Home | History | Annotate | Line # | Download | only in dist
      1 
      2  OpenSSL CHANGES
      3  _______________
      4 
      5  Changes between 1.0.1r and 1.0.1s [1 Mar 2016]
      6 
      7   * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
      8     Builds that are not configured with "enable-weak-ssl-ciphers" will not
      9     provide any "EXPORT" or "LOW" strength ciphers.
     10     [Viktor Dukhovni]
     11 
     12   * Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
     13     is by default disabled at build-time.  Builds that are not configured with
     14     "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
     15     users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
     16     will need to explicitly call either of:
     17 
     18         SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
     19     or
     20         SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
     21 
     22     as appropriate.  Even if either of those is used, or the application
     23     explicitly uses the version-specific SSLv2_method() or its client and
     24     server variants, SSLv2 ciphers vulnerable to exhaustive search key
     25     recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
     26     ciphers, and SSLv2 56-bit DES are no longer available.
     27     (CVE-2016-0800)
     28     [Viktor Dukhovni]
     29 
     30   *) Fix a double-free in DSA code
     31 
     32      A double free bug was discovered when OpenSSL parses malformed DSA private
     33      keys and could lead to a DoS attack or memory corruption for applications
     34      that receive DSA private keys from untrusted sources.  This scenario is
     35      considered rare.
     36 
     37      This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
     38      libFuzzer.
     39      (CVE-2016-0705)
     40      [Stephen Henson]
     41 
     42   *) Disable SRP fake user seed to address a server memory leak.
     43 
     44      Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
     45 
     46      SRP_VBASE_get_by_user had inconsistent memory management behaviour.
     47      In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
     48      was changed to ignore the "fake user" SRP seed, even if the seed
     49      is configured.
     50 
     51      Users should use SRP_VBASE_get1_by_user instead. Note that in
     52      SRP_VBASE_get1_by_user, caller must free the returned value. Note
     53      also that even though configuring the SRP seed attempts to hide
     54      invalid usernames by continuing the handshake with fake
     55      credentials, this behaviour is not constant time and no strong
     56      guarantees are made that the handshake is indistinguishable from
     57      that of a valid user.
     58      (CVE-2016-0798)
     59      [Emilia Ksper]
     60 
     61   *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
     62 
     63      In the BN_hex2bn function the number of hex digits is calculated using an
     64      int value |i|. Later |bn_expand| is called with a value of |i * 4|. For
     65      large values of |i| this can result in |bn_expand| not allocating any
     66      memory because |i * 4| is negative. This can leave the internal BIGNUM data
     67      field as NULL leading to a subsequent NULL ptr deref. For very large values
     68      of |i|, the calculation |i * 4| could be a positive value smaller than |i|.
     69      In this case memory is allocated to the internal BIGNUM data field, but it
     70      is insufficiently sized leading to heap corruption. A similar issue exists
     71      in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
     72      is ever called by user applications with very large untrusted hex/dec data.
     73      This is anticipated to be a rare occurrence.
     74 
     75      All OpenSSL internal usage of these functions use data that is not expected
     76      to be untrusted, e.g. config file data or application command line
     77      arguments. If user developed applications generate config file data based
     78      on untrusted data then it is possible that this could also lead to security
     79      consequences. This is also anticipated to be rare.
     80 
     81      This issue was reported to OpenSSL by Guido Vranken.
     82      (CVE-2016-0797)
     83      [Matt Caswell]
     84 
     85   *) Fix memory issues in BIO_*printf functions
     86 
     87      The internal |fmtstr| function used in processing a "%s" format string in
     88      the BIO_*printf functions could overflow while calculating the length of a
     89      string and cause an OOB read when printing very long strings.
     90 
     91      Additionally the internal |doapr_outch| function can attempt to write to an
     92      OOB memory location (at an offset from the NULL pointer) in the event of a
     93      memory allocation failure. In 1.0.2 and below this could be caused where
     94      the size of a buffer to be allocated is greater than INT_MAX. E.g. this
     95      could be in processing a very long "%s" format string. Memory leaks can
     96      also occur.
     97 
     98      The first issue may mask the second issue dependent on compiler behaviour.
     99      These problems could enable attacks where large amounts of untrusted data
    100      is passed to the BIO_*printf functions. If applications use these functions
    101      in this way then they could be vulnerable. OpenSSL itself uses these
    102      functions when printing out human-readable dumps of ASN.1 data. Therefore
    103      applications that print this data could be vulnerable if the data is from
    104      untrusted sources. OpenSSL command line applications could also be
    105      vulnerable where they print out ASN.1 data, or if untrusted data is passed
    106      as command line arguments.
    107 
    108      Libssl is not considered directly vulnerable. Additionally certificates etc
    109      received via remote connections via libssl are also unlikely to be able to
    110      trigger these issues because of message size limits enforced within libssl.
    111 
    112      This issue was reported to OpenSSL Guido Vranken.
    113      (CVE-2016-0799)
    114      [Matt Caswell]
    115 
    116   *) Side channel attack on modular exponentiation
    117 
    118      A side-channel attack was found which makes use of cache-bank conflicts on
    119      the Intel Sandy-Bridge microarchitecture which could lead to the recovery
    120      of RSA keys.  The ability to exploit this issue is limited as it relies on
    121      an attacker who has control of code in a thread running on the same
    122      hyper-threaded core as the victim thread which is performing decryptions.
    123 
    124      This issue was reported to OpenSSL by Yuval Yarom, The University of
    125      Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
    126      Nadia Heninger, University of Pennsylvania with more information at
    127      http://cachebleed.info.
    128      (CVE-2016-0702)
    129      [Andy Polyakov]
    130 
    131   *) Change the req app to generate a 2048-bit RSA/DSA key by default,
    132      if no keysize is specified with default_bits. This fixes an
    133      omission in an earlier change that changed all RSA/DSA key generation
    134      apps to use 2048 bits by default.
    135      [Emilia Ksper]
    136 
    137  Changes between 1.0.1q and 1.0.1r [28 Jan 2016]
    138 
    139   *) Protection for DH small subgroup attacks
    140 
    141      As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been
    142      switched on by default and cannot be disabled. This could have some
    143      performance impact.
    144      [Matt Caswell]
    145 
    146   *) SSLv2 doesn't block disabled ciphers
    147 
    148      A malicious client can negotiate SSLv2 ciphers that have been disabled on
    149      the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
    150      been disabled, provided that the SSLv2 protocol was not also disabled via
    151      SSL_OP_NO_SSLv2.
    152 
    153      This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
    154      and Sebastian Schinzel.
    155      (CVE-2015-3197)
    156      [Viktor Dukhovni]
    157 
    158   *) Reject DH handshakes with parameters shorter than 1024 bits.
    159      [Kurt Roeckx]
    160 
    161  Changes between 1.0.1p and 1.0.1q [3 Dec 2015]
    162 
    163   *) Certificate verify crash with missing PSS parameter
    164 
    165      The signature verification routines will crash with a NULL pointer
    166      dereference if presented with an ASN.1 signature using the RSA PSS
    167      algorithm and absent mask generation function parameter. Since these
    168      routines are used to verify certificate signature algorithms this can be
    169      used to crash any certificate verification operation and exploited in a
    170      DoS attack. Any application which performs certificate verification is
    171      vulnerable including OpenSSL clients and servers which enable client
    172      authentication.
    173 
    174      This issue was reported to OpenSSL by Loc Jonas Etienne (Qnective AG).
    175      (CVE-2015-3194)
    176      [Stephen Henson]
    177 
    178   *) X509_ATTRIBUTE memory leak
    179 
    180      When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
    181      memory. This structure is used by the PKCS#7 and CMS routines so any
    182      application which reads PKCS#7 or CMS data from untrusted sources is
    183      affected. SSL/TLS is not affected.
    184 
    185      This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
    186      libFuzzer.
    187      (CVE-2015-3195)
    188      [Stephen Henson]
    189 
    190   *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
    191      This changes the decoding behaviour for some invalid messages,
    192      though the change is mostly in the more lenient direction, and
    193      legacy behaviour is preserved as much as possible.
    194      [Emilia Ksper]
    195 
    196   *) In DSA_generate_parameters_ex, if the provided seed is too short,
    197      use a random seed, as already documented.
    198      [Rich Salz and Ismo Puustinen <ismo.puustinen (a] intel.com>]
    199 
    200  Changes between 1.0.1o and 1.0.1p [9 Jul 2015]
    201 
    202   *) Alternate chains certificate forgery
    203 
    204      During certificate verfification, OpenSSL will attempt to find an
    205      alternative certificate chain if the first attempt to build such a chain
    206      fails. An error in the implementation of this logic can mean that an
    207      attacker could cause certain checks on untrusted certificates to be
    208      bypassed, such as the CA flag, enabling them to use a valid leaf
    209      certificate to act as a CA and "issue" an invalid certificate.
    210 
    211      This issue was reported to OpenSSL by Adam Langley/David Benjamin
    212      (Google/BoringSSL).
    213      (CVE-2015-1793)
    214      [Matt Caswell]
    215 
    216   *) Race condition handling PSK identify hint
    217 
    218      If PSK identity hints are received by a multi-threaded client then
    219      the values are wrongly updated in the parent SSL_CTX structure. This can
    220      result in a race condition potentially leading to a double free of the
    221      identify hint data.
    222      (CVE-2015-3196)
    223      [Stephen Henson]
    224 
    225  Changes between 1.0.1n and 1.0.1o [12 Jun 2015]
    226   *) Fix HMAC ABI incompatibility. The previous version introduced an ABI
    227      incompatibility in the handling of HMAC. The previous ABI has now been
    228      restored.
    229 
    230  Changes between 1.0.1m and 1.0.1n [11 Jun 2015]
    231 
    232   *) Malformed ECParameters causes infinite loop
    233 
    234      When processing an ECParameters structure OpenSSL enters an infinite loop
    235      if the curve specified is over a specially malformed binary polynomial
    236      field.
    237 
    238      This can be used to perform denial of service against any
    239      system which processes public keys, certificate requests or
    240      certificates.  This includes TLS clients and TLS servers with
    241      client authentication enabled.
    242 
    243      This issue was reported to OpenSSL by Joseph Barr-Pixton.
    244      (CVE-2015-1788)
    245      [Andy Polyakov]
    246 
    247   *) Exploitable out-of-bounds read in X509_cmp_time
    248 
    249      X509_cmp_time does not properly check the length of the ASN1_TIME
    250      string and can read a few bytes out of bounds. In addition,
    251      X509_cmp_time accepts an arbitrary number of fractional seconds in the
    252      time string.
    253 
    254      An attacker can use this to craft malformed certificates and CRLs of
    255      various sizes and potentially cause a segmentation fault, resulting in
    256      a DoS on applications that verify certificates or CRLs. TLS clients
    257      that verify CRLs are affected. TLS clients and servers with client
    258      authentication enabled may be affected if they use custom verification
    259      callbacks.
    260 
    261      This issue was reported to OpenSSL by Robert Swiecki (Google), and
    262      independently by Hanno Bck.
    263      (CVE-2015-1789)
    264      [Emilia Ksper]
    265 
    266   *) PKCS7 crash with missing EnvelopedContent
    267 
    268      The PKCS#7 parsing code does not handle missing inner EncryptedContent
    269      correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
    270      with missing content and trigger a NULL pointer dereference on parsing.
    271 
    272      Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
    273      structures from untrusted sources are affected. OpenSSL clients and
    274      servers are not affected.
    275 
    276      This issue was reported to OpenSSL by Michal Zalewski (Google).
    277      (CVE-2015-1790)
    278      [Emilia Ksper]
    279 
    280   *) CMS verify infinite loop with unknown hash function
    281 
    282      When verifying a signedData message the CMS code can enter an infinite loop
    283      if presented with an unknown hash function OID. This can be used to perform
    284      denial of service against any system which verifies signedData messages using
    285      the CMS code.
    286      This issue was reported to OpenSSL by Johannes Bauer.
    287      (CVE-2015-1792)
    288      [Stephen Henson]
    289 
    290   *) Race condition handling NewSessionTicket
    291 
    292      If a NewSessionTicket is received by a multi-threaded client when attempting to
    293      reuse a previous ticket then a race condition can occur potentially leading to
    294      a double free of the ticket data.
    295      (CVE-2015-1791)
    296      [Matt Caswell]
    297 
    298   *) Reject DH handshakes with parameters shorter than 768 bits.
    299      [Kurt Roeckx and Emilia Kasper]
    300 
    301   *) dhparam: generate 2048-bit parameters by default.
