one-time blinding factors
3850      (this requires more computation than sharing rsa->blinding, but
3851      avoids excessive locking; and if an RSA object is not shared
3852      between threads, blinding will still be very fast).
3853      [Bodo Moeller]
3854 
3855   *) Fixed a typo bug that would cause ENGINE_set_default() to set an
3856      ENGINE as defaults for all supported algorithms irrespective of
3857      the 'flags' parameter. 'flags' is now honoured, so applications
3858      should make sure they are passing it correctly.
3859      [Geoff Thorpe]
3860 
3861   *) Target "mingw" now allows native Windows code to be generated in
3862      the Cygwin environment as well as with the MinGW compiler.
3863      [Ulf Moeller] 
3864 
3865  Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
3866 
3867   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
3868      via timing by performing a MAC computation even if incorrrect
3869      block cipher padding has been found.  This is a countermeasure
3870      against active attacks where the attacker has to distinguish
3871      between bad padding and a MAC verification error. (CVE-2003-0078)
3872 
3873      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
3874      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
3875      Martin Vuagnoux (EPFL, Ilion)]
3876 
3877   *) Make the no-err option work as intended.  The intention with no-err
3878      is not to have the whole error stack handling routines removed from
3879      libcrypto, it's only intended to remove all the function name and
3880      reason texts, thereby removing some of the footprint that may not
3881      be interesting if those errors aren't displayed anyway.
3882 
3883      NOTE: it's still possible for any application or module to have it's
3884      own set of error texts inserted.  The routines are there, just not
3885      used by default when no-err is given.
3886      [Richard Levitte]
3887 
3888   *) Add support for FreeBSD on IA64.
3889      [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
3890 
3891   *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
3892      Kerberos function mit_des_cbc_cksum().  Before this change,
3893      the value returned by DES_cbc_cksum() was like the one from
3894      mit_des_cbc_cksum(), except the bytes were swapped.
3895      [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
3896 
3897   *) Allow an application to disable the automatic SSL chain building.
3898      Before this a rather primitive chain build was always performed in
3899      ssl3_output_cert_chain(): an application had no way to send the 
3900      correct chain if the automatic operation produced an incorrect result.
3901 
3902      Now the chain builder is disabled if either:
3903 
3904      1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
3905 
3906      2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
3907 
3908      The reasoning behind this is that an application would not want the
3909      auto chain building to take place if extra chain certificates are
3910      present and it might also want a means of sending no additional
3911      certificates (for example the chain has two certificates and the
3912      root is omitted).
3913      [Steve Henson]
3914 
3915   *) Add the possibility to build without the ENGINE framework.
3916      [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
3917 
3918   *) Under Win32 gmtime() can return NULL: check return value in
3919      OPENSSL_gmtime(). Add error code for case where gmtime() fails.
3920      [Steve Henson]
3921 
3922   *) DSA routines: under certain error conditions uninitialized BN objects
3923      could be freed. Solution: make sure initialization is performed early
3924      enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
3925      Nils Larsch <nla@trustcenter.de> via PR#459)
3926      [Lutz Jaenicke]
3927 
3928   *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
3929      checked on reconnect on the client side, therefore session resumption
3930      could still fail with a "ssl session id is different" error. This
3931      behaviour is masked when SSL_OP_ALL is used due to
3932      SSL_OP_MICROSOFT_SESS_ID_BUG being set.
3933      Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
3934      followup to PR #377.
3935      [Lutz Jaenicke]
3936 
3937   *) IA-32 assembler support enhancements: unified ELF targets, support
3938      for SCO/Caldera platforms, fix for Cygwin shared build.
3939      [Andy Polyakov]
3940 
3941   *) Add support for FreeBSD on sparc64.  As a consequence, support for
3942      FreeBSD on non-x86 processors is separate from x86 processors on
3943      the config script, much like the NetBSD support.
3944      [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
3945 
3946  Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
3947 
3948   [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
3949   OpenSSL 0.9.7.]
3950 
3951   *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
3952      code (06) was taken as the first octet of the session ID and the last
3953      octet was ignored consequently. As a result SSLv2 client side session
3954      caching could not have worked due to the session ID mismatch between
3955      client and server.
3956      Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
3957      PR #377.
3958      [Lutz Jaenicke]
3959 
3960   *) Change the declaration of needed Kerberos libraries to use EX_LIBS
3961      instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
3962      removed entirely.
3963      [Richard Levitte]
3964 
3965   *) The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
3966      seems that in spite of existing for more than a year, many application
3967      author have done nothing to provide the necessary callbacks, which
3968      means that this particular engine will not work properly anywhere.
3969      This is a very unfortunate situation which forces us, in the name
3970      of usability, to give the hw_ncipher.c a static lock, which is part
3971      of libcrypto.
3972      NOTE: This is for the 0.9.7 series ONLY.  This hack will never
3973      appear in 0.9.8 or later.  We EXPECT application authors to have
3974      dealt properly with this when 0.9.8 is released (unless we actually
3975      make such changes in the libcrypto locking code that changes will
3976      have to be made anyway).
3977      [Richard Levitte]
3978 
3979   *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
3980      octets have been read, EOF or an error occurs. Without this change
3981      some truncated ASN1 structures will not produce an error.
3982      [Steve Henson]
3983 
3984   *) Disable Heimdal support, since it hasn't been fully implemented.
3985      Still give the possibility to force the use of Heimdal, but with
3986      warnings and a request that patches get sent to openssl-dev.
3987      [Richard Levitte]
3988 
3989   *) Add the VC-CE target, introduce the WINCE sysname, and add
3990      INSTALL.WCE and appropriate conditionals to make it build.
3991      [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
3992 
3993   *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
3994      cygssl-x.y.z.dll, where x, y and z are the major, minor and
3995      edit numbers of the version.
3996      [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
3997 
3998   *) Introduce safe string copy and catenation functions
3999      (BUF_strlcpy() and BUF_strlcat()).
4000      [Ben Laurie (CHATS) and Richard Levitte]
4001 
4002   *) Avoid using fixed-size buffers for one-line DNs.
4003      [Ben Laurie (CHATS)]
4004 
4005   *) Add BUF_MEM_grow_clean() to avoid information leakage when
4006      resizing buffers containing secrets, and use where appropriate.
4007      [Ben Laurie (CHATS)]
4008 
4009   *) Avoid using fixed size buffers for configuration file location.
4010      [Ben Laurie (CHATS)]
4011 
4012   *) Avoid filename truncation for various CA files.
4013      [Ben Laurie (CHATS)]
4014 
4015   *) Use sizeof in preference to magic numbers.
4016      [Ben Laurie (CHATS)]
4017 
4018   *) Avoid filename truncation in cert requests.
4019      [Ben Laurie (CHATS)]
4020 
4021   *) Add assertions to check for (supposedly impossible) buffer
4022      overflows.
4023      [Ben Laurie (CHATS)]
4024 
4025   *) Don't cache truncated DNS entries in the local cache (this could
4026      potentially lead to a spoofing attack).
4027      [Ben Laurie (CHATS)]
4028 
4029   *) Fix various buffers to be large enough for hex/decimal
4030      representations in a platform independent manner.
4031      [Ben Laurie (CHATS)]
4032 
4033   *) Add CRYPTO_realloc_clean() to avoid information leakage when
4034      resizing buffers containing secrets, and use where appropriate.
4035      [Ben Laurie (CHATS)]
4036 
4037   *) Add BIO_indent() to avoid much slightly worrying code to do
4038      indents.
4039      [Ben Laurie (CHATS)]
4040 
4041   *) Convert sprintf()/BIO_puts() to BIO_printf().
4042      [Ben Laurie (CHATS)]
4043 
4044   *) buffer_gets() could terminate with the buffer only half
4045      full. Fixed.
4046      [Ben Laurie (CHATS)]
4047 
4048   *) Add assertions to prevent user-supplied crypto functions from
4049      overflowing internal buffers by having large block sizes, etc.
4050      [Ben Laurie (CHATS)]
4051 
4052   *) New OPENSSL_assert() macro (similar to assert(), but enabled
4053      unconditionally).
4054      [Ben Laurie (CHATS)]
4055 
4056   *) Eliminate unused copy of key in RC4.
4057      [Ben Laurie (CHATS)]
4058 
4059   *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
4060      [Ben Laurie (CHATS)]
4061 
4062   *) Fix off-by-one error in EGD path.
4063      [Ben Laurie (CHATS)]
4064 
4065   *) If RANDFILE path is too long, ignore instead of truncating.
4066      [Ben Laurie (CHATS)]
4067 
4068   *) Eliminate unused and incorrectly sized X.509 structure
4069      CBCParameter.
4070      [Ben Laurie (CHATS)]
4071 
4072   *) Eliminate unused and dangerous function knumber().
4073      [Ben Laurie (CHATS)]
4074 
4075   *) Eliminate unused and dangerous structure, KSSL_ERR.
4076      [Ben Laurie (CHATS)]
4077 
4078   *) Protect against overlong session ID context length in an encoded
4079      session object. Since these are local, this does not appear to be
4080      exploitable.
4081      [Ben Laurie (CHATS)]
4082 
4083   *) Change from security patch (see 0.9.6e below) that did not affect
4084      the 0.9.6 release series:
4085 
4086      Remote buffer overflow in SSL3 protocol - an attacker could
4087      supply an oversized master key in Kerberos-enabled versions.
4088      (CVE-2002-0657)
4089      [Ben Laurie (CHATS)]
4090 
4091   *) Change the SSL kerb5 codes to match RFC 2712.
4092      [Richard Levitte]
4093 
4094   *) Make -nameopt work fully for req and add -reqopt switch.
4095      [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
4096 
4097   *) The "block size" for block ciphers in CFB and OFB mode should be 1.
4098      [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
4099 
4100   *) Make sure tests can be performed even if the corresponding algorithms
4101      have been removed entirely.  This was also the last step to make
4102      OpenSSL compilable with DJGPP under all reasonable conditions.
4103      [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
4104 
4105   *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
4106      to allow version independent disabling of normally unselected ciphers,
4107      which may be activated as a side-effect of selecting a single cipher.
4108 
4109      (E.g., cipher list string "RSA" enables ciphersuites that are left
4110      out of "ALL" because they do not provide symmetric encryption.
4111      "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
4112      [Lutz Jaenicke, Bodo Moeller]
4113 
4114   *) Add appropriate support for separate platform-dependent build
4115      directories.  The recommended way to make a platform-dependent
4116      build directory is the following (tested on Linux), maybe with
4117      some local tweaks:
4118 
4119         # Place yourself outside of the OpenSSL source tree.  In
4120         # this example, the environment variable OPENSSL_SOURCE
4121         # is assumed to contain the absolute OpenSSL source directory.
4122         mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
4123         cd objtree/"`uname -s`-`uname -r`-`uname -m`"
4124         (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
4125                 mkdir -p `dirname $F`
4126                 ln -s $OPENSSL_SOURCE/$F $F
4127         done
4128 
4129      To be absolutely sure not to disturb the source tree, a "make clean"
4130      is a good thing.  If it isn't successfull, don't worry about it,
4131      it probably means the source directory is very clean.
4132      [Richard Levitte]
4133 
4134   *) Make sure any ENGINE control commands make local copies of string
4135      pointers passed to them whenever necessary. Otherwise it is possible
4136      the caller may have overwritten (or deallocated) the original string
4137      data when a later ENGINE operation tries to use the stored values.
4138      [Götz Babin-Ebell <babinebell@trustcenter.de>]
4139 
4140   *) Improve diagnostics in file reading and command-line digests.
4141      [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
4142 
4143   *) Add AES modes CFB and OFB to the object database.  Correct an
4144      error in AES-CFB decryption.
4145      [Richard Levitte]
4146 
4147   *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 
4148      allows existing EVP_CIPHER_CTX structures to be reused after
4149      calling EVP_*Final(). This behaviour is used by encryption
4150      BIOs and some applications. This has the side effect that
4151      applications must explicitly clean up cipher contexts with
4152      EVP_CIPHER_CTX_cleanup() or they will leak memory.
4153      [Steve Henson]
4154 
4155   *) Check the values of dna and dnb in bn_mul_recursive before calling
4156      bn_mul_comba (a non zero value means the a or b arrays do not contain
4157      n2 elements) and fallback to bn_mul_normal if either is not zero.
4158      [Steve Henson]
4159 
4160   *) Fix escaping of non-ASCII characters when using the -subj option
4161      of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
4162      [Lutz Jaenicke]
4163 
4164   *) Make object definitions compliant to LDAP (RFC2256): SN is the short
4165      form for "surname", serialNumber has no short form.
4166      Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
4167      therefore remove "mail" short name for "internet 7".
4168      The OID for unique identifiers in X509 certificates is
4169      x500UniqueIdentifier, not uniqueIdentifier.
4170      Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
4171      [Lutz Jaenicke]
4172 
4173   *) Add an "init" command to the ENGINE config module and auto initialize
4174      ENGINEs. Without any "init" command the ENGINE will be initialized 
4175      after all ctrl commands have been executed on it. If init=1 the 
4176      ENGINE is initailized at that point (ctrls before that point are run
4177      on the uninitialized ENGINE and after on the initialized one). If
4178      init=0 then the ENGINE will not be iniatialized at all.
4179      [Steve Henson]
4180 
4181   *) Fix the 'app_verify_callback' interface so that the user-defined
4182      argument is actually passed to the callback: In the
4183      SSL_CTX_set_cert_verify_callback() prototype, the callback
4184      declaration has been changed from
4185           int (*cb)()
4186      into
4187           int (*cb)(X509_STORE_CTX *,void *);
4188      in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
4189           i=s->ctx->app_verify_callback(&ctx)
4190      has been changed into
4191           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
4192 
4193      To update applications using SSL_CTX_set_cert_verify_callback(),
4194      a dummy argument can be added to their callback functions.
4195      [D. K. Smetters <smetters@parc.xerox.com>]
4196 
4197   *) Added the '4758cca' ENGINE to support IBM 4758 cards.
4198      [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
4199 
4200   *) Add and OPENSSL_LOAD_CONF define which will cause
4201      OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
4202      This allows older applications to transparently support certain
4203      OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
4204      Two new functions OPENSSL_add_all_algorithms_noconf() which will never
4205      load the config file and OPENSSL_add_all_algorithms_conf() which will
4206      always load it have also been added.
4207      [Steve Henson]
4208 
4209   *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
4210      Adjust NIDs and EVP layer.
4211      [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
4212 
4213   *) Config modules support in openssl utility.
4214 
4215      Most commands now load modules from the config file,
4216      though in a few (such as version) this isn't done 
4217      because it couldn't be used for anything.
4218 
4219      In the case of ca and req the config file used is
4220      the same as the utility itself: that is the -config
4221      command line option can be used to specify an
4222      alternative file.
4223      [Steve Henson]
4224 
4225   *) Move default behaviour from OPENSSL_config(). If appname is NULL
4226      use "openssl_conf" if filename is NULL use default openssl config file.
4227      [Steve Henson]
4228 
4229   *) Add an argument to OPENSSL_config() to allow the use of an alternative
4230      config section name. Add a new flag to tolerate a missing config file
4231      and move code to CONF_modules_load_file().
4232      [Steve Henson]
4233 
4234   *) Support for crypto accelerator cards from Accelerated Encryption
4235      Processing, www.aep.ie.  (Use engine 'aep')
4236      The support was copied from 0.9.6c [engine] and adapted/corrected
4237      to work with the new engine framework.
4238      [AEP Inc. and Richard Levitte]
4239 
4240   *) Support for SureWare crypto accelerator cards from Baltimore
4241      Technologies.  (Use engine 'sureware')
4242      The support was copied from 0.9.6c [engine] and adapted
4243      to work with the new engine framework.
4244      [Richard Levitte]
4245 
4246   *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
4247      make the newer ENGINE framework commands for the CHIL engine work.
4248      [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
4249 
4250   *) Make it possible to produce shared libraries on ReliantUNIX.
4251      [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
4252 
4253   *) Add the configuration target debug-linux-ppro.
4254      Make 'openssl rsa' use the general key loading routines
4255      implemented in apps.c, and make those routines able to
4256      handle the key format FORMAT_NETSCAPE and the variant
4257      FORMAT_IISSGC.
4258      [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
4259 
4260  *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
4261      [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
4262 
4263   *) Add -keyform to rsautl, and document -engine.
4264      [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
4265 
4266   *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
4267      BIO_R_NO_SUCH_FILE error code rather than the generic
4268      ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
4269      [Ben Laurie]
4270 
4271   *) Add new functions
4272           ERR_peek_last_error
4273           ERR_peek_last_error_line
4274           ERR_peek_last_error_line_data.
4275      These are similar to
4276           ERR_peek_error
4277           ERR_peek_error_line
4278           ERR_peek_error_line_data,
4279      but report on the latest error recorded rather than the first one
4280      still in the error queue.
4281      [Ben Laurie, Bodo Moeller]
4282         
4283   *) default_algorithms option in ENGINE config module. This allows things
4284      like:
4285      default_algorithms = ALL
4286      default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
4287      [Steve Henson]
4288 
4289   *) Prelminary ENGINE config module.
4290      [Steve Henson]
4291 
4292   *) New experimental application configuration code.
4293      [Steve Henson]
4294 
4295   *) Change the AES code to follow the same name structure as all other
4296      symmetric ciphers, and behave the same way.  Move everything to
4297      the directory crypto/aes, thereby obsoleting crypto/rijndael.
4298      [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
4299 
4300   *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
4301      [Ben Laurie and Theo de Raadt]
4302 
4303   *) Add option to output public keys in req command.
4304      [Massimiliano Pala madwolf@openca.org]
4305 
4306   *) Use wNAFs in EC_POINTs_mul() for improved efficiency
4307      (up to about 10% better than before for P-192 and P-224).
4308      [Bodo Moeller]
4309 
4310   *) New functions/macros
4311 
4312           SSL_CTX_set_msg_callback(ctx, cb)
4313           SSL_CTX_set_msg_callback_arg(ctx, arg)
4314           SSL_set_msg_callback(ssl, cb)
4315           SSL_set_msg_callback_arg(ssl, arg)
4316 
4317      to request calling a callback function
4318 
4319           void cb(int write_p, int version, int content_type,
4320                   const void *buf, size_t len, SSL *ssl, void *arg)
4321 
4322      whenever a protocol message has been completely received
4323      (write_p == 0) or sent (write_p == 1).  Here 'version' is the
4324      protocol version  according to which the SSL library interprets
4325      the current protocol message (SSL2_VERSION, SSL3_VERSION, or
4326      TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
4327      the content type as defined in the SSL 3.0/TLS 1.0 protocol
4328      specification (change_cipher_spec(20), alert(21), handshake(22)).
4329      'buf' and 'len' point to the actual message, 'ssl' to the
4330      SSL object, and 'arg' is the application-defined value set by
4331      SSL[_CTX]_set_msg_callback_arg().
4332 
4333      'openssl s_client' and 'openssl s_server' have new '-msg' options
4334      to enable a callback that displays all protocol messages.
4335      [Bodo Moeller]
4336 
4337   *) Change the shared library support so shared libraries are built as
4338      soon as the corresponding static library is finished, and thereby get
4339      openssl and the test programs linked against the shared library.
4340      This still only happens when the keyword "shard" has been given to
4341      the configuration scripts.
4342 
4343      NOTE: shared library support is still an experimental thing, and
4344      backward binary compatibility is still not guaranteed.
4345      ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
4346 
4347   *) Add support for Subject Information Access extension.
4348      [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
4349 
4350   *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
4351      additional bytes when new memory had to be allocated, not just
4352      when reusing an existing buffer.
4353      [Bodo Moeller]
4354 
4355   *) New command line and configuration option 'utf8' for the req command.
4356      This allows field values to be specified as UTF8 strings.
4357      [Steve Henson]
4358 
4359   *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
4360      runs for the former and machine-readable output for the latter.
4361      [Ben Laurie]
4362 
4363   *) Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
4364      of the e-mail address in the DN (i.e., it will go into a certificate
4365      extension only).  The new configuration file option 'email_in_dn = no'
4366      has the same effect.
4367      [Massimiliano Pala madwolf@openca.org]
4368 
4369   *) Change all functions with names starting with des_ to be starting
4370      with DES_ instead.  Add wrappers that are compatible with libdes,
4371      but are named _ossl_old_des_*.  Finally, add macros that map the
4372      des_* symbols to the corresponding _ossl_old_des_* if libdes
4373      compatibility is desired.  If OpenSSL 0.9.6c compatibility is
4374      desired, the des_* symbols will be mapped to DES_*, with one
4375      exception.
4376 
4377      Since we provide two compatibility mappings, the user needs to
4378      define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
4379      compatibility is desired.  The default (i.e., when that macro
4380      isn't defined) is OpenSSL 0.9.6c compatibility.
4381 
4382      There are also macros that enable and disable the support of old
4383      des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
4384      and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
4385      are defined, the default will apply: to support the old des routines.
4386 
4387      In either case, one must include openssl/des.h to get the correct
4388      definitions.  Do not try to just include openssl/des_old.h, that
4389      won't work.
4390 
4391      NOTE: This is a major break of an old API into a new one.  Software
4392      authors are encouraged to switch to the DES_ style functions.  Some
4393      time in the future, des_old.h and the libdes compatibility functions
4394      will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
4395      default), and then completely removed.
4396      [Richard Levitte]
4397 
4398   *) Test for certificates which contain unsupported critical extensions.
4399      If such a certificate is found during a verify operation it is 
4400      rejected by default: this behaviour can be overridden by either
4401      handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
4402      by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
4403      X509_supported_extension() has also been added which returns 1 if a
4404      particular extension is supported.
4405      [Steve Henson]
4406 
4407   *) Modify the behaviour of EVP cipher functions in similar way to digests
4408      to retain compatibility with existing code.
4409      [Steve Henson]
4410 
4411   *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
4412      compatibility with existing code. In particular the 'ctx' parameter does
4413      not have to be to be initialized before the call to EVP_DigestInit() and
4414      it is tidied up after a call to EVP_DigestFinal(). New function
4415      EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
4416      EVP_MD_CTX_copy() changed to not require the destination to be
4417      initialized valid and new function EVP_MD_CTX_copy_ex() added which
4418      requires the destination to be valid.
4419 
4420      Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
4421      EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
4422      [Steve Henson]
4423 
4424   *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
4425      so that complete 'Handshake' protocol structures are kept in memory
4426      instead of overwriting 'msg_type' and 'length' with 'body' data.
4427      [Bodo Moeller]
4428 
4429   *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
4430      [Massimo Santin via Richard Levitte]
4431 
4432   *) Major restructuring to the underlying ENGINE code. This includes
4433      reduction of linker bloat, separation of pure "ENGINE" manipulation
4434      (initialisation, etc) from functionality dealing with implementations
4435      of specific crypto iterfaces. This change also introduces integrated
4436      support for symmetric ciphers and digest implementations - so ENGINEs
4437      can now accelerate these by providing EVP_CIPHER and EVP_MD
4438      implementations of their own. This is detailed in crypto/engine/README
4439      as it couldn't be adequately described here. However, there are a few
4440      API changes worth noting - some RSA, DSA, DH, and RAND functions that
4441      were changed in the original introduction of ENGINE code have now
4442      reverted back - the hooking from this code to ENGINE is now a good
4443      deal more passive and at run-time, operations deal directly with
4444      RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
4445      dereferencing through an ENGINE pointer any more. Also, the ENGINE
4446      functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
4447      they were not being used by the framework as there is no concept of a
4448      BIGNUM_METHOD and they could not be generalised to the new
4449      'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
4450      ENGINE_cpy() has been removed as it cannot be consistently defined in
4451      the new code.
4452      [Geoff Thorpe]
4453 
4454   *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
4455      [Steve Henson]
4456 
4457   *) Change mkdef.pl to sort symbols that get the same entry number,
4458      and make sure the automatically generated functions ERR_load_*
4459      become part of libeay.num as well.
4460      [Richard Levitte]
4461 
4462   *) New function SSL_renegotiate_pending().  This returns true once
4463      renegotiation has been requested (either SSL_renegotiate() call
4464      or HelloRequest/ClientHello receveived from the peer) and becomes
4465      false once a handshake has been completed.
4466      (For servers, SSL_renegotiate() followed by SSL_do_handshake()
4467      sends a HelloRequest, but does not ensure that a handshake takes
4468      place.  SSL_renegotiate_pending() is useful for checking if the
4469      client has followed the request.)
4470      [Bodo Moeller]
4471 
4472   *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
4473      By default, clients may request session resumption even during
4474      renegotiation (if session ID contexts permit); with this option,
4475      session resumption is possible only in the first handshake.
4476 
4477      SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
4478      more bits available for options that should not be part of
4479      SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
4480      [Bodo Moeller]
4481 
4482   *) Add some demos for certificate and certificate request creation.
4483      [Steve Henson]
4484 
4485   *) Make maximum certificate chain size accepted from the peer application
4486      settable (SSL*_get/set_max_cert_list()), as proposed by
4487      "Douglas E. Engert" <deengert@anl.gov>.
4488      [Lutz Jaenicke]
4489 
4490   *) Add support for shared libraries for Unixware-7
4491      (Boyd Lynn Gerber <gerberb@zenez.com>).
4492      [Lutz Jaenicke]
4493 
4494   *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
4495      be done prior to destruction. Use this to unload error strings from
4496      ENGINEs that load their own error strings. NB: This adds two new API
4497      functions to "get" and "set" this destroy handler in an ENGINE.
4498      [Geoff Thorpe]
4499 
4500   *) Alter all existing ENGINE implementations (except "openssl" and
4501      "openbsd") to dynamically instantiate their own error strings. This
4502      makes them more flexible to be built both as statically-linked ENGINEs
4503      and self-contained shared-libraries loadable via the "dynamic" ENGINE.
4504      Also, add stub code to each that makes building them as self-contained
4505      shared-libraries easier (see README.ENGINE).
4506      [Geoff Thorpe]
4507 
4508   *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
4509      implementations into applications that are completely implemented in
4510      self-contained shared-libraries. The "dynamic" ENGINE exposes control
4511      commands that can be used to configure what shared-library to load and
4512      to control aspects of the way it is handled. Also, made an update to
4513      the README.ENGINE file that brings its information up-to-date and
4514      provides some information and instructions on the "dynamic" ENGINE
4515      (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
4516      [Geoff Thorpe]
4517 
4518   *) Make it possible to unload ranges of ERR strings with a new
4519      "ERR_unload_strings" function.
4520      [Geoff Thorpe]
4521 
4522   *) Add a copy() function to EVP_MD.
4523      [Ben Laurie]
4524 
4525   *) Make EVP_MD routines take a context pointer instead of just the
4526      md_data void pointer.
4527      [Ben Laurie]
4528 
4529   *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
4530      that the digest can only process a single chunk of data
4531      (typically because it is provided by a piece of
4532      hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
4533      is only going to provide a single chunk of data, and hence the
4534      framework needn't accumulate the data for oneshot drivers.
4535      [Ben Laurie]
4536 
4537   *) As with "ERR", make it possible to replace the underlying "ex_data"
4538      functions. This change also alters the storage and management of global
4539      ex_data state - it's now all inside ex_data.c and all "class" code (eg.
4540      RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
4541      index counters. The API functions that use this state have been changed
4542      to take a "class_index" rather than pointers to the class's local STACK
4543      and counter, and there is now an API function to dynamically create new
4544      classes. This centralisation allows us to (a) plug a lot of the
4545      thread-safety problems that existed, and (b) makes it possible to clean
4546      up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
4547      such data would previously have always leaked in application code and
4548      workarounds were in place to make the memory debugging turn a blind eye
4549      to it. Application code that doesn't use this new function will still
4550      leak as before, but their memory debugging output will announce it now
4551      rather than letting it slide.
4552 
4553      Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
4554      induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
4555      has a return value to indicate success or failure.
4556      [Geoff Thorpe]
4557 
4558   *) Make it possible to replace the underlying "ERR" functions such that the
4559      global state (2 LHASH tables and 2 locks) is only used by the "default"
4560      implementation. This change also adds two functions to "get" and "set"
4561      the implementation prior to it being automatically set the first time
4562      any other ERR function takes place. Ie. an application can call "get",
4563      pass the return value to a module it has just loaded, and that module
4564      can call its own "set" function using that value. This means the
4565      module's "ERR" operations will use (and modify) the error state in the
4566      application and not in its own statically linked copy of OpenSSL code.
4567      [Geoff Thorpe]
4568 
4569   *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
4570      reference counts. This performs normal REF_PRINT/REF_CHECK macros on
4571      the operation, and provides a more encapsulated way for external code
4572      (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
4573      to use these functions rather than manually incrementing the counts.
4574 
4575      Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
4576      [Geoff Thorpe]
4577 
4578   *) Add EVP test program.
4579      [Ben Laurie]
4580 
4581   *) Add symmetric cipher support to ENGINE. Expect the API to change!
4582      [Ben Laurie]
4583 
4584   *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
4585      X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
4586      X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
4587      These allow a CRL to be built without having to access X509_CRL fields
4588      directly. Modify 'ca' application to use new functions.
4589      [Steve Henson]
4590 
4591   *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
4592      bug workarounds. Rollback attack detection is a security feature.
4593      The problem will only arise on OpenSSL servers when TLSv1 is not
4594      available (sslv3_server_method() or SSL_OP_NO_TLSv1).
4595      Software authors not wanting to support TLSv1 will have special reasons
4596      for their choice and can explicitly enable this option.
4597      [Bodo Moeller, Lutz Jaenicke]
4598 
4599   *) Rationalise EVP so it can be extended: don't include a union of
4600      cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
4601      (similar to those existing for EVP_CIPHER_CTX).
4602      Usage example:
4603 
4604          EVP_MD_CTX md;
4605 
4606          EVP_MD_CTX_init(&md);             /* new function call */
4607          EVP_DigestInit(&md, EVP_sha1());
4608          EVP_DigestUpdate(&md, in, len);
4609          EVP_DigestFinal(&md, out, NULL);
4610          EVP_MD_CTX_cleanup(&md);          /* new function call */
4611 
4612      [Ben Laurie]
4613 
4614   *) Make DES key schedule conform to the usual scheme, as well as
4615      correcting its structure. This means that calls to DES functions
4616      now have to pass a pointer to a des_key_schedule instead of a
4617      plain des_key_schedule (which was actually always a pointer
4618      anyway): E.g.,
4619 
4620          des_key_schedule ks;
4621 
4622          des_set_key_checked(..., &ks);
4623          des_ncbc_encrypt(..., &ks, ...);
4624 
4625      (Note that a later change renames 'des_...' into 'DES_...'.)
4626      [Ben Laurie]
4627 
4628   *) Initial reduction of linker bloat: the use of some functions, such as
4629      PEM causes large amounts of unused functions to be linked in due to
4630      poor organisation. For example pem_all.c contains every PEM function
4631      which has a knock on effect of linking in large amounts of (unused)
4632      ASN1 code. Grouping together similar functions and splitting unrelated
4633      functions prevents this.
4634      [Steve Henson]
4635 
4636   *) Cleanup of EVP macros.
4637      [Ben Laurie]
4638 
4639   *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
4640      correct _ecb suffix.