    302      [Kurt Roeckx and Emilia Kasper]
    303 
    304  Changes between 1.0.1l and 1.0.1m [19 Mar 2015]
    305 
    306   *) Segmentation fault in ASN1_TYPE_cmp fix
    307 
    308      The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
    309      made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
    310      certificate signature algorithm consistency this can be used to crash any
    311      certificate verification operation and exploited in a DoS attack. Any
    312      application which performs certificate verification is vulnerable including
    313      OpenSSL clients and servers which enable client authentication.
    314      (CVE-2015-0286)
    315      [Stephen Henson]
    316 
    317   *) ASN.1 structure reuse memory corruption fix
    318 
    319      Reusing a structure in ASN.1 parsing may allow an attacker to cause
    320      memory corruption via an invalid write. Such reuse is and has been
    321      strongly discouraged and is believed to be rare.
    322 
    323      Applications that parse structures containing CHOICE or ANY DEFINED BY
    324      components may be affected. Certificate parsing (d2i_X509 and related
    325      functions) are however not affected. OpenSSL clients and servers are
    326      not affected.
    327      (CVE-2015-0287)
    328      [Stephen Henson]
    329 
    330   *) PKCS7 NULL pointer dereferences fix
    331 
    332      The PKCS#7 parsing code does not handle missing outer ContentInfo
    333      correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
    334      missing content and trigger a NULL pointer dereference on parsing.
    335 
    336      Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
    337      otherwise parse PKCS#7 structures from untrusted sources are
    338      affected. OpenSSL clients and servers are not affected.
    339 
    340      This issue was reported to OpenSSL by Michal Zalewski (Google).
    341      (CVE-2015-0289)
    342      [Emilia Ksper]
    343 
    344   *) DoS via reachable assert in SSLv2 servers fix
    345 
    346      A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
    347      servers that both support SSLv2 and enable export cipher suites by sending
    348      a specially crafted SSLv2 CLIENT-MASTER-KEY message.
    349 
    350      This issue was discovered by Sean Burford (Google) and Emilia Ksper
    351      (OpenSSL development team).
    352      (CVE-2015-0293)
    353      [Emilia Ksper]
    354 
    355   *) Use After Free following d2i_ECPrivatekey error fix
    356 
    357      A malformed EC private key file consumed via the d2i_ECPrivateKey function
    358      could cause a use after free condition. This, in turn, could cause a double
    359      free in several private key parsing functions (such as d2i_PrivateKey
    360      or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
    361      for applications that receive EC private keys from untrusted
    362      sources. This scenario is considered rare.
    363 
    364      This issue was discovered by the BoringSSL project and fixed in their
    365      commit 517073cd4b.
    366      (CVE-2015-0209)
    367      [Matt Caswell]
    368 
    369   *) X509_to_X509_REQ NULL pointer deref fix
    370 
    371      The function X509_to_X509_REQ will crash with a NULL pointer dereference if
    372      the certificate key is invalid. This function is rarely used in practice.
    373 
    374      This issue was discovered by Brian Carpenter.
    375      (CVE-2015-0288)
    376      [Stephen Henson]
    377 
    378   *) Removed the export ciphers from the DEFAULT ciphers
    379      [Kurt Roeckx]
    380 
    381  Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
    382 
    383   *) Build fixes for the Windows and OpenVMS platforms
    384      [Matt Caswell and Richard Levitte]
    385 
    386  Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
    387 
    388   *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
    389      message can cause a segmentation fault in OpenSSL due to a NULL pointer
    390      dereference. This could lead to a Denial Of Service attack. Thanks to
    391      Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
    392      (CVE-2014-3571)
    393      [Steve Henson]
    394 
    395   *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
    396      dtls1_buffer_record function under certain conditions. In particular this
    397      could occur if an attacker sent repeated DTLS records with the same
    398      sequence number but for the next epoch. The memory leak could be exploited
    399      by an attacker in a Denial of Service attack through memory exhaustion.
    400      Thanks to Chris Mueller for reporting this issue.
    401      (CVE-2015-0206)
    402      [Matt Caswell]
    403 
    404   *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
    405      built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
    406      method would be set to NULL which could later result in a NULL pointer
    407      dereference. Thanks to Frank Schmirler for reporting this issue.
    408      (CVE-2014-3569)
    409      [Kurt Roeckx]
    410 
    411   *) Abort handshake if server key exchange message is omitted for ephemeral
    412      ECDH ciphersuites.
    413 
    414      Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
    415      reporting this issue.
    416      (CVE-2014-3572)
    417      [Steve Henson]
    418 
    419   *) Remove non-export ephemeral RSA code on client and server. This code
    420      violated the TLS standard by allowing the use of temporary RSA keys in
    421      non-export ciphersuites and could be used by a server to effectively
    422      downgrade the RSA key length used to a value smaller than the server
    423      certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
    424      INRIA or reporting this issue.
    425      (CVE-2015-0204)
    426      [Steve Henson]
    427 
    428   *) Fixed issue where DH client certificates are accepted without verification.
    429      An OpenSSL server will accept a DH certificate for client authentication
    430      without the certificate verify message. This effectively allows a client to
    431      authenticate without the use of a private key. This only affects servers
    432      which trust a client certificate authority which issues certificates
    433      containing DH keys: these are extremely rare and hardly ever encountered.
    434      Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
    435      this issue.
    436      (CVE-2015-0205)
    437      [Steve Henson]
    438 
    439   *) Ensure that the session ID context of an SSL is updated when its
    440      SSL_CTX is updated via SSL_set_SSL_CTX.
    441 
    442      The session ID context is typically set from the parent SSL_CTX,
    443      and can vary with the CTX.
    444      [Adam Langley]
    445 
    446   *) Fix various certificate fingerprint issues.
    447 
    448      By using non-DER or invalid encodings outside the signed portion of a
    449      certificate the fingerprint can be changed without breaking the signature.
    450      Although no details of the signed portion of the certificate can be changed
    451      this can cause problems with some applications: e.g. those using the
    452      certificate fingerprint for blacklists.
    453 
    454      1. Reject signatures with non zero unused bits.
    455 
    456      If the BIT STRING containing the signature has non zero unused bits reject
    457      the signature. All current signature algorithms require zero unused bits.
    458 
    459      2. Check certificate algorithm consistency.
    460 
    461      Check the AlgorithmIdentifier inside TBS matches the one in the
    462      certificate signature. NB: this will result in signature failure
    463      errors for some broken certificates.
    464 
    465      Thanks to Konrad Kraszewski from Google for reporting this issue.
    466 
    467      3. Check DSA/ECDSA signatures use DER.
    468 
    469      Reencode DSA/ECDSA signatures and compare with the original received
    470      signature. Return an error if there is a mismatch.
    471 
    472      This will reject various cases including garbage after signature
    473      (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
    474      program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
    475      (negative or with leading zeroes).
    476 
    477      Further analysis was conducted and fixes were developed by Stephen Henson
    478      of the OpenSSL core team.
    479 
    480      (CVE-2014-8275)
    481      [Steve Henson]
    482 
    483    *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
    484       results on some platforms, including x86_64. This bug occurs at random
    485       with a very low probability, and is not known to be exploitable in any
    486       way, though its exact impact is difficult to determine. Thanks to Pieter
    487       Wuille (Blockstream) who reported this issue and also suggested an initial
    488       fix. Further analysis was conducted by the OpenSSL development team and
    489       Adam Langley of Google. The final fix was developed by Andy Polyakov of
    490       the OpenSSL core team.
    491       (CVE-2014-3570)
    492       [Andy Polyakov]
    493 
    494    *) Do not resume sessions on the server if the negotiated protocol
    495       version does not match the session's version. Resuming with a different
    496       version, while not strictly forbidden by the RFC, is of questionable
    497       sanity and breaks all known clients.
    498       [David Benjamin, Emilia Ksper]
    499 
    500    *) Tighten handling of the ChangeCipherSpec (CCS) message: reject
    501       early CCS messages during renegotiation. (Note that because
    502       renegotiation is encrypted, this early CCS was not exploitable.)
    503       [Emilia Ksper]
    504 
    505    *) Tighten client-side session ticket handling during renegotiation:
    506       ensure that the client only accepts a session ticket if the server sends
    507       the extension anew in the ServerHello. Previously, a TLS client would
    508       reuse the old extension state and thus accept a session ticket if one was
    509       announced in the initial ServerHello.
    510 
    511       Similarly, ensure that the client requires a session ticket if one
    512       was advertised in the ServerHello. Previously, a TLS client would
    513       ignore a missing NewSessionTicket message.
    514       [Emilia Ksper]
    515 
    516  Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
    517 
    518   *) SRTP Memory Leak.
    519 
    520      A flaw in the DTLS SRTP extension parsing code allows an attacker, who
    521      sends a carefully crafted handshake message, to cause OpenSSL to fail
    522      to free up to 64k of memory causing a memory leak. This could be
    523      exploited in a Denial Of Service attack. This issue affects OpenSSL
    524      1.0.1 server implementations for both SSL/TLS and DTLS regardless of
    525      whether SRTP is used or configured. Implementations of OpenSSL that
    526      have been compiled with OPENSSL_NO_SRTP defined are not affected.
    527 
    528      The fix was developed by the OpenSSL team.
    529      (CVE-2014-3513)
    530      [OpenSSL team]
    531 
    532   *) Session Ticket Memory Leak.
    533 
    534      When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
    535      integrity of that ticket is first verified. In the event of a session
    536      ticket integrity check failing, OpenSSL will fail to free memory
    537      causing a memory leak. By sending a large number of invalid session
    538      tickets an attacker could exploit this issue in a Denial Of Service
    539      attack.
    540      (CVE-2014-3567)
    541      [Steve Henson]
    542 
    543   *) Build option no-ssl3 is incomplete.
    544 
    545      When OpenSSL is configured with "no-ssl3" as a build option, servers
    546      could accept and complete a SSL 3.0 handshake, and clients could be
    547      configured to send them.
    548      (CVE-2014-3568)
    549      [Akamai and the OpenSSL team]
    550 
    551   *) Add support for TLS_FALLBACK_SCSV.
    552      Client applications doing fallback retries should call
    553      SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
    554      (CVE-2014-3566)
    555      [Adam Langley, Bodo Moeller]
    556 
    557   *) Add additional DigestInfo checks.
    558  
    559      Reencode DigestInto in DER and check against the original when
    560      verifying RSA signature: this will reject any improperly encoded
    561      DigestInfo structures.
    562 
    563      Note: this is a precautionary measure and no attacks are currently known.
    564 
    565      [Steve Henson]
    566 
    567  Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
    568 
    569   *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
    570      SRP code can be overrun an internal buffer. Add sanity check that
    571      g, A, B < N to SRP code.
    572 
    573      Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
    574      Group for discovering this issue.
    575      (CVE-2014-3512)
    576      [Steve Henson]
    577 
    578   *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
    579      TLS 1.0 instead of higher protocol versions when the ClientHello message
    580      is badly fragmented. This allows a man-in-the-middle attacker to force a
    581      downgrade to TLS 1.0 even if both the server and the client support a
    582      higher protocol version, by modifying the client's TLS records.
    583 
    584      Thanks to David Benjamin and Adam Langley (Google) for discovering and
    585      researching this issue.
    586      (CVE-2014-3511)
    587      [David Benjamin]
    588 
    589   *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
    590      to a denial of service attack. A malicious server can crash the client
    591      with a null pointer dereference (read) by specifying an anonymous (EC)DH
    592      ciphersuite and sending carefully crafted handshake messages.
    593 
    594      Thanks to Felix Grbert (Google) for discovering and researching this
    595      issue.
    596      (CVE-2014-3510)
    597      [Emilia Ksper]
    598 
    599   *) By sending carefully crafted DTLS packets an attacker could cause openssl
    600      to leak memory. This can be exploited through a Denial of Service attack.
    601      Thanks to Adam Langley for discovering and researching this issue.
    602      (CVE-2014-3507)
    603      [Adam Langley]
    604 
    605   *) An attacker can force openssl to consume large amounts of memory whilst
    606      processing DTLS handshake messages. This can be exploited through a
    607      Denial of Service attack.
    608      Thanks to Adam Langley for discovering and researching this issue.
    609      (CVE-2014-3506)
    610      [Adam Langley]
    611 
    612   *) An attacker can force an error condition which causes openssl to crash
    613      whilst processing DTLS packets due to memory being freed twice. This
    614      can be exploited through a Denial of Service attack.
    615      Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
    616      this issue.
    617      (CVE-2014-3505)
    618      [Adam Langley]
    619 
    620   *) If a multithreaded client connects to a malicious server using a resumed
    621      session and the server sends an ec point format extension it could write
    622      up to 255 bytes to freed memory.
    623 
    624      Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
    625      issue.
    626      (CVE-2014-3509)
    627      [Gabor Tyukasz]
    628 
    629   *) A malicious server can crash an OpenSSL client with a null pointer
    630      dereference (read) by specifying an SRP ciphersuite even though it was not
    631      properly negotiated with the client. This can be exploited through a
    632      Denial of Service attack.
    633 
    634      Thanks to Joonas Kuorilehto and Riku Hietamki (Codenomicon) for
    635      discovering and researching this issue.
    636      (CVE-2014-5139)
    637      [Steve Henson]
    638 
    639   *) A flaw in OBJ_obj2txt may cause pretty printing functions such as
    640      X509_name_oneline, X509_name_print_ex et al. to leak some information
    641      from the stack. Applications may be affected if they echo pretty printing
    642      output to the attacker.
    643 
    644      Thanks to Ivan Fratric (Google) for discovering this issue.
    645      (CVE-2014-3508)
    646      [Emilia Ksper, and Steve Henson]
    647 
    648   *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
    649      for corner cases. (Certain input points at infinity could lead to
    650      bogus results, with non-infinity inputs mapped to infinity too.)
    651      [Bodo Moeller]
    652 
    653  Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
    654 
    655   *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
    656      handshake can force the use of weak keying material in OpenSSL
    657      SSL/TLS clients and servers.
    658 
    659      Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
    660      researching this issue. (CVE-2014-0224)
    661      [KIKUCHI Masashi, Steve Henson]
    662 
    663   *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
    664      OpenSSL DTLS client the code can be made to recurse eventually crashing
    665      in a DoS attack.
    666 
    667      Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
    668      (CVE-2014-0221)
    669      [Imre Rad, Steve Henson]
    670 
    671   *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
    672      be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
    673      client or server. This is potentially exploitable to run arbitrary
    674      code on a vulnerable client or server.
    675 
    676      Thanks to Jri Aedla for reporting this issue. (CVE-2014-0195)
    677      [Jri Aedla, Steve Henson]
    678 
    679   *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
    680      are subject to a denial of service attack.
    681 
    682      Thanks to Felix Grbert and Ivan Fratric at Google for discovering
    683      this issue. (CVE-2014-3470)
    684      [Felix Grbert, Ivan Fratric, Steve Henson]
    685 
    686   *) Harmonize version and its documentation. -f flag is used to display
    687      compilation flags.
    688      [mancha <mancha1 (a] zoho.com>]
    689 
    690   *) Fix eckey_priv_encode so it immediately returns an error upon a failure
    691      in i2d_ECPrivateKey.
    692      [mancha <mancha1 (a] zoho.com>]
    693 
    694   *) Fix some double frees. These are not thought to be exploitable.
    695      [mancha <mancha1 (a] zoho.com>]
    696 
    697  Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
    698 
    699   *) A missing bounds check in the handling of the TLS heartbeat extension
    700      can be used to reveal up to 64k of memory to a connected client or
    701      server.
    702 
    703      Thanks for Neel Mehta of Google Security for discovering this bug and to
    704      Adam Langley <agl (a] chromium.org> and Bodo Moeller <bmoeller (a] acm.org> for
    705      preparing the fix (CVE-2014-0160)
    706      [Adam Langley, Bodo Moeller]
    707 
    708   *) Fix for the attack described in the paper "Recovering OpenSSL
    709      ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
    710      by Yuval Yarom and Naomi Benger. Details can be obtained from:
    711      http://eprint.iacr.org/2014/140
    712 
    713      Thanks to Yuval Yarom and Naomi Benger for discovering this
    714      flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
    715      [Yuval Yarom and Naomi Benger]
    716 
    717   *) TLS pad extension: draft-agl-tls-padding-03
    718 
    719      Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
    720      TLS client Hello record length value would otherwise be > 255 and
    721      less that 512 pad with a dummy extension containing zeroes so it
    722      is at least 512 bytes long.
    723 
    724      [Adam Langley, Steve Henson]
    725 
    726  Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
    727 
    728   *) Fix for TLS record tampering bug. A carefully crafted invalid 
    729      handshake could crash OpenSSL with a NULL pointer exception.
    730      Thanks to Anton Johansson for reporting this issues.
    731      (CVE-2013-4353)
    732 
    733   *) Keep original DTLS digest and encryption contexts in retransmission
    734      structures so we can use the previous session parameters if they need
    735      to be resent. (CVE-2013-6450)
    736      [Steve Henson]
    737 
    738   *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
    739      avoids preferring ECDHE-ECDSA ciphers when the client appears to be
    740      Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
    741      several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
    742      is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
    743      10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
    744      [Rob Stradling, Adam Langley]
    745 
    746  Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
    747 
    748   *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI
    749      supporting platforms or when small records were transferred.
    750      [Andy Polyakov, Steve Henson]
    751 
    752  Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
    753 
    754   *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
    755 
    756      This addresses the flaw in CBC record processing discovered by 
    757      Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
    758      at: http://www.isg.rhul.ac.uk/tls/     
    759 
    760      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
    761      Security Group at Royal Holloway, University of London
    762      (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
    763      Emilia Ksper for the initial patch.
    764      (CVE-2013-0169)
    765      [Emilia Ksper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
    766 
    767   *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
    768      ciphersuites which can be exploited in a denial of service attack.
    769      Thanks go to and to Adam Langley <agl (a] chromium.org> for discovering
    770      and detecting this bug and to Wolfgang Ettlinger
    771      <wolfgang.ettlinger (a] gmail.com> for independently discovering this issue.
    772      (CVE-2012-2686)
    773      [Adam Langley]
    774 
    775   *) Return an error when checking OCSP signatures when key is NULL.
    776      This fixes a DoS attack. (CVE-2013-0166)
    777      [Steve Henson]
    778 
    779   *) Make openssl verify return errors.
    780      [Chris Palmer <palmer (a] google.com> and Ben Laurie]
    781 
    782   *) Call OCSP Stapling callback after ciphersuite has been chosen, so
    783      the right response is stapled. Also change SSL_get_certificate()
    784      so it returns the certificate actually sent.
    785      See http://rt.openssl.org/Ticket/Display.html?id=2836.
    786      [Rob Stradling <rob.stradling (a] comodo.com>]
    787 
    788   *) Fix possible deadlock when decoding public keys.
    789      [Steve Henson]
    790 
    791   *) Don't use TLS 1.0 record version number in initial client hello
    792      if renegotiating.
    793      [Steve Henson]
    794 
    795  Changes between 1.0.1b and 1.0.1c [10 May 2012]
    796 
    797   *) Sanity check record length before skipping explicit IV in TLS
    798      1.2, 1.1 and DTLS to fix DoS attack.
    799 
    800      Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
    801      fuzzing as a service testing platform.
    802      (CVE-2012-2333)
    803      [Steve Henson]
    804 
    805   *) Initialise tkeylen properly when encrypting CMS messages.
    806      Thanks to Solar Designer of Openwall for reporting this issue.
    807      [Steve Henson]
    808 
    809   *) In FIPS mode don't try to use composite ciphers as they are not
    810      approved.
    811      [Steve Henson]
    812 
    813  Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
    814 
    815   *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
    816      1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
    817      mean any application compiled against OpenSSL 1.0.0 headers setting
    818      SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng
    819      TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
    820      0x10000000L Any application which was previously compiled against
    821      OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
    822      will need to be recompiled as a result. Letting be results in
    823      inability to disable specifically TLS 1.1 and in client context,
    824      in unlike event, limit maximum offered version to TLS 1.0 [see below].
    825      [Steve Henson]
    826 
    827   *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not
    828      disable just protocol X, but all protocols above X *if* there are
    829      protocols *below* X still enabled. In more practical terms it means
    830      that if application wants to disable TLS1.0 in favor of TLS1.1 and
    831      above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass
    832      SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to
    833      client side.
    834      [Andy Polyakov]
    835 
    836  Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
    837 
    838   *) Check for potentially exploitable overflows in asn1_d2i_read_bio
    839      BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
    840      in CRYPTO_realloc_clean.
    841 
    842      Thanks to Tavis Ormandy, Google Security Team, for discovering this
    843      issue and to Adam Langley <agl (a] chromium.org> for fixing it.
    844      (CVE-2012-2110)
    845      [Adam Langley (Google), Tavis Ormandy, Google Security Team]
    846 
    847   *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
    848      [Adam Langley]
    849 
    850   *) Workarounds for some broken servers that "hang" if a client hello
    851      record length exceeds 255 bytes.
    852 
    853      1. Do not use record version number > TLS 1.0 in initial client
    854         hello: some (but not all) hanging servers will now work.
    855      2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
    856 	the number of ciphers sent in the client hello. This should be
    857         set to an even number, such as 50, for example by passing:
    858         -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
    859         Most broken servers should now work.
    860      3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
    861 	TLS 1.2 client support entirely.
    862      [Steve Henson]
    863 
    864   *) Fix SEGV in Vector Permutation AES module observed in OpenSSH.
    865      [Andy Polyakov]
    866 
    867  Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
    868 
    869   *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET
    870      STRING form instead of a DigestInfo.
    871      [Steve Henson]
    872 
    873   *) The format used for MDC2 RSA signatures is inconsistent between EVP
    874      and the RSA_sign/RSA_verify functions. This was made more apparent when
    875      OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
    876      those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 
    877      the correct format in RSA_verify so both forms transparently work.
    878      [Steve Henson]
    879 
    880   *) Some servers which support TLS 1.0 can choke if we initially indicate
    881      support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
    882      encrypted premaster secret. As a workaround use the maximum pemitted
    883      client version in client hello, this should keep such servers happy
    884      and still work with previous versions of OpenSSL.
    885      [Steve Henson]
    886 
    887   *) Add support for TLS/DTLS heartbeats.
    888      [Robin Seggelmann <seggelmann (a] fh-muenster.de>]
    889 
    890   *) Add support for SCTP.
    891      [Robin Seggelmann <seggelmann (a] fh-muenster.de>]
    892 
    893   *) Improved PRNG seeding for VOS.
    894      [Paul Green <Paul.Green (a] stratus.com>]
    895 
    896   *) Extensive assembler packs updates, most notably:
    897 
    898 	- x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
    899 	- x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
    900 	- x86_64:       bit-sliced AES implementation;
    901 	- ARM:          NEON support, contemporary platforms optimizations;
    902 	- s390x:        z196 support;
    903 	- *:            GHASH and GF(2^m) multiplication implementations;
    904 
    905      [Andy Polyakov]
    906 
    907   *) Make TLS-SRP code conformant with RFC 5054 API cleanup
    908      (removal of unnecessary code)
    909      [Peter Sylvester <peter.sylvester (a] edelweb.fr>]
    910 
    911   *) Add TLS key material exporter from RFC 5705.
    912      [Eric Rescorla]
    913 
    914   *) Add DTLS-SRTP negotiation from RFC 5764.
    915      [Eric Rescorla]
    916 
    917   *) Add Next Protocol Negotiation,
    918      http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
    919      disabled with a no-npn flag to config or Configure. Code donated
    920      by Google.
    921      [Adam Langley <agl (a] google.com> and Ben Laurie]
    922 
    923   *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
    924      NIST-P256, NIST-P521, with constant-time single point multiplication on
    925      typical inputs. Compiler support for the nonstandard type __uint128_t is
    926      required to use this (present in gcc 4.4 and later, for 64-bit builds).
    927      Code made available under Apache License version 2.0.
    928 
    929      Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
    930      line to include this in your build of OpenSSL, and run "make depend" (or
    931      "make update"). This enables the following EC_METHODs:
    932 
    933          EC_GFp_nistp224_method()
    934          EC_GFp_nistp256_method()
    935          EC_GFp_nistp521_method()
    936 
    937      EC_GROUP_new_by_curve_name() will automatically use these (while
    938      EC_GROUP_new_curve_GFp() currently prefers the more flexible
    939      implementations).
    940      [Emilia Ksper, Adam Langley, Bodo Moeller (Google)]
    941 
    942   *) Use type ossl_ssize_t instad of ssize_t which isn't available on
    943      all platforms. Move ssize_t definition from e_os.h to the public
    944      header file e_os2.h as it now appears in public header file cms.h
    945      [Steve Henson]
    946 
    947   *) New -sigopt option to the ca, req and x509 utilities. Additional
    948      signature parameters can be passed using this option and in
    949      particular PSS. 
    950      [Steve Henson]
    951 
    952   *) Add RSA PSS signing function. This will generate and set the
    953      appropriate AlgorithmIdentifiers for PSS based on those in the
    954      corresponding EVP_MD_CTX structure. No application support yet.
    955      [Steve Henson]
    956 
    957   *) Support for companion algorithm specific ASN1 signing routines.
    958      New function ASN1_item_sign_ctx() signs a pre-initialised
    959      EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
    960      the appropriate parameters.
    961      [Steve Henson]
    962 
    963   *) Add new algorithm specific ASN1 verification initialisation function
    964      to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
    965      handling will be the same no matter what EVP_PKEY_METHOD is used.
    966      Add a PSS handler to support verification of PSS signatures: checked
    967      against a number of sample certificates.
    968      [Steve Henson]
    969 
    970   *) Add signature printing for PSS. Add PSS OIDs.