4641      [Ben Laurie]
4642 
4643   *) Add initial OCSP responder support to ocsp application. The
4644      revocation information is handled using the text based index
4645      use by the ca application. The responder can either handle
4646      requests generated internally, supplied in files (for example
4647      via a CGI script) or using an internal minimal server.
4648      [Steve Henson]
4649 
4650   *) Add configuration choices to get zlib compression for TLS.
4651      [Richard Levitte]
4652 
4653   *) Changes to Kerberos SSL for RFC 2712 compliance:
4654      1.  Implemented real KerberosWrapper, instead of just using
4655          KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
4656      2.  Implemented optional authenticator field of KerberosWrapper.
4657 
4658      Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
4659      and authenticator structs; see crypto/krb5/.
4660 
4661      Generalized Kerberos calls to support multiple Kerberos libraries.
4662      [Vern Staats <staatsvr@asc.hpc.mil>,
4663       Jeffrey Altman <jaltman@columbia.edu>
4664       via Richard Levitte]
4665 
4666   *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
4667      already does with RSA. testdsa.h now has 'priv_key/pub_key'
4668      values for each of the key sizes rather than having just
4669      parameters (and 'speed' generating keys each time).
4670      [Geoff Thorpe]
4671 
4672   *) Speed up EVP routines.
4673      Before:
4674 encrypt
4675 type              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
4676 des-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
4677 des-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
4678 des-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
4679 decrypt
4680 des-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
4681 des-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
4682 des-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
4683      After:
4684 encrypt
4685 des-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
4686 decrypt
4687 des-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
4688      [Ben Laurie]
4689 
4690   *) Added the OS2-EMX target.
4691      ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
4692 
4693   *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
4694      to support NCONF routines in extension code. New function CONF_set_nconf()
4695      to allow functions which take an NCONF to also handle the old LHASH
4696      structure: this means that the old CONF compatible routines can be
4697      retained (in particular wrt extensions) without having to duplicate the
4698      code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
4699      [Steve Henson]
4700 
4701   *) Enhance the general user interface with mechanisms for inner control
4702      and with possibilities to have yes/no kind of prompts.
4703      [Richard Levitte]
4704 
4705   *) Change all calls to low level digest routines in the library and
4706      applications to use EVP. Add missing calls to HMAC_cleanup() and
4707      don't assume HMAC_CTX can be copied using memcpy().
4708      [Verdon Walker <VWalker@novell.com>, Steve Henson]
4709 
4710   *) Add the possibility to control engines through control names but with
4711      arbitrary arguments instead of just a string.
4712      Change the key loaders to take a UI_METHOD instead of a callback
4713      function pointer.  NOTE: this breaks binary compatibility with earlier
4714      versions of OpenSSL [engine].
4715      Adapt the nCipher code for these new conditions and add a card insertion
4716      callback.
4717      [Richard Levitte]
4718 
4719   *) Enhance the general user interface with mechanisms to better support
4720      dialog box interfaces, application-defined prompts, the possibility
4721      to use defaults (for example default passwords from somewhere else)
4722      and interrupts/cancellations.
4723      [Richard Levitte]
4724 
4725   *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
4726      attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
4727      [Steve Henson]
4728 
4729   *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
4730      tidy up some unnecessarily weird code in 'sk_new()').
4731      [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
4732 
4733   *) Change the key loading routines for ENGINEs to use the same kind
4734      callback (pem_password_cb) as all other routines that need this
4735      kind of callback.
4736      [Richard Levitte]
4737 
4738   *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
4739      256 bit (=32 byte) keys. Of course seeding with more entropy bytes
4740      than this minimum value is recommended.
4741      [Lutz Jaenicke]
4742 
4743   *) New random seeder for OpenVMS, using the system process statistics
4744      that are easily reachable.
4745      [Richard Levitte]
4746 
4747   *) Windows apparently can't transparently handle global
4748      variables defined in DLLs. Initialisations such as:
4749 
4750         const ASN1_ITEM *it = &ASN1_INTEGER_it;
4751 
4752      wont compile. This is used by the any applications that need to
4753      declare their own ASN1 modules. This was fixed by adding the option
4754      EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
4755      needed for static libraries under Win32.
4756      [Steve Henson]
4757 
4758   *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
4759      setting of purpose and trust fields. New X509_STORE trust and
4760      purpose functions and tidy up setting in other SSL functions.
4761      [Steve Henson]
4762 
4763   *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
4764      structure. These are inherited by X509_STORE_CTX when it is 
4765      initialised. This allows various defaults to be set in the
4766      X509_STORE structure (such as flags for CRL checking and custom
4767      purpose or trust settings) for functions which only use X509_STORE_CTX
4768      internally such as S/MIME.
4769 
4770      Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
4771      trust settings if they are not set in X509_STORE. This allows X509_STORE
4772      purposes and trust (in S/MIME for example) to override any set by default.
4773 
4774      Add command line options for CRL checking to smime, s_client and s_server
4775      applications.
4776      [Steve Henson]
4777 
4778   *) Initial CRL based revocation checking. If the CRL checking flag(s)
4779      are set then the CRL is looked up in the X509_STORE structure and
4780      its validity and signature checked, then if the certificate is found
4781      in the CRL the verify fails with a revoked error.
4782 
4783      Various new CRL related callbacks added to X509_STORE_CTX structure.
4784 
4785      Command line options added to 'verify' application to support this.
4786 
4787      This needs some additional work, such as being able to handle multiple
4788      CRLs with different times, extension based lookup (rather than just
4789      by subject name) and ultimately more complete V2 CRL extension
4790      handling.
4791      [Steve Henson]
4792 
4793   *) Add a general user interface API (crypto/ui/).  This is designed
4794      to replace things like des_read_password and friends (backward
4795      compatibility functions using this new API are provided).
4796      The purpose is to remove prompting functions from the DES code
4797      section as well as provide for prompting through dialog boxes in
4798      a window system and the like.
4799      [Richard Levitte]
4800 
4801   *) Add "ex_data" support to ENGINE so implementations can add state at a
4802      per-structure level rather than having to store it globally.
4803      [Geoff]
4804 
4805   *) Make it possible for ENGINE structures to be copied when retrieved by
4806      ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
4807      This causes the "original" ENGINE structure to act like a template,
4808      analogous to the RSA vs. RSA_METHOD type of separation. Because of this
4809      operational state can be localised to each ENGINE structure, despite the
4810      fact they all share the same "methods". New ENGINE structures returned in
4811      this case have no functional references and the return value is the single
4812      structural reference. This matches the single structural reference returned
4813      by ENGINE_by_id() normally, when it is incremented on the pre-existing
4814      ENGINE structure.
4815      [Geoff]
4816 
4817   *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
4818      needs to match any other type at all we need to manually clear the
4819      tag cache.
4820      [Steve Henson]
4821 
4822   *) Changes to the "openssl engine" utility to include;
4823      - verbosity levels ('-v', '-vv', and '-vvv') that provide information
4824        about an ENGINE's available control commands.
4825      - executing control commands from command line arguments using the
4826        '-pre' and '-post' switches. '-post' is only used if '-t' is
4827        specified and the ENGINE is successfully initialised. The syntax for
4828        the individual commands are colon-separated, for example;
4829          openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
4830      [Geoff]
4831 
4832   *) New dynamic control command support for ENGINEs. ENGINEs can now
4833      declare their own commands (numbers), names (strings), descriptions,
4834      and input types for run-time discovery by calling applications. A
4835      subset of these commands are implicitly classed as "executable"
4836      depending on their input type, and only these can be invoked through
4837      the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
4838      can be based on user input, config files, etc). The distinction is
4839      that "executable" commands cannot return anything other than a boolean
4840      result and can only support numeric or string input, whereas some
4841      discoverable commands may only be for direct use through
4842      ENGINE_ctrl(), eg. supporting the exchange of binary data, function
4843      pointers, or other custom uses. The "executable" commands are to
4844      support parameterisations of ENGINE behaviour that can be
4845      unambiguously defined by ENGINEs and used consistently across any
4846      OpenSSL-based application. Commands have been added to all the
4847      existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
4848      control over shared-library paths without source code alterations.
4849      [Geoff]
4850 
4851   *) Changed all ENGINE implementations to dynamically allocate their
4852      ENGINEs rather than declaring them statically. Apart from this being
4853      necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
4854      this also allows the implementations to compile without using the
4855      internal engine_int.h header.
4856      [Geoff]
4857 
4858   *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
4859      'const' value. Any code that should be able to modify a RAND_METHOD
4860      should already have non-const pointers to it (ie. they should only
4861      modify their own ones).
4862      [Geoff]
4863 
4864   *) Made a variety of little tweaks to the ENGINE code.
4865      - "atalla" and "ubsec" string definitions were moved from header files
4866        to C code. "nuron" string definitions were placed in variables
4867        rather than hard-coded - allowing parameterisation of these values
4868        later on via ctrl() commands.
4869      - Removed unused "#if 0"'d code.
4870      - Fixed engine list iteration code so it uses ENGINE_free() to release
4871        structural references.
4872      - Constified the RAND_METHOD element of ENGINE structures.
4873      - Constified various get/set functions as appropriate and added
4874        missing functions (including a catch-all ENGINE_cpy that duplicates
4875        all ENGINE values onto a new ENGINE except reference counts/state).
4876      - Removed NULL parameter checks in get/set functions. Setting a method
4877        or function to NULL is a way of cancelling out a previously set
4878        value.  Passing a NULL ENGINE parameter is just plain stupid anyway
4879        and doesn't justify the extra error symbols and code.
4880      - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
4881        flags from engine_int.h to engine.h.
4882      - Changed prototypes for ENGINE handler functions (init(), finish(),
4883        ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
4884      [Geoff]
4885 
4886   *) Implement binary inversion algorithm for BN_mod_inverse in addition
4887      to the algorithm using long division.  The binary algorithm can be
4888      used only if the modulus is odd.  On 32-bit systems, it is faster
4889      only for relatively small moduli (roughly 20-30% for 128-bit moduli,
4890      roughly 5-15% for 256-bit moduli), so we use it only for moduli
4891      up to 450 bits.  In 64-bit environments, the binary algorithm
4892      appears to be advantageous for much longer moduli; here we use it
4893      for moduli up to 2048 bits.
4894      [Bodo Moeller]
4895 
4896   *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
4897      could not support the combine flag in choice fields.
4898      [Steve Henson]
4899 
4900   *) Add a 'copy_extensions' option to the 'ca' utility. This copies
4901      extensions from a certificate request to the certificate.
4902      [Steve Henson]
4903 
4904   *) Allow multiple 'certopt' and 'nameopt' options to be separated
4905      by commas. Add 'namopt' and 'certopt' options to the 'ca' config
4906      file: this allows the display of the certificate about to be
4907      signed to be customised, to allow certain fields to be included
4908      or excluded and extension details. The old system didn't display
4909      multicharacter strings properly, omitted fields not in the policy
4910      and couldn't display additional details such as extensions.
4911      [Steve Henson]
4912 
4913   *) Function EC_POINTs_mul for multiple scalar multiplication
4914      of an arbitrary number of elliptic curve points
4915           \sum scalars[i]*points[i],
4916      optionally including the generator defined for the EC_GROUP:
4917           scalar*generator +  \sum scalars[i]*points[i].
4918 
4919      EC_POINT_mul is a simple wrapper function for the typical case
4920      that the point list has just one item (besides the optional
4921      generator).
4922      [Bodo Moeller]
4923 
4924   *) First EC_METHODs for curves over GF(p):
4925 
4926      EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
4927      operations and provides various method functions that can also
4928      operate with faster implementations of modular arithmetic.     
4929 
4930      EC_GFp_mont_method() reuses most functions that are part of
4931      EC_GFp_simple_method, but uses Montgomery arithmetic.
4932 
4933      [Bodo Moeller; point addition and point doubling
4934      implementation directly derived from source code provided by
4935      Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
4936 
4937   *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
4938      crypto/ec/ec_lib.c):
4939 
4940      Curves are EC_GROUP objects (with an optional group generator)
4941      based on EC_METHODs that are built into the library.
4942 
4943      Points are EC_POINT objects based on EC_GROUP objects.
4944 
4945      Most of the framework would be able to handle curves over arbitrary
4946      finite fields, but as there are no obvious types for fields other
4947      than GF(p), some functions are limited to that for now.
4948      [Bodo Moeller]
4949 
4950   *) Add the -HTTP option to s_server.  It is similar to -WWW, but requires
4951      that the file contains a complete HTTP response.
4952      [Richard Levitte]
4953 
4954   *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
4955      change the def and num file printf format specifier from "%-40sXXX"
4956      to "%-39s XXX". The latter will always guarantee a space after the
4957      field while the former will cause them to run together if the field
4958      is 40 of more characters long.
4959      [Steve Henson]
4960 
4961   *) Constify the cipher and digest 'method' functions and structures
4962      and modify related functions to take constant EVP_MD and EVP_CIPHER
4963      pointers.
4964      [Steve Henson]
4965 
4966   *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
4967      in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
4968      [Bodo Moeller]
4969 
4970   *) Modify EVP_Digest*() routines so they now return values. Although the
4971      internal software routines can never fail additional hardware versions
4972      might.
4973      [Steve Henson]
4974 
4975   *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
4976 
4977      Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
4978      (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
4979 
4980      ASN1 error codes
4981           ERR_R_NESTED_ASN1_ERROR
4982           ...
4983           ERR_R_MISSING_ASN1_EOS
4984      were 4 .. 9, conflicting with
4985           ERR_LIB_RSA (= ERR_R_RSA_LIB)
4986           ...
4987           ERR_LIB_PEM (= ERR_R_PEM_LIB).
4988      They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
4989 
4990      Add new error code 'ERR_R_INTERNAL_ERROR'.
4991      [Bodo Moeller]
4992 
4993   *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
4994      suffices.
4995      [Bodo Moeller]
4996 
4997   *) New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
4998      sets the subject name for a new request or supersedes the
4999      subject name in a given request. Formats that can be parsed are
5000           'CN=Some Name, OU=myOU, C=IT'
5001      and
5002           'CN=Some Name/OU=myOU/C=IT'.
5003 
5004      Add options '-batch' and '-verbose' to 'openssl req'.
5005      [Massimiliano Pala <madwolf@hackmasters.net>]
5006 
5007   *) Introduce the possibility to access global variables through
5008      functions on platform were that's the best way to handle exporting
5009      global variables in shared libraries.  To enable this functionality,
5010      one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
5011      "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
5012      is normally done by Configure or something similar).
5013 
5014      To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
5015      in the source file (foo.c) like this:
5016 
5017         OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
5018         OPENSSL_IMPLEMENT_GLOBAL(double,bar);
5019 
5020      To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
5021      and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
5022 
5023         OPENSSL_DECLARE_GLOBAL(int,foo);
5024         #define foo OPENSSL_GLOBAL_REF(foo)
5025         OPENSSL_DECLARE_GLOBAL(double,bar);
5026         #define bar OPENSSL_GLOBAL_REF(bar)
5027 
5028      The #defines are very important, and therefore so is including the
5029      header file everywhere where the defined globals are used.
5030 
5031      The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
5032      of ASN.1 items, but that structure is a bit different.
5033 
5034      The largest change is in util/mkdef.pl which has been enhanced with
5035      better and easier to understand logic to choose which symbols should
5036      go into the Windows .def files as well as a number of fixes and code
5037      cleanup (among others, algorithm keywords are now sorted
5038      lexicographically to avoid constant rewrites).
5039      [Richard Levitte]
5040 
5041   *) In BN_div() keep a copy of the sign of 'num' before writing the
5042      result to 'rm' because if rm==num the value will be overwritten
5043      and produce the wrong result if 'num' is negative: this caused
5044      problems with BN_mod() and BN_nnmod().
5045      [Steve Henson]
5046 
5047   *) Function OCSP_request_verify(). This checks the signature on an
5048      OCSP request and verifies the signer certificate. The signer
5049      certificate is just checked for a generic purpose and OCSP request
5050      trust settings.
5051      [Steve Henson]
5052 
5053   *) Add OCSP_check_validity() function to check the validity of OCSP
5054      responses. OCSP responses are prepared in real time and may only
5055      be a few seconds old. Simply checking that the current time lies
5056      between thisUpdate and nextUpdate max reject otherwise valid responses
5057      caused by either OCSP responder or client clock inaccuracy. Instead
5058      we allow thisUpdate and nextUpdate to fall within a certain period of
5059      the current time. The age of the response can also optionally be
5060      checked. Two new options -validity_period and -status_age added to
5061      ocsp utility.
5062      [Steve Henson]
5063 
5064   *) If signature or public key algorithm is unrecognized print out its
5065      OID rather that just UNKNOWN.
5066      [Steve Henson]
5067 
5068   *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
5069      OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
5070      ID to be generated from the issuer certificate alone which can then be
5071      passed to OCSP_id_issuer_cmp().
5072      [Steve Henson]
5073 
5074   *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
5075      ASN1 modules to export functions returning ASN1_ITEM pointers
5076      instead of the ASN1_ITEM structures themselves. This adds several
5077      new macros which allow the underlying ASN1 function/structure to
5078      be accessed transparently. As a result code should not use ASN1_ITEM
5079      references directly (such as &X509_it) but instead use the relevant
5080      macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
5081      use of the new ASN1 code on platforms where exporting structures
5082      is problematical (for example in shared libraries) but exporting
5083      functions returning pointers to structures is not.
5084      [Steve Henson]
5085 
5086   *) Add support for overriding the generation of SSL/TLS session IDs.
5087      These callbacks can be registered either in an SSL_CTX or per SSL.
5088      The purpose of this is to allow applications to control, if they wish,
5089      the arbitrary values chosen for use as session IDs, particularly as it
5090      can be useful for session caching in multiple-server environments. A
5091      command-line switch for testing this (and any client code that wishes
5092      to use such a feature) has been added to "s_server".
5093      [Geoff Thorpe, Lutz Jaenicke]
5094 
5095   *) Modify mkdef.pl to recognise and parse preprocessor conditionals
5096      of the form '#if defined(...) || defined(...) || ...' and
5097      '#if !defined(...) && !defined(...) && ...'.  This also avoids
5098      the growing number of special cases it was previously handling.
5099      [Richard Levitte]
5100 
5101   *) Make all configuration macros available for application by making
5102      sure they are available in opensslconf.h, by giving them names starting
5103      with "OPENSSL_" to avoid conflicts with other packages and by making
5104      sure e_os2.h will cover all platform-specific cases together with
5105      opensslconf.h.
5106      Additionally, it is now possible to define configuration/platform-
5107      specific names (called "system identities").  In the C code, these
5108      are prefixed with "OPENSSL_SYSNAME_".  e_os2.h will create another
5109      macro with the name beginning with "OPENSSL_SYS_", which is determined
5110      from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
5111      what is available.
5112      [Richard Levitte]
5113 
5114   *) New option -set_serial to 'req' and 'x509' this allows the serial
5115      number to use to be specified on the command line. Previously self
5116      signed certificates were hard coded with serial number 0 and the 
5117      CA options of 'x509' had to use a serial number in a file which was
5118      auto incremented.
5119      [Steve Henson]
5120 
5121   *) New options to 'ca' utility to support V2 CRL entry extensions.
5122      Currently CRL reason, invalidity date and hold instruction are
5123      supported. Add new CRL extensions to V3 code and some new objects.
5124      [Steve Henson]
5125 
5126   *) New function EVP_CIPHER_CTX_set_padding() this is used to
5127      disable standard block padding (aka PKCS#5 padding) in the EVP
5128      API, which was previously mandatory. This means that the data is
5129      not padded in any way and so the total length much be a multiple
5130      of the block size, otherwise an error occurs.
5131      [Steve Henson]
5132 
5133   *) Initial (incomplete) OCSP SSL support.
5134      [Steve Henson]
5135 
5136   *) New function OCSP_parse_url(). This splits up a URL into its host,
5137      port and path components: primarily to parse OCSP URLs. New -url
5138      option to ocsp utility.
5139      [Steve Henson]
5140 
5141   *) New nonce behavior. The return value of OCSP_check_nonce() now 
5142      reflects the various checks performed. Applications can decide
5143      whether to tolerate certain situations such as an absent nonce
5144      in a response when one was present in a request: the ocsp application
5145      just prints out a warning. New function OCSP_add1_basic_nonce()
5146      this is to allow responders to include a nonce in a response even if
5147      the request is nonce-less.
5148      [Steve Henson]
5149 
5150   *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
5151      skipped when using openssl x509 multiple times on a single input file,
5152      e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
5153      [Bodo Moeller]
5154 
5155   *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
5156      set string type: to handle setting ASN1_TIME structures. Fix ca
5157      utility to correctly initialize revocation date of CRLs.
5158      [Steve Henson]
5159 
5160   *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
5161      the clients preferred ciphersuites and rather use its own preferences.
5162      Should help to work around M$ SGC (Server Gated Cryptography) bug in
5163      Internet Explorer by ensuring unchanged hash method during stepup.
5164      (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
5165      [Lutz Jaenicke]
5166 
5167   *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
5168      to aes and add a new 'exist' option to print out symbols that don't
5169      appear to exist.
5170      [Steve Henson]
5171 
5172   *) Additional options to ocsp utility to allow flags to be set and
5173      additional certificates supplied.
5174      [Steve Henson]
5175 
5176   *) Add the option -VAfile to 'openssl ocsp', so the user can give the
5177      OCSP client a number of certificate to only verify the response
5178      signature against.
5179      [Richard Levitte]
5180 
5181   *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
5182      handle the new API. Currently only ECB, CBC modes supported. Add new
5183      AES OIDs.
5184 
5185      Add TLS AES ciphersuites as described in RFC3268, "Advanced
5186      Encryption Standard (AES) Ciphersuites for Transport Layer
5187      Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
5188      not enabled by default and were not part of the "ALL" ciphersuite
5189      alias because they were not yet official; they could be
5190      explicitly requested by specifying the "AESdraft" ciphersuite
5191      group alias.  In the final release of OpenSSL 0.9.7, the group
5192      alias is called "AES" and is part of "ALL".)
5193      [Ben Laurie, Steve  Henson, Bodo Moeller]
5194 
5195   *) New function OCSP_copy_nonce() to copy nonce value (if present) from
5196      request to response.
5197      [Steve Henson]
5198 
5199   *) Functions for OCSP responders. OCSP_request_onereq_count(),
5200      OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
5201      extract information from a certificate request. OCSP_response_create()
5202      creates a response and optionally adds a basic response structure.
5203      OCSP_basic_add1_status() adds a complete single response to a basic
5204      response and returns the OCSP_SINGLERESP structure just added (to allow
5205      extensions to be included for example). OCSP_basic_add1_cert() adds a
5206      certificate to a basic response and OCSP_basic_sign() signs a basic
5207      response with various flags. New helper functions ASN1_TIME_check()
5208      (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
5209      (converts ASN1_TIME to GeneralizedTime).
5210      [Steve Henson]
5211 
5212   *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
5213      in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
5214      structure from a certificate. X509_pubkey_digest() digests the public_key
5215      contents: this is used in various key identifiers. 
5216      [Steve Henson]
5217 
5218   *) Make sk_sort() tolerate a NULL argument.
5219      [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
5220 
5221   *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
5222      passed by the function are trusted implicitly. If any of them signed the
5223      response then it is assumed to be valid and is not verified.
5224      [Steve Henson]
5225 
5226   *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
5227      to data. This was previously part of the PKCS7 ASN1 code. This
5228      was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
5229      [Steve Henson, reported by Kenneth R. Robinette
5230                                 <support@securenetterm.com>]
5231 
5232   *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
5233      routines: without these tracing memory leaks is very painful.
5234      Fix leaks in PKCS12 and PKCS7 routines.
5235      [Steve Henson]
5236 
5237   *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
5238      Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
5239      effectively meant GeneralizedTime would never be used. Now it
5240      is initialised to -1 but X509_time_adj() now has to check the value
5241      and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
5242      V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
5243      [Steve Henson, reported by Kenneth R. Robinette
5244                                 <support@securenetterm.com>]
5245 
5246   *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
5247      result in a zero length in the ASN1_INTEGER structure which was
5248      not consistent with the structure when d2i_ASN1_INTEGER() was used
5249      and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
5250      to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
5251      where it did not print out a minus for negative ASN1_INTEGER.
5252      [Steve Henson]
5253 
5254   *) Add summary printout to ocsp utility. The various functions which
5255      convert status values to strings have been renamed to:
5256      OCSP_response_status_str(), OCSP_cert_status_str() and
5257      OCSP_crl_reason_str() and are no longer static. New options
5258      to verify nonce values and to disable verification. OCSP response
5259      printout format cleaned up.
5260      [Steve Henson]
5261 
5262   *) Add additional OCSP certificate checks. These are those specified
5263      in RFC2560. This consists of two separate checks: the CA of the
5264      certificate being checked must either be the OCSP signer certificate
5265      or the issuer of the OCSP signer certificate. In the latter case the
5266      OCSP signer certificate must contain the OCSP signing extended key
5267      usage. This check is performed by attempting to match the OCSP
5268      signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
5269      in the OCSP_CERTID structures of the response.
5270      [Steve Henson]
5271 
5272   *) Initial OCSP certificate verification added to OCSP_basic_verify()
5273      and related routines. This uses the standard OpenSSL certificate
5274      verify routines to perform initial checks (just CA validity) and
5275      to obtain the certificate chain. Then additional checks will be
5276      performed on the chain. Currently the root CA is checked to see
5277      if it is explicitly trusted for OCSP signing. This is used to set
5278      a root CA as a global signing root: that is any certificate that
5279      chains to that CA is an acceptable OCSP signing certificate.
5280      [Steve Henson]
5281 
5282   *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
5283      extensions from a separate configuration file.
5284      As when reading extensions from the main configuration file,
5285      the '-extensions ...' option may be used for specifying the
5286      section to use.
5287      [Massimiliano Pala <madwolf@comune.modena.it>]
5288 
5289   *) New OCSP utility. Allows OCSP requests to be generated or
5290      read. The request can be sent to a responder and the output
5291      parsed, outputed or printed in text form. Not complete yet:
5292      still needs to check the OCSP response validity.
5293      [Steve Henson]
5294 
5295   *) New subcommands for 'openssl ca':
5296      'openssl ca -status <serial>' prints the status of the cert with
5297      the given serial number (according to the index file).
5298      'openssl ca -updatedb' updates the expiry status of certificates
5299      in the index file.
5300      [Massimiliano Pala <madwolf@comune.modena.it>]
5301 
5302   *) New '-newreq-nodes' command option to CA.pl.  This is like
5303      '-newreq', but calls 'openssl req' with the '-nodes' option
5304      so that the resulting key is not encrypted.
5305      [Damien Miller <djm@mindrot.org>]
5306 
5307   *) New configuration for the GNU Hurd.
5308      [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
5309 
5310   *) Initial code to implement OCSP basic response verify. This
5311      is currently incomplete. Currently just finds the signer's
5312      certificate and verifies the signature on the response.
5313      [Steve Henson]
5314 
5315   *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
5316      value of OPENSSLDIR.  This is available via the new '-d' option
5317      to 'openssl version', and is also included in 'openssl version -a'.
5318      [Bodo Moeller]
5319 
5320   *) Allowing defining memory allocation callbacks that will be given
5321      file name and line number information in additional arguments
5322      (a const char* and an int).  The basic functionality remains, as
5323      well as the original possibility to just replace malloc(),
5324      realloc() and free() by functions that do not know about these
5325      additional arguments.  To register and find out the current
5326      settings for extended allocation functions, the following
5327      functions are provided:
5328 
5329         CRYPTO_set_mem_ex_functions
5330         CRYPTO_set_locked_mem_ex_functions
5331         CRYPTO_get_mem_ex_functions
5332         CRYPTO_get_locked_mem_ex_functions
5333 
5334      These work the same way as CRYPTO_set_mem_functions and friends.
5335      CRYPTO_get_[locked_]mem_functions now writes 0 where such an
5336      extended allocation function is enabled.
5337      Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
5338      a conventional allocation function is enabled.
5339      [Richard Levitte, Bodo Moeller]
5340 
5341   *) Finish off removing the remaining LHASH function pointer casts.
5342      There should no longer be any prototype-casting required when using
5343      the LHASH abstraction, and any casts that remain are "bugs". See
5344      the callback types and macros at the head of lhash.h for details
5345      (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
5346      [Geoff Thorpe]
5347 
5348   *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
5349      If /dev/[u]random devices are not available or do not return enough
5350      entropy, EGD style sockets (served by EGD or PRNGD) will automatically
5351      be queried.
5352      The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
5353      /etc/entropy will be queried once each in this sequence, quering stops
5354      when enough entropy was collected without querying more sockets.
5355      [Lutz Jaenicke]
5356 
5357   *) Change the Unix RAND_poll() variant to be able to poll several
5358      random devices, as specified by DEVRANDOM, until a sufficient amount
5359      of data has been collected.   We spend at most 10 ms on each file
5360      (select timeout) and read in non-blocking mode.  DEVRANDOM now
5361      defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
5362      (previously it was just the string "/dev/urandom"), so on typical
5363      platforms the 10 ms delay will never occur.
5364      Also separate out the Unix variant to its own file, rand_unix.c.
5365      For VMS, there's a currently-empty rand_vms.c.
5366      [Richard Levitte]
5367 
5368   *) Move OCSP client related routines to ocsp_cl.c. These
5369      provide utility functions which an application needing
5370      to issue a request to an OCSP responder and analyse the
5371      response will typically need: as opposed to those which an
5372      OCSP responder itself would need which will be added later.
5373 
5374      OCSP_request_sign() signs an OCSP request with an API similar
5375      to PKCS7_sign(). OCSP_response_status() returns status of OCSP
5376      response. OCSP_response_get1_basic() extracts basic response
5377      from response. OCSP_resp_find_status(): finds and extracts status
5378      information from an OCSP_CERTID structure (which will be created
5379      when the request structure is built). These are built from lower
5380      level functions which work on OCSP_SINGLERESP structures but
5381      wont normally be used unless the application wishes to examine
5382      extensions in the OCSP response for example.
5383 
5384      Replace nonce routines with a pair of functions.
5385      OCSP_request_add1_nonce() adds a nonce value and optionally
5386      generates a random value. OCSP_check_nonce() checks the
5387      validity of the nonce in an OCSP response.
5388      [Steve Henson]
5389 
5390   *) Change function OCSP_request_add() to OCSP_request_add0_id().
5391      This doesn't copy the supplied OCSP_CERTID and avoids the
5392      need to free up the newly created id. Change return type
5393      to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
5394      This can then be used to add extensions to the request.
5395      Deleted OCSP_request_new(), since most of its functionality
5396      is now in OCSP_REQUEST_new() (and the case insensitive name
5397      clash) apart from the ability to set the request name which
5398      will be added elsewhere.
5399      [Steve Henson]
5400 
5401   *) Update OCSP API. Remove obsolete extensions argument from
5402      various functions. Extensions are now handled using the new
5403      OCSP extension code. New simple OCSP HTTP function which 
5404      can be used to send requests and parse the response.