    971      [Steve Henson, Martin Kaiser <lists (a] kaiser.cx>]
    972 
    973   *) Add algorithm specific signature printing. An individual ASN1 method
    974      can now print out signatures instead of the standard hex dump. 
    975 
    976      More complex signatures (e.g. PSS) can print out more meaningful
    977      information. Include DSA version that prints out the signature
    978      parameters r, s.
    979      [Steve Henson]
    980 
    981   *) Password based recipient info support for CMS library: implementing
    982      RFC3211.
    983      [Steve Henson]
    984 
    985   *) Split password based encryption into PBES2 and PBKDF2 functions. This
    986      neatly separates the code into cipher and PBE sections and is required
    987      for some algorithms that split PBES2 into separate pieces (such as
    988      password based CMS).
    989      [Steve Henson]
    990 
    991   *) Session-handling fixes:
    992      - Fix handling of connections that are resuming with a session ID,
    993        but also support Session Tickets.
    994      - Fix a bug that suppressed issuing of a new ticket if the client
    995        presented a ticket with an expired session.
    996      - Try to set the ticket lifetime hint to something reasonable.
    997      - Make tickets shorter by excluding irrelevant information.
    998      - On the client side, don't ignore renewed tickets.
    999      [Adam Langley, Bodo Moeller (Google)]
   1000 
   1001   *) Fix PSK session representation.
   1002      [Bodo Moeller]
   1003 
   1004   *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
   1005 
   1006      This work was sponsored by Intel.
   1007      [Andy Polyakov]
   1008 
   1009   *) Add GCM support to TLS library. Some custom code is needed to split
   1010      the IV between the fixed (from PRF) and explicit (from TLS record)
   1011      portions. This adds all GCM ciphersuites supported by RFC5288 and 
   1012      RFC5289. Generalise some AES* cipherstrings to inlclude GCM and
   1013      add a special AESGCM string for GCM only.
   1014      [Steve Henson]
   1015 
   1016   *) Expand range of ctrls for AES GCM. Permit setting invocation
   1017      field on decrypt and retrieval of invocation field only on encrypt.
   1018      [Steve Henson]
   1019 
   1020   *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
   1021      As required by RFC5289 these ciphersuites cannot be used if for
   1022      versions of TLS earlier than 1.2.
   1023      [Steve Henson]
   1024 
   1025   *) For FIPS capable OpenSSL interpret a NULL default public key method
   1026      as unset and return the appopriate default but do *not* set the default.
   1027      This means we can return the appopriate method in applications that
   1028      swicth between FIPS and non-FIPS modes.
   1029      [Steve Henson]
   1030 
   1031   *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
   1032      ENGINE is used then we cannot handle that in the FIPS module so we
   1033      keep original code iff non-FIPS operations are allowed.
   1034      [Steve Henson]
   1035 
   1036   *) Add -attime option to openssl utilities.
   1037      [Peter Eckersley <pde (a] eff.org>, Ben Laurie and Steve Henson]
   1038 
   1039   *) Redirect DSA and DH operations to FIPS module in FIPS mode.
   1040      [Steve Henson]
   1041 
   1042   *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
   1043      FIPS EC methods unconditionally for now.
   1044      [Steve Henson]
   1045 
   1046   *) New build option no-ec2m to disable characteristic 2 code.
   1047      [Steve Henson]
   1048 
   1049   *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
   1050      all cases can be covered as some introduce binary incompatibilities.
   1051      [Steve Henson]
   1052 
   1053   *) Redirect RSA operations to FIPS module including keygen,
   1054      encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
   1055      [Steve Henson]
   1056 
   1057   *) Add similar low level API blocking to ciphers.
   1058      [Steve Henson]
   1059 
   1060   *) Low level digest APIs are not approved in FIPS mode: any attempt
   1061      to use these will cause a fatal error. Applications that *really* want
   1062      to use them can use the private_* version instead.
   1063      [Steve Henson]
   1064 
   1065   *) Redirect cipher operations to FIPS module for FIPS builds. 
   1066      [Steve Henson]
   1067 
   1068   *) Redirect digest operations to FIPS module for FIPS builds. 
   1069      [Steve Henson]
   1070 
   1071   *) Update build system to add "fips" flag which will link in fipscanister.o
   1072      for static and shared library builds embedding a signature if needed.
   1073      [Steve Henson]
   1074 
   1075   *) Output TLS supported curves in preference order instead of numerical
   1076      order. This is currently hardcoded for the highest order curves first.
   1077      This should be configurable so applications can judge speed vs strength.
   1078      [Steve Henson]
   1079 
   1080   *) Add TLS v1.2 server support for client authentication. 
   1081      [Steve Henson]
   1082 
   1083   *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
   1084      and enable MD5.
   1085      [Steve Henson]
   1086 
   1087   *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
   1088      FIPS modules versions.
   1089      [Steve Henson]
   1090 
   1091   *) Add TLS v1.2 client side support for client authentication. Keep cache
   1092      of handshake records longer as we don't know the hash algorithm to use
   1093      until after the certificate request message is received.
   1094      [Steve Henson]
   1095 
   1096   *) Initial TLS v1.2 client support. Add a default signature algorithms
   1097      extension including all the algorithms we support. Parse new signature
   1098      format in client key exchange. Relax some ECC signing restrictions for
   1099      TLS v1.2 as indicated in RFC5246.
   1100      [Steve Henson]
   1101 
   1102   *) Add server support for TLS v1.2 signature algorithms extension. Switch
   1103      to new signature format when needed using client digest preference.
   1104      All server ciphersuites should now work correctly in TLS v1.2. No client
   1105      support yet and no support for client certificates.
   1106      [Steve Henson]
   1107 
   1108   *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
   1109      to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
   1110      ciphersuites. At present only RSA key exchange ciphersuites work with
   1111      TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
   1112      SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
   1113      and version checking.
   1114      [Steve Henson]
   1115 
   1116   *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
   1117      with this defined it will not be affected by any changes to ssl internal
   1118      structures. Add several utility functions to allow openssl application
   1119      to work with OPENSSL_NO_SSL_INTERN defined.
   1120      [Steve Henson]
   1121 
   1122   *) Add SRP support.
   1123      [Tom Wu <tjw (a] cs.stanford.edu> and Ben Laurie]
   1124 
   1125   *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
   1126      [Steve Henson]
   1127 
   1128   *) Permit abbreviated handshakes when renegotiating using the function
   1129      SSL_renegotiate_abbreviated().
   1130      [Robin Seggelmann <seggelmann (a] fh-muenster.de>]
   1131 
   1132   *) Add call to ENGINE_register_all_complete() to
   1133      ENGINE_load_builtin_engines(), so some implementations get used
   1134      automatically instead of needing explicit application support.
   1135      [Steve Henson]
   1136 
   1137   *) Add support for TLS key exporter as described in RFC5705.
   1138      [Robin Seggelmann <seggelmann (a] fh-muenster.de>, Steve Henson]
   1139 
   1140   *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
   1141      a few changes are required:
   1142 
   1143        Add SSL_OP_NO_TLSv1_1 flag.
   1144        Add TLSv1_1 methods.
   1145        Update version checking logic to handle version 1.1.
   1146        Add explicit IV handling (ported from DTLS code).
   1147        Add command line options to s_client/s_server.
   1148      [Steve Henson]
   1149 
   1150  Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
   1151 
   1152   *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
   1153      in CMS and PKCS7 code. When RSA decryption fails use a random key for
   1154      content decryption and always return the same error. Note: this attack
   1155      needs on average 2^20 messages so it only affects automated senders. The
   1156      old behaviour can be reenabled in the CMS code by setting the
   1157      CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
   1158      an MMA defence is not necessary.
   1159      Thanks to Ivan Nestlerode <inestlerode (a] us.ibm.com> for discovering
   1160      this issue. (CVE-2012-0884)
   1161      [Steve Henson]
   1162 
   1163   *) Fix CVE-2011-4619: make sure we really are receiving a 
   1164      client hello before rejecting multiple SGC restarts. Thanks to
   1165      Ivan Nestlerode <inestlerode (a] us.ibm.com> for discovering this bug.
   1166      [Steve Henson]
   1167 
   1168  Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
   1169 
   1170   *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
   1171      Thanks to Antonio Martin, Enterprise Secure Access Research and
   1172      Development, Cisco Systems, Inc. for discovering this bug and
   1173      preparing a fix. (CVE-2012-0050)
   1174      [Antonio Martin]
   1175 
   1176  Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
   1177 
   1178   *) Nadhem Alfardan and Kenny Paterson have discovered an extension
   1179      of the Vaudenay padding oracle attack on CBC mode encryption
   1180      which enables an efficient plaintext recovery attack against
   1181      the OpenSSL implementation of DTLS. Their attack exploits timing
   1182      differences arising during decryption processing. A research
   1183      paper describing this attack can be found at:
   1184                   http://www.isg.rhul.ac.uk/~kp/dtls.pdf
   1185      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
   1186      Security Group at Royal Holloway, University of London
   1187      (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
   1188      <seggelmann (a] fh-muenster.de> and Michael Tuexen <tuexen (a] fh-muenster.de>
   1189      for preparing the fix. (CVE-2011-4108)
   1190      [Robin Seggelmann, Michael Tuexen]
   1191 
   1192   *) Clear bytes used for block padding of SSL 3.0 records.
   1193      (CVE-2011-4576)
   1194      [Adam Langley (Google)]
   1195 
   1196   *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
   1197      Kadianakis <desnacked (a] gmail.com> for discovering this issue and
   1198      Adam Langley for preparing the fix. (CVE-2011-4619)
   1199      [Adam Langley (Google)]
   1200 
   1201   *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027)
   1202      [Andrey Kulikov <amdeich (a] gmail.com>]
   1203 
   1204   *) Prevent malformed RFC3779 data triggering an assertion failure.
   1205      Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
   1206      and Rob Austein <sra (a] hactrn.net> for fixing it. (CVE-2011-4577)
   1207      [Rob Austein <sra (a] hactrn.net>]
   1208 
   1209   *) Improved PRNG seeding for VOS.
   1210      [Paul Green <Paul.Green (a] stratus.com>]
   1211 
   1212   *) Fix ssl_ciph.c set-up race.
   1213      [Adam Langley (Google)]
   1214 
   1215   *) Fix spurious failures in ecdsatest.c.
   1216      [Emilia Ksper (Google)]
   1217 
   1218   *) Fix the BIO_f_buffer() implementation (which was mixing different
   1219      interpretations of the '..._len' fields).
   1220      [Adam Langley (Google)]
   1221 
   1222   *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
   1223      BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
   1224      threads won't reuse the same blinding coefficients.
   1225 
   1226      This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
   1227      lock to call BN_BLINDING_invert_ex, and avoids one use of
   1228      BN_BLINDING_update for each BN_BLINDING structure (previously,
   1229      the last update always remained unused).
   1230      [Emilia Ksper (Google)]
   1231 
   1232   *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
   1233      [Bob Buckholz (Google)]
   1234 
   1235  Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
   1236 
   1237   *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
   1238      by initialising X509_STORE_CTX properly. (CVE-2011-3207)
   1239      [Kaspar Brand <ossl (a] velox.ch>]
   1240 
   1241   *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
   1242      for multi-threaded use of ECDH. (CVE-2011-3210)
   1243      [Adam Langley (Google)]
   1244 
   1245   *) Fix x509_name_ex_d2i memory leak on bad inputs.
   1246      [Bodo Moeller]
   1247 
   1248   *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
   1249      signature public key algorithm by using OID xref utilities instead.
   1250      Before this you could only use some ECC ciphersuites with SHA1 only.
   1251      [Steve Henson]
   1252 
   1253   *) Add protection against ECDSA timing attacks as mentioned in the paper
   1254      by Billy Bob Brumley and Nicola Tuveri, see:
   1255 
   1256 	http://eprint.iacr.org/2011/232.pdf
   1257 
   1258      [Billy Bob Brumley and Nicola Tuveri]
   1259 
   1260  Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
   1261 
   1262   *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
   1263      [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
   1264 
   1265   *) Fix bug in string printing code: if *any* escaping is enabled we must
   1266      escape the escape character (backslash) or the resulting string is
   1267      ambiguous.
   1268      [Steve Henson]
   1269 
   1270  Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
   1271 
   1272   *) Disable code workaround for ancient and obsolete Netscape browsers
   1273      and servers: an attacker can use it in a ciphersuite downgrade attack.
   1274      Thanks to Martin Rex for discovering this bug. CVE-2010-4180
   1275      [Steve Henson]
   1276 
   1277   *) Fixed J-PAKE implementation error, originally discovered by
   1278      Sebastien Martini, further info and confirmation from Stefan
   1279      Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
   1280      [Ben Laurie]
   1281 
   1282  Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
   1283 
   1284   *) Fix extension code to avoid race conditions which can result in a buffer
   1285      overrun vulnerability: resumed sessions must not be modified as they can
   1286      be shared by multiple threads. CVE-2010-3864
   1287      [Steve Henson]
   1288 
   1289   *) Fix WIN32 build system to correctly link an ENGINE directory into
   1290      a DLL. 
   1291      [Steve Henson]
   1292 
   1293  Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
   1294 
   1295   *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 
   1296      (CVE-2010-1633)
   1297      [Steve Henson, Peter-Michael Hager <hager (a] dortmund.net>]
   1298 
   1299  Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
   1300 
   1301   *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
   1302      context. The operation can be customised via the ctrl mechanism in
   1303      case ENGINEs want to include additional functionality.
   1304      [Steve Henson]
   1305 
   1306   *) Tolerate yet another broken PKCS#8 key format: private key value negative.
   1307      [Steve Henson]
   1308 
   1309   *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
   1310      output hashes compatible with older versions of OpenSSL.
   1311      [Willy Weisz <weisz (a] vcpc.univie.ac.at>]
   1312 
   1313   *) Fix compression algorithm handling: if resuming a session use the
   1314      compression algorithm of the resumed session instead of determining
   1315      it from client hello again. Don't allow server to change algorithm.
   1316      [Steve Henson]
   1317 
   1318   *) Add load_crls() function to apps tidying load_certs() too. Add option
   1319      to verify utility to allow additional CRLs to be included.
   1320      [Steve Henson]
   1321 
   1322   *) Update OCSP request code to permit adding custom headers to the request:
   1323      some responders need this.
   1324      [Steve Henson]
   1325 
   1326   *) The function EVP_PKEY_sign() returns <=0 on error: check return code
   1327      correctly.
   1328      [Julia Lawall <julia (a] diku.dk>]
   1329 
   1330   *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
   1331      needlessly dereferenced structures, used obsolete functions and
   1332      didn't handle all updated verify codes correctly.
   1333      [Steve Henson]
   1334 
   1335   *) Disable MD2 in the default configuration.
   1336      [Steve Henson]
   1337 
   1338   *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
   1339      indicate the initial BIO being pushed or popped. This makes it possible
   1340      to determine whether the BIO is the one explicitly called or as a result
   1341      of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
   1342      it handles reference counts correctly and doesn't zero out the I/O bio
   1343      when it is not being explicitly popped. WARNING: applications which
   1344      included workarounds for the old buggy behaviour will need to be modified
   1345      or they could free up already freed BIOs.
   1346      [Steve Henson]
   1347 
   1348   *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
   1349      renaming to all platforms (within the 0.9.8 branch, this was
   1350      done conditionally on Netware platforms to avoid a name clash).
   1351      [Guenter <lists (a] gknw.net>]
   1352 
   1353   *) Add ECDHE and PSK support to DTLS.
   1354      [Michael Tuexen <tuexen (a] fh-muenster.de>]
   1355 
   1356   *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
   1357      be used on C++.
   1358      [Steve Henson]
   1359 
   1360   *) Add "missing" function EVP_MD_flags() (without this the only way to
   1361      retrieve a digest flags is by accessing the structure directly. Update
   1362      EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
   1363      or cipher is registered as in the "from" argument. Print out all
   1364      registered digests in the dgst usage message instead of manually 
   1365      attempting to work them out.
   1366      [Steve Henson]
   1367 
   1368   *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
   1369      this allows the use of compression and extensions. Change default cipher
   1370      string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
   1371      by default unless an application cipher string requests it.
   1372      [Steve Henson]
   1373 
   1374   *) Alter match criteria in PKCS12_parse(). It used to try to use local
   1375      key ids to find matching certificates and keys but some PKCS#12 files
   1376      don't follow the (somewhat unwritten) rules and this strategy fails.
   1377      Now just gather all certificates together and the first private key
   1378      then look for the first certificate that matches the key.
   1379      [Steve Henson]
   1380 
   1381   *) Support use of registered digest and cipher names for dgst and cipher
   1382      commands instead of having to add each one as a special case. So now
   1383      you can do:
   1384 
   1385         openssl sha256 foo
   1386 
   1387      as well as:
   1388 
   1389         openssl dgst -sha256 foo
   1390 
   1391      and this works for ENGINE based algorithms too.
   1392 
   1393      [Steve Henson]
   1394 
   1395   *) Update Gost ENGINE to support parameter files.
   1396      [Victor B. Wagner <vitus (a] cryptocom.ru>]
   1397 
   1398   *) Support GeneralizedTime in ca utility. 
   1399      [Oliver Martin <oliver (a] volatilevoid.net>, Steve Henson]
   1400 
   1401   *) Enhance the hash format used for certificate directory links. The new
   1402      form uses the canonical encoding (meaning equivalent names will work
   1403      even if they aren't identical) and uses SHA1 instead of MD5. This form
   1404      is incompatible with the older format and as a result c_rehash should
   1405      be used to rebuild symbolic links.
   1406      [Steve Henson]
   1407 
   1408   *) Make PKCS#8 the default write format for private keys, replacing the
   1409      traditional format. This form is standardised, more secure and doesn't
   1410      include an implicit MD5 dependency.
   1411      [Steve Henson]
   1412 
   1413   *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
   1414      committed to OpenSSL should pass this lot as a minimum.
   1415      [Steve Henson]
   1416 
   1417   *) Add session ticket override functionality for use by EAP-FAST.
   1418      [Jouni Malinen <j (a] w1.fi>]
   1419 
   1420   *) Modify HMAC functions to return a value. Since these can be implemented
   1421      in an ENGINE errors can occur.
   1422      [Steve Henson]
   1423 
   1424   *) Type-checked OBJ_bsearch_ex.
   1425      [Ben Laurie]
   1426 
   1427   *) Type-checked OBJ_bsearch. Also some constification necessitated
   1428      by type-checking.  Still to come: TXT_DB, bsearch(?),
   1429      OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
   1430      CONF_VALUE.
   1431      [Ben Laurie]
   1432 
   1433   *) New function OPENSSL_gmtime_adj() to add a specific number of days and
   1434      seconds to a tm structure directly, instead of going through OS
   1435      specific date routines. This avoids any issues with OS routines such
   1436      as the year 2038 bug. New *_adj() functions for ASN1 time structures
   1437      and X509_time_adj_ex() to cover the extended range. The existing
   1438      X509_time_adj() is still usable and will no longer have any date issues.
   1439      [Steve Henson]
   1440 
   1441   *) Delta CRL support. New use deltas option which will attempt to locate
   1442      and search any appropriate delta CRLs available.
   1443 
   1444      This work was sponsored by Google.
   1445      [Steve Henson]
   1446 
   1447   *) Support for CRLs partitioned by reason code. Reorganise CRL processing
   1448      code and add additional score elements. Validate alternate CRL paths
   1449      as part of the CRL checking and indicate a new error "CRL path validation
   1450      error" in this case. Applications wanting additional details can use
   1451      the verify callback and check the new "parent" field. If this is not
   1452      NULL CRL path validation is taking place. Existing applications wont
   1453      see this because it requires extended CRL support which is off by
   1454      default.
   1455 
   1456      This work was sponsored by Google.
   1457      [Steve Henson]
   1458 
   1459   *) Support for freshest CRL extension.
   1460 
   1461      This work was sponsored by Google.
   1462      [Steve Henson]
   1463 
   1464   *) Initial indirect CRL support. Currently only supported in the CRLs
   1465      passed directly and not via lookup. Process certificate issuer
   1466      CRL entry extension and lookup CRL entries by bother issuer name
   1467      and serial number. Check and process CRL issuer entry in IDP extension.
   1468 
   1469      This work was sponsored by Google.
   1470      [Steve Henson]
   1471 
   1472   *) Add support for distinct certificate and CRL paths. The CRL issuer
   1473      certificate is validated separately in this case. Only enabled if
   1474      an extended CRL support flag is set: this flag will enable additional
   1475      CRL functionality in future.
   1476 
   1477      This work was sponsored by Google.
   1478      [Steve Henson]
   1479 
   1480   *) Add support for policy mappings extension.
   1481 
   1482      This work was sponsored by Google.
   1483      [Steve Henson]
   1484 
   1485   *) Fixes to pathlength constraint, self issued certificate handling,
   1486      policy processing to align with RFC3280 and PKITS tests.
   1487 
   1488      This work was sponsored by Google.
   1489      [Steve Henson]
   1490 
   1491   *) Support for name constraints certificate extension. DN, email, DNS
   1492      and URI types are currently supported.
   1493 
   1494      This work was sponsored by Google.
   1495      [Steve Henson]
   1496 
   1497   *) To cater for systems that provide a pointer-based thread ID rather
   1498      than numeric, deprecate the current numeric thread ID mechanism and
   1499      replace it with a structure and associated callback type. This
   1500      mechanism allows a numeric "hash" to be extracted from a thread ID in
   1501      either case, and on platforms where pointers are larger than 'long',
   1502      mixing is done to help ensure the numeric 'hash' is usable even if it
   1503      can't be guaranteed unique. The default mechanism is to use "&errno"
   1504      as a pointer-based thread ID to distinguish between threads.
   1505 
   1506      Applications that want to provide their own thread IDs should now use
   1507      CRYPTO_THREADID_set_callback() to register a callback that will call
   1508      either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
   1509 
   1510      Note that ERR_remove_state() is now deprecated, because it is tied
   1511      to the assumption that thread IDs are numeric.  ERR_remove_state(0)
   1512      to free the current thread's error state should be replaced by
   1513      ERR_remove_thread_state(NULL).
   1514 
   1515      (This new approach replaces the functions CRYPTO_set_idptr_callback(),
   1516      CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
   1517      OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
   1518      application was previously providing a numeric thread callback that
   1519      was inappropriate for distinguishing threads, then uniqueness might
   1520      have been obtained with &errno that happened immediately in the
   1521      intermediate development versions of OpenSSL; this is no longer the
   1522      case, the numeric thread callback will now override the automatic use
   1523      of &errno.)
   1524      [Geoff Thorpe, with help from Bodo Moeller]
   1525 
   1526   *) Initial support for different CRL issuing certificates. This covers a
   1527      simple case where the self issued certificates in the chain exist and
   1528      the real CRL issuer is higher in the existing chain.
   1529 
   1530      This work was sponsored by Google.
   1531      [Steve Henson]
   1532 
   1533   *) Removed effectively defunct crypto/store from the build.
   1534      [Ben Laurie]
   1535 
   1536   *) Revamp of STACK to provide stronger type-checking. Still to come:
   1537      TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
   1538      ASN1_STRING, CONF_VALUE.
   1539      [Ben Laurie]
   1540 
   1541   *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
   1542      RAM on SSL connections.  This option can save about 34k per idle SSL.
   1543      [Nick Mathewson]
   1544 
   1545   *) Revamp of LHASH to provide stronger type-checking. Still to come:
   1546      STACK, TXT_DB, bsearch, qsort.
   1547      [Ben Laurie]
   1548 
   1549   *) Initial support for Cryptographic Message Syntax (aka CMS) based
   1550      on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
   1551      support for data, signedData, compressedData, digestedData and
   1552      encryptedData, envelopedData types included. Scripts to check against
   1553      RFC4134 examples draft and interop and consistency checks of many
   1554      content types and variants.
   1555      [Steve Henson]
   1556 
   1557   *) Add options to enc utility to support use of zlib compression BIO.
   1558      [Steve Henson]
   1559 
   1560   *) Extend mk1mf to support importing of options and assembly language
   1561      files from Configure script, currently only included in VC-WIN32.
   1562      The assembly language rules can now optionally generate the source
   1563      files from the associated perl scripts.
   1564      [Steve Henson]
   1565 
   1566   *) Implement remaining functionality needed to support GOST ciphersuites.
   1567      Interop testing has been performed using CryptoPro implementations.
   1568      [Victor B. Wagner <vitus (a] cryptocom.ru>]
   1569 
   1570   *) s390x assembler pack.
   1571      [Andy Polyakov]
   1572 
   1573   *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
   1574      "family."
   1575      [Andy Polyakov]
   1576 
   1577   *) Implement Opaque PRF Input TLS extension as specified in
   1578      draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
   1579      official specification yet and no extension type assignment by
   1580      IANA exists, this extension (for now) will have to be explicitly
   1581      enabled when building OpenSSL by providing the extension number
   1582      to use.  For example, specify an option
   1583 
   1584          -DTLSEXT_TYPE_opaque_prf_input=0x9527
   1585 
   1586      to the "config" or "Configure" script to enable the extension,
   1587      assuming extension number 0x9527 (which is a completely arbitrary
   1588      and unofficial assignment based on the MD5 hash of the Internet
   1589      Draft).  Note that by doing so, you potentially lose
   1590      interoperability with other TLS implementations since these might
   1591      be using the same extension number for other purposes.