5405      [Steve Henson]
5406 
5407   *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
5408      ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
5409      uses the special reorder version of SET OF to sort the attributes
5410      and reorder them to match the encoded order. This resolves a long
5411      standing problem: a verify on a PKCS7 structure just after signing
5412      it used to fail because the attribute order did not match the
5413      encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
5414      it uses the received order. This is necessary to tolerate some broken
5415      software that does not order SET OF. This is handled by encoding
5416      as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
5417      to produce the required SET OF.
5418      [Steve Henson]
5419 
5420   *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
5421      OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
5422      files to get correct declarations of the ASN.1 item variables.
5423      [Richard Levitte]
5424 
5425   *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
5426      PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
5427      asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
5428      NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
5429      New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
5430      ASN1_ITEM and no wrapper functions.
5431      [Steve Henson]
5432 
5433   *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
5434      replace the old function pointer based I/O routines. Change most of
5435      the *_d2i_bio() and *_d2i_fp() functions to use these.
5436      [Steve Henson]
5437 
5438   *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
5439      lines, recognice more "algorithms" that can be deselected, and make
5440      it complain about algorithm deselection that isn't recognised.
5441      [Richard Levitte]
5442 
5443   *) New ASN1 functions to handle dup, sign, verify, digest, pack and
5444      unpack operations in terms of ASN1_ITEM. Modify existing wrappers
5445      to use new functions. Add NO_ASN1_OLD which can be set to remove
5446      some old style ASN1 functions: this can be used to determine if old
5447      code will still work when these eventually go away.
5448      [Steve Henson]
5449 
5450   *) New extension functions for OCSP structures, these follow the
5451      same conventions as certificates and CRLs.
5452      [Steve Henson]
5453 
5454   *) New function X509V3_add1_i2d(). This automatically encodes and
5455      adds an extension. Its behaviour can be customised with various
5456      flags to append, replace or delete. Various wrappers added for
5457      certifcates and CRLs.
5458      [Steve Henson]
5459 
5460   *) Fix to avoid calling the underlying ASN1 print routine when
5461      an extension cannot be parsed. Correct a typo in the
5462      OCSP_SERVICELOC extension. Tidy up print OCSP format.
5463      [Steve Henson]
5464 
5465   *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
5466      entries for variables.
5467      [Steve Henson]
5468 
5469   *) Add functionality to apps/openssl.c for detecting locking
5470      problems: As the program is single-threaded, all we have
5471      to do is register a locking callback using an array for
5472      storing which locks are currently held by the program.
5473      [Bodo Moeller]
5474 
5475   *) Use a lock around the call to CRYPTO_get_ex_new_index() in
5476      SSL_get_ex_data_X509_STORE_idx(), which is used in
5477      ssl_verify_cert_chain() and thus can be called at any time
5478      during TLS/SSL handshakes so that thread-safety is essential.
5479      Unfortunately, the ex_data design is not at all suited
5480      for multi-threaded use, so it probably should be abolished.
5481      [Bodo Moeller]
5482 
5483   *) Added Broadcom "ubsec" ENGINE to OpenSSL.
5484      [Broadcom, tweaked and integrated by Geoff Thorpe]
5485 
5486   *) Move common extension printing code to new function
5487      X509V3_print_extensions(). Reorganise OCSP print routines and
5488      implement some needed OCSP ASN1 functions. Add OCSP extensions.
5489      [Steve Henson]
5490 
5491   *) New function X509_signature_print() to remove duplication in some
5492      print routines.
5493      [Steve Henson]
5494 
5495   *) Add a special meaning when SET OF and SEQUENCE OF flags are both
5496      set (this was treated exactly the same as SET OF previously). This
5497      is used to reorder the STACK representing the structure to match the
5498      encoding. This will be used to get round a problem where a PKCS7
5499      structure which was signed could not be verified because the STACK
5500      order did not reflect the encoded order.
5501      [Steve Henson]
5502 
5503   *) Reimplement the OCSP ASN1 module using the new code.
5504      [Steve Henson]
5505 
5506   *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
5507      for its ASN1 operations. The old style function pointers still exist
5508      for now but they will eventually go away.
5509      [Steve Henson]
5510 
5511   *) Merge in replacement ASN1 code from the ASN1 branch. This almost
5512      completely replaces the old ASN1 functionality with a table driven
5513      encoder and decoder which interprets an ASN1_ITEM structure describing
5514      the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
5515      largely maintained. Almost all of the old asn1_mac.h macro based ASN1
5516      has also been converted to the new form.
5517      [Steve Henson]
5518 
5519   *) Change BN_mod_exp_recp so that negative moduli are tolerated
5520      (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
5521      so that BN_mod_exp_mont and BN_mod_exp_mont_word work
5522      for negative moduli.
5523      [Bodo Moeller]
5524 
5525   *) Fix BN_uadd and BN_usub: Always return non-negative results instead
5526      of not touching the result's sign bit.
5527      [Bodo Moeller]
5528 
5529   *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
5530      set.
5531      [Bodo Moeller]
5532 
5533   *) Changed the LHASH code to use prototypes for callbacks, and created
5534      macros to declare and implement thin (optionally static) functions
5535      that provide type-safety and avoid function pointer casting for the
5536      type-specific callbacks.
5537      [Geoff Thorpe]
5538 
5539   *) Added Kerberos Cipher Suites to be used with TLS, as written in
5540      RFC 2712.
5541      [Veers Staats <staatsvr@asc.hpc.mil>,
5542       Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
5543 
5544   *) Reformat the FAQ so the different questions and answers can be divided
5545      in sections depending on the subject.
5546      [Richard Levitte]
5547 
5548   *) Have the zlib compression code load ZLIB.DLL dynamically under
5549      Windows.
5550      [Richard Levitte]
5551 
5552   *) New function BN_mod_sqrt for computing square roots modulo a prime
5553      (using the probabilistic Tonelli-Shanks algorithm unless
5554      p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
5555      be handled deterministically).
5556      [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
5557 
5558   *) Make BN_mod_inverse faster by explicitly handling small quotients
5559      in the Euclid loop. (Speed gain about 20% for small moduli [256 or
5560      512 bits], about 30% for larger ones [1024 or 2048 bits].)
5561      [Bodo Moeller]
5562 
5563   *) New function BN_kronecker.
5564      [Bodo Moeller]
5565 
5566   *) Fix BN_gcd so that it works on negative inputs; the result is
5567      positive unless both parameters are zero.
5568      Previously something reasonably close to an infinite loop was
5569      possible because numbers could be growing instead of shrinking
5570      in the implementation of Euclid's algorithm.
5571      [Bodo Moeller]
5572 
5573   *) Fix BN_is_word() and BN_is_one() macros to take into account the
5574      sign of the number in question.
5575 
5576      Fix BN_is_word(a,w) to work correctly for w == 0.
5577 
5578      The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
5579      because its test if the absolute value of 'a' equals 'w'.
5580      Note that BN_abs_is_word does *not* handle w == 0 reliably;
5581      it exists mostly for use in the implementations of BN_is_zero(),
5582      BN_is_one(), and BN_is_word().
5583      [Bodo Moeller]
5584 
5585   *) New function BN_swap.
5586      [Bodo Moeller]
5587 
5588   *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
5589      the exponentiation functions are more likely to produce reasonable
5590      results on negative inputs.
5591      [Bodo Moeller]
5592 
5593   *) Change BN_mod_mul so that the result is always non-negative.
5594      Previously, it could be negative if one of the factors was negative;
5595      I don't think anyone really wanted that behaviour.
5596      [Bodo Moeller]
5597 
5598   *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
5599      (except for exponentiation, which stays in crypto/bn/bn_exp.c,
5600      and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
5601      and add new functions:
5602 
5603           BN_nnmod
5604           BN_mod_sqr
5605           BN_mod_add
5606           BN_mod_add_quick
5607           BN_mod_sub
5608           BN_mod_sub_quick
5609           BN_mod_lshift1
5610           BN_mod_lshift1_quick
5611           BN_mod_lshift
5612           BN_mod_lshift_quick
5613 
5614      These functions always generate non-negative results.
5615 
5616      BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder  r
5617      such that  |m| < r < 0,  BN_nnmod will output  rem + |m|  instead).
5618 
5619      BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
5620      BN_mod_XXX(r, a, [b,] m, ctx), but requires that  a  [and  b]
5621      be reduced modulo  m.
5622      [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
5623 
5624 #if 0
5625      The following entry accidentily appeared in the CHANGES file
5626      distributed with OpenSSL 0.9.7.  The modifications described in
5627      it do *not* apply to OpenSSL 0.9.7.
5628 
5629   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
5630      was actually never needed) and in BN_mul().  The removal in BN_mul()
5631      required a small change in bn_mul_part_recursive() and the addition
5632      of the functions bn_cmp_part_words(), bn_sub_part_words() and
5633      bn_add_part_words(), which do the same thing as bn_cmp_words(),
5634      bn_sub_words() and bn_add_words() except they take arrays with
5635      differing sizes.
5636      [Richard Levitte]
5637 #endif
5638 
5639   *) In 'openssl passwd', verify passwords read from the terminal
5640      unless the '-salt' option is used (which usually means that
5641      verification would just waste user's time since the resulting
5642      hash is going to be compared with some given password hash)
5643      or the new '-noverify' option is used.
5644 
5645      This is an incompatible change, but it does not affect
5646      non-interactive use of 'openssl passwd' (passwords on the command
5647      line, '-stdin' option, '-in ...' option) and thus should not
5648      cause any problems.
5649      [Bodo Moeller]
5650 
5651   *) Remove all references to RSAref, since there's no more need for it.
5652      [Richard Levitte]
5653 
5654   *) Make DSO load along a path given through an environment variable
5655      (SHLIB_PATH) with shl_load().
5656      [Richard Levitte]
5657 
5658   *) Constify the ENGINE code as a result of BIGNUM constification.
5659      Also constify the RSA code and most things related to it.  In a
5660      few places, most notable in the depth of the ASN.1 code, ugly
5661      casts back to non-const were required (to be solved at a later
5662      time)
5663      [Richard Levitte]
5664 
5665   *) Make it so the openssl application has all engines loaded by default.
5666      [Richard Levitte]
5667 
5668   *) Constify the BIGNUM routines a little more.
5669      [Richard Levitte]
5670 
5671   *) Add the following functions:
5672 
5673         ENGINE_load_cswift()
5674         ENGINE_load_chil()
5675         ENGINE_load_atalla()
5676         ENGINE_load_nuron()
5677         ENGINE_load_builtin_engines()
5678 
5679      That way, an application can itself choose if external engines that
5680      are built-in in OpenSSL shall ever be used or not.  The benefit is
5681      that applications won't have to be linked with libdl or other dso
5682      libraries unless it's really needed.
5683 
5684      Changed 'openssl engine' to load all engines on demand.
5685      Changed the engine header files to avoid the duplication of some
5686      declarations (they differed!).
5687      [Richard Levitte]
5688 
5689   *) 'openssl engine' can now list capabilities.
5690      [Richard Levitte]
5691 
5692   *) Better error reporting in 'openssl engine'.
5693      [Richard Levitte]
5694 
5695   *) Never call load_dh_param(NULL) in s_server.
5696      [Bodo Moeller]
5697 
5698   *) Add engine application.  It can currently list engines by name and
5699      identity, and test if they are actually available.
5700      [Richard Levitte]
5701 
5702   *) Improve RPM specification file by forcing symbolic linking and making
5703      sure the installed documentation is also owned by root.root.
5704      [Damien Miller <djm@mindrot.org>]
5705 
5706   *) Give the OpenSSL applications more possibilities to make use of
5707      keys (public as well as private) handled by engines.
5708      [Richard Levitte]
5709 
5710   *) Add OCSP code that comes from CertCo.
5711      [Richard Levitte]
5712 
5713   *) Add VMS support for the Rijndael code.
5714      [Richard Levitte]
5715 
5716   *) Added untested support for Nuron crypto accelerator.
5717      [Ben Laurie]
5718 
5719   *) Add support for external cryptographic devices.  This code was
5720      previously distributed separately as the "engine" branch.
5721      [Geoff Thorpe, Richard Levitte]
5722 
5723   *) Rework the filename-translation in the DSO code. It is now possible to
5724      have far greater control over how a "name" is turned into a filename
5725      depending on the operating environment and any oddities about the
5726      different shared library filenames on each system.
5727      [Geoff Thorpe]
5728 
5729   *) Support threads on FreeBSD-elf in Configure.
5730      [Richard Levitte]
5731 
5732   *) Fix for SHA1 assembly problem with MASM: it produces
5733      warnings about corrupt line number information when assembling
5734      with debugging information. This is caused by the overlapping
5735      of two sections.
5736      [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
5737 
5738   *) NCONF changes.
5739      NCONF_get_number() has no error checking at all.  As a replacement,
5740      NCONF_get_number_e() is defined (_e for "error checking") and is
5741      promoted strongly.  The old NCONF_get_number is kept around for
5742      binary backward compatibility.
5743      Make it possible for methods to load from something other than a BIO,
5744      by providing a function pointer that is given a name instead of a BIO.
5745      For example, this could be used to load configuration data from an
5746      LDAP server.
5747      [Richard Levitte]
5748 
5749   *) Fix for non blocking accept BIOs. Added new I/O special reason
5750      BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
5751      with non blocking I/O was not possible because no retry code was
5752      implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
5753      this case.
5754      [Steve Henson]
5755 
5756   *) Added the beginnings of Rijndael support.
5757      [Ben Laurie]
5758 
5759   *) Fix for bug in DirectoryString mask setting. Add support for
5760      X509_NAME_print_ex() in 'req' and X509_print_ex() function
5761      to allow certificate printing to more controllable, additional
5762      'certopt' option to 'x509' to allow new printing options to be
5763      set.
5764      [Steve Henson]
5765 
5766   *) Clean old EAY MD5 hack from e_os.h.
5767      [Richard Levitte]
5768 
5769  Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
5770 
5771   *) Fix null-pointer assignment in do_change_cipher_spec() revealed
5772      by using the Codenomicon TLS Test Tool (CVE-2004-0079)
5773      [Joe Orton, Steve Henson]
5774 
5775  Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
5776 
5777   *) Fix additional bug revealed by the NISCC test suite:
5778 
5779      Stop bug triggering large recursion when presented with
5780      certain ASN.1 tags (CVE-2003-0851)
5781      [Steve Henson]
5782 
5783  Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
5784 
5785   *) Fix various bugs revealed by running the NISCC test suite:
5786 
5787      Stop out of bounds reads in the ASN1 code when presented with
5788      invalid tags (CVE-2003-0543 and CVE-2003-0544).
5789      
5790      If verify callback ignores invalid public key errors don't try to check
5791      certificate signature with the NULL public key.
5792 
5793      [Steve Henson]
5794 
5795   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
5796      if the server requested one: as stated in TLS 1.0 and SSL 3.0
5797      specifications.
5798      [Steve Henson]
5799 
5800   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
5801      extra data after the compression methods not only for TLS 1.0
5802      but also for SSL 3.0 (as required by the specification).
5803      [Bodo Moeller; problem pointed out by Matthias Loepfe]
5804 
5805   *) Change X509_certificate_type() to mark the key as exported/exportable
5806      when it's 512 *bits* long, not 512 bytes.
5807      [Richard Levitte]
5808 
5809  Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
5810 
5811   *) Countermeasure against the Klima-Pokorny-Rosa extension of
5812      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
5813      a protocol version number mismatch like a decryption error
5814      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
5815      [Bodo Moeller]
5816 
5817   *) Turn on RSA blinding by default in the default implementation
5818      to avoid a timing attack. Applications that don't want it can call
5819      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
5820      They would be ill-advised to do so in most cases.
5821      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
5822 
5823   *) Change RSA blinding code so that it works when the PRNG is not
5824      seeded (in this case, the secret RSA exponent is abused as
5825      an unpredictable seed -- if it is not unpredictable, there
5826      is no point in blinding anyway).  Make RSA blinding thread-safe
5827      by remembering the creator's thread ID in rsa->blinding and
5828      having all other threads use local one-time blinding factors
5829      (this requires more computation than sharing rsa->blinding, but
5830      avoids excessive locking; and if an RSA object is not shared
5831      between threads, blinding will still be very fast).
5832      [Bodo Moeller]
5833 
5834  Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
5835 
5836   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
5837      via timing by performing a MAC computation even if incorrrect
5838      block cipher padding has been found.  This is a countermeasure
5839      against active attacks where the attacker has to distinguish
5840      between bad padding and a MAC verification error. (CVE-2003-0078)
5841 
5842      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
5843      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
5844      Martin Vuagnoux (EPFL, Ilion)]
5845 
5846  Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
5847 
5848   *) New function OPENSSL_cleanse(), which is used to cleanse a section of
5849      memory from it's contents.  This is done with a counter that will
5850      place alternating values in each byte.  This can be used to solve
5851      two issues: 1) the removal of calls to memset() by highly optimizing
5852      compilers, and 2) cleansing with other values than 0, since those can
5853      be read through on certain media, for example a swap space on disk.
5854      [Geoff Thorpe]
5855 
5856   *) Bugfix: client side session caching did not work with external caching,
5857      because the session->cipher setting was not restored when reloading
5858      from the external cache. This problem was masked, when
5859      SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
5860      (Found by Steve Haslam <steve@araqnid.ddts.net>.)
5861      [Lutz Jaenicke]
5862 
5863   *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
5864      length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
5865      [Zeev Lieber <zeev-l@yahoo.com>]
5866 
5867   *) Undo an undocumented change introduced in 0.9.6e which caused
5868      repeated calls to OpenSSL_add_all_ciphers() and 
5869      OpenSSL_add_all_digests() to be ignored, even after calling
5870      EVP_cleanup().
5871      [Richard Levitte]
5872 
5873   *) Change the default configuration reader to deal with last line not
5874      being properly terminated.
5875      [Richard Levitte]
5876 
5877   *) Change X509_NAME_cmp() so it applies the special rules on handling
5878      DN values that are of type PrintableString, as well as RDNs of type
5879      emailAddress where the value has the type ia5String.
5880      [stefank@valicert.com via Richard Levitte]
5881 
5882   *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
5883      the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
5884      doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
5885      the bitwise-OR of the two for use by the majority of applications
5886      wanting this behaviour, and update the docs. The documented
5887      behaviour and actual behaviour were inconsistent and had been
5888      changing anyway, so this is more a bug-fix than a behavioural
5889      change.
5890      [Geoff Thorpe, diagnosed by Nadav Har'El]
5891 
5892   *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
5893      (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
5894      [Bodo Moeller]
5895 
5896   *) Fix initialization code race conditions in
5897         SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
5898         SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
5899         SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
5900         TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
5901         ssl2_get_cipher_by_char(),
5902         ssl3_get_cipher_by_char().
5903      [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
5904 
5905   *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
5906      the cached sessions are flushed, as the remove_cb() might use ex_data
5907      contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
5908      (see [openssl.org #212]).
5909      [Geoff Thorpe, Lutz Jaenicke]
5910 
5911   *) Fix typo in OBJ_txt2obj which incorrectly passed the content
5912      length, instead of the encoding length to d2i_ASN1_OBJECT.
5913      [Steve Henson]
5914 
5915  Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
5916 
5917   *) [In 0.9.6g-engine release:]
5918      Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
5919      [Lynn Gazis <lgazis@rainbow.com>]
5920 
5921  Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
5922 
5923   *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
5924      and get fix the header length calculation.
5925      [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
5926         Alon Kantor <alonk@checkpoint.com> (and others),
5927         Steve Henson]
5928 
5929   *) Use proper error handling instead of 'assertions' in buffer
5930      overflow checks added in 0.9.6e.  This prevents DoS (the
5931      assertions could call abort()).
5932      [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
5933 
5934  Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
5935 
5936   *) Add various sanity checks to asn1_get_length() to reject
5937      the ASN1 length bytes if they exceed sizeof(long), will appear
5938      negative or the content length exceeds the length of the
5939      supplied buffer.
5940      [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
5941 
5942   *) Fix cipher selection routines: ciphers without encryption had no flags
5943      for the cipher strength set and where therefore not handled correctly
5944      by the selection routines (PR #130).
5945      [Lutz Jaenicke]
5946 
5947   *) Fix EVP_dsa_sha macro.
5948      [Nils Larsch]
5949 
5950   *) New option
5951           SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
5952      for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
5953      that was added in OpenSSL 0.9.6d.
5954 
5955      As the countermeasure turned out to be incompatible with some
5956      broken SSL implementations, the new option is part of SSL_OP_ALL.
5957      SSL_OP_ALL is usually employed when compatibility with weird SSL
5958      implementations is desired (e.g. '-bugs' option to 's_client' and
5959      's_server'), so the new option is automatically set in many
5960      applications.
5961      [Bodo Moeller]
5962 
5963   *) Changes in security patch:
5964 
5965      Changes marked "(CHATS)" were sponsored by the Defense Advanced
5966      Research Projects Agency (DARPA) and Air Force Research Laboratory,
5967      Air Force Materiel Command, USAF, under agreement number
5968      F30602-01-2-0537.
5969 
5970   *) Add various sanity checks to asn1_get_length() to reject
5971      the ASN1 length bytes if they exceed sizeof(long), will appear
5972      negative or the content length exceeds the length of the
5973      supplied buffer. (CVE-2002-0659)
5974      [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
5975 
5976   *) Assertions for various potential buffer overflows, not known to
5977      happen in practice.
5978      [Ben Laurie (CHATS)]
5979 
5980   *) Various temporary buffers to hold ASCII versions of integers were
5981      too small for 64 bit platforms. (CVE-2002-0655)
5982      [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
5983 
5984   *) Remote buffer overflow in SSL3 protocol - an attacker could
5985      supply an oversized session ID to a client. (CVE-2002-0656)
5986      [Ben Laurie (CHATS)]
5987 
5988   *) Remote buffer overflow in SSL2 protocol - an attacker could
5989      supply an oversized client master key. (CVE-2002-0656)
5990      [Ben Laurie (CHATS)]
5991 
5992  Changes between 0.9.6c and 0.9.6d  [9 May 2002]
5993 
5994   *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
5995      encoded as NULL) with id-dsa-with-sha1.
5996      [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
5997 
5998   *) Check various X509_...() return values in apps/req.c.
5999      [Nils Larsch <nla@trustcenter.de>]
6000 
6001   *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
6002      an end-of-file condition would erronously be flagged, when the CRLF
6003      was just at the end of a processed block. The bug was discovered when
6004      processing data through a buffering memory BIO handing the data to a
6005      BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
6006      <ptsekov@syntrex.com> and Nedelcho Stanev.
6007      [Lutz Jaenicke]
6008 
6009   *) Implement a countermeasure against a vulnerability recently found
6010      in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
6011      before application data chunks to avoid the use of known IVs
6012      with data potentially chosen by the attacker.
6013      [Bodo Moeller]
6014 
6015   *) Fix length checks in ssl3_get_client_hello().
6016      [Bodo Moeller]
6017 
6018   *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
6019      to prevent ssl3_read_internal() from incorrectly assuming that
6020      ssl3_read_bytes() found application data while handshake
6021      processing was enabled when in fact s->s3->in_read_app_data was
6022      merely automatically cleared during the initial handshake.
6023      [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
6024 
6025   *) Fix object definitions for Private and Enterprise: they were not
6026      recognized in their shortname (=lowercase) representation. Extend
6027      obj_dat.pl to issue an error when using undefined keywords instead
6028      of silently ignoring the problem (Svenning Sorensen
6029      <sss@sss.dnsalias.net>).
6030      [Lutz Jaenicke]
6031 
6032   *) Fix DH_generate_parameters() so that it works for 'non-standard'
6033      generators, i.e. generators other than 2 and 5.  (Previously, the
6034      code did not properly initialise the 'add' and 'rem' values to
6035      BN_generate_prime().)
6036 
6037      In the new general case, we do not insist that 'generator' is
6038      actually a primitive root: This requirement is rather pointless;
6039      a generator of the order-q subgroup is just as good, if not
6040      better.
6041      [Bodo Moeller]
6042  
6043   *) Map new X509 verification errors to alerts. Discovered and submitted by
6044      Tom Wu <tom@arcot.com>.
6045      [Lutz Jaenicke]
6046 
6047   *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
6048      returning non-zero before the data has been completely received
6049      when using non-blocking I/O.
6050      [Bodo Moeller; problem pointed out by John Hughes]
6051 
6052   *) Some of the ciphers missed the strength entry (SSL_LOW etc).
6053      [Ben Laurie, Lutz Jaenicke]
6054 
6055   *) Fix bug in SSL_clear(): bad sessions were not removed (found by
6056      Yoram Zahavi <YoramZ@gilian.com>).
6057      [Lutz Jaenicke]
6058 
6059   *) Add information about CygWin 1.3 and on, and preserve proper
6060      configuration for the versions before that.
6061      [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
6062 
6063   *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
6064      check whether we deal with a copy of a session and do not delete from
6065      the cache in this case. Problem reported by "Izhar Shoshani Levi"
6066      <izhar@checkpoint.com>.
6067      [Lutz Jaenicke]
6068 
6069   *) Do not store session data into the internal session cache, if it
6070      is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
6071      flag is set). Proposed by Aslam <aslam@funk.com>.
6072      [Lutz Jaenicke]
6073 
6074   *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
6075      value is 0.
6076      [Richard Levitte]
6077 
6078   *) [In 0.9.6d-engine release:]
6079      Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
6080      [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
6081 
6082   *) Add the configuration target linux-s390x.
6083      [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
6084 
6085   *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
6086      ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
6087      variable as an indication that a ClientHello message has been
6088      received.  As the flag value will be lost between multiple
6089      invocations of ssl3_accept when using non-blocking I/O, the
6090      function may not be aware that a handshake has actually taken
6091      place, thus preventing a new session from being added to the
6092      session cache.
6093 
6094      To avoid this problem, we now set s->new_session to 2 instead of
6095      using a local variable.
6096      [Lutz Jaenicke, Bodo Moeller]
6097 
6098   *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
6099      if the SSL_R_LENGTH_MISMATCH error is detected.
6100      [Geoff Thorpe, Bodo Moeller]
6101 
6102   *) New 'shared_ldflag' column in Configure platform table.
6103      [Richard Levitte]
6104 
6105   *) Fix EVP_CIPHER_mode macro.
6106      ["Dan S. Camper" <dan@bti.net>]
6107 
6108   *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
6109      type, we must throw them away by setting rr->length to 0.
6110      [D P Chang <dpc@qualys.com>]
6111 
6112  Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
6113 
6114   *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
6115      <Dominikus.Scherkl@biodata.com>.  (The previous implementation
6116      worked incorrectly for those cases where  range = 10..._2  and
6117      3*range  is two bits longer than  range.)
6118      [Bodo Moeller]
6119 
6120   *) Only add signing time to PKCS7 structures if it is not already
6121      present.
6122      [Steve Henson]
6123 
6124   *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
6125      OBJ_ld_ce should be OBJ_id_ce.
6126      Also some ip-pda OIDs in crypto/objects/objects.txt were
6127      incorrect (cf. RFC 3039).
6128      [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
6129 
6130   *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
6131      returns early because it has nothing to do.
6132      [Andy Schneider <andy.schneider@bjss.co.uk>]
6133 
6134   *) [In 0.9.6c-engine release:]
6135      Fix mutex callback return values in crypto/engine/hw_ncipher.c.
6136      [Andy Schneider <andy.schneider@bjss.co.uk>]
6137 
6138   *) [In 0.9.6c-engine release:]
6139      Add support for Cryptographic Appliance's keyserver technology.
6140      (Use engine 'keyclient')
6141      [Cryptographic Appliances and Geoff Thorpe]
6142 
6143   *) Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
6144      is called via tools/c89.sh because arguments have to be
6145      rearranged (all '-L' options must appear before the first object
6146      modules).
6147      [Richard Shapiro <rshapiro@abinitio.com>]
6148 
6149   *) [In 0.9.6c-engine release:]
6150      Add support for Broadcom crypto accelerator cards, backported
6151      from 0.9.7.
6152      [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
6153 
6154   *) [In 0.9.6c-engine release:]
6155      Add support for SureWare crypto accelerator cards from 
6156      Baltimore Technologies.  (Use engine 'sureware')
6157      [Baltimore Technologies and Mark Cox]
6158 
6159   *) [In 0.9.6c-engine release:]
6160      Add support for crypto accelerator cards from Accelerated
6161      Encryption Processing, www.aep.ie.  (Use engine 'aep')
6162      [AEP Inc. and Mark Cox]
6163 
6164   *) Add a configuration entry for gcc on UnixWare.
6165      [Gary Benson <gbenson@redhat.com>]
6166 
6167   *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
6168      messages are stored in a single piece (fixed-length part and
6169      variable-length part combined) and fix various bugs found on the way.
6170      [Bodo Moeller]
6171 
6172   *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
6173      instead.  BIO_gethostbyname() does not know what timeouts are
6174      appropriate, so entries would stay in cache even when they have
6175      become invalid.
6176      [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
6177 
6178   *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
6179      faced with a pathologically small ClientHello fragment that does
6180      not contain client_version: Instead of aborting with an error,
6181      simply choose the highest available protocol version (i.e.,
6182      TLS 1.0 unless it is disabled).  In practice, ClientHello
6183      messages are never sent like this, but this change gives us
6184      strictly correct behaviour at least for TLS.
6185      [Bodo Moeller]
6186 
6187   *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
6188      never resets s->method to s->ctx->method when called from within
6189      one of the SSL handshake functions.
6190      [Bodo Moeller; problem pointed out by Niko Baric]
6191 
6192   *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
6193      (sent using the client's version number) if client_version is
6194      smaller than the protocol version in use.  Also change
6195      ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
6196      the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
6197      the client will at least see that alert.
6198      [Bodo Moeller]
6199 
6200   *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
6201      correctly.
6202      [Bodo Moeller]
6203 
6204   *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
6205      client receives HelloRequest while in a handshake.
6206      [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
6207 
6208   *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
6209      should end in 'break', not 'goto end' which circuments various
6210      cleanups done in state SSL_ST_OK.   But session related stuff
6211      must be disabled for SSL_ST_OK in the case that we just sent a
6212      HelloRequest.
6213 
6214      Also avoid some overhead by not calling ssl_init_wbio_buffer()
6215      before just sending a HelloRequest.
6216      [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
6217 
6218   *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
6219      reveal whether illegal block cipher padding was found or a MAC
6220      verification error occured.  (Neither SSLerr() codes nor alerts
6221      are directly visible to potential attackers, but the information
6222      may leak via logfiles.)
6223 
6224      Similar changes are not required for the SSL 2.0 implementation
6225      because the number of padding bytes is sent in clear for SSL 2.0,
6226      and the extra bytes are just ignored.  However ssl/s2_pkt.c
6227      failed to verify that the purported number of padding bytes is in
6228      the legal range.
6229      [Bodo Moeller]
6230 
6231   *) Add OpenUNIX-8 support including shared libraries
6232      (Boyd Lynn Gerber <gerberb@zenez.com>).
6233      [Lutz Jaenicke]
6234 
6235   *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
6236      'wristwatch attack' using huge encoding parameters (cf.