   1592 
   1593      SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
   1594      opaque PRF input value to use in the handshake.  This will create
   1595      an interal copy of the length-'len' string at 'src', and will
   1596      return non-zero for success.
   1597 
   1598      To get more control and flexibility, provide a callback function
   1599      by using
   1600 
   1601           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
   1602           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
   1603 
   1604      where
   1605 
   1606           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
   1607           void *arg;
   1608 
   1609      Callback function 'cb' will be called in handshakes, and is
   1610      expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
   1611      Argument 'arg' is for application purposes (the value as given to
   1612      SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
   1613      be provided to the callback function).  The callback function
   1614      has to return non-zero to report success: usually 1 to use opaque
   1615      PRF input just if possible, or 2 to enforce use of the opaque PRF
   1616      input.  In the latter case, the library will abort the handshake
   1617      if opaque PRF input is not successfully negotiated.
   1618 
   1619      Arguments 'peerinput' and 'len' given to the callback function
   1620      will always be NULL and 0 in the case of a client.  A server will
   1621      see the client's opaque PRF input through these variables if
   1622      available (NULL and 0 otherwise).  Note that if the server
   1623      provides an opaque PRF input, the length must be the same as the
   1624      length of the client's opaque PRF input.
   1625 
   1626      Note that the callback function will only be called when creating
   1627      a new session (session resumption can resume whatever was
   1628      previously negotiated), and will not be called in SSL 2.0
   1629      handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
   1630      SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
   1631      for applications that need to enforce opaque PRF input.
   1632 
   1633      [Bodo Moeller]
   1634 
   1635   *) Update ssl code to support digests other than SHA1+MD5 for handshake
   1636      MAC. 
   1637 
   1638      [Victor B. Wagner <vitus (a] cryptocom.ru>]
   1639 
   1640   *) Add RFC4507 support to OpenSSL. This includes the corrections in
   1641      RFC4507bis. The encrypted ticket format is an encrypted encoded
   1642      SSL_SESSION structure, that way new session features are automatically
   1643      supported.
   1644 
   1645      If a client application caches session in an SSL_SESSION structure
   1646      support is transparent because tickets are now stored in the encoded
   1647      SSL_SESSION.
   1648      
   1649      The SSL_CTX structure automatically generates keys for ticket
   1650      protection in servers so again support should be possible
   1651      with no application modification.
   1652 
   1653      If a client or server wishes to disable RFC4507 support then the option
   1654      SSL_OP_NO_TICKET can be set.
   1655 
   1656      Add a TLS extension debugging callback to allow the contents of any client
   1657      or server extensions to be examined.
   1658 
   1659      This work was sponsored by Google.
   1660      [Steve Henson]
   1661 
   1662   *) Final changes to avoid use of pointer pointer casts in OpenSSL.
   1663      OpenSSL should now compile cleanly on gcc 4.2
   1664      [Peter Hartley <pdh (a] utter.chaos.org.uk>, Steve Henson]
   1665 
   1666   *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
   1667      support including streaming MAC support: this is required for GOST
   1668      ciphersuite support.
   1669      [Victor B. Wagner <vitus (a] cryptocom.ru>, Steve Henson]
   1670 
   1671   *) Add option -stream to use PKCS#7 streaming in smime utility. New
   1672      function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
   1673      to output in BER and PEM format.
   1674      [Steve Henson]
   1675 
   1676   *) Experimental support for use of HMAC via EVP_PKEY interface. This
   1677      allows HMAC to be handled via the EVP_DigestSign*() interface. The
   1678      EVP_PKEY "key" in this case is the HMAC key, potentially allowing
   1679      ENGINE support for HMAC keys which are unextractable. New -mac and
   1680      -macopt options to dgst utility.
   1681      [Steve Henson]
   1682 
   1683   *) New option -sigopt to dgst utility. Update dgst to use
   1684      EVP_Digest{Sign,Verify}*. These two changes make it possible to use
   1685      alternative signing paramaters such as X9.31 or PSS in the dgst 
   1686      utility.
   1687      [Steve Henson]
   1688 
   1689   *) Change ssl_cipher_apply_rule(), the internal function that does
   1690      the work each time a ciphersuite string requests enabling
   1691      ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
   1692      removing ("!foo+bar") a class of ciphersuites: Now it maintains
   1693      the order of disabled ciphersuites such that those ciphersuites
   1694      that most recently went from enabled to disabled not only stay
   1695      in order with respect to each other, but also have higher priority
   1696      than other disabled ciphersuites the next time ciphersuites are
   1697      enabled again.
   1698 
   1699      This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
   1700      the same ciphersuites as with "HIGH" alone, but in a specific
   1701      order where the PSK ciphersuites come first (since they are the
   1702      most recently disabled ciphersuites when "HIGH" is parsed).
   1703 
   1704      Also, change ssl_create_cipher_list() (using this new
   1705      funcionality) such that between otherwise identical
   1706      cihpersuites, ephemeral ECDH is preferred over ephemeral DH in
   1707      the default order.
   1708      [Bodo Moeller]
   1709 
   1710   *) Change ssl_create_cipher_list() so that it automatically
   1711      arranges the ciphersuites in reasonable order before starting
   1712      to process the rule string.  Thus, the definition for "DEFAULT"
   1713      (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
   1714      remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH".
   1715      This makes it much easier to arrive at a reasonable default order
   1716      in applications for which anonymous ciphers are OK (meaning
   1717      that you can't actually use DEFAULT).
   1718      [Bodo Moeller; suggested by Victor Duchovni]
   1719 
   1720   *) Split the SSL/TLS algorithm mask (as used for ciphersuite string
   1721      processing) into multiple integers instead of setting
   1722      "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
   1723      "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
   1724      (These masks as well as the individual bit definitions are hidden
   1725      away into the non-exported interface ssl/ssl_locl.h, so this
   1726      change to the definition of the SSL_CIPHER structure shouldn't
   1727      affect applications.)  This give us more bits for each of these
   1728      categories, so there is no longer a need to coagulate AES128 and
   1729      AES256 into a single algorithm bit, and to coagulate Camellia128
   1730      and Camellia256 into a single algorithm bit, which has led to all
   1731      kinds of kludges.
   1732 
   1733      Thus, among other things, the kludge introduced in 0.9.7m and
   1734      0.9.8e for masking out AES256 independently of AES128 or masking
   1735      out Camellia256 independently of AES256 is not needed here in 0.9.9.
   1736 
   1737      With the change, we also introduce new ciphersuite aliases that
   1738      so far were missing: "AES128", "AES256", "CAMELLIA128", and
   1739      "CAMELLIA256".
   1740      [Bodo Moeller]
   1741 
   1742   *) Add support for dsa-with-SHA224 and dsa-with-SHA256.
   1743      Use the leftmost N bytes of the signature input if the input is
   1744      larger than the prime q (with N being the size in bytes of q).
   1745      [Nils Larsch]
   1746 
   1747   *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
   1748      it yet and it is largely untested.
   1749      [Steve Henson]
   1750 
   1751   *) Add support for the ecdsa-with-SHA224/256/384/512 signature types.
   1752      [Nils Larsch]
   1753 
   1754   *) Initial incomplete changes to avoid need for function casts in OpenSSL
   1755      some compilers (gcc 4.2 and later) reject their use. Safestack is
   1756      reimplemented.  Update ASN1 to avoid use of legacy functions. 
   1757      [Steve Henson]
   1758 
   1759   *) Win32/64 targets are linked with Winsock2.
   1760      [Andy Polyakov]
   1761 
   1762   *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
   1763      to external functions. This can be used to increase CRL handling 
   1764      efficiency especially when CRLs are very large by (for example) storing
   1765      the CRL revoked certificates in a database.
   1766      [Steve Henson]
   1767 
   1768   *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so
   1769      new CRLs added to a directory can be used. New command line option
   1770      -verify_return_error to s_client and s_server. This causes real errors
   1771      to be returned by the verify callback instead of carrying on no matter
   1772      what. This reflects the way a "real world" verify callback would behave.
   1773      [Steve Henson]
   1774 
   1775   *) GOST engine, supporting several GOST algorithms and public key formats.
   1776      Kindly donated by Cryptocom.
   1777      [Cryptocom]
   1778 
   1779   *) Partial support for Issuing Distribution Point CRL extension. CRLs
   1780      partitioned by DP are handled but no indirect CRL or reason partitioning
   1781      (yet). Complete overhaul of CRL handling: now the most suitable CRL is
   1782      selected via a scoring technique which handles IDP and AKID in CRLs.
   1783      [Steve Henson]
   1784 
   1785   *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
   1786      will ultimately be used for all verify operations: this will remove the
   1787      X509_STORE dependency on certificate verification and allow alternative
   1788      lookup methods.  X509_STORE based implementations of these two callbacks.
   1789      [Steve Henson]
   1790 
   1791   *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
   1792      Modify get_crl() to find a valid (unexpired) CRL if possible.
   1793      [Steve Henson]
   1794 
   1795   *) New function X509_CRL_match() to check if two CRLs are identical. Normally
   1796      this would be called X509_CRL_cmp() but that name is already used by
   1797      a function that just compares CRL issuer names. Cache several CRL 
   1798      extensions in X509_CRL structure and cache CRLDP in X509.
   1799      [Steve Henson]
   1800 
   1801   *) Store a "canonical" representation of X509_NAME structure (ASN1 Name)
   1802      this maps equivalent X509_NAME structures into a consistent structure.
   1803      Name comparison can then be performed rapidly using memcmp().
   1804      [Steve Henson]
   1805 
   1806   *) Non-blocking OCSP request processing. Add -timeout option to ocsp 
   1807      utility.
   1808      [Steve Henson]
   1809 
   1810   *) Allow digests to supply their own micalg string for S/MIME type using
   1811      the ctrl EVP_MD_CTRL_MICALG.
   1812      [Steve Henson]
   1813 
   1814   *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the
   1815      EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
   1816      ctrl. It can then customise the structure before and/or after signing
   1817      if necessary.
   1818      [Steve Henson]
   1819 
   1820   *) New function OBJ_add_sigid() to allow application defined signature OIDs
   1821      to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
   1822      to free up any added signature OIDs.
   1823      [Steve Henson]
   1824 
   1825   *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
   1826      EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
   1827      digest and cipher tables. New options added to openssl utility:
   1828      list-message-digest-algorithms and list-cipher-algorithms.
   1829      [Steve Henson]
   1830 
   1831   *) Change the array representation of binary polynomials: the list
   1832      of degrees of non-zero coefficients is now terminated with -1.
   1833      Previously it was terminated with 0, which was also part of the
   1834      value; thus, the array representation was not applicable to
   1835      polynomials where t^0 has coefficient zero.  This change makes
   1836      the array representation useful in a more general context.
   1837      [Douglas Stebila]
   1838 
   1839   *) Various modifications and fixes to SSL/TLS cipher string
   1840      handling.  For ECC, the code now distinguishes between fixed ECDH
   1841      with RSA certificates on the one hand and with ECDSA certificates
   1842      on the other hand, since these are separate ciphersuites.  The
   1843      unused code for Fortezza ciphersuites has been removed.
   1844 
   1845      For consistency with EDH, ephemeral ECDH is now called "EECDH"
   1846      (not "ECDHE").  For consistency with the code for DH
   1847      certificates, use of ECDH certificates is now considered ECDH
   1848      authentication, not RSA or ECDSA authentication (the latter is
   1849      merely the CA's signing algorithm and not actively used in the
   1850      protocol).
   1851 
   1852      The temporary ciphersuite alias "ECCdraft" is no longer
   1853      available, and ECC ciphersuites are no longer excluded from "ALL"
   1854      and "DEFAULT".  The following aliases now exist for RFC 4492
   1855      ciphersuites, most of these by analogy with the DH case:
   1856 
   1857          kECDHr   - ECDH cert, signed with RSA
   1858          kECDHe   - ECDH cert, signed with ECDSA
   1859          kECDH    - ECDH cert (signed with either RSA or ECDSA)
   1860          kEECDH   - ephemeral ECDH
   1861          ECDH     - ECDH cert or ephemeral ECDH
   1862 
   1863          aECDH    - ECDH cert
   1864          aECDSA   - ECDSA cert
   1865          ECDSA    - ECDSA cert
   1866 
   1867          AECDH    - anonymous ECDH
   1868          EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
   1869 
   1870      [Bodo Moeller]
   1871 
   1872   *) Add additional S/MIME capabilities for AES and GOST ciphers if supported.
   1873      Use correct micalg parameters depending on digest(s) in signed message.
   1874      [Steve Henson]
   1875 
   1876   *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
   1877      an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
   1878      [Steve Henson]
   1879 
   1880   *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
   1881      an engine to register a method. Add ENGINE lookups for methods and
   1882      functional reference processing.
   1883      [Steve Henson]
   1884 
   1885   *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of
   1886      EVP_{Sign,Verify}* which allow an application to customise the signature
   1887      process.
   1888      [Steve Henson]
   1889 
   1890   *) New -resign option to smime utility. This adds one or more signers
   1891      to an existing PKCS#7 signedData structure. Also -md option to use an
   1892      alternative message digest algorithm for signing.
   1893      [Steve Henson]
   1894 
   1895   *) Tidy up PKCS#7 routines and add new functions to make it easier to
   1896      create PKCS7 structures containing multiple signers. Update smime
   1897      application to support multiple signers.
   1898      [Steve Henson]
   1899 
   1900   *) New -macalg option to pkcs12 utility to allow setting of an alternative
   1901      digest MAC.
   1902      [Steve Henson]
   1903 
   1904   *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
   1905      Reorganize PBE internals to lookup from a static table using NIDs,
   1906      add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
   1907      EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
   1908      PRF which will be automatically used with PBES2.
   1909      [Steve Henson]
   1910 
   1911   *) Replace the algorithm specific calls to generate keys in "req" with the
   1912      new API.
   1913      [Steve Henson]
   1914 
   1915   *) Update PKCS#7 enveloped data routines to use new API. This is now
   1916      supported by any public key method supporting the encrypt operation. A
   1917      ctrl is added to allow the public key algorithm to examine or modify
   1918      the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
   1919      a no op.
   1920      [Steve Henson]
   1921 
   1922   *) Add a ctrl to asn1 method to allow a public key algorithm to express
   1923      a default digest type to use. In most cases this will be SHA1 but some
   1924      algorithms (such as GOST) need to specify an alternative digest. The
   1925      return value indicates how strong the prefernce is 1 means optional and
   1926      2 is mandatory (that is it is the only supported type). Modify
   1927      ASN1_item_sign() to accept a NULL digest argument to indicate it should
   1928      use the default md. Update openssl utilities to use the default digest
   1929      type for signing if it is not explicitly indicated.
   1930      [Steve Henson]
   1931 
   1932   *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 
   1933      EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
   1934      signing method from the key type. This effectively removes the link
   1935      between digests and public key types.
   1936      [Steve Henson]
   1937 
   1938   *) Add an OID cross reference table and utility functions. Its purpose is to
   1939      translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
   1940      rsaEncryption. This will allow some of the algorithm specific hackery
   1941      needed to use the correct OID to be removed. 
   1942      [Steve Henson]
   1943 
   1944   *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
   1945      structures for PKCS7_sign(). They are now set up by the relevant public
   1946      key ASN1 method.
   1947      [Steve Henson]
   1948 
   1949   *) Add provisional EC pkey method with support for ECDSA and ECDH.
   1950      [Steve Henson]
   1951 
   1952   *) Add support for key derivation (agreement) in the API, DH method and
   1953      pkeyutl.
   1954      [Steve Henson]
   1955 
   1956   *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
   1957      public and private key formats. As a side effect these add additional 
   1958      command line functionality not previously available: DSA signatures can be
   1959      generated and verified using pkeyutl and DH key support and generation in
   1960      pkey, genpkey.
   1961      [Steve Henson]
   1962 
   1963   *) BeOS support.
   1964      [Oliver Tappe <zooey (a] hirschkaefer.de>]
   1965 
   1966   *) New make target "install_html_docs" installs HTML renditions of the
   1967      manual pages.
   1968      [Oliver Tappe <zooey (a] hirschkaefer.de>]
   1969 
   1970   *) New utility "genpkey" this is analagous to "genrsa" etc except it can
   1971      generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
   1972      support key and parameter generation and add initial key generation
   1973      functionality for RSA.
   1974      [Steve Henson]
   1975 
   1976   *) Add functions for main EVP_PKEY_method operations. The undocumented
   1977      functions EVP_PKEY_{encrypt,decrypt} have been renamed to
   1978      EVP_PKEY_{encrypt,decrypt}_old. 
   1979      [Steve Henson]
   1980 
   1981   *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public
   1982      key API, doesn't do much yet.
   1983      [Steve Henson]
   1984 
   1985   *) New function EVP_PKEY_asn1_get0_info() to retrieve information about
   1986      public key algorithms. New option to openssl utility:
   1987      "list-public-key-algorithms" to print out info.
   1988      [Steve Henson]
   1989 
   1990   *) Implement the Supported Elliptic Curves Extension for
   1991      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
   1992      [Douglas Stebila]
   1993 
   1994   *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
   1995      EVP_CIPHER structures to avoid later problems in EVP_cleanup().
   1996      [Steve Henson]
   1997 
   1998   *) New utilities pkey and pkeyparam. These are similar to algorithm specific
   1999      utilities such as rsa, dsa, dsaparam etc except they process any key
   2000      type.
   2001      [Steve Henson]
   2002 
   2003   *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 
   2004      functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
   2005      EVP_PKEY_print_param() to print public key data from an EVP_PKEY
   2006      structure.
   2007      [Steve Henson]
   2008 
   2009   *) Initial support for pluggable public key ASN1.
   2010      De-spaghettify the public key ASN1 handling. Move public and private
   2011      key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
   2012      algorithm specific handling to a single module within the relevant
   2013      algorithm directory. Add functions to allow (near) opaque processing
   2014      of public and private key structures.
   2015      [Steve Henson]
   2016 
   2017   *) Implement the Supported Point Formats Extension for
   2018      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
   2019      [Douglas Stebila]
   2020 
   2021   *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
   2022      for the psk identity [hint] and the psk callback functions to the
   2023      SSL_SESSION, SSL and SSL_CTX structure.
   2024      
   2025      New ciphersuites:
   2026          PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
   2027          PSK-AES256-CBC-SHA
   2028  
   2029      New functions:
   2030          SSL_CTX_use_psk_identity_hint
   2031          SSL_get_psk_identity_hint
   2032          SSL_get_psk_identity
   2033          SSL_use_psk_identity_hint
   2034 
   2035      [Mika Kousa and Pasi Eronen of Nokia Corporation]
   2036 
   2037   *) Add RFC 3161 compliant time stamp request creation, response generation
   2038      and response verification functionality.
   2039      [Zoltn Glzik <zglozik (a] opentsa.org>, The OpenTSA Project]
   2040 
   2041   *) Add initial support for TLS extensions, specifically for the server_name
   2042      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
   2043      have new members for a host name.  The SSL data structure has an
   2044      additional member SSL_CTX *initial_ctx so that new sessions can be
   2045      stored in that context to allow for session resumption, even after the
   2046      SSL has been switched to a new SSL_CTX in reaction to a client's
   2047      server_name extension.
   2048 
   2049      New functions (subject to change):
   2050 
   2051          SSL_get_servername()
   2052          SSL_get_servername_type()
   2053          SSL_set_SSL_CTX()
   2054 
   2055      New CTRL codes and macros (subject to change):
   2056 
   2057          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
   2058                                  - SSL_CTX_set_tlsext_servername_callback()
   2059          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
   2060                                       - SSL_CTX_set_tlsext_servername_arg()
   2061          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
   2062 
   2063      openssl s_client has a new '-servername ...' option.
   2064 
   2065      openssl s_server has new options '-servername_host ...', '-cert2 ...',
   2066      '-key2 ...', '-servername_fatal' (subject to change).  This allows
   2067      testing the HostName extension for a specific single host name ('-cert'
   2068      and '-key' remain fallbacks for handshakes without HostName
   2069      negotiation).  If the unrecogninzed_name alert has to be sent, this by
   2070      default is a warning; it becomes fatal with the '-servername_fatal'
   2071      option.
   2072 
   2073      [Peter Sylvester,  Remy Allais, Christophe Renou]
   2074 
   2075   *) Whirlpool hash implementation is added.
   2076      [Andy Polyakov]
   2077 
   2078   *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
   2079      bn(64,32). Because of instruction set limitations it doesn't have
   2080      any negative impact on performance. This was done mostly in order
   2081      to make it possible to share assembler modules, such as bn_mul_mont
   2082      implementations, between 32- and 64-bit builds without hassle.
   2083      [Andy Polyakov]
   2084 
   2085   *) Move code previously exiled into file crypto/ec/ec2_smpt.c
   2086      to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
   2087      macro.
   2088      [Bodo Moeller]
   2089 
   2090   *) New candidate for BIGNUM assembler implementation, bn_mul_mont,
   2091      dedicated Montgomery multiplication procedure, is introduced.
   2092      BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
   2093      "64-bit" performance on certain 32-bit targets.
   2094      [Andy Polyakov]
   2095 
   2096   *) New option SSL_OP_NO_COMP to disable use of compression selectively
   2097      in SSL structures. New SSL ctrl to set maximum send fragment size. 
   2098      Save memory by seeting the I/O buffer sizes dynamically instead of
   2099      using the maximum available value.
   2100      [Steve Henson]
   2101 
   2102   *) New option -V for 'openssl ciphers'. This prints the ciphersuite code
   2103      in addition to the text details.
   2104      [Bodo Moeller]
   2105 
   2106   *) Very, very preliminary EXPERIMENTAL support for printing of general
   2107      ASN1 structures. This currently produces rather ugly output and doesn't
   2108      handle several customised structures at all.
   2109      [Steve Henson]
   2110 
   2111   *) Integrated support for PVK file format and some related formats such
   2112      as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
   2113      these in the 'rsa' and 'dsa' utilities.
   2114      [Steve Henson]
   2115 
   2116   *) Support for PKCS#1 RSAPublicKey format on rsa utility command line.
   2117      [Steve Henson]
   2118 
   2119   *) Remove the ancient ASN1_METHOD code. This was only ever used in one
   2120      place for the (very old) "NETSCAPE" format certificates which are now
   2121      handled using new ASN1 code equivalents.
   2122      [Steve Henson]
   2123 
   2124   *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
   2125      pointer and make the SSL_METHOD parameter in SSL_CTX_new,
   2126      SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
   2127      [Nils Larsch]
   2128 
   2129   *) Modify CRL distribution points extension code to print out previously
   2130      unsupported fields. Enhance extension setting code to allow setting of
   2131      all fields.
   2132      [Steve Henson]
   2133 
   2134   *) Add print and set support for Issuing Distribution Point CRL extension.
   2135      [Steve Henson]
   2136 
   2137   *) Change 'Configure' script to enable Camellia by default.
   2138      [NTT]
   2139 
   2140  Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
   2141 
   2142   *) When rejecting SSL/TLS records due to an incorrect version number, never
   2143      update s->server with a new major version number.  As of
   2144      - OpenSSL 0.9.8m if 'short' is a 16-bit type,
   2145      - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
   2146      the previous behavior could result in a read attempt at NULL when
   2147      receiving specific incorrect SSL/TLS records once record payload
   2148      protection is active.  (CVE-2010-0740)
   2149      [Bodo Moeller, Adam Langley <agl (a] chromium.org>]
   2150 
   2151   *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 
   2152      could be crashed if the relevant tables were not present (e.g. chrooted).
   2153      [Tomas Hoger <thoger (a] redhat.com>]
   2154 
   2155  Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
   2156 
   2157   *) Always check bn_wexpend() return values for failure.  (CVE-2009-3245)
   2158      [Martin Olsson, Neel Mehta]
   2159 
   2160   *) Fix X509_STORE locking: Every 'objs' access requires a lock (to
   2161      accommodate for stack sorting, always a write lock!).
   2162      [Bodo Moeller]
   2163 
   2164   *) On some versions of WIN32 Heap32Next is very slow. This can cause
   2165      excessive delays in the RAND_poll(): over a minute. As a workaround
   2166      include a time check in the inner Heap32Next loop too.
   2167      [Steve Henson]
   2168 
   2169   *) The code that handled flushing of data in SSL/TLS originally used the
   2170      BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
   2171      the problem outlined in PR#1949. The fix suggested there however can
   2172      trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
   2173      of Apache). So instead simplify the code to flush unconditionally.
   2174      This should be fine since flushing with no data to flush is a no op.
   2175      [Steve Henson]
   2176 
   2177   *) Handle TLS versions 2.0 and later properly and correctly use the
   2178      highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
   2179      off ancient servers have a habit of sticking around for a while...
   2180      [Steve Henson]
   2181 
   2182   *) Modify compression code so it frees up structures without using the
   2183      ex_data callbacks. This works around a problem where some applications
   2184      call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
   2185      restarting) then use compression (e.g. SSL with compression) later.
   2186      This results in significant per-connection memory leaks and
   2187      has caused some security issues including CVE-2008-1678 and
   2188      CVE-2009-4355.
   2189      [Steve Henson]
   2190 
   2191   *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
   2192      change when encrypting or decrypting.
   2193      [Bodo Moeller]
   2194 
   2195   *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
   2196      connect and renegotiate with servers which do not support RI.