6237      James H. Manger's CRYPTO 2001 paper).  Note that the
6238      RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
6239      encoding parameters and hence was not vulnerable.
6240      [Bodo Moeller]
6241 
6242   *) BN_sqr() bug fix.
6243      [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
6244 
6245   *) Rabin-Miller test analyses assume uniformly distributed witnesses,
6246      so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
6247      followed by modular reduction.
6248      [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
6249 
6250   *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
6251      equivalent based on BN_pseudo_rand() instead of BN_rand().
6252      [Bodo Moeller]
6253 
6254   *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
6255      This function was broken, as the check for a new client hello message
6256      to handle SGC did not allow these large messages.
6257      (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
6258      [Lutz Jaenicke]
6259 
6260   *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
6261      [Lutz Jaenicke]
6262 
6263   *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
6264      for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
6265      [Lutz Jaenicke]
6266 
6267   *) Rework the configuration and shared library support for Tru64 Unix.
6268      The configuration part makes use of modern compiler features and
6269      still retains old compiler behavior for those that run older versions
6270      of the OS.  The shared library support part includes a variant that
6271      uses the RPATH feature, and is available through the special
6272      configuration target "alpha-cc-rpath", which will never be selected
6273      automatically.
6274      [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
6275 
6276   *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
6277      with the same message size as in ssl3_get_certificate_request().
6278      Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
6279      messages might inadvertently be reject as too long.
6280      [Petr Lampa <lampa@fee.vutbr.cz>]
6281 
6282   *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
6283      [Andy Polyakov]
6284 
6285   *) Modified SSL library such that the verify_callback that has been set
6286      specificly for an SSL object with SSL_set_verify() is actually being
6287      used. Before the change, a verify_callback set with this function was
6288      ignored and the verify_callback() set in the SSL_CTX at the time of
6289      the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
6290      to allow the necessary settings.
6291      [Lutz Jaenicke]
6292 
6293   *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
6294      explicitly to NULL, as at least on Solaris 8 this seems not always to be
6295      done automatically (in contradiction to the requirements of the C
6296      standard). This made problems when used from OpenSSH.
6297      [Lutz Jaenicke]
6298 
6299   *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
6300      dh->length and always used
6301 
6302           BN_rand_range(priv_key, dh->p).
6303 
6304      BN_rand_range() is not necessary for Diffie-Hellman, and this
6305      specific range makes Diffie-Hellman unnecessarily inefficient if
6306      dh->length (recommended exponent length) is much smaller than the
6307      length of dh->p.  We could use BN_rand_range() if the order of
6308      the subgroup was stored in the DH structure, but we only have
6309      dh->length.
6310 
6311      So switch back to
6312 
6313           BN_rand(priv_key, l, ...)
6314 
6315      where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
6316      otherwise.
6317      [Bodo Moeller]
6318 
6319   *) In
6320 
6321           RSA_eay_public_encrypt
6322           RSA_eay_private_decrypt
6323           RSA_eay_private_encrypt (signing)
6324           RSA_eay_public_decrypt (signature verification)
6325 
6326      (default implementations for RSA_public_encrypt,
6327      RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
6328      always reject numbers >= n.
6329      [Bodo Moeller]
6330 
6331   *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
6332      to synchronize access to 'locking_thread'.  This is necessary on
6333      systems where access to 'locking_thread' (an 'unsigned long'
6334      variable) is not atomic.
6335      [Bodo Moeller]
6336 
6337   *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
6338      *before* setting the 'crypto_lock_rand' flag.  The previous code had
6339      a race condition if 0 is a valid thread ID.
6340      [Travis Vitek <vitek@roguewave.com>]
6341 
6342   *) Add support for shared libraries under Irix.
6343      [Albert Chin-A-Young <china@thewrittenword.com>]
6344 
6345   *) Add configuration option to build on Linux on both big-endian and
6346      little-endian MIPS.
6347      [Ralf Baechle <ralf@uni-koblenz.de>]
6348 
6349   *) Add the possibility to create shared libraries on HP-UX.
6350      [Richard Levitte]
6351 
6352  Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
6353 
6354   *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
6355      to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
6356      Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
6357      PRNG state recovery was possible based on the output of
6358      one PRNG request appropriately sized to gain knowledge on
6359      'md' followed by enough consecutive 1-byte PRNG requests
6360      to traverse all of 'state'.
6361 
6362      1. When updating 'md_local' (the current thread's copy of 'md')
6363         during PRNG output generation, hash all of the previous
6364         'md_local' value, not just the half used for PRNG output.
6365 
6366      2. Make the number of bytes from 'state' included into the hash
6367         independent from the number of PRNG bytes requested.
6368 
6369      The first measure alone would be sufficient to avoid
6370      Markku-Juhani's attack.  (Actually it had never occurred
6371      to me that the half of 'md_local' used for chaining was the
6372      half from which PRNG output bytes were taken -- I had always
6373      assumed that the secret half would be used.)  The second
6374      measure makes sure that additional data from 'state' is never
6375      mixed into 'md_local' in small portions; this heuristically
6376      further strengthens the PRNG.
6377      [Bodo Moeller]
6378 
6379   *) Fix crypto/bn/asm/mips3.s.
6380      [Andy Polyakov]
6381 
6382   *) When only the key is given to "enc", the IV is undefined. Print out
6383      an error message in this case.
6384      [Lutz Jaenicke]
6385 
6386   *) Handle special case when X509_NAME is empty in X509 printing routines.
6387      [Steve Henson]
6388 
6389   *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
6390      positive and less than q.
6391      [Bodo Moeller]
6392 
6393   *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
6394      used: it isn't thread safe and the add_lock_callback should handle
6395      that itself.
6396      [Paul Rose <Paul.Rose@bridge.com>]
6397 
6398   *) Verify that incoming data obeys the block size in
6399      ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
6400      [Bodo Moeller]
6401 
6402   *) Fix OAEP check.
6403      [Ulf Möller, Bodo Möller]
6404 
6405   *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
6406      RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
6407      when fixing the server behaviour for backwards-compatible 'client
6408      hello' messages.  (Note that the attack is impractical against
6409      SSL 3.0 and TLS 1.0 anyway because length and version checking
6410      means that the probability of guessing a valid ciphertext is
6411      around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
6412      paper.)
6413 
6414      Before 0.9.5, the countermeasure (hide the error by generating a
6415      random 'decryption result') did not work properly because
6416      ERR_clear_error() was missing, meaning that SSL_get_error() would
6417      detect the supposedly ignored error.
6418 
6419      Both problems are now fixed.
6420      [Bodo Moeller]
6421 
6422   *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
6423      (previously it was 1024).
6424      [Bodo Moeller]
6425 
6426   *) Fix for compatibility mode trust settings: ignore trust settings
6427      unless some valid trust or reject settings are present.
6428      [Steve Henson]
6429 
6430   *) Fix for blowfish EVP: its a variable length cipher.
6431      [Steve Henson]
6432 
6433   *) Fix various bugs related to DSA S/MIME verification. Handle missing
6434      parameters in DSA public key structures and return an error in the
6435      DSA routines if parameters are absent.
6436      [Steve Henson]
6437 
6438   *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
6439      in the current directory if neither $RANDFILE nor $HOME was set.
6440      RAND_file_name() in 0.9.6a returned NULL in this case.  This has
6441      caused some confusion to Windows users who haven't defined $HOME.
6442      Thus RAND_file_name() is changed again: e_os.h can define a
6443      DEFAULT_HOME, which will be used if $HOME is not set.
6444      For Windows, we use "C:"; on other platforms, we still require
6445      environment variables.
6446 
6447   *) Move 'if (!initialized) RAND_poll()' into regions protected by
6448      CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
6449      having multiple threads call RAND_poll() concurrently.
6450      [Bodo Moeller]
6451 
6452   *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
6453      combination of a flag and a thread ID variable.
6454      Otherwise while one thread is in ssleay_rand_bytes (which sets the
6455      flag), *other* threads can enter ssleay_add_bytes without obeying
6456      the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
6457      that they do not hold after the first thread unsets add_do_not_lock).
6458      [Bodo Moeller]
6459 
6460   *) Change bctest again: '-x' expressions are not available in all
6461      versions of 'test'.
6462      [Bodo Moeller]
6463 
6464  Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
6465 
6466   *) Fix a couple of memory leaks in PKCS7_dataDecode()
6467      [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
6468 
6469   *) Change Configure and Makefiles to provide EXE_EXT, which will contain
6470      the default extension for executables, if any.  Also, make the perl
6471      scripts that use symlink() to test if it really exists and use "cp"
6472      if it doesn't.  All this made OpenSSL compilable and installable in
6473      CygWin.
6474      [Richard Levitte]
6475 
6476   *) Fix for asn1_GetSequence() for indefinite length constructed data.
6477      If SEQUENCE is length is indefinite just set c->slen to the total
6478      amount of data available.
6479      [Steve Henson, reported by shige@FreeBSD.org]
6480      [This change does not apply to 0.9.7.]
6481 
6482   *) Change bctest to avoid here-documents inside command substitution
6483      (workaround for FreeBSD /bin/sh bug).
6484      For compatibility with Ultrix, avoid shell functions (introduced
6485      in the bctest version that searches along $PATH).
6486      [Bodo Moeller]
6487 
6488   *) Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
6489      with des_encrypt() defined on some operating systems, like Solaris
6490      and UnixWare.
6491      [Richard Levitte]
6492 
6493   *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
6494      On the Importance of Eliminating Errors in Cryptographic
6495      Computations, J. Cryptology 14 (2001) 2, 101-119,
6496      http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
6497      [Ulf Moeller]
6498   
6499   *) MIPS assembler BIGNUM division bug fix. 
6500      [Andy Polyakov]
6501 
6502   *) Disabled incorrect Alpha assembler code.
6503      [Richard Levitte]
6504 
6505   *) Fix PKCS#7 decode routines so they correctly update the length
6506      after reading an EOC for the EXPLICIT tag.
6507      [Steve Henson]
6508      [This change does not apply to 0.9.7.]
6509 
6510   *) Fix bug in PKCS#12 key generation routines. This was triggered
6511      if a 3DES key was generated with a 0 initial byte. Include
6512      PKCS12_BROKEN_KEYGEN compilation option to retain the old
6513      (but broken) behaviour.
6514      [Steve Henson]
6515 
6516   *) Enhance bctest to search for a working bc along $PATH and print
6517      it when found.
6518      [Tim Rice <tim@multitalents.net> via Richard Levitte]
6519 
6520   *) Fix memory leaks in err.c: free err_data string if necessary;
6521      don't write to the wrong index in ERR_set_error_data.
6522      [Bodo Moeller]
6523 
6524   *) Implement ssl23_peek (analogous to ssl23_read), which previously
6525      did not exist.
6526      [Bodo Moeller]
6527 
6528   *) Replace rdtsc with _emit statements for VC++ version 5.
6529      [Jeremy Cooper <jeremy@baymoo.org>]
6530 
6531   *) Make it possible to reuse SSLv2 sessions.
6532      [Richard Levitte]
6533 
6534   *) In copy_email() check for >= 0 as a return value for
6535      X509_NAME_get_index_by_NID() since 0 is a valid index.
6536      [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
6537 
6538   *) Avoid coredump with unsupported or invalid public keys by checking if
6539      X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
6540      PKCS7_verify() fails with non detached data.
6541      [Steve Henson]
6542 
6543   *) Don't use getenv in library functions when run as setuid/setgid.
6544      New function OPENSSL_issetugid().
6545      [Ulf Moeller]
6546 
6547   *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
6548      due to incorrect handling of multi-threading:
6549 
6550      1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
6551 
6552      2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
6553 
6554      3. Count how many times MemCheck_off() has been called so that
6555         nested use can be treated correctly.  This also avoids 
6556         inband-signalling in the previous code (which relied on the
6557         assumption that thread ID 0 is impossible).
6558      [Bodo Moeller]
6559 
6560   *) Add "-rand" option also to s_client and s_server.
6561      [Lutz Jaenicke]
6562 
6563   *) Fix CPU detection on Irix 6.x.
6564      [Kurt Hockenbury <khockenb@stevens-tech.edu> and
6565       "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
6566 
6567   *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
6568      was empty.
6569      [Steve Henson]
6570      [This change does not apply to 0.9.7.]
6571 
6572   *) Use the cached encoding of an X509_NAME structure rather than
6573      copying it. This is apparently the reason for the libsafe "errors"
6574      but the code is actually correct.
6575      [Steve Henson]
6576 
6577   *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
6578      Bleichenbacher's DSA attack.
6579      Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
6580      to be set and top=0 forces the highest bit to be set; top=-1 is new
6581      and leaves the highest bit random.
6582      [Ulf Moeller, Bodo Moeller]
6583 
6584   *) In the NCONF_...-based implementations for CONF_... queries
6585      (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
6586      a temporary CONF structure with the data component set to NULL
6587      (which gives segmentation faults in lh_retrieve).
6588      Instead, use NULL for the CONF pointer in CONF_get_string and
6589      CONF_get_number (which may use environment variables) and directly
6590      return NULL from CONF_get_section.
6591      [Bodo Moeller]
6592 
6593   *) Fix potential buffer overrun for EBCDIC.
6594      [Ulf Moeller]
6595 
6596   *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
6597      keyUsage if basicConstraints absent for a CA.
6598      [Steve Henson]
6599 
6600   *) Make SMIME_write_PKCS7() write mail header values with a format that
6601      is more generally accepted (no spaces before the semicolon), since
6602      some programs can't parse those values properly otherwise.  Also make
6603      sure BIO's that break lines after each write do not create invalid
6604      headers.
6605      [Richard Levitte]
6606 
6607   *) Make the CRL encoding routines work with empty SEQUENCE OF. The
6608      macros previously used would not encode an empty SEQUENCE OF
6609      and break the signature.
6610      [Steve Henson]
6611      [This change does not apply to 0.9.7.]
6612 
6613   *) Zero the premaster secret after deriving the master secret in
6614      DH ciphersuites.
6615      [Steve Henson]
6616 
6617   *) Add some EVP_add_digest_alias registrations (as found in
6618      OpenSSL_add_all_digests()) to SSL_library_init()
6619      aka OpenSSL_add_ssl_algorithms().  This provides improved
6620      compatibility with peers using X.509 certificates
6621      with unconventional AlgorithmIdentifier OIDs.
6622      [Bodo Moeller]
6623 
6624   *) Fix for Irix with NO_ASM.
6625      ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
6626 
6627   *) ./config script fixes.
6628      [Ulf Moeller, Richard Levitte]
6629 
6630   *) Fix 'openssl passwd -1'.
6631      [Bodo Moeller]
6632 
6633   *) Change PKCS12_key_gen_asc() so it can cope with non null
6634      terminated strings whose length is passed in the passlen
6635      parameter, for example from PEM callbacks. This was done
6636      by adding an extra length parameter to asc2uni().
6637      [Steve Henson, reported by <oddissey@samsung.co.kr>]
6638 
6639   *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
6640      call failed, free the DSA structure.
6641      [Bodo Moeller]
6642 
6643   *) Fix to uni2asc() to cope with zero length Unicode strings.
6644      These are present in some PKCS#12 files.
6645      [Steve Henson]
6646 
6647   *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
6648      Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
6649      when writing a 32767 byte record.
6650      [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
6651 
6652   *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
6653      obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
6654 
6655      (RSA objects have a reference count access to which is protected
6656      by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
6657      so they are meant to be shared between threads.)
6658      [Bodo Moeller, Geoff Thorpe; original patch submitted by
6659      "Reddie, Steven" <Steven.Reddie@ca.com>]
6660 
6661   *) Fix a deadlock in CRYPTO_mem_leaks().
6662      [Bodo Moeller]
6663 
6664   *) Use better test patterns in bntest.
6665      [Ulf Möller]
6666 
6667   *) rand_win.c fix for Borland C.
6668      [Ulf Möller]
6669  
6670   *) BN_rshift bugfix for n == 0.
6671      [Bodo Moeller]
6672 
6673   *) Add a 'bctest' script that checks for some known 'bc' bugs
6674      so that 'make test' does not abort just because 'bc' is broken.
6675      [Bodo Moeller]
6676 
6677   *) Store verify_result within SSL_SESSION also for client side to
6678      avoid potential security hole. (Re-used sessions on the client side
6679      always resulted in verify_result==X509_V_OK, not using the original
6680      result of the server certificate verification.)
6681      [Lutz Jaenicke]
6682 
6683   *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
6684      SSL3_RT_APPLICATION_DATA, return 0.
6685      Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
6686      [Bodo Moeller]
6687 
6688   *) Fix SSL_peek:
6689      Both ssl2_peek and ssl3_peek, which were totally broken in earlier
6690      releases, have been re-implemented by renaming the previous
6691      implementations of ssl2_read and ssl3_read to ssl2_read_internal
6692      and ssl3_read_internal, respectively, and adding 'peek' parameters
6693      to them.  The new ssl[23]_{read,peek} functions are calls to
6694      ssl[23]_read_internal with the 'peek' flag set appropriately.
6695      A 'peek' parameter has also been added to ssl3_read_bytes, which
6696      does the actual work for ssl3_read_internal.
6697      [Bodo Moeller]
6698 
6699   *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
6700      the method-specific "init()" handler. Also clean up ex_data after
6701      calling the method-specific "finish()" handler. Previously, this was
6702      happening the other way round.
6703      [Geoff Thorpe]
6704 
6705   *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
6706      The previous value, 12, was not always sufficient for BN_mod_exp().
6707      [Bodo Moeller]
6708 
6709   *) Make sure that shared libraries get the internal name engine with
6710      the full version number and not just 0.  This should mark the
6711      shared libraries as not backward compatible.  Of course, this should
6712      be changed again when we can guarantee backward binary compatibility.
6713      [Richard Levitte]
6714 
6715   *) Fix typo in get_cert_by_subject() in by_dir.c
6716      [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
6717 
6718   *) Rework the system to generate shared libraries:
6719 
6720      - Make note of the expected extension for the shared libraries and
6721        if there is a need for symbolic links from for example libcrypto.so.0
6722        to libcrypto.so.0.9.7.  There is extended info in Configure for
6723        that.
6724 
6725      - Make as few rebuilds of the shared libraries as possible.
6726 
6727      - Still avoid linking the OpenSSL programs with the shared libraries.
6728 
6729      - When installing, install the shared libraries separately from the
6730        static ones.
6731      [Richard Levitte]
6732 
6733   *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
6734 
6735      Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
6736      and not in SSL_clear because the latter is also used by the
6737      accept/connect functions; previously, the settings made by
6738      SSL_set_read_ahead would be lost during the handshake.
6739      [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]     
6740 
6741   *) Correct util/mkdef.pl to be selective about disabled algorithms.
6742      Previously, it would create entries for disableed algorithms no
6743      matter what.
6744      [Richard Levitte]
6745 
6746   *) Added several new manual pages for SSL_* function.
6747      [Lutz Jaenicke]
6748 
6749  Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
6750 
6751   *) In ssl23_get_client_hello, generate an error message when faced
6752      with an initial SSL 3.0/TLS record that is too small to contain the
6753      first two bytes of the ClientHello message, i.e. client_version.
6754      (Note that this is a pathologic case that probably has never happened
6755      in real life.)  The previous approach was to use the version number
6756      from the record header as a substitute; but our protocol choice
6757      should not depend on that one because it is not authenticated
6758      by the Finished messages.
6759      [Bodo Moeller]
6760 
6761   *) More robust randomness gathering functions for Windows.
6762      [Jeffrey Altman <jaltman@columbia.edu>]
6763 
6764   *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
6765      not set then we don't setup the error code for issuer check errors
6766      to avoid possibly overwriting other errors which the callback does
6767      handle. If an application does set the flag then we assume it knows
6768      what it is doing and can handle the new informational codes
6769      appropriately.
6770      [Steve Henson]
6771 
6772   *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
6773      a general "ANY" type, as such it should be able to decode anything
6774      including tagged types. However it didn't check the class so it would
6775      wrongly interpret tagged types in the same way as their universal
6776      counterpart and unknown types were just rejected. Changed so that the
6777      tagged and unknown types are handled in the same way as a SEQUENCE:
6778      that is the encoding is stored intact. There is also a new type
6779      "V_ASN1_OTHER" which is used when the class is not universal, in this
6780      case we have no idea what the actual type is so we just lump them all
6781      together.
6782      [Steve Henson]
6783 
6784   *) On VMS, stdout may very well lead to a file that is written to
6785      in a record-oriented fashion.  That means that every write() will
6786      write a separate record, which will be read separately by the
6787      programs trying to read from it.  This can be very confusing.
6788 
6789      The solution is to put a BIO filter in the way that will buffer
6790      text until a linefeed is reached, and then write everything a
6791      line at a time, so every record written will be an actual line,
6792      not chunks of lines and not (usually doesn't happen, but I've
6793      seen it once) several lines in one record.  BIO_f_linebuffer() is
6794      the answer.
6795 
6796      Currently, it's a VMS-only method, because that's where it has
6797      been tested well enough.
6798      [Richard Levitte]
6799 
6800   *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
6801      it can return incorrect results.
6802      (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
6803      but it was in 0.9.6-beta[12].)
6804      [Bodo Moeller]
6805 
6806   *) Disable the check for content being present when verifying detached
6807      signatures in pk7_smime.c. Some versions of Netscape (wrongly)
6808      include zero length content when signing messages.
6809      [Steve Henson]
6810 
6811   *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
6812      BIO_ctrl (for BIO pairs).
6813      [Bodo Möller]
6814 
6815   *) Add DSO method for VMS.
6816      [Richard Levitte]
6817 
6818   *) Bug fix: Montgomery multiplication could produce results with the
6819      wrong sign.
6820      [Ulf Möller]
6821 
6822   *) Add RPM specification openssl.spec and modify it to build three
6823      packages.  The default package contains applications, application
6824      documentation and run-time libraries.  The devel package contains
6825      include files, static libraries and function documentation.  The
6826      doc package contains the contents of the doc directory.  The original
6827      openssl.spec was provided by Damien Miller <djm@mindrot.org>.
6828      [Richard Levitte]
6829      
6830   *) Add a large number of documentation files for many SSL routines.
6831      [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
6832 
6833   *) Add a configuration entry for Sony News 4.
6834      [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
6835 
6836   *) Don't set the two most significant bits to one when generating a
6837      random number < q in the DSA library.
6838      [Ulf Möller]
6839 
6840   *) New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
6841      behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
6842      the underlying transport is blocking) if a handshake took place.
6843      (The default behaviour is needed by applications such as s_client
6844      and s_server that use select() to determine when to use SSL_read;
6845      but for applications that know in advance when to expect data, it
6846      just makes things more complicated.)
6847      [Bodo Moeller]
6848 
6849   *) Add RAND_egd_bytes(), which gives control over the number of bytes read
6850      from EGD.
6851      [Ben Laurie]
6852 
6853   *) Add a few more EBCDIC conditionals that make `req' and `x509'
6854      work better on such systems.
6855      [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
6856 
6857   *) Add two demo programs for PKCS12_parse() and PKCS12_create().
6858      Update PKCS12_parse() so it copies the friendlyName and the
6859      keyid to the certificates aux info.
6860      [Steve Henson]
6861 
6862   *) Fix bug in PKCS7_verify() which caused an infinite loop
6863      if there was more than one signature.
6864      [Sven Uszpelkat <su@celocom.de>]
6865 
6866   *) Major change in util/mkdef.pl to include extra information
6867      about each symbol, as well as presentig variables as well
6868      as functions.  This change means that there's n more need
6869      to rebuild the .num files when some algorithms are excluded.
6870      [Richard Levitte]
6871 
6872   *) Allow the verify time to be set by an application,
6873      rather than always using the current time.
6874      [Steve Henson]
6875   
6876   *) Phase 2 verify code reorganisation. The certificate
6877      verify code now looks up an issuer certificate by a
6878      number of criteria: subject name, authority key id
6879      and key usage. It also verifies self signed certificates
6880      by the same criteria. The main comparison function is
6881      X509_check_issued() which performs these checks.
6882  
6883      Lot of changes were necessary in order to support this
6884      without completely rewriting the lookup code.
6885  
6886      Authority and subject key identifier are now cached.
6887  
6888      The LHASH 'certs' is X509_STORE has now been replaced
6889      by a STACK_OF(X509_OBJECT). This is mainly because an
6890      LHASH can't store or retrieve multiple objects with
6891      the same hash value.
6892 
6893      As a result various functions (which were all internal
6894      use only) have changed to handle the new X509_STORE
6895      structure. This will break anything that messed round
6896      with X509_STORE internally.
6897  
6898      The functions X509_STORE_add_cert() now checks for an
6899      exact match, rather than just subject name.
6900  
6901      The X509_STORE API doesn't directly support the retrieval
6902      of multiple certificates matching a given criteria, however
6903      this can be worked round by performing a lookup first
6904      (which will fill the cache with candidate certificates)
6905      and then examining the cache for matches. This is probably
6906      the best we can do without throwing out X509_LOOKUP
6907      entirely (maybe later...).
6908  
6909      The X509_VERIFY_CTX structure has been enhanced considerably.
6910  
6911      All certificate lookup operations now go via a get_issuer()
6912      callback. Although this currently uses an X509_STORE it
6913      can be replaced by custom lookups. This is a simple way
6914      to bypass the X509_STORE hackery necessary to make this
6915      work and makes it possible to use more efficient techniques
6916      in future. A very simple version which uses a simple
6917      STACK for its trusted certificate store is also provided
6918      using X509_STORE_CTX_trusted_stack().
6919  
6920      The verify_cb() and verify() callbacks now have equivalents
6921      in the X509_STORE_CTX structure.
6922  
6923      X509_STORE_CTX also has a 'flags' field which can be used
6924      to customise the verify behaviour.
6925      [Steve Henson]
6926  
6927   *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 
6928      excludes S/MIME capabilities.
6929      [Steve Henson]
6930 
6931   *) When a certificate request is read in keep a copy of the
6932      original encoding of the signed data and use it when outputing
6933      again. Signatures then use the original encoding rather than
6934      a decoded, encoded version which may cause problems if the
6935      request is improperly encoded.
6936      [Steve Henson]
6937 
6938   *) For consistency with other BIO_puts implementations, call
6939      buffer_write(b, ...) directly in buffer_puts instead of calling
6940      BIO_write(b, ...).
6941 
6942      In BIO_puts, increment b->num_write as in BIO_write.
6943      [Peter.Sylvester@EdelWeb.fr]
6944 
6945   *) Fix BN_mul_word for the case where the word is 0. (We have to use
6946      BN_zero, we may not return a BIGNUM with an array consisting of
6947      words set to zero.)
6948      [Bodo Moeller]
6949 
6950   *) Avoid calling abort() from within the library when problems are
6951      detected, except if preprocessor symbols have been defined
6952      (such as REF_CHECK, BN_DEBUG etc.).
6953      [Bodo Moeller]
6954 
6955   *) New openssl application 'rsautl'. This utility can be
6956      used for low level RSA operations. DER public key
6957      BIO/fp routines also added.
6958      [Steve Henson]
6959 
6960   *) New Configure entry and patches for compiling on QNX 4.
6961      [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
6962 
6963   *) A demo state-machine implementation was sponsored by
6964      Nuron (http://www.nuron.com/) and is now available in
6965      demos/state_machine.
6966      [Ben Laurie]
6967 
6968   *) New options added to the 'dgst' utility for signature
6969      generation and verification.
6970      [Steve Henson]
6971 
6972   *) Unrecognized PKCS#7 content types are now handled via a
6973      catch all ASN1_TYPE structure. This allows unsupported
6974      types to be stored as a "blob" and an application can
6975      encode and decode it manually.
6976      [Steve Henson]
6977 
6978   *) Fix various signed/unsigned issues to make a_strex.c
6979      compile under VC++.
6980      [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
6981 
6982   *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
6983      length if passed a buffer. ASN1_INTEGER_to_BN failed
6984      if passed a NULL BN and its argument was negative.
6985      [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
6986 
6987   *) Modification to PKCS#7 encoding routines to output definite
6988      length encoding. Since currently the whole structures are in
6989      memory there's not real point in using indefinite length 
6990      constructed encoding. However if OpenSSL is compiled with
6991      the flag PKCS7_INDEFINITE_ENCODING the old form is used.
6992      [Steve Henson]
6993 
6994   *) Added BIO_vprintf() and BIO_vsnprintf().
6995      [Richard Levitte]
6996 
6997   *) Added more prefixes to parse for in the the strings written
6998      through a logging bio, to cover all the levels that are available
6999      through syslog.  The prefixes are now:
7000 
7001         PANIC, EMERG, EMR       =>      LOG_EMERG
7002         ALERT, ALR              =>      LOG_ALERT
7003         CRIT, CRI               =>      LOG_CRIT
7004         ERROR, ERR              =>      LOG_ERR
7005         WARNING, WARN, WAR      =>      LOG_WARNING
7006         NOTICE, NOTE, NOT       =>      LOG_NOTICE
7007         INFO, INF               =>      LOG_INFO
7008         DEBUG, DBG              =>      LOG_DEBUG
7009 
7010      and as before, if none of those prefixes are present at the
7011      beginning of the string, LOG_ERR is chosen.
7012 
7013      On Win32, the LOG_* levels are mapped according to this:
7014 
7015         LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
7016         LOG_WARNING                             => EVENTLOG_WARNING_TYPE
7017         LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE
7018 
7019      [Richard Levitte]
7020 
7021   *) Made it possible to reconfigure with just the configuration
7022      argument "reconf" or "reconfigure".  The command line arguments
7023      are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
7024      and are retrieved from there when reconfiguring.
7025      [Richard Levitte]
7026 
7027   *) MD4 implemented.
7028      [Assar Westerlund <assar@sics.se>, Richard Levitte]
7029 
7030   *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
7031      [Richard Levitte]
7032 
7033   *) The obj_dat.pl script was messing up the sorting of object
7034      names. The reason was that it compared the quoted version
7035      of strings as a result "OCSP" > "OCSP Signing" because
7036      " > SPACE. Changed script to store unquoted versions of
7037      names and add quotes on output. It was also omitting some
7038      names from the lookup table if they were given a default
7039      value (that is if SN is missing it is given the same
7040      value as LN and vice versa), these are now added on the
7041      grounds that if an object has a name we should be able to
7042      look it up. Finally added warning output when duplicate
7043      short or long names are found.
7044      [Steve Henson]
7045 
7046   *) Changes needed for Tandem NSK.
7047      [Scott Uroff <scott@xypro.com>]
7048 
7049   *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
7050      RSA_padding_check_SSLv23(), special padding was never detected
7051      and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
7052      version rollback attacks was not effective.
7053 
7054      In s23_clnt.c, don't use special rollback-attack detection padding
7055      (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
7056      client; similarly, in s23_srvr.c, don't do the rollback check if
7057      SSL 2.0 is the only protocol enabled in the server.
7058      [Bodo Moeller]
7059 
7060   *) Make it possible to get hexdumps of unprintable data with 'openssl
7061      asn1parse'.  By implication, the functions ASN1_parse_dump() and
7062      BIO_dump_indent() are added.
7063      [Richard Levitte]
7064 
7065   *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
7066      these print out strings and name structures based on various
7067      flags including RFC2253 support and proper handling of
7068      multibyte characters. Added options to the 'x509' utility 
7069      to allow the various flags to be set.
7070      [Steve Henson]
7071 
7072   *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
7073      Also change the functions X509_cmp_current_time() and
7074      X509_gmtime_adj() work with an ASN1_TIME structure,
7075      this will enable certificates using GeneralizedTime in validity
7076      dates to be checked.
7077      [Steve Henson]
7078 
7079   *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
7080      negative public key encodings) on by default,
7081      NO_NEG_PUBKEY_BUG can be set to disable it.
7082      [Steve Henson]
7083 
7084   *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
7085      content octets. An i2c_ASN1_OBJECT is unnecessary because
7086      the encoding can be trivially obtained from the structure.
7087      [Steve Henson]
7088 
7089   *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
7090      not read locks (CRYPTO_r_[un]lock).
7091      [Bodo Moeller]
7092 
7093   *) A first attempt at creating official support for shared
7094      libraries through configuration.  I've kept it so the
7095      default is static libraries only, and the OpenSSL programs
7096      are always statically linked for now, but there are
7097      preparations for dynamic linking in place.
7098      This has been tested on Linux and Tru64.
7099      [Richard Levitte]
7100 
7101   *) Randomness polling function for Win9x, as described in:
7102      Peter Gutmann, Software Generation of Practically Strong
7103      Random Numbers.
7104      [Ulf Möller]
7105 
7106   *) Fix so PRNG is seeded in req if using an already existing
7107      DSA key.
7108      [Steve Henson]
7109 
7110   *) New options to smime application. -inform and -outform
7111      allow alternative formats for the S/MIME message including
7112      PEM and DER. The -content option allows the content to be
7113      specified separately. This should allow things like Netscape
7114      form signing output easier to verify.
7115      [Steve Henson]
7116 
7117   *) Fix the ASN1 encoding of tags using the 'long form'.
7118      [Steve Henson]
7119 
7120   *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
7121      STRING types. These convert content octets to and from the
7122      underlying type. The actual tag and length octets are
7123      already assumed to have been read in and checked. These
7124      are needed because all other string types have virtually
7125      identical handling apart from the tag. By having versions
7126      of the ASN1 functions that just operate on content octets
7127      IMPLICIT tagging can be handled properly. It also allows
7128      the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
7129      and ASN1_INTEGER are identical apart from the tag.
7130      [Steve Henson]
7131 
7132   *) Change the handling of OID objects as follows:
7133 
7134      - New object identifiers are inserted in objects.txt, following
7135        the syntax given in objects.README.
7136      - objects.pl is used to process obj_mac.num and create a new
7137        obj_mac.h.
7138      - obj_dat.pl is used to create a new obj_dat.h, using the data in
7139        obj_mac.h.
7140 
7141      This is currently kind of a hack, and the perl code in objects.pl
7142      isn't very elegant, but it works as I intended.  The simplest way
7143      to check that it worked correctly is to look in obj_dat.h and
7144      check the array nid_objs and make sure the objects haven't moved
7145      around (this is important!).  Additions are OK, as well as
7146      consistent name changes. 
7147      [Richard Levitte]
7148 
7149   *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
7150      [Bodo Moeller]
7151 
7152   *) Addition of the command line parameter '-rand file' to 'openssl req'.
7153      The given file adds to whatever has already been seeded into the
7154      random pool through the RANDFILE configuration file option or
7155      environment variable, or the default random state file.
7156      [Richard Levitte]
7157 
7158   *) mkstack.pl now sorts each macro group into lexical order.
7159      Previously the output order depended on the order the files
7160      appeared in the directory, resulting in needless rewriting
7161      of safestack.h .
7162      [Steve Henson]
7163 
7164   *) Patches to make OpenSSL compile under Win32 again. Mostly
7165      work arounds for the VC++ problem that it treats func() as
7166      func(void). Also stripped out the parts of mkdef.pl that
7167      added extra typesafe functions: these no longer exist.
7168      [Steve Henson]
7169 
7170   *) Reorganisation of the stack code. The macros are now all 
7171      collected in safestack.h . Each macro is defined in terms of
7172      a "stack macro" of the form SKM_<name>(type, a, b). The 
7173      DEBUG_SAFESTACK is now handled in terms of function casts,
7174      this has the advantage of retaining type safety without the
7175      use of additional functions. If DEBUG_SAFESTACK is not defined
7176      then the non typesafe macros are used instead. Also modified the
7177      mkstack.pl script to handle the new form. Needs testing to see
7178      if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
7179      the default if no major problems. Similar behaviour for ASN1_SET_OF
7180      and PKCS12_STACK_OF.
7181      [Steve Henson]
7182 
7183   *) When some versions of IIS use the 'NET' form of private key the
7184      key derivation algorithm is different. Normally MD5(password) is
7185      used as a 128 bit RC4 key. In the modified case
7186      MD5(MD5(password) + "SGCKEYSALT")  is used insted. Added some
7187      new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
7188      as the old Netscape_RSA functions except they have an additional
7189      'sgckey' parameter which uses the modified algorithm. Also added
7190      an -sgckey command line option to the rsa utility. Thanks to 
7191      Adrian Peck <bertie@ncipher.com> for posting details of the modified
7192      algorithm to openssl-dev.
7193      [Steve Henson]
7194 
7195   *) The evp_local.h macros were using 'c.##kname' which resulted in
7196      invalid expansion on some systems (SCO 5.0.5 for example).
7197      Corrected to 'c.kname'.
7198      [Phillip Porch <root@theporch.com>]
7199 
7200   *) New X509_get1_email() and X509_REQ_get1_email() functions that return
7201      a STACK of email addresses from a certificate or request, these look
7202      in the subject name and the subject alternative name extensions and 
7203      omit any duplicate addresses.
7204      [Steve Henson]
7205 
7206   *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
7207      This makes DSA verification about 2 % faster.
7208      [Bodo Moeller]
7209 
7210   *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
7211      (meaning that now 2^5 values will be precomputed, which is only 4 KB
7212      plus overhead for 1024 bit moduli).
7213      This makes exponentiations about 0.5 % faster for 1024 bit
7214      exponents (as measured by "openssl speed rsa2048").
7215      [Bodo Moeller]
7216 
7217   *) Rename memory handling macros to avoid conflicts with other
7218      software:
7219           Malloc         =>  OPENSSL_malloc
7220           Malloc_locked  =>  OPENSSL_malloc_locked
7221           Realloc        =>  OPENSSL_realloc
7222           Free           =>  OPENSSL_free
7223      [Richard Levitte]
7224 
7225   *) New function BN_mod_exp_mont_word for small bases (roughly 15%
7226      faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
7227      [Bodo Moeller]
7228 
7229   *) CygWin32 support.
7230      [John Jarvie <jjarvie@newsguy.com>]
7231 
7232   *) The type-safe stack code has been rejigged. It is now only compiled
7233      in when OpenSSL is configured with the DEBUG_SAFESTACK option and
7234      by default all type-specific stack functions are "#define"d back to
7235      standard stack functions. This results in more streamlined output
7236      but retains the type-safety checking possibilities of the original
7237      approach.
7238      [Geoff Thorpe]
7239 
7240   *) The STACK code has been cleaned up, and certain type declarations
7241      that didn't make a lot of sense have been brought in line. This has
7242      also involved a cleanup of sorts in safestack.h to more correctly
7243      map type-safe stack functions onto their plain stack counterparts.
7244      This work has also resulted in a variety of "const"ifications of
7245      lots of the code, especially "_cmp" operations which should normally
7246      be prototyped with "const" parameters anyway.
7247      [Geoff Thorpe]
7248 
7249   *) When generating bytes for the first time in md_rand.c, 'stir the pool'
7250      by seeding with STATE_SIZE dummy bytes (with zero entropy count).
7251      (The PRNG state consists of two parts, the large pool 'state' and 'md',
7252      where all of 'md' is used each time the PRNG is used, but 'state'
7253      is used only indexed by a cyclic counter. As entropy may not be
7254      well distributed from the beginning, 'md' is important as a
7255      chaining variable. However, the output function chains only half
7256      of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
7257      all of 'md', and seeding with STATE_SIZE dummy bytes will result
7258      in all of 'state' being rewritten, with the new values depending
7259      on virtually all of 'md'.  This overcomes the 80 bit limitation.)
7260      [Bodo Moeller]
7261 
7262   *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
7263      the handshake is continued after ssl_verify_cert_chain();
7264      otherwise, if SSL_VERIFY_NONE is set, remaining error codes
7265      can lead to 'unexplainable' connection aborts later.
7266      [Bodo Moeller; problem tracked down by Lutz Jaenicke]
7267 
7268   *) Major EVP API cipher revision.
7269      Add hooks for extra EVP features. This allows various cipher
7270      parameters to be set in the EVP interface. Support added for variable
7271      key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
7272      setting of RC2 and RC5 parameters.
7273 
7274      Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
7275      ciphers.
7276 
7277      Remove lots of duplicated code from the EVP library. For example *every*
7278      cipher init() function handles the 'iv' in the same way according to the
7279      cipher mode. They also all do nothing if the 'key' parameter is NULL and
7280      for CFB and OFB modes they zero ctx->num.
7281 
7282      New functionality allows removal of S/MIME code RC2 hack.
7283 
7284      Most of the routines have the same form and so can be declared in terms
7285      of macros.
7286 
7287      By shifting this to the top level EVP_CipherInit() it can be removed from
7288      all individual ciphers. If the cipher wants to handle IVs or keys
7289      differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
7290      flags.
7291 
7292      Change lots of functions like EVP_EncryptUpdate() to now return a
7293      value: although software versions of the algorithms cannot fail
7294      any installed hardware versions can.
7295      [Steve Henson]
7296 
7297   *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
7298      this option is set, tolerate broken clients that send the negotiated
7299      protocol version number instead of the requested protocol version
7300      number.
7301      [Bodo Moeller]
7302 
7303   *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
7304      i.e. non-zero for export ciphersuites, zero otherwise.
7305      Previous versions had this flag inverted, inconsistent with
7306      rsa_tmp_cb (..._TMP_RSA_CB).
7307      [Bodo Moeller; problem reported by Amit Chopra]
7308 
7309   *) Add missing DSA library text string. Work around for some IIS
7310      key files with invalid SEQUENCE encoding.
7311      [Steve Henson]
7312 
7313   *) Add a document (doc/standards.txt) that list all kinds of standards
7314      and so on that are implemented in OpenSSL.
7315      [Richard Levitte]
7316 
7317   *) Enhance c_rehash script. Old version would mishandle certificates
7318      with the same subject name hash and wouldn't handle CRLs at all.
7319      Added -fingerprint option to crl utility, to support new c_rehash
7320      features.
7321      [Steve Henson]
7322 
7323   *) Eliminate non-ANSI declarations in crypto.h and stack.h.
7324      [Ulf Möller]
7325 
7326   *) Fix for SSL server purpose checking. Server checking was
7327      rejecting certificates which had extended key usage present
7328      but no ssl client purpose.
7329      [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
7330 
7331   *) Make PKCS#12 code work with no password. The PKCS#12 spec
7332      is a little unclear about how a blank password is handled.
7333      Since the password in encoded as a BMPString with terminating
7334      double NULL a zero length password would end up as just the
7335      double NULL. However no password at all is different and is
7336      handled differently in the PKCS#12 key generation code. NS
7337      treats a blank password as zero length. MSIE treats it as no
7338      password on export: but it will try both on import. We now do
7339      the same: PKCS12_parse() tries zero length and no password if
7340      the password is set to "" or NULL (NULL is now a valid password:
7341      it wasn't before) as does the pkcs12 application.
7342      [Steve Henson]
7343 
7344   *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
7345      perror when PEM_read_bio_X509_REQ fails, the error message must
7346      be obtained from the error queue.
7347      [Bodo Moeller]
7348 
7349   *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
7350      it in ERR_remove_state if appropriate, and change ERR_get_state
7351      accordingly to avoid race conditions (this is necessary because
7352      thread_hash is no longer constant once set).
7353      [Bodo Moeller]
7354 
7355   *) Bugfix for linux-elf makefile.one.
7356      [Ulf Möller]
7357 
7358   *) RSA_get_default_method() will now cause a default
7359      RSA_METHOD to be chosen if one doesn't exist already.
7360      Previously this was only set during a call to RSA_new()
7361      or RSA_new_method(NULL) meaning it was possible for
7362      RSA_get_default_method() to return NULL.
7363      [Geoff Thorpe]
7364 
7365   *) Added native name translation to the existing DSO code
7366      that will convert (if the flag to do so is set) filenames
7367      that are sufficiently small and have no path information
7368      into a canonical native form. Eg. "blah" converted to
7369      "libblah.so" or "blah.dll" etc.
7370      [Geoff Thorpe]
7371 
7372   *) New function ERR_error_string_n(e, buf, len) which is like
7373      ERR_error_string(e, buf), but writes at most 'len' bytes
7374      including the 0 terminator.  For ERR_error_string_n, 'buf'
7375      may not be NULL.
7376      [Damien Miller <djm@mindrot.org>, Bodo Moeller]
7377 
7378   *) CONF library reworked to become more general.  A new CONF
7379      configuration file reader "class" is implemented as well as a
7380      new functions (NCONF_*, for "New CONF") to handle it.  The now
7381      old CONF_* functions are still there, but are reimplemented to
7382      work in terms of the new functions.  Also, a set of functions
7383      to handle the internal storage of the configuration data is
7384      provided to make it easier to write new configuration file
7385      reader "classes" (I can definitely see something reading a
7386      configuration file in XML format, for example), called _CONF_*,
7387      or "the configuration storage API"...
7388 
7389      The new configuration file reading functions are:
7390 
7391         NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
7392         NCONF_get_section, NCONF_get_string, NCONF_get_numbre
7393 
7394         NCONF_default, NCONF_WIN32
7395 
7396         NCONF_dump_fp, NCONF_dump_bio
7397 
7398      NCONF_default and NCONF_WIN32 are method (or "class") choosers,
7399      NCONF_new creates a new CONF object.  This works in the same way
7400      as other interfaces in OpenSSL, like the BIO interface.
7401      NCONF_dump_* dump the internal storage of the configuration file,
7402      which is useful for debugging.  All other functions take the same
7403      arguments as the old CONF_* functions wth the exception of the
7404      first that must be a `CONF *' instead of a `LHASH *'.
7405 
7406      To make it easer to use the new classes with the old CONF_* functions,
7407      the function CONF_set_default_method is provided.
7408      [Richard Levitte]
7409 
7410   *) Add '-tls1' option to 'openssl ciphers', which was already
7411      mentioned in the documentation but had not been implemented.
7412      (This option is not yet really useful because even the additional
7413      experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
7414      [Bodo Moeller]
7415 
7416   *) Initial DSO code added into libcrypto for letting OpenSSL (and
7417      OpenSSL-based applications) load shared libraries and bind to
7418      them in a portable way.
7419      [Geoff Thorpe, with contributions from Richard Levitte]
7420 
7421  Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
7422 
7423   *) Make sure _lrotl and _lrotr are only used with MSVC.
7424 
7425   *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
7426      (the default implementation of RAND_status).
7427 
7428   *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
7429      to '-clrext' (= clear extensions), as intended and documented.
7430      [Bodo Moeller; inconsistency pointed out by Michael Attili
7431      <attili@amaxo.com>]
7432 
7433   *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
7434      was larger than the MD block size.      
7435      [Steve Henson, pointed out by Yost William <YostW@tce.com>]
7436 
7437   *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
7438      fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
7439      using the passed key: if the passed key was a private key the result
7440      of X509_print(), for example, would be to print out all the private key
7441      components.
7442      [Steve Henson]
7443 
7444   *) des_quad_cksum() byte order bug fix.
7445      [Ulf Möller, using the problem description in krb4-0.9.7, where
7446       the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
7447 
7448   *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
7449      discouraged.
7450      [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
7451 
7452   *) For easily testing in shell scripts whether some command
7453      'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
7454      returns with exit code 0 iff no command of the given name is available.
7455      'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
7456      the output goes to stdout and nothing is printed to stderr.
7457      Additional arguments are always ignored.
7458 
7459      Since for each cipher there is a command of the same name,
7460      the 'no-cipher' compilation switches can be tested this way.
7461 
7462      ('openssl no-XXX' is not able to detect pseudo-commands such
7463      as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
7464      [Bodo Moeller]
7465 
7466   *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
7467      [Bodo Moeller]
7468 
7469   *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
7470      is set; it will be thrown away anyway because each handshake creates
7471      its own key.
7472      ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
7473      to parameters -- in previous versions (since OpenSSL 0.9.3) the
7474      'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
7475      you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
7476      [Bodo Moeller]
7477 
7478   *) New s_client option -ign_eof: EOF at stdin is ignored, and
7479      'Q' and 'R' lose their special meanings (quit/renegotiate).
7480      This is part of what -quiet does; unlike -quiet, -ign_eof
7481      does not suppress any output.
7482      [Richard Levitte]
7483 
7484   *) Add compatibility options to the purpose and trust code. The
7485      purpose X509_PURPOSE_ANY is "any purpose" which automatically
7486      accepts a certificate or CA, this was the previous behaviour,
7487      with all the associated security issues.
7488 
7489      X509_TRUST_COMPAT is the old trust behaviour: only and
7490      automatically trust self signed roots in certificate store. A
7491      new trust setting X509_TRUST_DEFAULT is used to specify that
7492      a purpose has no associated trust setting and it should instead
7493      use the value in the default purpose.
7494      [Steve Henson]
7495 
7496   *) Fix the PKCS#8 DSA private key code so it decodes keys again
7497      and fix a memory leak.
7498      [Steve Henson]
7499 
7500   *) In util/mkerr.pl (which implements 'make errors'), preserve
7501      reason strings from the previous version of the .c file, as
7502      the default to have only downcase letters (and digits) in
7503      automatically generated reasons codes is not always appropriate.
7504      [Bodo Moeller]
7505 
7506   *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
7507      using strerror.  Previously, ERR_reason_error_string() returned
7508      library names as reason strings for SYSerr; but SYSerr is a special
7509      case where small numbers are errno values, not library numbers.
7510      [Bodo Moeller]
7511 
7512   *) Add '-dsaparam' option to 'openssl dhparam' application.  This
7513      converts DSA parameters into DH parameters. (When creating parameters,
7514      DSA_generate_parameters is used.)
7515      [Bodo Moeller]
7516 
7517   *) Include 'length' (recommended exponent length) in C code generated
7518      by 'openssl dhparam -C'.
7519      [Bodo Moeller]
7520 
7521   *) The second argument to set_label in perlasm was already being used
7522      so couldn't be used as a "file scope" flag. Moved to third argument
7523      which was free.
7524      [Steve Henson]
7525 
7526   *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
7527      instead of RAND_bytes for encryption IVs and salts.
7528      [Bodo Moeller]
7529 
7530   *) Include RAND_status() into RAND_METHOD instead of implementing
7531      it only for md_rand.c  Otherwise replacing the PRNG by calling
7532      RAND_set_rand_method would be impossible.
7533      [Bodo Moeller]
7534 
7535   *) Don't let DSA_generate_key() enter an infinite loop if the random
7536      number generation fails.
7537      [Bodo Moeller]
7538 
7539   *) New 'rand' application for creating pseudo-random output.
7540      [Bodo Moeller]
7541 
7542   *) Added configuration support for Linux/IA64
7543      [Rolf Haberrecker <rolf@suse.de>]
7544 
7545   *) Assembler module support for Mingw32.
7546      [Ulf Möller]
7547 
7548   *) Shared library support for HPUX (in shlib/).
7549      [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
7550 
7551   *) Shared library support for Solaris gcc.
7552      [Lutz Behnke <behnke@trustcenter.de>]
7553 
7554  Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
7555 
7556   *) PKCS7_encrypt() was adding text MIME headers twice because they
7557      were added manually and by SMIME_crlf_copy().
7558      [Steve Henson]
7559 
7560   *) In bntest.c don't call BN_rand with zero bits argument.
7561      [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
7562 
7563   *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
7564      case was implemented. This caused BN_div_recp() to fail occasionally.
7565      [Ulf Möller]
7566 
7567   *) Add an optional second argument to the set_label() in the perl
7568      assembly language builder. If this argument exists and is set
7569      to 1 it signals that the assembler should use a symbol whose 
7570      scope is the entire file, not just the current function. This
7571      is needed with MASM which uses the format label:: for this scope.
7572      [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
7573 
7574   *) Change the ASN1 types so they are typedefs by default. Before
7575      almost all types were #define'd to ASN1_STRING which was causing
7576      STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
7577      for example.
7578      [Steve Henson]
7579 
7580   *) Change names of new functions to the new get1/get0 naming
7581      convention: After 'get1', the caller owns a reference count
7582      and has to call ..._free; 'get0' returns a pointer to some
7583      data structure without incrementing reference counters.
7584      (Some of the existing 'get' functions increment a reference
7585      counter, some don't.)
7586      Similarly, 'set1' and 'add1' functions increase reference
7587      counters or duplicate objects.
7588      [Steve Henson]
7589 
7590   *) Allow for the possibility of temp RSA key generation failure:
7591      the code used to assume it always worked and crashed on failure.
7592      [Steve Henson]
7593 
7594   *) Fix potential buffer overrun problem in BIO_printf().
7595      [Ulf Möller, using public domain code by Patrick Powell; problem
7596       pointed out by David Sacerdote <das33@cornell.edu>]
7597 
7598   *) Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
7599      RAND_egd() and RAND_status().  In the command line application,
7600      the EGD socket can be specified like a seed file using RANDFILE
7601      or -rand.
7602      [Ulf Möller]
7603 
7604   *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
7605      Some CAs (e.g. Verisign) distribute certificates in this form.
7606      [Steve Henson]
7607 
7608   *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
7609      list to exclude them. This means that no special compilation option
7610      is needed to use anonymous DH: it just needs to be included in the
7611      cipher list.
7612      [Steve Henson]
7613 
7614   *) Change the EVP_MD_CTX_type macro so its meaning consistent with
7615      EVP_MD_type. The old functionality is available in a new macro called
7616      EVP_MD_md(). Change code that uses it and update docs.
7617      [Steve Henson]
7618 
7619   *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
7620      where the 'void *' argument is replaced by a function pointer argument.
7621      Previously 'void *' was abused to point to functions, which works on
7622      many platforms, but is not correct.  As these functions are usually
7623      called by macros defined in OpenSSL header files, most source code
7624      should work without changes.
7625      [Richard Levitte]
7626 
7627   *) <openssl/opensslconf.h> (which is created by Configure) now contains
7628      sections with information on -D... compiler switches used for
7629      compiling the library so that applications can see them.  To enable
7630      one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
7631      must be defined.  E.g.,
7632         #define OPENSSL_ALGORITHM_DEFINES
7633         #include <openssl/opensslconf.h>
7634      defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
7635      [Richard Levitte, Ulf and Bodo Möller]
7636 
7637   *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
7638      record layer.
7639      [Bodo Moeller]
7640 
7641   *) Change the 'other' type in certificate aux info to a STACK_OF
7642      X509_ALGOR. Although not an AlgorithmIdentifier as such it has
7643      the required ASN1 format: arbitrary types determined by an OID.
7644      [Steve Henson]
7645 
7646   *) Add some PEM_write_X509_REQ_NEW() functions and a command line
7647      argument to 'req'. This is not because the function is newer or
7648      better than others it just uses the work 'NEW' in the certificate
7649      request header lines. Some software needs this.
7650      [Steve Henson]
7651 
7652   *) Reorganise password command line arguments: now passwords can be
7653      obtained from various sources. Delete the PEM_cb function and make
7654      it the default behaviour: i.e. if the callback is NULL and the
7655      usrdata argument is not NULL interpret it as a null terminated pass
7656      phrase. If usrdata and the callback are NULL then the pass phrase
7657      is prompted for as usual.
7658      [Steve Henson]
7659 
7660   *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
7661      the support is automatically enabled. The resulting binaries will
7662      autodetect the card and use it if present.
7663      [Ben Laurie and Compaq Inc.]
7664 
7665   *) Work around for Netscape hang bug. This sends certificate request
7666      and server done in one record. Since this is perfectly legal in the
7667      SSL/TLS protocol it isn't a "bug" option and is on by default. See
7668      the bugs/SSLv3 entry for more info.
7669      [Steve Henson]
7670 
7671   *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
7672      [Andy Polyakov]
7673 
7674   *) Add -rand argument to smime and pkcs12 applications and read/write
7675      of seed file.
7676      [Steve Henson]
7677 
7678   *) New 'passwd' tool for crypt(3) and apr1 password hashes.
7679      [Bodo Moeller]
7680 
7681   *) Add command line password options to the remaining applications.
7682      [Steve Henson]
7683 
7684   *) Bug fix for BN_div_recp() for numerators with an even number of
7685      bits.
7686      [Ulf Möller]
7687 
7688   *) More tests in bntest.c, and changed test_bn output.
7689      [Ulf Möller]
7690 
7691   *) ./config recognizes MacOS X now.
7692      [Andy Polyakov]
7693 
7694   *) Bug fix for BN_div() when the first words of num and divsor are
7695      equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
7696      [Ulf Möller]
7697 
7698   *) Add support for various broken PKCS#8 formats, and command line
7699      options to produce them.
7700      [Steve Henson]
7701 
7702   *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
7703      get temporary BIGNUMs from a BN_CTX.
7704      [Ulf Möller]
7705 
7706   *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
7707      for p == 0.
7708      [Ulf Möller]
7709 
7710   *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
7711      include a #define from the old name to the new. The original intent
7712      was that statically linked binaries could for example just call
7713      SSLeay_add_all_ciphers() to just add ciphers to the table and not
7714      link with digests. This never worked becayse SSLeay_add_all_digests()
7715      and SSLeay_add_all_ciphers() were in the same source file so calling
7716      one would link with the other. They are now in separate source files.
7717      [Steve Henson]
7718 
7719   *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
7720      [Steve Henson]
7721 
7722   *) Use a less unusual form of the Miller-Rabin primality test (it used
7723      a binary algorithm for exponentiation integrated into the Miller-Rabin
7724      loop, our standard modexp algorithms are faster).
7725      [Bodo Moeller]
7726 
7727   *) Support for the EBCDIC character set completed.
7728      [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
7729 
7730   *) Source code cleanups: use const where appropriate, eliminate casts,
7731      use void * instead of char * in lhash.
7732      [Ulf Möller] 
7733 
7734   *) Bugfix: ssl3_send_server_key_exchange was not restartable
7735      (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
7736      this the server could overwrite ephemeral keys that the client
7737      has already seen).
7738      [Bodo Moeller]
7739 
7740   *) Turn DSA_is_prime into a macro that calls BN_is_prime,
7741      using 50 iterations of the Rabin-Miller test.
7742 
7743      DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
7744      iterations of the Rabin-Miller test as required by the appendix
7745      to FIPS PUB 186[-1]) instead of DSA_is_prime.
7746      As BN_is_prime_fasttest includes trial division, DSA parameter
7747      generation becomes much faster.
7748 
7749      This implies a change for the callback functions in DSA_is_prime
7750      and DSA_generate_parameters: The callback function is called once
7751      for each positive witness in the Rabin-Miller test, not just
7752      occasionally in the inner loop; and the parameters to the
7753      callback function now provide an iteration count for the outer
7754      loop rather than for the current invocation of the inner loop.
7755      DSA_generate_parameters additionally can call the callback
7756      function with an 'iteration count' of -1, meaning that a
7757      candidate has passed the trial division test (when q is generated 
7758      from an application-provided seed, trial division is skipped).
7759      [Bodo Moeller]
7760 
7761   *) New function BN_is_prime_fasttest that optionally does trial
7762      division before starting the Rabin-Miller test and has
7763      an additional BN_CTX * argument (whereas BN_is_prime always
7764      has to allocate at least one BN_CTX).
7765      'callback(1, -1, cb_arg)' is called when a number has passed the
7766      trial division stage.
7767      [Bodo Moeller]
7768 
7769   *) Fix for bug in CRL encoding. The validity dates weren't being handled
7770      as ASN1_TIME.
7771      [Steve Henson]
7772 
7773   *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
7774      [Steve Henson]
7775 
7776   *) New function BN_pseudo_rand().
7777      [Ulf Möller]
7778 
7779   *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
7780      bignum version of BN_from_montgomery() with the working code from
7781      SSLeay 0.9.0 (the word based version is faster anyway), and clean up
7782      the comments.
7783      [Ulf Möller]
7784 
7785   *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
7786      made it impossible to use the same SSL_SESSION data structure in
7787      SSL2 clients in multiple threads.
7788      [Bodo Moeller]
7789 
7790   *) The return value of RAND_load_file() no longer counts bytes obtained
7791      by stat().  RAND_load_file(..., -1) is new and uses the complete file
7792      to seed the PRNG (previously an explicit byte count was required).
7793      [Ulf Möller, Bodo Möller]
7794 
7795   *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
7796      used (char *) instead of (void *) and had casts all over the place.
7797      [Steve Henson]
7798 
7799   *) Make BN_generate_prime() return NULL on error if ret!=NULL.
7800      [Ulf Möller]
7801 
7802   *) Retain source code compatibility for BN_prime_checks macro:
7803      BN_is_prime(..., BN_prime_checks, ...) now uses
7804      BN_prime_checks_for_size to determine the appropriate number of
7805      Rabin-Miller iterations.
7806      [Ulf Möller]
7807 
7808   *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
7809      DH_CHECK_P_NOT_SAFE_PRIME.
7810      (Check if this is true? OpenPGP calls them "strong".)
7811      [Ulf Möller]
7812 
7813   *) Merge the functionality of "dh" and "gendh" programs into a new program
7814      "dhparam". The old programs are retained for now but will handle DH keys
7815      (instead of parameters) in future.
7816      [Steve Henson]
7817 
7818   *) Make the ciphers, s_server and s_client programs check the return values
7819      when a new cipher list is set.
7820      [Steve Henson]
7821 
7822   *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
7823      ciphers. Before when the 56bit ciphers were enabled the sorting was
7824      wrong.
7825 
7826      The syntax for the cipher sorting has been extended to support sorting by
7827      cipher-strength (using the strength_bits hard coded in the tables).
7828      The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
7829 
7830      Fix a bug in the cipher-command parser: when supplying a cipher command
7831      string with an "undefined" symbol (neither command nor alphanumeric
7832      [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
7833      an error is flagged.
7834 
7835      Due to the strength-sorting extension, the code of the
7836      ssl_create_cipher_list() function was completely rearranged. I hope that
7837      the readability was also increased :-)
7838      [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
7839 
7840   *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
7841      for the first serial number and places 2 in the serial number file. This
7842      avoids problems when the root CA is created with serial number zero and
7843      the first user certificate has the same issuer name and serial number
7844      as the root CA.
7845      [Steve Henson]
7846 
7847   *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
7848      the new code. Add documentation for this stuff.
7849      [Steve Henson]
7850 
7851   *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
7852      X509_*() to X509at_*() on the grounds that they don't handle X509
7853      structures and behave in an analagous way to the X509v3 functions:
7854      they shouldn't be called directly but wrapper functions should be used
7855      instead.
7856 
7857      So we also now have some wrapper functions that call the X509at functions
7858      when passed certificate requests. (TO DO: similar things can be done with
7859      PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
7860      things. Some of these need some d2i or i2d and print functionality
7861      because they handle more complex structures.)
7862      [Steve Henson]
7863 
7864   *) Add missing #ifndefs that caused missing symbols when building libssl
7865      as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
7866      NO_RSA in ssl/s2*.c. 
7867      [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
7868 
7869   *) Precautions against using the PRNG uninitialized: RAND_bytes() now
7870      has a return value which indicates the quality of the random data
7871      (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
7872      error queue. New function RAND_pseudo_bytes() generates output that is
7873      guaranteed to be unique but not unpredictable. RAND_add is like
7874      RAND_seed, but takes an extra argument for an entropy estimate
7875      (RAND_seed always assumes full entropy).
7876      [Ulf Möller]
7877 
7878   *) Do more iterations of Rabin-Miller probable prime test (specifically,
7879      3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
7880      instead of only 2 for all lengths; see BN_prime_checks_for_size definition
7881      in crypto/bn/bn_prime.c for the complete table).  This guarantees a
7882      false-positive rate of at most 2^-80 for random input.
7883      [Bodo Moeller]
7884 
7885   *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
7886      [Bodo Moeller]
7887 
7888   *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
7889      in the 0.9.5 release), this returns the chain
7890      from an X509_CTX structure with a dup of the stack and all
7891      the X509 reference counts upped: so the stack will exist
7892      after X509_CTX_cleanup() has been called. Modify pkcs12.c
7893      to use this.
7894 
7895      Also make SSL_SESSION_print() print out the verify return
7896      code.
7897      [Steve Henson]
7898 
7899   *) Add manpage for the pkcs12 command. Also change the default
7900      behaviour so MAC iteration counts are used unless the new
7901      -nomaciter option is used. This improves file security and
7902      only older versions of MSIE (4.0 for example) need it.
7903      [Steve Henson]
7904 
7905   *) Honor the no-xxx Configure options when creating .DEF files.
7906      [Ulf Möller]
7907 
7908   *) Add PKCS#10 attributes to field table: challengePassword, 
7909      unstructuredName and unstructuredAddress. These are taken from
7910      draft PKCS#9 v2.0 but are compatible with v1.2 provided no 
7911      international characters are used.
7912 
7913      More changes to X509_ATTRIBUTE code: allow the setting of types
7914      based on strings. Remove the 'loc' parameter when adding
7915      attributes because these will be a SET OF encoding which is sorted
7916      in ASN1 order.
7917      [Steve Henson]
7918 
7919   *) Initial changes to the 'req' utility to allow request generation
7920      automation. This will allow an application to just generate a template
7921      file containing all the field values and have req construct the
7922      request.
7923 
7924      Initial support for X509_ATTRIBUTE handling. Stacks of these are
7925      used all over the place including certificate requests and PKCS#7
7926      structures. They are currently handled manually where necessary with
7927      some primitive wrappers for PKCS#7. The new functions behave in a
7928      manner analogous to the X509 extension functions: they allow
7929      attributes to be looked up by NID and added.
7930 
7931      Later something similar to the X509V3 code would be desirable to
7932      automatically handle the encoding, decoding and printing of the
7933      more complex types. The string types like challengePassword can
7934      be handled by the string table functions.
7935 
7936      Also modified the multi byte string table handling. Now there is
7937      a 'global mask' which masks out certain types. The table itself
7938      can use the flag STABLE_NO_MASK to ignore the mask setting: this
7939      is useful when for example there is only one permissible type
7940      (as in countryName) and using the mask might result in no valid
7941      types at all.
7942      [Steve Henson]
7943 
7944   *) Clean up 'Finished' handling, and add functions SSL_get_finished and
7945      SSL_get_peer_finished to allow applications to obtain the latest
7946      Finished messages sent to the peer or expected from the peer,
7947      respectively.  (SSL_get_peer_finished is usually the Finished message
7948      actually received from the peer, otherwise the protocol will be aborted.)
7949 
7950      As the Finished message are message digests of the complete handshake
7951      (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
7952      be used for external authentication procedures when the authentication
7953      provided by SSL/TLS is not desired or is not enough.
7954      [Bodo Moeller]
7955 
7956   *) Enhanced support for Alpha Linux is added. Now ./config checks if
7957      the host supports BWX extension and if Compaq C is present on the
7958      $PATH. Just exploiting of the BWX extension results in 20-30%
7959      performance kick for some algorithms, e.g. DES and RC4 to mention
7960      a couple. Compaq C in turn generates ~20% faster code for MD5 and
7961      SHA1.
7962      [Andy Polyakov]
7963 
7964   *) Add support for MS "fast SGC". This is arguably a violation of the
7965      SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
7966      weak crypto and after checking the certificate is SGC a second one
7967      with strong crypto. MS SGC stops the first handshake after receiving
7968      the server certificate message and sends a second client hello. Since
7969      a server will typically do all the time consuming operations before
7970      expecting any further messages from the client (server key exchange
7971      is the most expensive) there is little difference between the two.
7972 
7973      To get OpenSSL to support MS SGC we have to permit a second client
7974      hello message after we have sent server done. In addition we have to
7975      reset the MAC if we do get this second client hello.
7976      [Steve Henson]
7977 
7978   *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
7979      if a DER encoded private key is RSA or DSA traditional format. Changed
7980      d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
7981      format DER encoded private key. Newer code should use PKCS#8 format which
7982      has the key type encoded in the ASN1 structure. Added DER private key
7983      support to pkcs8 application.
7984      [Steve Henson]
7985 
7986   *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
7987      ciphersuites has been selected (as required by the SSL 3/TLS 1
7988      specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
7989      is set, we interpret this as a request to violate the specification
7990      (the worst that can happen is a handshake failure, and 'correct'
7991      behaviour would result in a handshake failure anyway).
7992      [Bodo Moeller]
7993 
7994   *) In SSL_CTX_add_session, take into account that there might be multiple
7995      SSL_SESSION structures with the same session ID (e.g. when two threads
7996      concurrently obtain them from an external cache).
7997      The internal cache can handle only one SSL_SESSION with a given ID,
7998      so if there's a conflict, we now throw out the old one to achieve
7999      consistency.
8000      [Bodo Moeller]
8001 
8002   *) Add OIDs for idea and blowfish in CBC mode. This will allow both
8003      to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
8004      some routines that use cipher OIDs: some ciphers do not have OIDs
8005      defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
8006      example.
8007      [Steve Henson]
8008 
8009   *) Simplify the trust setting structure and code. Now we just have
8010      two sequences of OIDs for trusted and rejected settings. These will
8011      typically have values the same as the extended key usage extension
8012      and any application specific purposes.
8013 
8014      The trust checking code now has a default behaviour: it will just
8015      check for an object with the same NID as the passed id. Functions can
8016      be provided to override either the default behaviour or the behaviour
8017      for a given id. SSL client, server and email already have functions
8018      in place for compatibility: they check the NID and also return "trusted"
8019      if the certificate is self signed.
8020      [Steve Henson]
8021 
8022   *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
8023      traditional format into an EVP_PKEY structure.
8024      [Steve Henson]
8025 
8026   *) Add a password callback function PEM_cb() which either prompts for
8027      a password if usr_data is NULL or otherwise assumes it is a null
8028      terminated password. Allow passwords to be passed on command line
8029      environment or config files in a few more utilities.
8030      [Steve Henson]
8031 
8032   *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
8033      keys. Add some short names for PKCS#8 PBE algorithms and allow them
8034      to be specified on the command line for the pkcs8 and pkcs12 utilities.
8035      Update documentation.
8036      [Steve Henson]
8037 
8038   *) Support for ASN1 "NULL" type. This could be handled before by using
8039      ASN1_TYPE but there wasn't any function that would try to read a NULL
8040      and produce an error if it couldn't. For compatibility we also have
8041      ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
8042      don't allocate anything because they don't need to.
8043      [Steve Henson]
8044 
8045   *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
8046      for details.
8047      [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
8048 
8049   *) Rebuild of the memory allocation routines used by OpenSSL code and
8050      possibly others as well.  The purpose is to make an interface that
8051      provide hooks so anyone can build a separate set of allocation and
8052      deallocation routines to be used by OpenSSL, for example memory
8053      pool implementations, or something else, which was previously hard
8054      since Malloc(), Realloc() and Free() were defined as macros having
8055      the values malloc, realloc and free, respectively (except for Win32
8056      compilations).  The same is provided for memory debugging code.
8057      OpenSSL already comes with functionality to find memory leaks, but
8058      this gives people a chance to debug other memory problems.
8059 
8060      With these changes, a new set of functions and macros have appeared:
8061 
8062        CRYPTO_set_mem_debug_functions()         [F]
8063        CRYPTO_get_mem_debug_functions()         [F]
8064        CRYPTO_dbg_set_options()                 [F]
8065        CRYPTO_dbg_get_options()                 [F]
8066        CRYPTO_malloc_debug_init()               [M]
8067 
8068      The memory debug functions are NULL by default, unless the library
8069      is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
8070      wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
8071      gives the standard debugging functions that come with OpenSSL) or
8072      CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
8073      provided by the library user) must be used.  When the standard
8074      debugging functions are used, CRYPTO_dbg_set_options can be used to
8075      request additional information:
8076      CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
8077      the CRYPTO_MDEBUG_xxx macro when compiling the library.   
8078 
8079      Also, things like CRYPTO_set_mem_functions will always give the
8080      expected result (the new set of functions is used for allocation
8081      and deallocation) at all times, regardless of platform and compiler
8082      options.
8083 
8084      To finish it up, some functions that were never use in any other
8085      way than through macros have a new API and new semantic:
8086 
8087        CRYPTO_dbg_malloc()
8088        CRYPTO_dbg_realloc()
8089        CRYPTO_dbg_free()
8090 
8091      All macros of value have retained their old syntax.
8092      [Richard Levitte and Bodo Moeller]
8093 
8094   *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
8095      ordering of SMIMECapabilities wasn't in "strength order" and there
8096      was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
8097      algorithm.
8098      [Steve Henson]
8099 
8100   *) Some ASN1 types with illegal zero length encoding (INTEGER,
8101      ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
8102      [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
8103 
8104   *) Merge in my S/MIME library for OpenSSL. This provides a simple
8105      S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
8106      functionality to handle multipart/signed properly) and a utility
8107      called 'smime' to call all this stuff. This is based on code I
8108      originally wrote for Celo who have kindly allowed it to be
8109      included in OpenSSL.
8110      [Steve Henson]
8111 
8112   *) Add variants des_set_key_checked and des_set_key_unchecked of
8113      des_set_key (aka des_key_sched).  Global variable des_check_key
8114      decides which of these is called by des_set_key; this way
8115      des_check_key behaves as it always did, but applications and
8116      the library itself, which was buggy for des_check_key == 1,
8117      have a cleaner way to pick the version they need.
8118      [Bodo Moeller]
8119 
8120   *) New function PKCS12_newpass() which changes the password of a
8121      PKCS12 structure.
8122      [Steve Henson]
8123 
8124   *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
8125      dynamic mix. In both cases the ids can be used as an index into the
8126      table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
8127      functions so they accept a list of the field values and the
8128      application doesn't need to directly manipulate the X509_TRUST
8129      structure.
8130      [Steve Henson]
8131 
8132   *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
8133      need initialising.
8134      [Steve Henson]
8135 
8136   *) Modify the way the V3 extension code looks up extensions. This now
8137      works in a similar way to the object code: we have some "standard"
8138      extensions in a static table which is searched with OBJ_bsearch()
8139      and the application can add dynamic ones if needed. The file
8140      crypto/x509v3/ext_dat.h now has the info: this file needs to be
8141      updated whenever a new extension is added to the core code and kept
8142      in ext_nid order. There is a simple program 'tabtest.c' which checks
8143      this. New extensions are not added too often so this file can readily
8144      be maintained manually.
8145 
8146      There are two big advantages in doing things this way. The extensions
8147      can be looked up immediately and no longer need to be "added" using
8148      X509V3_add_standard_extensions(): this function now does nothing.
8149      [Side note: I get *lots* of email saying the extension code doesn't
8150       work because people forget to call this function]
8151      Also no dynamic allocation is done unless new extensions are added:
8152      so if we don't add custom extensions there is no need to call
8153      X509V3_EXT_cleanup().
8154      [Steve Henson]
8155 
8156   *) Modify enc utility's salting as follows: make salting the default. Add a
8157      magic header, so unsalted files fail gracefully instead of just decrypting
8158      to garbage. This is because not salting is a big security hole, so people
8159      should be discouraged from doing it.
8160      [Ben Laurie]
8161 
8162   *) Fixes and enhancements to the 'x509' utility. It allowed a message
8163      digest to be passed on the command line but it only used this
8164      parameter when signing a certificate. Modified so all relevant
8165      operations are affected by the digest parameter including the
8166      -fingerprint and -x509toreq options. Also -x509toreq choked if a
8167      DSA key was used because it didn't fix the digest.
8168      [Steve Henson]
8169 
8170   *) Initial certificate chain verify code. Currently tests the untrusted
8171      certificates for consistency with the verify purpose (which is set
8172      when the X509_STORE_CTX structure is set up) and checks the pathlength.
8173 
8174      There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
8175      this is because it will reject chains with invalid extensions whereas
8176      every previous version of OpenSSL and SSLeay made no checks at all.
8177 
8178      Trust code: checks the root CA for the relevant trust settings. Trust
8179      settings have an initial value consistent with the verify purpose: e.g.
8180      if the verify purpose is for SSL client use it expects the CA to be
8181      trusted for SSL client use. However the default value can be changed to
8182      permit custom trust settings: one example of this would be to only trust
8183      certificates from a specific "secure" set of CAs.
8184 
8185      Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
8186      which should be used for version portability: especially since the
8187      verify structure is likely to change more often now.
8188 
8189      SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
8190      to set them. If not set then assume SSL clients will verify SSL servers
8191      and vice versa.
8192 
8193      Two new options to the verify program: -untrusted allows a set of
8194      untrusted certificates to be passed in and -purpose which sets the
8195      intended purpose of the certificate. If a purpose is set then the
8196      new chain verify code is used to check extension consistency.
8197      [Steve Henson]
8198 
8199   *) Support for the authority information access extension.
8200      [Steve Henson]
8201 
8202   *) Modify RSA and DSA PEM read routines to transparently handle
8203      PKCS#8 format private keys. New *_PUBKEY_* functions that handle
8204      public keys in a format compatible with certificate
8205      SubjectPublicKeyInfo structures. Unfortunately there were already
8206      functions called *_PublicKey_* which used various odd formats so
8207      these are retained for compatibility: however the DSA variants were
8208      never in a public release so they have been deleted. Changed dsa/rsa
8209      utilities to handle the new format: note no releases ever handled public
8210      keys so we should be OK.
8211 
8212      The primary motivation for this change is to avoid the same fiasco
8213      that dogs private keys: there are several incompatible private key
8214      formats some of which are standard and some OpenSSL specific and
8215      require various evil hacks to allow partial transparent handling and
8216      even then it doesn't work with DER formats. Given the option anything
8217      other than PKCS#8 should be dumped: but the other formats have to
8218      stay in the name of compatibility.
8219 
8220      With public keys and the benefit of hindsight one standard format 
8221      is used which works with EVP_PKEY, RSA or DSA structures: though
8222      it clearly returns an error if you try to read the wrong kind of key.
8223 
8224      Added a -pubkey option to the 'x509' utility to output the public key.
8225      Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
8226      (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
8227      EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
8228      that do the same as the EVP_PKEY_assign_*() except they up the
8229      reference count of the added key (they don't "swallow" the
8230      supplied key).
8231      [Steve Henson]
8232 
8233   *) Fixes to crypto/x509/by_file.c the code to read in certificates and
8234      CRLs would fail if the file contained no certificates or no CRLs:
8235      added a new function to read in both types and return the number
8236      read: this means that if none are read it will be an error. The
8237      DER versions of the certificate and CRL reader would always fail
8238      because it isn't possible to mix certificates and CRLs in DER format
8239      without choking one or the other routine. Changed this to just read
8240      a certificate: this is the best we can do. Also modified the code
8241      in apps/verify.c to take notice of return codes: it was previously
8242      attempting to read in certificates from NULL pointers and ignoring
8243      any errors: this is one reason why the cert and CRL reader seemed
8244      to work. It doesn't check return codes from the default certificate
8245      routines: these may well fail if the certificates aren't installed.
8246      [Steve Henson]
8247 
8248   *) Code to support otherName option in GeneralName.
8249      [Steve Henson]
8250 
8251   *) First update to verify code. Change the verify utility
8252      so it warns if it is passed a self signed certificate:
8253      for consistency with the normal behaviour. X509_verify
8254      has been modified to it will now verify a self signed
8255      certificate if *exactly* the same certificate appears
8256      in the store: it was previously impossible to trust a
8257      single self signed certificate. This means that:
8258      openssl verify ss.pem
8259      now gives a warning about a self signed certificate but
8260      openssl verify -CAfile ss.pem ss.pem
8261      is OK.
8262      [Steve Henson]
8263 
8264   *) For servers, store verify_result in SSL_SESSION data structure
8265      (and add it to external session representation).
8266      This is needed when client certificate verifications fails,
8267      but an application-provided verification callback (set by
8268      SSL_CTX_set_cert_verify_callback) allows accepting the session
8269      anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
8270      but returns 1): When the session is reused, we have to set
8271      ssl->verify_result to the appropriate error code to avoid
8272      security holes.
8273      [Bodo Moeller, problem pointed out by Lutz Jaenicke]
8274 
8275   *) Fix a bug in the new PKCS#7 code: it didn't consider the
8276      case in PKCS7_dataInit() where the signed PKCS7 structure
8277      didn't contain any existing data because it was being created.
8278      [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
8279 
8280   *) Add a salt to the key derivation routines in enc.c. This
8281      forms the first 8 bytes of the encrypted file. Also add a
8282      -S option to allow a salt to be input on the command line.
8283      [Steve Henson]
8284 
8285   *) New function X509_cmp(). Oddly enough there wasn't a function
8286      to compare two certificates. We do this by working out the SHA1
8287      hash and comparing that. X509_cmp() will be needed by the trust
8288      code.
8289      [Steve Henson]
8290 
8291   *) SSL_get1_session() is like SSL_get_session(), but increments
8292      the reference count in the SSL_SESSION returned.
8293      [Geoff Thorpe <geoff@eu.c2.net>]
8294 
8295   *) Fix for 'req': it was adding a null to request attributes.
8296      Also change the X509_LOOKUP and X509_INFO code to handle
8297      certificate auxiliary information.
8298      [Steve Henson]
8299 
8300   *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
8301      the 'enc' command.
8302      [Steve Henson]
8303 
8304   *) Add the possibility to add extra information to the memory leak
8305      detecting output, to form tracebacks, showing from where each
8306      allocation was originated: CRYPTO_push_info("constant string") adds
8307      the string plus current file name and line number to a per-thread
8308      stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
8309      is like calling CYRPTO_pop_info() until the stack is empty.
8310      Also updated memory leak detection code to be multi-thread-safe.
8311      [Richard Levitte]
8312 
8313   *) Add options -text and -noout to pkcs7 utility and delete the
8314      encryption options which never did anything. Update docs.
8315      [Steve Henson]
8316 
8317   *) Add options to some of the utilities to allow the pass phrase
8318      to be included on either the command line (not recommended on
8319      OSes like Unix) or read from the environment. Update the
8320      manpages and fix a few bugs.
8321      [Steve Henson]
8322 
8323   *) Add a few manpages for some of the openssl commands.
8324      [Steve Henson]
8325 
8326   *) Fix the -revoke option in ca. It was freeing up memory twice,
8327      leaking and not finding already revoked certificates.
8328      [Steve Henson]
8329 
8330   *) Extensive changes to support certificate auxiliary information.
8331      This involves the use of X509_CERT_AUX structure and X509_AUX
8332      functions. An X509_AUX function such as PEM_read_X509_AUX()
8333      can still read in a certificate file in the usual way but it
8334      will also read in any additional "auxiliary information". By
8335      doing things this way a fair degree of compatibility can be
8336      retained: existing certificates can have this information added
8337      using the new 'x509' options. 
8338 
8339      Current auxiliary information includes an "alias" and some trust
8340      settings. The trust settings will ultimately be used in enhanced
8341      certificate chain verification routines: currently a certificate
8342      can only be trusted if it is self signed and then it is trusted
8343      for all purposes.
8344      [Steve Henson]
8345 
8346   *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
8347      The problem was that one of the replacement routines had not been working
8348      since SSLeay releases.  For now the offending routine has been replaced
8349      with non-optimised assembler.  Even so, this now gives around 95%
8350      performance improvement for 1024 bit RSA signs.
8351      [Mark Cox]
8352 
8353   *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 
8354      handling. Most clients have the effective key size in bits equal to
8355      the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
8356      A few however don't do this and instead use the size of the decrypted key
8357      to determine the RC2 key length and the AlgorithmIdentifier to determine
8358      the effective key length. In this case the effective key length can still
8359      be 40 bits but the key length can be 168 bits for example. This is fixed
8360      by manually forcing an RC2 key into the EVP_PKEY structure because the
8361      EVP code can't currently handle unusual RC2 key sizes: it always assumes
8362      the key length and effective key length are equal.
8363      [Steve Henson]
8364 
8365   *) Add a bunch of functions that should simplify the creation of 
8366      X509_NAME structures. Now you should be able to do:
8367      X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
8368      and have it automatically work out the correct field type and fill in
8369      the structures. The more adventurous can try:
8370      X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
8371      and it will (hopefully) work out the correct multibyte encoding.
8372      [Steve Henson]
8373 
8374   *) Change the 'req' utility to use the new field handling and multibyte
8375      copy routines. Before the DN field creation was handled in an ad hoc
8376      way in req, ca, and x509 which was rather broken and didn't support
8377      BMPStrings or UTF8Strings. Since some software doesn't implement
8378      BMPStrings or UTF8Strings yet, they can be enabled using the config file
8379      using the dirstring_type option. See the new comment in the default
8380      openssl.cnf for more info.
8381      [Steve Henson]
8382 
8383   *) Make crypto/rand/md_rand.c more robust:
8384      - Assure unique random numbers after fork().
8385      - Make sure that concurrent threads access the global counter and
8386        md serializably so that we never lose entropy in them
8387        or use exactly the same state in multiple threads.
8388        Access to the large state is not always serializable because
8389        the additional locking could be a performance killer, and
8390        md should be large enough anyway.
8391      [Bodo Moeller]
8392 
8393   *) New file apps/app_rand.c with commonly needed functionality
8394      for handling the random seed file.
8395 
8396      Use the random seed file in some applications that previously did not:
8397           ca,
8398           dsaparam -genkey (which also ignored its '-rand' option), 
8399           s_client,
8400           s_server,
8401           x509 (when signing).
8402      Except on systems with /dev/urandom, it is crucial to have a random
8403      seed file at least for key creation, DSA signing, and for DH exchanges;
8404      for RSA signatures we could do without one.
8405 
8406      gendh and gendsa (unlike genrsa) used to read only the first byte
8407      of each file listed in the '-rand' option.  The function as previously
8408      found in genrsa is now in app_rand.c and is used by all programs
8409      that support '-rand'.
8410      [Bodo Moeller]
8411 
8412   *) In RAND_write_file, use mode 0600 for creating files;
8413      don't just chmod when it may be too late.
8414      [Bodo Moeller]
8415 
8416   *) Report an error from X509_STORE_load_locations
8417      when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
8418      [Bill Perry]
8419 
8420   *) New function ASN1_mbstring_copy() this copies a string in either
8421      ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
8422      into an ASN1_STRING type. A mask of permissible types is passed
8423      and it chooses the "minimal" type to use or an error if not type
8424      is suitable.
8425      [Steve Henson]
8426 
8427   *) Add function equivalents to the various macros in asn1.h. The old
8428      macros are retained with an M_ prefix. Code inside the library can
8429      use the M_ macros. External code (including the openssl utility)
8430      should *NOT* in order to be "shared library friendly".
8431      [Steve Henson]
8432 
8433   *) Add various functions that can check a certificate's extensions
8434      to see if it usable for various purposes such as SSL client,
8435      server or S/MIME and CAs of these types. This is currently 
8436      VERY EXPERIMENTAL but will ultimately be used for certificate chain
8437      verification. Also added a -purpose flag to x509 utility to
8438      print out all the purposes.
8439      [Steve Henson]
8440 
8441   *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
8442      functions.
8443      [Steve Henson]
8444 
8445   *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
8446      for, obtain and decode and extension and obtain its critical flag.
8447      This allows all the necessary extension code to be handled in a
8448      single function call.
8449      [Steve Henson]
8450 
8451   *) RC4 tune-up featuring 30-40% performance improvement on most RISC
8452      platforms. See crypto/rc4/rc4_enc.c for further details.
8453      [Andy Polyakov]
8454 
8455   *) New -noout option to asn1parse. This causes no output to be produced
8456      its main use is when combined with -strparse and -out to extract data
8457      from a file (which may not be in ASN.1 format).
8458      [Steve Henson]
8459 
8460   *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
8461      when producing the local key id.
8462      [Richard Levitte <levitte@stacken.kth.se>]
8463 
8464   *) New option -dhparam in s_server. This allows a DH parameter file to be
8465      stated explicitly. If it is not stated then it tries the first server
8466      certificate file. The previous behaviour hard coded the filename
8467      "server.pem".
8468      [Steve Henson]
8469 
8470   *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
8471      a public key to be input or output. For example:
8472      openssl rsa -in key.pem -pubout -out pubkey.pem
8473      Also added necessary DSA public key functions to handle this.
8474      [Steve Henson]
8475 
8476   *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
8477      in the message. This was handled by allowing
8478      X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
8479      [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
8480 
8481   *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
8482      to the end of the strings whereas this didn't. This would cause problems
8483      if strings read with d2i_ASN1_bytes() were later modified.
8484      [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
8485 
8486   *) Fix for base64 decode bug. When a base64 bio reads only one line of
8487      data and it contains EOF it will end up returning an error. This is
8488      caused by input 46 bytes long. The cause is due to the way base64
8489      BIOs find the start of base64 encoded data. They do this by trying a
8490      trial decode on each line until they find one that works. When they
8491      do a flag is set and it starts again knowing it can pass all the
8492      data directly through the decoder. Unfortunately it doesn't reset
8493      the context it uses. This means that if EOF is reached an attempt
8494      is made to pass two EOFs through the context and this causes the
8495      resulting error. This can also cause other problems as well. As is
8496      usual with these problems it takes *ages* to find and the fix is
8497      trivial: move one line.
8498      [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
8499 
8500   *) Ugly workaround to get s_client and s_server working under Windows. The
8501      old code wouldn't work because it needed to select() on sockets and the
8502      tty (for keypresses and to see if data could be written). Win32 only
8503      supports select() on sockets so we select() with a 1s timeout on the
8504      sockets and then see if any characters are waiting to be read, if none
8505      are present then we retry, we also assume we can always write data to
8506      the tty. This isn't nice because the code then blocks until we've
8507      received a complete line of data and it is effectively polling the
8508      keyboard at 1s intervals: however it's quite a bit better than not
8509      working at all :-) A dedicated Windows application might handle this
8510      with an event loop for example.
8511      [Steve Henson]
8512 
8513   *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
8514      and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
8515      will be called when RSA_sign() and RSA_verify() are used. This is useful
8516      if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
8517      For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
8518      should *not* be used: RSA_sign() and RSA_verify() must be used instead.
8519      This necessitated the support of an extra signature type NID_md5_sha1
8520      for SSL signatures and modifications to the SSL library to use it instead
8521      of calling RSA_public_decrypt() and RSA_private_encrypt().
8522      [Steve Henson]
8523 
8524   *) Add new -verify -CAfile and -CApath options to the crl program, these
8525      will lookup a CRL issuers certificate and verify the signature in a
8526      similar way to the verify program. Tidy up the crl program so it
8527      no longer accesses structures directly. Make the ASN1 CRL parsing a bit
8528      less strict. It will now permit CRL extensions even if it is not
8529      a V2 CRL: this will allow it to tolerate some broken CRLs.
8530      [Steve Henson]
8531 
8532   *) Initialize all non-automatic variables each time one of the openssl
8533      sub-programs is started (this is necessary as they may be started
8534      multiple times from the "OpenSSL>" prompt).
8535      [Lennart Bang, Bodo Moeller]
8536 
8537   *) Preliminary compilation option RSA_NULL which disables RSA crypto without
8538      removing all other RSA functionality (this is what NO_RSA does). This
8539      is so (for example) those in the US can disable those operations covered
8540      by the RSA patent while allowing storage and parsing of RSA keys and RSA
8541      key generation.
8542      [Steve Henson]
8543 
8544   *) Non-copying interface to BIO pairs.
8545      (still largely untested)
8546      [Bodo Moeller]
8547 
8548   *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
8549      ASCII string. This was handled independently in various places before.
8550      [Steve Henson]
8551 
8552   *) New functions UTF8_getc() and UTF8_putc() that parse and generate
8553      UTF8 strings a character at a time.
8554      [Steve Henson]
8555 
8556   *) Use client_version from client hello to select the protocol
8557      (s23_srvr.c) and for RSA client key exchange verification
8558      (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
8559      [Bodo Moeller]
8560 
8561   *) Add various utility functions to handle SPKACs, these were previously
8562      handled by poking round in the structure internals. Added new function
8563      NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
8564      print, verify and generate SPKACs. Based on an original idea from
8565      Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
8566      [Steve Henson]
8567 
8568   *) RIPEMD160 is operational on all platforms and is back in 'make test'.
8569      [Andy Polyakov]
8570 
8571   *) Allow the config file extension section to be overwritten on the
8572      command line. Based on an original idea from Massimiliano Pala
8573      <madwolf@comune.modena.it>. The new option is called -extensions
8574      and can be applied to ca, req and x509. Also -reqexts to override
8575      the request extensions in req and -crlexts to override the crl extensions
8576      in ca.
8577      [Steve Henson]
8578 
8579   *) Add new feature to the SPKAC handling in ca.  Now you can include
8580      the same field multiple times by preceding it by "XXXX." for example:
8581      1.OU="Unit name 1"
8582      2.OU="Unit name 2"
8583      this is the same syntax as used in the req config file.
8584      [Steve Henson]
8585 
8586   *) Allow certificate extensions to be added to certificate requests. These
8587      are specified in a 'req_extensions' option of the req section of the
8588      config file. They can be printed out with the -text option to req but
8589      are otherwise ignored at present.
8590      [Steve Henson]
8591 
8592   *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
8593      data read consists of only the final block it would not decrypted because
8594      EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
8595      A misplaced 'break' also meant the decrypted final block might not be
8596      copied until the next read.
8597      [Steve Henson]
8598 
8599   *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
8600      a few extra parameters to the DH structure: these will be useful if
8601      for example we want the value of 'q' or implement X9.42 DH.
8602      [Steve Henson]
8603 
8604   *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
8605      provides hooks that allow the default DSA functions or functions on a
8606      "per key" basis to be replaced. This allows hardware acceleration and
8607      hardware key storage to be handled without major modification to the
8608      library. Also added low level modexp hooks and CRYPTO_EX structure and 
8609      associated functions.
8610      [Steve Henson]
8611 
8612   *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
8613      as "read only": it can't be written to and the buffer it points to will
8614      not be freed. Reading from a read only BIO is much more efficient than
8615      a normal memory BIO. This was added because there are several times when
8616      an area of memory needs to be read from a BIO. The previous method was
8617      to create a memory BIO and write the data to it, this results in two
8618      copies of the data and an O(n^2) reading algorithm. There is a new
8619      function BIO_new_mem_buf() which creates a read only memory BIO from
8620      an area of memory. Also modified the PKCS#7 routines to use read only
8621      memory BIOs.
8622      [Steve Henson]
8623 
8624   *) Bugfix: ssl23_get_client_hello did not work properly when called in
8625      state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
8626      a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
8627      but a retry condition occured while trying to read the rest.
8628      [Bodo Moeller]
8629 
8630   *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
8631      NID_pkcs7_encrypted by default: this was wrong since this should almost
8632      always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
8633      the encrypted data type: this is a more sensible place to put it and it
8634      allows the PKCS#12 code to be tidied up that duplicated this
8635      functionality.
8636      [Steve Henson]
8637 
8638   *) Changed obj_dat.pl script so it takes its input and output files on
8639      the command line. This should avoid shell escape redirection problems
8640      under Win32.
8641      [Steve Henson]
8642 
8643   *) Initial support for certificate extension requests, these are included
8644      in things like Xenroll certificate requests. Included functions to allow
8645      extensions to be obtained and added.
8646      [Steve Henson]
8647 
8648   *) -crlf option to s_client and s_server for sending newlines as
8649      CRLF (as required by many protocols).
8650      [Bodo Moeller]
8651 
8652  Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
8653   
8654   *) Install libRSAglue.a when OpenSSL is built with RSAref.
8655      [Ralf S. Engelschall]
8656 
8657   *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
8658      [Andrija Antonijevic <TheAntony2@bigfoot.com>]
8659 
8660   *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
8661      program.
8662      [Steve Henson]
8663 
8664   *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
8665      DH parameters/keys (q is lost during that conversion, but the resulting
8666      DH parameters contain its length).
8667 
8668      For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
8669      much faster than DH_generate_parameters (which creates parameters
8670      where p = 2*q + 1), and also the smaller q makes DH computations
8671      much more efficient (160-bit exponentiation instead of 1024-bit
8672      exponentiation); so this provides a convenient way to support DHE
8673      ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
8674      utter importance to use
8675          SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
8676      or
8677          SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
8678      when such DH parameters are used, because otherwise small subgroup
8679      attacks may become possible!
8680      [Bodo Moeller]
8681 
8682   *) Avoid memory leak in i2d_DHparams.
8683      [Bodo Moeller]
8684 
8685   *) Allow the -k option to be used more than once in the enc program:
8686      this allows the same encrypted message to be read by multiple recipients.
8687      [Steve Henson]
8688 
8689   *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
8690      an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
8691      it will always use the numerical form of the OID, even if it has a short
8692      or long name.
8693      [Steve Henson]
8694 
8695   *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
8696      method only got called if p,q,dmp1,dmq1,iqmp components were present,
8697      otherwise bn_mod_exp was called. In the case of hardware keys for example
8698      no private key components need be present and it might store extra data
8699      in the RSA structure, which cannot be accessed from bn_mod_exp.
8700      By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
8701      private key operations.
8702      [Steve Henson]
8703 
8704   *) Added support for SPARC Linux.
8705      [Andy Polyakov]
8706 
8707   *) pem_password_cb function type incompatibly changed from
8708           typedef int pem_password_cb(char *buf, int size, int rwflag);
8709      to
8710           ....(char *buf, int size, int rwflag, void *userdata);
8711      so that applications can pass data to their callbacks:
8712      The PEM[_ASN1]_{read,write}... functions and macros now take an
8713      additional void * argument, which is just handed through whenever
8714      the password callback is called.
8715      [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
8716 
8717      New function SSL_CTX_set_default_passwd_cb_userdata.
8718 
8719      Compatibility note: As many C implementations push function arguments
8720      onto the stack in reverse order, the new library version is likely to
8721      interoperate with programs that have been compiled with the old
8722      pem_password_cb definition (PEM_whatever takes some data that
8723      happens to be on the stack as its last argument, and the callback
8724      just ignores this garbage); but there is no guarantee whatsoever that
8725      this will work.
8726 
8727   *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
8728      (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
8729      problems not only on Windows, but also on some Unix platforms.
8730      To avoid problematic command lines, these definitions are now in an
8731      auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
8732      for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
8733      [Bodo Moeller]
8734 
8735   *) MIPS III/IV assembler module is reimplemented.
8736      [Andy Polyakov]
8737 
8738   *) More DES library cleanups: remove references to srand/rand and
8739      delete an unused file.
8740      [Ulf Möller]
8741 
8742   *) Add support for the the free Netwide assembler (NASM) under Win32,
8743      since not many people have MASM (ml) and it can be hard to obtain.
8744      This is currently experimental but it seems to work OK and pass all
8745      the tests. Check out INSTALL.W32 for info.
8746      [Steve Henson]
8747 
8748   *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
8749      without temporary keys kept an extra copy of the server key,
8750      and connections with temporary keys did not free everything in case
8751      of an error.
8752      [Bodo Moeller]
8753 
8754   *) New function RSA_check_key and new openssl rsa option -check
8755      for verifying the consistency of RSA keys.
8756      [Ulf Moeller, Bodo Moeller]
8757 
8758   *) Various changes to make Win32 compile work: 
8759      1. Casts to avoid "loss of data" warnings in p5_crpt2.c
8760      2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
8761         comparison" warnings.
8762      3. Add sk_<TYPE>_sort to DEF file generator and do make update.
8763      [Steve Henson]
8764 
8765   *) Add a debugging option to PKCS#5 v2 key generation function: when
8766      you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
8767      derived keys are printed to stderr.
8768      [Steve Henson]
8769 
8770   *) Copy the flags in ASN1_STRING_dup().
8771      [Roman E. Pavlov <pre@mo.msk.ru>]
8772 
8773   *) The x509 application mishandled signing requests containing DSA
8774      keys when the signing key was also DSA and the parameters didn't match.
8775 
8776      It was supposed to omit the parameters when they matched the signing key:
8777      the verifying software was then supposed to automatically use the CA's
8778      parameters if they were absent from the end user certificate.
8779 
8780      Omitting parameters is no longer recommended. The test was also
8781      the wrong way round! This was probably due to unusual behaviour in
8782      EVP_cmp_parameters() which returns 1 if the parameters match. 
8783      This meant that parameters were omitted when they *didn't* match and
8784      the certificate was useless. Certificates signed with 'ca' didn't have
8785      this bug.
8786      [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
8787 
8788   *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
8789      The interface is as follows:
8790      Applications can use
8791          CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
8792          CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
8793      "off" is now the default.
8794      The library internally uses
8795          CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
8796          CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
8797      to disable memory-checking temporarily.
8798 
8799      Some inconsistent states that previously were possible (and were
8800      even the default) are now avoided.
8801 
8802      -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
8803      with each memory chunk allocated; this is occasionally more helpful
8804      than just having a counter.
8805 
8806      -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
8807 
8808      -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
8809      extensions.
8810      [Bodo Moeller]
8811 
8812   *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
8813      which largely parallels "options", but is for changing API behaviour,
8814      whereas "options" are about protocol behaviour.
8815      Initial "mode" flags are:
8816 
8817      SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
8818                                      a single record has been written.
8819      SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
8820                                      retries use the same buffer location.
8821                                      (But all of the contents must be
8822                                      copied!)
8823      [Bodo Moeller]
8824 
8825   *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
8826      worked.
8827 
8828   *) Fix problems with no-hmac etc.
8829      [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
8830 
8831   *) New functions RSA_get_default_method(), RSA_set_method() and
8832      RSA_get_method(). These allows replacement of RSA_METHODs without having
8833      to mess around with the internals of an RSA structure.
8834      [Steve Henson]
8835 
8836   *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
8837      Also really enable memory leak checks in openssl.c and in some
8838      test programs.
8839      [Chad C. Mulligan, Bodo Moeller]
8840 
8841   *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
8842      up the length of negative integers. This has now been simplified to just
8843      store the length when it is first determined and use it later, rather
8844      than trying to keep track of where data is copied and updating it to
8845      point to the end.
8846      [Steve Henson, reported by Brien Wheeler
8847       <bwheeler@authentica-security.com>]
8848 
8849   *) Add a new function PKCS7_signatureVerify. This allows the verification
8850      of a PKCS#7 signature but with the signing certificate passed to the
8851      function itself. This contrasts with PKCS7_dataVerify which assumes the
8852      certificate is present in the PKCS#7 structure. This isn't always the
8853      case: certificates can be omitted from a PKCS#7 structure and be
8854      distributed by "out of band" means (such as a certificate database).
8855      [Steve Henson]
8856 
8857   *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
8858      function prototypes in pem.h, also change util/mkdef.pl to add the
8859      necessary function names. 
8860      [Steve Henson]
8861 
8862   *) mk1mf.pl (used by Windows builds) did not properly read the
8863      options set by Configure in the top level Makefile, and Configure
8864      was not even able to write more than one option correctly.
8865      Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
8866      [Bodo Moeller]
8867 
8868   *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
8869      file to be loaded from a BIO or FILE pointer. The BIO version will
8870      for example allow memory BIOs to contain config info.
8871      [Steve Henson]
8872 
8873   *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
8874      Whoever hopes to achieve shared-library compatibility across versions
8875      must use this, not the compile-time macro.
8876      (Exercise 0.9.4: Which is the minimum library version required by
8877      such programs?)
8878      Note: All this applies only to multi-threaded programs, others don't
8879      need locks.
8880      [Bodo Moeller]
8881 
8882   *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
8883      through a BIO pair triggered the default case, i.e.
8884      SSLerr(...,SSL_R_UNKNOWN_STATE).
8885      [Bodo Moeller]
8886 
8887   *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
8888      can use the SSL library even if none of the specific BIOs is
8889      appropriate.
8890      [Bodo Moeller]
8891 
8892   *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
8893      for the encoded length.
8894      [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
8895 
8896   *) Add initial documentation of the X509V3 functions.
8897      [Steve Henson]
8898 
8899   *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and 
8900      PEM_write_bio_PKCS8PrivateKey() that are equivalent to
8901      PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
8902      secure PKCS#8 private key format with a high iteration count.
8903      [Steve Henson]
8904 
8905   *) Fix determination of Perl interpreter: A perl or perl5
8906      _directory_ in $PATH was also accepted as the interpreter.
8907      [Ralf S. Engelschall]
8908 
8909   *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
8910      wrong with it but it was very old and did things like calling
8911      PEM_ASN1_read() directly and used MD5 for the hash not to mention some
8912      unusual formatting.
8913      [Steve Henson]
8914 
8915   *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
8916      to use the new extension code.
8917      [Steve Henson]
8918 
8919   *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
8920      with macros. This should make it easier to change their form, add extra
8921      arguments etc. Fix a few PEM prototypes which didn't have cipher as a
8922      constant.
8923      [Steve Henson]
8924 
8925   *) Add to configuration table a new entry that can specify an alternative
8926      name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
8927      according to Mark Crispin <MRC@Panda.COM>.
8928      [Bodo Moeller]
8929 
8930 #if 0
8931   *) DES CBC did not update the IV. Weird.
8932      [Ben Laurie]
8933 #else
8934      des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
8935      Changing the behaviour of the former might break existing programs --
8936      where IV updating is needed, des_ncbc_encrypt can be used.
8937 #endif
8938 
8939   *) When bntest is run from "make test" it drives bc to check its
8940      calculations, as well as internally checking them. If an internal check
8941      fails, it needs to cause bc to give a non-zero result or make test carries
8942      on without noticing the failure. Fixed.
8943      [Ben Laurie]
8944 
8945   *) DES library cleanups.
8946      [Ulf Möller]
8947 
8948   *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
8949      used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
8950      ciphers. NOTE: although the key derivation function has been verified
8951      against some published test vectors it has not been extensively tested
8952      yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
8953      of v2.0.
8954      [Steve Henson]
8955 
8956   *) Instead of "mkdir -p", which is not fully portable, use new
8957      Perl script "util/mkdir-p.pl".
8958      [Bodo Moeller]
8959 
8960   *) Rewrite the way password based encryption (PBE) is handled. It used to
8961      assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
8962      structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
8963      but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
8964      the 'parameter' field of the AlgorithmIdentifier is passed to the
8965      underlying key generation function so it must do its own ASN1 parsing.
8966      This has also changed the EVP_PBE_CipherInit() function which now has a
8967      'parameter' argument instead of literal salt and iteration count values
8968      and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
8969      [Steve Henson]
8970 
8971   *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
8972      and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
8973      Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
8974      KEY" because this clashed with PKCS#8 unencrypted string. Since this
8975      value was just used as a "magic string" and not used directly its
8976      value doesn't matter.
8977      [Steve Henson]
8978 
8979   *) Introduce some semblance of const correctness to BN. Shame C doesn't
8980      support mutable.
8981      [Ben Laurie]
8982 
8983   *) "linux-sparc64" configuration (ultrapenguin).
8984      [Ray Miller <ray.miller@oucs.ox.ac.uk>]
8985      "linux-sparc" configuration.
8986      [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
8987 
8988   *) config now generates no-xxx options for missing ciphers.
8989      [Ulf Möller]
8990 
8991   *) Support the EBCDIC character set (work in progress).
8992      File ebcdic.c not yet included because it has a different license.
8993      [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
8994 
8995   *) Support BS2000/OSD-POSIX.
8996      [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
8997 
8998   *) Make callbacks for key generation use void * instead of char *.
8999      [Ben Laurie]
9000 
9001   *) Make S/MIME samples compile (not yet tested).
9002      [Ben Laurie]
9003 
9004   *) Additional typesafe stacks.
9005      [Ben Laurie]
9006 
9007   *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
9008      [Bodo Moeller]
9009 
9010 
9011  Changes between 0.9.3 and 0.9.3a  [29 May 1999]
9012 
9013   *) New configuration variant "sco5-gcc".
9014 
9015   *) Updated some demos.
9016      [Sean O Riordain, Wade Scholine]
9017 
9018   *) Add missing BIO_free at exit of pkcs12 application.
9019      [Wu Zhigang]
9020 
9021   *) Fix memory leak in conf.c.
9022      [Steve Henson]
9023 
9024   *) Updates for Win32 to assembler version of MD5.
9025      [Steve Henson]
9026 
9027   *) Set #! path to perl in apps/der_chop to where we found it
9028      instead of using a fixed path.
9029      [Bodo Moeller]
9030 
9031   *) SHA library changes for irix64-mips4-cc.
9032      [Andy Polyakov]
9033 
9034   *) Improvements for VMS support.
9035      [Richard Levitte]
9036 
9037 
9038  Changes between 0.9.2b and 0.9.3  [24 May 1999]
9039 
9040   *) Bignum library bug fix. IRIX 6 passes "make test" now!
9041      This also avoids the problems with SC4.2 and unpatched SC5.  
9042      [Andy Polyakov <appro@fy.chalmers.se>]
9043 
9044   *) New functions sk_num, sk_value and sk_set to replace the previous macros.
9045      These are required because of the typesafe stack would otherwise break 
9046      existing code. If old code used a structure member which used to be STACK
9047      and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
9048      sk_num or sk_value it would produce an error because the num, data members
9049      are not present in STACK_OF. Now it just produces a warning. sk_set
9050      replaces the old method of assigning a value to sk_value
9051      (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
9052      that does this will no longer work (and should use sk_set instead) but
9053      this could be regarded as a "questionable" behaviour anyway.
9054      [Steve Henson]
9055 
9056   *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
9057      correctly handle encrypted S/MIME data.
9058      [Steve Henson]
9059 
9060   *) Change type of various DES function arguments from des_cblock
9061      (which means, in function argument declarations, pointer to char)
9062      to des_cblock * (meaning pointer to array with 8 char elements),
9063      which allows the compiler to do more typechecking; it was like
9064      that back in SSLeay, but with lots of ugly casts.
9065 
9066      Introduce new type const_des_cblock.
9067      [Bodo Moeller]
9068 
9069   *) Reorganise the PKCS#7 library and get rid of some of the more obvious
9070      problems: find RecipientInfo structure that matches recipient certificate
9071      and initialise the ASN1 structures properly based on passed cipher.
9072      [Steve Henson]
9073 
9074   *) Belatedly make the BN tests actually check the results.
9075      [Ben Laurie]
9076 
9077   *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
9078      to and from BNs: it was completely broken. New compilation option
9079      NEG_PUBKEY_BUG to allow for some broken certificates that encode public
9080      key elements as negative integers.
9081      [Steve Henson]
9082 
9083   *) Reorganize and speed up MD5.
9084      [Andy Polyakov <appro@fy.chalmers.se>]
9085 
9086   *) VMS support.
9087      [Richard Levitte <richard@levitte.org>]
9088 
9089   *) New option -out to asn1parse to allow the parsed structure to be
9090      output to a file. This is most useful when combined with the -strparse
9091      option to examine the output of things like OCTET STRINGS.
9092      [Steve Henson]
9093 
9094   *) Make SSL library a little more fool-proof by not requiring any longer
9095      that SSL_set_{accept,connect}_state be called before
9096      SSL_{accept,connect} may be used (SSL_set_..._state is omitted
9097      in many applications because usually everything *appeared* to work as
9098      intended anyway -- now it really works as intended).
9099      [Bodo Moeller]
9100 
9101   *) Move openssl.cnf out of lib/.
9102      [Ulf Möller]
9103 
9104   *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
9105      -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
9106      -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+ 
9107      [Ralf S. Engelschall]
9108 
9109   *) Various fixes to the EVP and PKCS#7 code. It may now be able to
9110      handle PKCS#7 enveloped data properly.
9111      [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
9112 
9113   *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
9114      copying pointers.  The cert_st handling is changed by this in
9115      various ways (and thus what used to be known as ctx->default_cert
9116      is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
9117      any longer when s->cert does not give us what we need).
9118      ssl_cert_instantiate becomes obsolete by this change.
9119      As soon as we've got the new code right (possibly it already is?),
9120      we have solved a couple of bugs of the earlier code where s->cert
9121      was used as if it could not have been shared with other SSL structures.
9122 
9123      Note that using the SSL API in certain dirty ways now will result
9124      in different behaviour than observed with earlier library versions:
9125      Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
9126      does not influence s as it used to.
9127      
9128      In order to clean up things more thoroughly, inside SSL_SESSION
9129      we don't use CERT any longer, but a new structure SESS_CERT
9130      that holds per-session data (if available); currently, this is
9131      the peer's certificate chain and, for clients, the server's certificate
9132      and temporary key.  CERT holds only those values that can have
9133      meaningful defaults in an SSL_CTX.
9134      [Bodo Moeller]
9135 
9136   *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
9137      from the internal representation. Various PKCS#7 fixes: remove some
9138      evil casts and set the enc_dig_alg field properly based on the signing
9139      key type.
9140      [Steve Henson]
9141 
9142   *) Allow PKCS#12 password to be set from the command line or the
9143      environment. Let 'ca' get its config file name from the environment
9144      variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
9145      and 'x509').
9146      [Steve Henson]
9147 
9148   *) Allow certificate policies extension to use an IA5STRING for the
9149      organization field. This is contrary to the PKIX definition but
9150      VeriSign uses it and IE5 only recognises this form. Document 'x509'
9151      extension option.
9152      [Steve Henson]
9153 
9154   *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
9155      without disallowing inline assembler and the like for non-pedantic builds.
9156      [Ben Laurie]
9157 
9158   *) Support Borland C++ builder.
9159      [Janez Jere <jj@void.si>, modified by Ulf Möller]
9160 
9161   *) Support Mingw32.
9162      [Ulf Möller]
9163 
9164   *) SHA-1 cleanups and performance enhancements.
9165      [Andy Polyakov <appro@fy.chalmers.se>]
9166 
9167   *) Sparc v8plus assembler for the bignum library.
9168      [Andy Polyakov <appro@fy.chalmers.se>]
9169 
9170   *) Accept any -xxx and +xxx compiler options in Configure.
9171      [Ulf Möller]
9172 
9173   *) Update HPUX configuration.
9174      [Anonymous]
9175   
9176   *) Add missing sk_<type>_unshift() function to safestack.h
9177      [Ralf S. Engelschall]
9178 
9179   *) New function SSL_CTX_use_certificate_chain_file that sets the
9180      "extra_cert"s in addition to the certificate.  (This makes sense
9181      only for "PEM" format files, as chains as a whole are not
9182      DER-encoded.)
9183      [Bodo Moeller]
9184 
9185   *) Support verify_depth from the SSL API.
9186      x509_vfy.c had what can be considered an off-by-one-error:
9187      Its depth (which was not part of the external interface)
9188      was actually counting the number of certificates in a chain;
9189      now it really counts the depth.
9190      [Bodo Moeller]
9191 
9192   *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
9193      instead of X509err, which often resulted in confusing error
9194      messages since the error codes are not globally unique
9195      (e.g. an alleged error in ssl3_accept when a certificate
9196      didn't match the private key).
9197 
9198   *) New function SSL_CTX_set_session_id_context that allows to set a default
9199      value (so that you don't need SSL_set_session_id_context for each
9200      connection using the SSL_CTX).
9201      [Bodo Moeller]
9202 
9203   *) OAEP decoding bug fix.
9204      [Ulf Möller]
9205 
9206   *) Support INSTALL_PREFIX for package builders, as proposed by
9207      David Harris.
9208      [Bodo Moeller]
9209 
9210   *) New Configure options "threads" and "no-threads".  For systems
9211      where the proper compiler options are known (currently Solaris
9212      and Linux), "threads" is the default.
9213      [Bodo Moeller]
9214 
9215   *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
9216      [Bodo Moeller]
9217 
9218   *) Install various scripts to $(OPENSSLDIR)/misc, not to
9219      $(INSTALLTOP)/bin -- they shouldn't clutter directories
9220      such as /usr/local/bin.
9221      [Bodo Moeller]
9222 
9223   *) "make linux-shared" to build shared libraries.
9224      [Niels Poppe <niels@netbox.org>]
9225 
9226   *) New Configure option no-<cipher> (rsa, idea, rc5, ...).
9227      [Ulf Möller]
9228 
9229   *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
9230      extension adding in x509 utility.
9231      [Steve Henson]
9232 
9233   *) Remove NOPROTO sections and error code comments.
9234      [Ulf Möller]
9235 
9236   *) Partial rewrite of the DEF file generator to now parse the ANSI
9237      prototypes.
9238      [Steve Henson]
9239 
9240   *) New Configure options --prefix=DIR and --openssldir=DIR.
9241      [Ulf Möller]
9242 
9243   *) Complete rewrite of the error code script(s). It is all now handled
9244      by one script at the top level which handles error code gathering,
9245      header rewriting and C source file generation. It should be much better
9246      than the old method: it now uses a modified version of Ulf's parser to
9247      read the ANSI prototypes in all header files (thus the old K&R definitions
9248      aren't needed for error creation any more) and do a better job of
9249      translating function codes into names. The old 'ASN1 error code imbedded
9250      in a comment' is no longer necessary and it doesn't use .err files which
9251      have now been deleted. Also the error code call doesn't have to appear all
9252      on one line (which resulted in some large lines...).
9253      [Steve Henson]
9254 
9255   *) Change #include filenames from <foo.h> to <openssl/foo.h>.
9256      [Bodo Moeller]
9257 
9258   *) Change behaviour of ssl2_read when facing length-0 packets: Don't return
9259      0 (which usually indicates a closed connection), but continue reading.
9260      [Bodo Moeller]
9261 
9262   *) Fix some race conditions.
9263      [Bodo Moeller]
9264 
9265   *) Add support for CRL distribution points extension. Add Certificate
9266      Policies and CRL distribution points documentation.
9267      [Steve Henson]
9268 
9269   *) Move the autogenerated header file parts to crypto/opensslconf.h.
9270      [Ulf Möller]
9271 
9272   *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
9273      8 of keying material. Merlin has also confirmed interop with this fix
9274      between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
9275      [Merlin Hughes <merlin@baltimore.ie>]
9276 
9277   *) Fix lots of warnings.
9278      [Richard Levitte <levitte@stacken.kth.se>]
9279  
9280   *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
9281      the directory spec didn't end with a LIST_SEPARATOR_CHAR.
9282      [Richard Levitte <levitte@stacken.kth.se>]
9283  
9284   *) Fix problems with sizeof(long) == 8.
9285      [Andy Polyakov <appro@fy.chalmers.se>]
9286 
9287   *) Change functions to ANSI C.
9288      [Ulf Möller]
9289 
9290   *) Fix typos in error codes.
9291      [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller]
9292 
9293   *) Remove defunct assembler files from Configure.
9294      [Ulf Möller]
9295 
9296   *) SPARC v8 assembler BIGNUM implementation.
9297      [Andy Polyakov <appro@fy.chalmers.se>]
9298 
9299   *) Support for Certificate Policies extension: both print and set.
9300      Various additions to support the r2i method this uses.
9301      [Steve Henson]
9302 
9303   *) A lot of constification, and fix a bug in X509_NAME_oneline() that could
9304      return a const string when you are expecting an allocated buffer.
9305      [Ben Laurie]
9306 
9307   *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
9308      types DirectoryString and DisplayText.
9309      [Steve Henson]
9310 
9311   *) Add code to allow r2i extensions to access the configuration database,
9312      add an LHASH database driver and add several ctx helper functions.
9313      [Steve Henson]
9314 
9315   *) Fix an evil bug in bn_expand2() which caused various BN functions to
9316      fail when they extended the size of a BIGNUM.
9317      [Steve Henson]
9318 
9319   *) Various utility functions to handle SXNet extension. Modify mkdef.pl to
9320      support typesafe stack.
9321      [Steve Henson]
9322 
9323   *) Fix typo in SSL_[gs]et_options().
9324      [Nils Frostberg <nils@medcom.se>]
9325 
9326   *) Delete various functions and files that belonged to the (now obsolete)
9327      old X509V3 handling code.
9328      [Steve Henson]
9329 
9330   *) New Configure option "rsaref".
9331      [Ulf Möller]
9332 
9333   *) Don't auto-generate pem.h.
9334      [Bodo Moeller]
9335 
9336   *) Introduce type-safe ASN.1 SETs.
9337      [Ben Laurie]
9338 
9339   *) Convert various additional casted stacks to type-safe STACK_OF() variants.
9340      [Ben Laurie, Ralf S. Engelschall, Steve Henson]
9341 
9342   *) Introduce type-safe STACKs. This will almost certainly break lots of code
9343      that links with OpenSSL (well at least cause lots of warnings), but fear
9344      not: the conversion is trivial, and it eliminates loads of evil casts. A
9345      few STACKed things have been converted already. Feel free to convert more.
9346      In the fullness of time, I'll do away with the STACK type altogether.
9347      [Ben Laurie]
9348 
9349   *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate
9350      specified in <certfile> by updating the entry in the index.txt file.
9351      This way one no longer has to edit the index.txt file manually for
9352      revoking a certificate. The -revoke option does the gory details now.
9353      [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall]
9354 
9355   *) Fix `openssl crl -noout -text' combination where `-noout' killed the
9356      `-text' option at all and this way the `-noout -text' combination was
9357      inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'.
9358      [Ralf S. Engelschall]
9359 
9360   *) Make sure a corresponding plain text error message exists for the
9361      X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
9362      verify callback function determined that a certificate was revoked.
9363      [Ralf S. Engelschall]
9364 
9365   *) Bugfix: In test/testenc, don't test "openssl <cipher>" for
9366      ciphers that were excluded, e.g. by -DNO_IDEA.  Also, test
9367      all available cipers including rc5, which was forgotten until now.
9368      In order to let the testing shell script know which algorithms
9369      are available, a new (up to now undocumented) command
9370      "openssl list-cipher-commands" is used.
9371      [Bodo Moeller]
9372 
9373   *) Bugfix: s_client occasionally would sleep in select() when
9374      it should have checked SSL_pending() first.
9375      [Bodo Moeller]
9376 
9377   *) New functions DSA_do_sign and DSA_do_verify to provide access to
9378      the raw DSA values prior to ASN.1 encoding.
9379      [Ulf Möller]
9380 
9381   *) Tweaks to Configure
9382      [Niels Poppe <niels@netbox.org>]
9383 
9384   *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
9385      yet...
9386      [Steve Henson]
9387 
9388   *) New variables $(RANLIB) and $(PERL) in the Makefiles.
9389      [Ulf Möller]
9390 
9391   *) New config option to avoid instructions that are illegal on the 80386.
9392      The default code is faster, but requires at least a 486.
9393      [Ulf Möller]
9394   
9395   *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
9396      SSL2_SERVER_VERSION (not used at all) macros, which are now the
9397      same as SSL2_VERSION anyway.
9398      [Bodo Moeller]
9399 
9400   *) New "-showcerts" option for s_client.
9401      [Bodo Moeller]
9402 
9403   *) Still more PKCS#12 integration. Add pkcs12 application to openssl
9404      application. Various cleanups and fixes.
9405      [Steve Henson]
9406 
9407   *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
9408      modify error routines to work internally. Add error codes and PBE init
9409      to library startup routines.
9410      [Steve Henson]
9411 
9412   *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and
9413      packing functions to asn1 and evp. Changed function names and error
9414      codes along the way.
9415      [Steve Henson]
9416 
9417   *) PKCS12 integration: and so it begins... First of several patches to
9418      slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
9419      objects to objects.h
9420      [Steve Henson]
9421 
9422   *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1
9423      and display support for Thawte strong extranet extension.
9424      [Steve Henson]
9425 
9426   *) Add LinuxPPC support.
9427      [Jeff Dubrule <igor@pobox.org>]
9428 
9429   *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
9430      bn_div_words in alpha.s.
9431      [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie]
9432 
9433   *) Make sure the RSA OAEP test is skipped under -DRSAref because
9434      OAEP isn't supported when OpenSSL is built with RSAref.
9435      [Ulf Moeller <ulf@fitug.de>]
9436 
9437   *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h 
9438      so they no longer are missing under -DNOPROTO. 
9439      [Soren S. Jorvang <soren@t.dk>]
9440 
9441 
9442  Changes between 0.9.1c and 0.9.2b  [22 Mar 1999]
9443 
9444   *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
9445      doesn't work when the session is reused. Coming soon!
9446      [Ben Laurie]
9447 
9448   *) Fix a security hole, that allows sessions to be reused in the wrong
9449      context thus bypassing client cert protection! All software that uses
9450      client certs and session caches in multiple contexts NEEDS PATCHING to
9451      allow session reuse! A fuller solution is in the works.
9452      [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)]
9453 
9454   *) Some more source tree cleanups (removed obsolete files
9455      crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
9456      permission on "config" script to be executable) and a fix for the INSTALL
9457      document.
9458      [Ulf Moeller <ulf@fitug.de>]
9459 
9460   *) Remove some legacy and erroneous uses of malloc, free instead of
9461      Malloc, Free.
9462      [Lennart Bang <lob@netstream.se>, with minor changes by Steve]
9463 
9464   *) Make rsa_oaep_test return non-zero on error.
9465      [Ulf Moeller <ulf@fitug.de>]
9466 
9467   *) Add support for native Solaris shared libraries. Configure
9468      solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
9469      if someone would make that last step automatic.
9470      [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>]
9471 
9472   *) ctx_size was not built with the right compiler during "make links". Fixed.
9473      [Ben Laurie]
9474 
9475   *) Change the meaning of 'ALL' in the cipher list. It now means "everything
9476      except NULL ciphers". This means the default cipher list will no longer
9477      enable NULL ciphers. They need to be specifically enabled e.g. with
9478      the string "DEFAULT:eNULL".
9479      [Steve Henson]
9480 
9481   *) Fix to RSA private encryption routines: if p < q then it would
9482      occasionally produce an invalid result. This will only happen with
9483      externally generated keys because OpenSSL (and SSLeay) ensure p > q.
9484      [Steve Henson]
9485 
9486   *) Be less restrictive and allow also `perl util/perlpath.pl
9487      /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin',
9488      because this way one can also use an interpreter named `perl5' (which is
9489      usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
9490      installed as `perl').
9491      [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
9492 
9493   *) Let util/clean-depend.pl work also with older Perl 5.00x versions.
9494      [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
9495 
9496   *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
9497      advapi32.lib to Win32 build and change the pem test comparision
9498      to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
9499      suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
9500      and crypto/des/ede_cbcm_enc.c.
9501      [Steve Henson]
9502 
9503   *) DES quad checksum was broken on big-endian architectures. Fixed.
9504      [Ben Laurie]
9505 
9506   *) Comment out two functions in bio.h that aren't implemented. Fix up the
9507      Win32 test batch file so it (might) work again. The Win32 test batch file
9508      is horrible: I feel ill....
9509      [Steve Henson]
9510 
9511   *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
9512      in e_os.h. Audit of header files to check ANSI and non ANSI
9513      sections: 10 functions were absent from non ANSI section and not exported
9514      from Windows DLLs. Fixed up libeay.num for new functions.
9515      [Steve Henson]
9516 
9517   *) Make `openssl version' output lines consistent.
9518      [Ralf S. Engelschall]
9519 
9520   *) Fix Win32 symbol export lists for BIO functions: Added
9521      BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
9522      to ms/libeay{16,32}.def.
9523      [Ralf S. Engelschall]
9524 
9525   *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
9526      fine under Unix and passes some trivial tests I've now added. But the
9527      whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
9528      added to make sure no one expects that this stuff really works in the
9529      OpenSSL 0.9.2 release.  Additionally I've started to clean the XS sources
9530      up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
9531      openssl_bio.xs.
9532      [Ralf S. Engelschall]
9533 
9534   *) Fix the generation of two part addresses in perl.
9535      [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie]
9536 
9537   *) Add config entry for Linux on MIPS.
9538      [John Tobey <jtobey@channel1.com>]
9539 
9540   *) Make links whenever Configure is run, unless we are on Windoze.
9541      [Ben Laurie]
9542 
9543   *) Permit extensions to be added to CRLs using crl_section in openssl.cnf.
9544      Currently only issuerAltName and AuthorityKeyIdentifier make any sense
9545      in CRLs.
9546      [Steve Henson]
9547 
9548   *) Add a useful kludge to allow package maintainers to specify compiler and
9549      other platforms details on the command line without having to patch the
9550      Configure script everytime: One now can use ``perl Configure
9551      <id>:<details>'', i.e. platform ids are allowed to have details appended
9552      to them (seperated by colons). This is treated as there would be a static
9553      pre-configured entry in Configure's %table under key <id> with value
9554      <details> and ``perl Configure <id>'' is called.  So, when you want to
9555      perform a quick test-compile under FreeBSD 3.1 with pgcc and without
9556      assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"''
9557      now, which overrides the FreeBSD-elf entry on-the-fly.
9558      [Ralf S. Engelschall]
9559 
9560   *) Disable new TLS1 ciphersuites by default: they aren't official yet.
9561      [Ben Laurie]
9562 
9563   *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
9564      on the `perl Configure ...' command line. This way one can compile
9565      OpenSSL libraries with Position Independent Code (PIC) which is needed
9566      for linking it into DSOs.
9567      [Ralf S. Engelschall]
9568 
9569   *) Remarkably, export ciphers were totally broken and no-one had noticed!
9570      Fixed.
9571      [Ben Laurie]
9572 
9573   *) Cleaned up the LICENSE document: The official contact for any license
9574      questions now is the OpenSSL core team under openssl-core@openssl.org.
9575      And add a paragraph about the dual-license situation to make sure people
9576      recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
9577      to the OpenSSL toolkit.
9578      [Ralf S. Engelschall]
9579 
9580   *) General source tree makefile cleanups: Made `making xxx in yyy...'
9581      display consistent in the source tree and replaced `/bin/rm' by `rm'.
9582      Additonally cleaned up the `make links' target: Remove unnecessary
9583      semicolons, subsequent redundant removes, inline point.sh into mklink.sh
9584      to speed processing and no longer clutter the display with confusing
9585      stuff. Instead only the actually done links are displayed.
9586      [Ralf S. Engelschall]
9587 
9588   *) Permit null encryption ciphersuites, used for authentication only. It used
9589      to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
9590      It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
9591      encryption.
9592      [Ben Laurie]
9593 
9594   *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
9595      signed attributes when verifying signatures (this would break them), 
9596      the detached data encoding was wrong and public keys obtained using
9597      X509_get_pubkey() weren't freed.
9598      [Steve Henson]
9599 
9600   *) Add text documentation for the BUFFER functions. Also added a work around
9601      to a Win95 console bug. This was triggered by the password read stuff: the
9602      last character typed gets carried over to the next fread(). If you were 
9603      generating a new cert request using 'req' for example then the last
9604      character of the passphrase would be CR which would then enter the first
9605      field as blank.
9606      [Steve Henson]
9607 
9608   *) Added the new `Includes OpenSSL Cryptography Software' button as
9609      doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
9610      button and can be used by applications based on OpenSSL to show the
9611      relationship to the OpenSSL project.  
9612      [Ralf S. Engelschall]
9613 
9614   *) Remove confusing variables in function signatures in files
9615      ssl/ssl_lib.c and ssl/ssl.h.
9616      [Lennart Bong <lob@kulthea.stacken.kth.se>]
9617 
9618   *) Don't install bss_file.c under PREFIX/include/
9619      [Lennart Bong <lob@kulthea.stacken.kth.se>]
9620 
9621   *) Get the Win32 compile working again. Modify mkdef.pl so it can handle
9622      functions that return function pointers and has support for NT specific
9623      stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
9624      #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
9625      unsigned to signed types: this was killing the Win32 compile.
9626      [Steve Henson]
9627 
9628   *) Add new certificate file to stack functions,
9629      SSL_add_dir_cert_subjects_to_stack() and
9630      SSL_add_file_cert_subjects_to_stack().  These largely supplant
9631      SSL_load_client_CA_file(), and can be used to add multiple certs easily
9632      to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
9633      This means that Apache-SSL and similar packages don't have to mess around
9634      to add as many CAs as they want to the preferred list.
9635      [Ben Laurie]
9636 
9637   *) Experiment with doxygen documentation. Currently only partially applied to
9638      ssl/ssl_lib.c.
9639      See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with
9640      openssl.doxy as the configuration file.
9641      [Ben Laurie]
9642   
9643   *) Get rid of remaining C++-style comments which strict C compilers hate.
9644      [Ralf S. Engelschall, pointed out by Carlos Amengual]
9645 
9646   *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
9647      compiled in by default: it has problems with large keys.
9648      [Steve Henson]
9649 
9650   *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
9651      DH private keys and/or callback functions which directly correspond to
9652      their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
9653      is needed for applications which have to configure certificates on a
9654      per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
9655      (e.g. s_server). 
9656         For the RSA certificate situation is makes no difference, but
9657      for the DSA certificate situation this fixes the "no shared cipher"
9658      problem where the OpenSSL cipher selection procedure failed because the
9659      temporary keys were not overtaken from the context and the API provided
9660      no way to reconfigure them. 
9661         The new functions now let applications reconfigure the stuff and they
9662      are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
9663      SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback.  Additionally a new
9664      non-public-API function ssl_cert_instantiate() is used as a helper
9665      function and also to reduce code redundancy inside ssl_rsa.c.
9666      [Ralf S. Engelschall]
9667 
9668   *) Move s_server -dcert and -dkey options out of the undocumented feature
9669      area because they are useful for the DSA situation and should be
9670      recognized by the users.
9671      [Ralf S. Engelschall]
9672 
9673   *) Fix the cipher decision scheme for export ciphers: the export bits are
9674      *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
9675      SSL_EXP_MASK.  So, the original variable has to be used instead of the
9676      already masked variable.
9677      [Richard Levitte <levitte@stacken.kth.se>]
9678 
9679   *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c
9680      [Richard Levitte <levitte@stacken.kth.se>]
9681 
9682   *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
9683      from `int' to `unsigned int' because it's a length and initialized by
9684      EVP_DigestFinal() which expects an `unsigned int *'.
9685      [Richard Levitte <levitte@stacken.kth.se>]
9686 
9687   *) Don't hard-code path to Perl interpreter on shebang line of Configure
9688      script. Instead use the usual Shell->Perl transition trick.
9689      [Ralf S. Engelschall]
9690 
9691   *) Make `openssl x509 -noout -modulus' functional also for DSA certificates
9692      (in addition to RSA certificates) to match the behaviour of `openssl dsa
9693      -noout -modulus' as it's already the case for `openssl rsa -noout
9694      -modulus'.  For RSA the -modulus is the real "modulus" while for DSA
9695      currently the public key is printed (a decision which was already done by
9696      `openssl dsa -modulus' in the past) which serves a similar purpose.
9697      Additionally the NO_RSA no longer completely removes the whole -modulus
9698      option; it now only avoids using the RSA stuff. Same applies to NO_DSA
9699      now, too.
9700      [Ralf S.  Engelschall]
9701 
9702   *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
9703      BIO. See the source (crypto/evp/bio_ok.c) for more info.
9704      [Arne Ansper <arne@ats.cyber.ee>]
9705 
9706   *) Dump the old yucky req code that tried (and failed) to allow raw OIDs
9707      to be added. Now both 'req' and 'ca' can use new objects defined in the
9708      config file.
9709      [Steve Henson]
9710 
9711   *) Add cool BIO that does syslog (or event log on NT).
9712      [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie]
9713 
9714   *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
9715      TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
9716      TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
9717      Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
9718      [Ben Laurie]
9719 
9720   *) Add preliminary config info for new extension code.
9721      [Steve Henson]
9722 
9723   *) Make RSA_NO_PADDING really use no padding.
9724      [Ulf Moeller <ulf@fitug.de>]
9725 
9726   *) Generate errors when private/public key check is done.
9727      [Ben Laurie]
9728 
9729   *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
9730      for some CRL extensions and new objects added.
9731      [Steve Henson]
9732 
9733   *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private
9734      key usage extension and fuller support for authority key id.
9735      [Steve Henson]
9736 
9737   *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
9738      padding method for RSA, which is recommended for new applications in PKCS
9739      #1 v2.0 (RFC 2437, October 1998).
9740      OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
9741      foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
9742      against Bleichbacher's attack on RSA.
9743      [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
9744       Ben Laurie]
9745 
9746   *) Updates to the new SSL compression code
9747      [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9748 
9749   *) Fix so that the version number in the master secret, when passed
9750      via RSA, checks that if TLS was proposed, but we roll back to SSLv3
9751      (because the server will not accept higher), that the version number
9752      is 0x03,0x01, not 0x03,0x00
9753      [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9754 
9755   *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory
9756      leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes
9757      in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c
9758      [Steve Henson]
9759 
9760   *) Support for RAW extensions where an arbitrary extension can be
9761      created by including its DER encoding. See apps/openssl.cnf for
9762      an example.
9763      [Steve Henson]
9764 
9765   *) Make sure latest Perl versions don't interpret some generated C array
9766      code as Perl array code in the crypto/err/err_genc.pl script.
9767      [Lars Weber <3weber@informatik.uni-hamburg.de>]
9768 
9769   *) Modify ms/do_ms.bat to not generate assembly language makefiles since
9770      not many people have the assembler. Various Win32 compilation fixes and
9771      update to the INSTALL.W32 file with (hopefully) more accurate Win32
9772      build instructions.
9773      [Steve Henson]
9774 
9775   *) Modify configure script 'Configure' to automatically create crypto/date.h
9776      file under Win32 and also build pem.h from pem.org. New script
9777      util/mkfiles.pl to create the MINFO file on environments that can't do a
9778      'make files': perl util/mkfiles.pl >MINFO should work.
9779      [Steve Henson]
9780 
9781   *) Major rework of DES function declarations, in the pursuit of correctness
9782      and purity. As a result, many evil casts evaporated, and some weirdness,
9783      too. You may find this causes warnings in your code. Zapping your evil
9784      casts will probably fix them. Mostly.
9785      [Ben Laurie]
9786 
9787   *) Fix for a typo in asn1.h. Bug fix to object creation script
9788      obj_dat.pl. It considered a zero in an object definition to mean
9789      "end of object": none of the objects in objects.h have any zeros
9790      so it wasn't spotted.
9791      [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>]
9792 
9793   *) Add support for Triple DES Cipher Block Chaining with Output Feedback
9794      Masking (CBCM). In the absence of test vectors, the best I have been able
9795      to do is check that the decrypt undoes the encrypt, so far. Send me test
9796      vectors if you have them.
9797      [Ben Laurie]
9798 
9799   *) Correct calculation of key length for export ciphers (too much space was
9800      allocated for null ciphers). This has not been tested!
9801      [Ben Laurie]
9802 
9803   *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage
9804      message is now correct (it understands "crypto" and "ssl" on its
9805      command line). There is also now an "update" option. This will update
9806      the util/ssleay.num and util/libeay.num files with any new functions.
9807      If you do a: 
9808      perl util/mkdef.pl crypto ssl update
9809      it will update them.
9810      [Steve Henson]
9811 
9812   *) Overhauled the Perl interface (perl/*):
9813      - ported BN stuff to OpenSSL's different BN library
9814      - made the perl/ source tree CVS-aware
9815      - renamed the package from SSLeay to OpenSSL (the files still contain
9816        their history because I've copied them in the repository)
9817      - removed obsolete files (the test scripts will be replaced
9818        by better Test::Harness variants in the future)
9819      [Ralf S. Engelschall]
9820 
9821   *) First cut for a very conservative source tree cleanup:
9822      1. merge various obsolete readme texts into doc/ssleay.txt
9823      where we collect the old documents and readme texts.
9824      2. remove the first part of files where I'm already sure that we no
9825      longer need them because of three reasons: either they are just temporary
9826      files which were left by Eric or they are preserved original files where
9827      I've verified that the diff is also available in the CVS via "cvs diff
9828      -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
9829      the crypto/md/ stuff).
9830      [Ralf S. Engelschall]
9831 
9832   *) More extension code. Incomplete support for subject and issuer alt
9833      name, issuer and authority key id. Change the i2v function parameters
9834      and add an extra 'crl' parameter in the X509V3_CTX structure: guess
9835      what that's for :-) Fix to ASN1 macro which messed up
9836      IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
9837      [Steve Henson]
9838 
9839   *) Preliminary support for ENUMERATED type. This is largely copied from the
9840      INTEGER code.
9841      [Steve Henson]
9842 
9843   *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
9844      [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9845 
9846   *) Make sure `make rehash' target really finds the `openssl' program.
9847      [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
9848 
9849   *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
9850      like to hear about it if this slows down other processors.
9851      [Ben Laurie]
9852 
9853   *) Add CygWin32 platform information to Configure script.
9854      [Alan Batie <batie@aahz.jf.intel.com>]
9855 
9856   *) Fixed ms/32all.bat script: `no_asm' -> `no-asm'
9857      [Rainer W. Gerling <gerling@mpg-gv.mpg.de>]
9858   
9859   *) New program nseq to manipulate netscape certificate sequences
9860      [Steve Henson]
9861 
9862   *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
9863      few typos.
9864      [Steve Henson]
9865 
9866   *) Fixes to BN code.  Previously the default was to define BN_RECURSION
9867      but the BN code had some problems that would cause failures when
9868      doing certificate verification and some other functions.
9869      [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
9870 
9871   *) Add ASN1 and PEM code to support netscape certificate sequences.
9872      [Steve Henson]
9873 
9874   *) Add ASN1 and PEM code to support netscape certificate sequences.
9875      [Steve Henson]
9876 
9877   *) Add several PKIX and private extended key usage OIDs.
9878      [Steve Henson]
9879 
9880   *) Modify the 'ca' program to handle the new extension code. Modify
9881      openssl.cnf for new extension format, add comments.
9882      [Steve Henson]
9883 
9884   *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
9885      and add a sample to openssl.cnf so req -x509 now adds appropriate
9886      CA extensions.
9887      [Steve Henson]
9888 
9889   *) Continued X509 V3 changes. Add to other makefiles, integrate with the
9890      error code, add initial support to X509_print() and x509 application.
9891      [Steve Henson]
9892 
9893   *) Takes a deep breath and start addding X509 V3 extension support code. Add
9894      files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
9895      stuff is currently isolated and isn't even compiled yet.
9896      [Steve Henson]
9897 
9898   *) Continuing patches for GeneralizedTime. Fix up certificate and CRL
9899      ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
9900      Removed the versions check from X509 routines when loading extensions:
9901      this allows certain broken certificates that don't set the version
9902      properly to be processed.
9903      [Steve Henson]
9904 
9905   *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
9906      Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
9907      can still be regenerated with "make depend".
9908      [Ben Laurie]
9909 
9910   *) Spelling mistake in C version of CAST-128.
9911      [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>]
9912 
9913   *) Changes to the error generation code. The perl script err-code.pl 
9914      now reads in the old error codes and retains the old numbers, only
9915      adding new ones if necessary. It also only changes the .err files if new
9916      codes are added. The makefiles have been modified to only insert errors
9917      when needed (to avoid needlessly modifying header files). This is done
9918      by only inserting errors if the .err file is newer than the auto generated
9919      C file. To rebuild all the error codes from scratch (the old behaviour)
9920      either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
9921      or delete all the .err files.
9922      [Steve Henson]
9923 
9924   *) CAST-128 was incorrectly implemented for short keys. The C version has
9925      been fixed, but is untested. The assembler versions are also fixed, but
9926      new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
9927      to regenerate it if needed.
9928      [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
9929       Hagino <itojun@kame.net>]
9930 
9931   *) File was opened incorrectly in randfile.c.
9932      [Ulf Möller <ulf@fitug.de>]
9933 
9934   *) Beginning of support for GeneralizedTime. d2i, i2d, check and print
9935      functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
9936      GeneralizedTime. ASN1_TIME is the proper type used in certificates et
9937      al: it's just almost always a UTCTime. Note this patch adds new error
9938      codes so do a "make errors" if there are problems.
9939      [Steve Henson]
9940 
9941   *) Correct Linux 1 recognition in config.
9942      [Ulf Möller <ulf@fitug.de>]
9943 
9944   *) Remove pointless MD5 hash when using DSA keys in ca.
9945      [Anonymous <nobody@replay.com>]
9946 
9947   *) Generate an error if given an empty string as a cert directory. Also
9948      generate an error if handed NULL (previously returned 0 to indicate an
9949      error, but didn't set one).
9950      [Ben Laurie, reported by Anonymous <nobody@replay.com>]
9951 
9952   *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
9953      [Ben Laurie]
9954 
9955   *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
9956      parameters. This was causing a warning which killed off the Win32 compile.
9957      [Steve Henson]
9958 
9959   *) Remove C++ style comments from crypto/bn/bn_local.h.
9960      [Neil Costigan <neil.costigan@celocom.com>]
9961 
9962   *) The function OBJ_txt2nid was broken. It was supposed to return a nid
9963      based on a text string, looking up short and long names and finally
9964      "dot" format. The "dot" format stuff didn't work. Added new function
9965      OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote 
9966      OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
9967      OID is not part of the table.
9968      [Steve Henson]
9969 
9970   *) Add prototypes to X509 lookup/verify methods, fixing a bug in
9971      X509_LOOKUP_by_alias().
9972      [Ben Laurie]
9973 
9974   *) Sort openssl functions by name.
9975      [Ben Laurie]
9976 
9977   *) Get the gendsa program working (hopefully) and add it to app list. Remove
9978      encryption from sample DSA keys (in case anyone is interested the password
9979      was "1234").
9980      [Steve Henson]
9981 
9982   *) Make _all_ *_free functions accept a NULL pointer.
9983      [Frans Heymans <fheymans@isaserver.be>]
9984 
9985   *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use
9986      NULL pointers.
9987      [Anonymous <nobody@replay.com>]
9988 
9989   *) s_server should send the CAfile as acceptable CAs, not its own cert.
9990      [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
9991 
9992   *) Don't blow it for numeric -newkey arguments to apps/req.
9993      [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
9994 
9995   *) Temp key "for export" tests were wrong in s3_srvr.c.
9996      [Anonymous <nobody@replay.com>]
9997 
9998   *) Add prototype for temp key callback functions
9999      SSL_CTX_set_tmp_{rsa,dh}_callback().
10000      [Ben Laurie]
10001 
10002   *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
10003      DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
10004      [Steve Henson]
10005 
10006   *) X509_name_add_entry() freed the wrong thing after an error.
10007      [Arne Ansper <arne@ats.cyber.ee>]
10008 
10009   *) rsa_eay.c would attempt to free a NULL context.
10010      [Arne Ansper <arne@ats.cyber.ee>]
10011 
10012   *) BIO_s_socket() had a broken should_retry() on Windoze.
10013      [Arne Ansper <arne@ats.cyber.ee>]
10014 
10015   *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
10016      [Arne Ansper <arne@ats.cyber.ee>]
10017 
10018   *) Make sure the already existing X509_STORE->depth variable is initialized
10019      in X509_STORE_new(), but document the fact that this variable is still
10020      unused in the certificate verification process.
10021      [Ralf S. Engelschall]
10022 
10023   *) Fix the various library and apps files to free up pkeys obtained from
10024      X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
10025      [Steve Henson]
10026 
10027   *) Fix reference counting in X509_PUBKEY_get(). This makes
10028      demos/maurice/example2.c work, amongst others, probably.
10029      [Steve Henson and Ben Laurie]
10030 
10031   *) First cut of a cleanup for apps/. First the `ssleay' program is now named
10032      `openssl' and second, the shortcut symlinks for the `openssl <command>'
10033      are no longer created. This way we have a single and consistent command
10034      line interface `openssl <command>', similar to `cvs <command>'.
10035      [Ralf S. Engelschall, Paul Sutton and Ben Laurie]
10036 
10037   *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
10038      BIT STRING wrapper always have zero unused bits.
10039      [Steve Henson]
10040 
10041   *) Add CA.pl, perl version of CA.sh, add extended key usage OID.
10042      [Steve Henson]
10043 
10044   *) Make the top-level INSTALL documentation easier to understand.
10045      [Paul Sutton]
10046 
10047   *) Makefiles updated to exit if an error occurs in a sub-directory
10048      make (including if user presses ^C) [Paul Sutton]
10049 
10050   *) Make Montgomery context stuff explicit in RSA data structure.
10051      [Ben Laurie]
10052 
10053   *) Fix build order of pem and err to allow for generated pem.h.
10054      [Ben Laurie]
10055 
10056   *) Fix renumbering bug in X509_NAME_delete_entry().
10057      [Ben Laurie]
10058 
10059   *) Enhanced the err-ins.pl script so it makes the error library number 
10060      global and can add a library name. This is needed for external ASN1 and
10061      other error libraries.
10062      [Steve Henson]
10063 
10064   *) Fixed sk_insert which never worked properly.
10065      [Steve Henson]
10066 
10067   *) Fix ASN1 macros so they can handle indefinite length construted 
10068      EXPLICIT tags. Some non standard certificates use these: they can now
10069      be read in.
10070      [Steve Henson]
10071 
10072   *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
10073      into a single doc/ssleay.txt bundle. This way the information is still
10074      preserved but no longer messes up this directory. Now it's new room for
10075      the new set of documenation files.
10076      [Ralf S. Engelschall]
10077 
10078   *) SETs were incorrectly DER encoded. This was a major pain, because they
10079      shared code with SEQUENCEs, which aren't coded the same. This means that
10080      almost everything to do with SETs or SEQUENCEs has either changed name or
10081      number of arguments.
10082      [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>]
10083 
10084   *) Fix test data to work with the above.
10085      [Ben Laurie]
10086 
10087   *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
10088      was already fixed by Eric for 0.9.1 it seems.
10089      [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>]
10090 
10091   *) Autodetect FreeBSD3.
10092      [Ben Laurie]
10093 
10094   *) Fix various bugs in Configure. This affects the following platforms:
10095      nextstep
10096      ncr-scde
10097      unixware-2.0
10098      unixware-2.0-pentium
10099      sco5-cc.
10100      [Ben Laurie]
10101 
10102   *) Eliminate generated files from CVS. Reorder tests to regenerate files
10103      before they are needed.
10104      [Ben Laurie]
10105 
10106   *) Generate Makefile.ssl from Makefile.org (to keep CVS happy).
10107      [Ben Laurie]
10108 
10109 
10110  Changes between 0.9.1b and 0.9.1c  [23-Dec-1998]
10111 
10112   *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and 
10113      changed SSLeay to OpenSSL in version strings.
10114      [Ralf S. Engelschall]
10115   
10116   *) Some fixups to the top-level documents.
10117      [Paul Sutton]
10118 
10119   *) Fixed the nasty bug where rsaref.h was not found under compile-time
10120      because the symlink to include/ was missing.
10121      [Ralf S. Engelschall]
10122 
10123   *) Incorporated the popular no-RSA/DSA-only patches 
10124      which allow to compile a RSA-free SSLeay.
10125      [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall]
10126 
10127   *) Fixed nasty rehash problem under `make -f Makefile.ssl links'
10128      when "ssleay" is still not found.
10129      [Ralf S. Engelschall]
10130 
10131   *) Added more platforms to Configure: Cray T3E, HPUX 11, 
10132      [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>]
10133 
10134   *) Updated the README file.
10135      [Ralf S. Engelschall]
10136 
10137   *) Added various .cvsignore files in the CVS repository subdirs
10138      to make a "cvs update" really silent.
10139      [Ralf S. Engelschall]
10140 
10141   *) Recompiled the error-definition header files and added
10142      missing symbols to the Win32 linker tables.
10143      [Ralf S. Engelschall]
10144 
10145   *) Cleaned up the top-level documents;
10146      o new files: CHANGES and LICENSE
10147      o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay 
10148      o merged COPYRIGHT into LICENSE
10149      o removed obsolete TODO file
10150      o renamed MICROSOFT to INSTALL.W32
10151      [Ralf S. Engelschall]
10152 
10153   *) Removed dummy files from the 0.9.1b source tree: 
10154      crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
10155      crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
10156      crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
10157      crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
10158      util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
10159      [Ralf S. Engelschall]
10160 
10161   *) Added various platform portability fixes.
10162      [Mark J. Cox]
10163 
10164   *) The Genesis of the OpenSSL rpject:
10165      We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
10166      Young and Tim J. Hudson created while they were working for C2Net until
10167      summer 1998.
10168      [The OpenSSL Project]
10169  
10170 
10171  Changes between 0.9.0b and 0.9.1b  [not released]
10172 
10173   *) Updated a few CA certificates under certs/
10174      [Eric A. Young]
10175 
10176   *) Changed some BIGNUM api stuff.
10177      [Eric A. Young]
10178 
10179   *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD, 
10180      DGUX x86, Linux Alpha, etc.
10181      [Eric A. Young]
10182 
10183   *) New COMP library [crypto/comp/] for SSL Record Layer Compression: 
10184      RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
10185      available).
10186      [Eric A. Young]
10187 
10188   *) Add -strparse option to asn1pars program which parses nested 
10189      binary structures 
10190      [Dr Stephen Henson <shenson@bigfoot.com>]
10191 
10192   *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
10193      [Eric A. Young]
10194 
10195   *) DSA fix for "ca" program.
10196      [Eric A. Young]
10197 
10198   *) Added "-genkey" option to "dsaparam" program.
10199      [Eric A. Young]
10200 
10201   *) Added RIPE MD160 (rmd160) message digest.
10202      [Eric A. Young]
10203 
10204   *) Added -a (all) option to "ssleay version" command.
10205      [Eric A. Young]
10206 
10207   *) Added PLATFORM define which is the id given to Configure.
10208      [Eric A. Young]
10209 
10210   *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
10211      [Eric A. Young]
10212 
10213   *) Extended the ASN.1 parser routines.
10214      [Eric A. Young]
10215 
10216   *) Extended BIO routines to support REUSEADDR, seek, tell, etc.
10217      [Eric A. Young]
10218 
10219   *) Added a BN_CTX to the BN library.
10220      [Eric A. Young]
10221 
10222   *) Fixed the weak key values in DES library
10223      [Eric A. Young]
10224 
10225   *) Changed API in EVP library for cipher aliases.
10226      [Eric A. Young]
10227 
10228   *) Added support for RC2/64bit cipher.
10229      [Eric A. Young]
10230 
10231   *) Converted the lhash library to the crypto/mem.c functions.
10232      [Eric A. Young]
10233 
10234   *) Added more recognized ASN.1 object ids.
10235      [Eric A. Young]
10236 
10237   *) Added more RSA padding checks for SSL/TLS.
10238      [Eric A. Young]
10239 
10240   *) Added BIO proxy/filter functionality.
10241      [Eric A. Young]
10242 
10243   *) Added extra_certs to SSL_CTX which can be used
10244      send extra CA certificates to the client in the CA cert chain sending
10245      process. It can be configured with SSL_CTX_add_extra_chain_cert().
10246      [Eric A. Young]
10247 
10248   *) Now Fortezza is denied in the authentication phase because
10249      this is key exchange mechanism is not supported by SSLeay at all.
10250      [Eric A. Young]
10251 
10252   *) Additional PKCS1 checks.
10253      [Eric A. Young]
10254 
10255   *) Support the string "TLSv1" for all TLS v1 ciphers.
10256      [Eric A. Young]
10257 
10258   *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
10259      ex_data index of the SSL context in the X509_STORE_CTX ex_data.
10260      [Eric A. Young]
10261 
10262   *) Fixed a few memory leaks.
10263      [Eric A. Young]
10264 
10265   *) Fixed various code and comment typos.
10266      [Eric A. Young]
10267 
10268   *) A minor bug in ssl/s3_clnt.c where there would always be 4 0 
10269      bytes sent in the client random.
10270      [Edward Bishop <ebishop@spyglass.com>]
10271