   2197      Until RI is more widely deployed this option is enabled by default.
   2198      [Steve Henson]
   2199 
   2200   *) Add "missing" ssl ctrls to clear options and mode.
   2201      [Steve Henson]
   2202 
   2203   *) If client attempts to renegotiate and doesn't support RI respond with
   2204      a no_renegotiation alert as required by RFC5746.  Some renegotiating
   2205      TLS clients will continue a connection gracefully when they receive
   2206      the alert. Unfortunately OpenSSL mishandled this alert and would hang
   2207      waiting for a server hello which it will never receive. Now we treat a
   2208      received no_renegotiation alert as a fatal error. This is because
   2209      applications requesting a renegotiation might well expect it to succeed
   2210      and would have no code in place to handle the server denying it so the
   2211      only safe thing to do is to terminate the connection.
   2212      [Steve Henson]
   2213 
   2214   *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
   2215      peer supports secure renegotiation and 0 otherwise. Print out peer
   2216      renegotiation support in s_client/s_server.
   2217      [Steve Henson]
   2218 
   2219   *) Replace the highly broken and deprecated SPKAC certification method with
   2220      the updated NID creation version. This should correctly handle UTF8.
   2221      [Steve Henson]
   2222 
   2223   *) Implement RFC5746. Re-enable renegotiation but require the extension
   2224      as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
   2225      turns out to be a bad idea. It has been replaced by
   2226      SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
   2227      SSL_CTX_set_options(). This is really not recommended unless you
   2228      know what you are doing.
   2229      [Eric Rescorla <ekr (a] networkresonance.com>, Ben Laurie, Steve Henson]
   2230 
   2231   *) Fixes to stateless session resumption handling. Use initial_ctx when
   2232      issuing and attempting to decrypt tickets in case it has changed during
   2233      servername handling. Use a non-zero length session ID when attempting
   2234      stateless session resumption: this makes it possible to determine if
   2235      a resumption has occurred immediately after receiving server hello
   2236      (several places in OpenSSL subtly assume this) instead of later in
   2237      the handshake.
   2238      [Steve Henson]
   2239 
   2240   *) The functions ENGINE_ctrl(), OPENSSL_isservice(),
   2241      CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
   2242      fixes for a few places where the return code is not checked
   2243      correctly.
   2244      [Julia Lawall <julia (a] diku.dk>]
   2245 
   2246   *) Add --strict-warnings option to Configure script to include devteam
   2247      warnings in other configurations.
   2248      [Steve Henson]
   2249 
   2250   *) Add support for --libdir option and LIBDIR variable in makefiles. This
   2251      makes it possible to install openssl libraries in locations which
   2252      have names other than "lib", for example "/usr/lib64" which some
   2253      systems need.
   2254      [Steve Henson, based on patch from Jeremy Utley]
   2255 
   2256   *) Don't allow the use of leading 0x80 in OIDs. This is a violation of
   2257      X690 8.9.12 and can produce some misleading textual output of OIDs.
   2258      [Steve Henson, reported by Dan Kaminsky]
   2259 
   2260   *) Delete MD2 from algorithm tables. This follows the recommendation in
   2261      several standards that it is not used in new applications due to
   2262      several cryptographic weaknesses. For binary compatibility reasons
   2263      the MD2 API is still compiled in by default.
   2264      [Steve Henson]
   2265 
   2266   *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
   2267      and restored.
   2268      [Steve Henson]
   2269 
   2270   *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
   2271      OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
   2272      clash.
   2273      [Guenter <lists (a] gknw.net>]
   2274 
   2275   *) Fix the server certificate chain building code to use X509_verify_cert(),
   2276      it used to have an ad-hoc builder which was unable to cope with anything
   2277      other than a simple chain.
   2278      [David Woodhouse <dwmw2 (a] infradead.org>, Steve Henson]
   2279 
   2280   *) Don't check self signed certificate signatures in X509_verify_cert()
   2281      by default (a flag can override this): it just wastes time without
   2282      adding any security. As a useful side effect self signed root CAs
   2283      with non-FIPS digests are now usable in FIPS mode.
   2284      [Steve Henson]
   2285 
   2286   *) In dtls1_process_out_of_seq_message() the check if the current message
   2287      is already buffered was missing. For every new message was memory
   2288      allocated, allowing an attacker to perform an denial of service attack
   2289      with sending out of seq handshake messages until there is no memory
   2290      left. Additionally every future messege was buffered, even if the
   2291      sequence number made no sense and would be part of another handshake.
   2292      So only messages with sequence numbers less than 10 in advance will be
   2293      buffered.  (CVE-2009-1378)
   2294      [Robin Seggelmann, discovered by Daniel Mentz] 	
   2295 
   2296   *) Records are buffered if they arrive with a future epoch to be
   2297      processed after finishing the corresponding handshake. There is
   2298      currently no limitation to this buffer allowing an attacker to perform
   2299      a DOS attack with sending records with future epochs until there is no
   2300      memory left. This patch adds the pqueue_size() function to detemine
   2301      the size of a buffer and limits the record buffer to 100 entries.
   2302      (CVE-2009-1377)
   2303      [Robin Seggelmann, discovered by Daniel Mentz] 	
   2304 
   2305   *) Keep a copy of frag->msg_header.frag_len so it can be used after the
   2306      parent structure is freed.  (CVE-2009-1379)
   2307      [Daniel Mentz] 	
   2308 
   2309   *) Handle non-blocking I/O properly in SSL_shutdown() call.
   2310      [Darryl Miles <darryl-mailinglists (a] netbauds.net>]
   2311 
   2312   *) Add 2.5.4.* OIDs
   2313      [Ilya O. <vrghost (a] gmail.com>]
   2314 
   2315  Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
   2316 
   2317   *) Disable renegotiation completely - this fixes a severe security
   2318      problem (CVE-2009-3555) at the cost of breaking all
   2319      renegotiation. Renegotiation can be re-enabled by setting
   2320      SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
   2321      run-time. This is really not recommended unless you know what
   2322      you're doing.
   2323      [Ben Laurie]
   2324 
   2325  Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
   2326 
   2327   *) Don't set val to NULL when freeing up structures, it is freed up by
   2328      underlying code. If sizeof(void *) > sizeof(long) this can result in
   2329      zeroing past the valid field. (CVE-2009-0789)
   2330      [Paolo Ganci <Paolo.Ganci (a] AdNovum.CH>]
   2331 
   2332   *) Fix bug where return value of CMS_SignerInfo_verify_content() was not
   2333      checked correctly. This would allow some invalid signed attributes to
   2334      appear to verify correctly. (CVE-2009-0591)
   2335      [Ivan Nestlerode <inestlerode (a] us.ibm.com>]
   2336 
   2337   *) Reject UniversalString and BMPString types with invalid lengths. This
   2338      prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
   2339      a legal length. (CVE-2009-0590)
   2340      [Steve Henson]
   2341 
   2342   *) Set S/MIME signing as the default purpose rather than setting it 
   2343      unconditionally. This allows applications to override it at the store
   2344      level.
   2345      [Steve Henson]
   2346 
   2347   *) Permit restricted recursion of ASN1 strings. This is needed in practice
   2348      to handle some structures.
   2349      [Steve Henson]
   2350 
   2351   *) Improve efficiency of mem_gets: don't search whole buffer each time
   2352      for a '\n'
   2353      [Jeremy Shapiro <jnshapir (a] us.ibm.com>]
   2354 
   2355   *) New -hex option for openssl rand.
   2356      [Matthieu Herrb]
   2357 
   2358   *) Print out UTF8String and NumericString when parsing ASN1.
   2359      [Steve Henson]
   2360 
   2361   *) Support NumericString type for name components.
   2362      [Steve Henson]
   2363 
   2364   *) Allow CC in the environment to override the automatically chosen
   2365      compiler. Note that nothing is done to ensure flags work with the
   2366      chosen compiler.
   2367      [Ben Laurie]
   2368 
   2369  Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
   2370 
   2371   *) Properly check EVP_VerifyFinal() and similar return values
   2372      (CVE-2008-5077).
   2373      [Ben Laurie, Bodo Moeller, Google Security Team]
   2374 
   2375   *) Enable TLS extensions by default.
   2376      [Ben Laurie]
   2377 
   2378   *) Allow the CHIL engine to be loaded, whether the application is
   2379      multithreaded or not. (This does not release the developer from the
   2380      obligation to set up the dynamic locking callbacks.)
   2381      [Sander Temme <sander (a] temme.net>]
   2382 
   2383   *) Use correct exit code if there is an error in dgst command.
   2384      [Steve Henson; problem pointed out by Roland Dirlewanger]
   2385 
   2386   *) Tweak Configure so that you need to say "experimental-jpake" to enable
   2387      JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
   2388      [Bodo Moeller]
   2389 
   2390   *) Add experimental JPAKE support, including demo authentication in
   2391      s_client and s_server.
   2392      [Ben Laurie]
   2393 
   2394   *) Set the comparison function in v3_addr_canonize().
   2395      [Rob Austein <sra (a] hactrn.net>]
   2396 
   2397   *) Add support for XMPP STARTTLS in s_client.
   2398      [Philip Paeps <philip (a] freebsd.org>]
   2399 
   2400   *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
   2401      to ensure that even with this option, only ciphersuites in the
   2402      server's preference list will be accepted.  (Note that the option
   2403      applies only when resuming a session, so the earlier behavior was
   2404      just about the algorithm choice for symmetric cryptography.)
   2405      [Bodo Moeller]
   2406 
   2407  Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
   2408 
   2409   *) Fix NULL pointer dereference if a DTLS server received
   2410      ChangeCipherSpec as first record (CVE-2009-1386).
   2411      [PR #1679]
   2412 
   2413   *) Fix a state transitition in s3_srvr.c and d1_srvr.c
   2414      (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
   2415      [Nagendra Modadugu]
   2416 
   2417   *) The fix in 0.9.8c that supposedly got rid of unsafe
   2418      double-checked locking was incomplete for RSA blinding,
   2419      addressing just one layer of what turns out to have been
   2420      doubly unsafe triple-checked locking.
   2421 
   2422      So now fix this for real by retiring the MONT_HELPER macro
   2423      in crypto/rsa/rsa_eay.c.
   2424 
   2425      [Bodo Moeller; problem pointed out by Marius Schilder]
   2426 
   2427   *) Various precautionary measures:
   2428 
   2429      - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
   2430 
   2431      - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
   2432        (NB: This would require knowledge of the secret session ticket key
   2433        to exploit, in which case you'd be SOL either way.)
   2434 
   2435      - Change bn_nist.c so that it will properly handle input BIGNUMs
   2436        outside the expected range.
   2437 
   2438      - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
   2439        builds.
   2440 
   2441      [Neel Mehta, Bodo Moeller]
   2442 
   2443   *) Allow engines to be "soft loaded" - i.e. optionally don't die if
   2444      the load fails. Useful for distros.
   2445      [Ben Laurie and the FreeBSD team]
   2446 
   2447   *) Add support for Local Machine Keyset attribute in PKCS#12 files.
   2448      [Steve Henson]
   2449 
   2450   *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
   2451      [Huang Ying]
   2452 
   2453   *) Expand ENGINE to support engine supplied SSL client certificate functions.
   2454 
   2455      This work was sponsored by Logica.
   2456      [Steve Henson]
   2457 
   2458   *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
   2459      keystores. Support for SSL/TLS client authentication too.
   2460      Not compiled unless enable-capieng specified to Configure.
   2461 
   2462      This work was sponsored by Logica.
   2463      [Steve Henson]
   2464 
   2465   *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using
   2466      ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
   2467      attribute creation routines such as certifcate requests and PKCS#12
   2468      files.
   2469      [Steve Henson]
   2470 
   2471  Changes between 0.9.8g and 0.9.8h  [28 May 2008]
   2472 
   2473   *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
   2474      handshake which could lead to a cilent crash as found using the
   2475      Codenomicon TLS test suite (CVE-2008-1672) 
   2476      [Steve Henson, Mark Cox]
   2477 
   2478   *) Fix double free in TLS server name extensions which could lead to
   2479      a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 
   2480      [Joe Orton]
   2481 
   2482   *) Clear error queue in SSL_CTX_use_certificate_chain_file()
   2483 
   2484      Clear the error queue to ensure that error entries left from
   2485      older function calls do not interfere with the correct operation.
   2486      [Lutz Jaenicke, Erik de Castro Lopo]
   2487 
   2488   *) Remove root CA certificates of commercial CAs:
   2489 
   2490      The OpenSSL project does not recommend any specific CA and does not
   2491      have any policy with respect to including or excluding any CA.
   2492      Therefore it does not make any sense to ship an arbitrary selection
   2493      of root CA certificates with the OpenSSL software.
   2494      [Lutz Jaenicke]
   2495 
   2496   *) RSA OAEP patches to fix two separate invalid memory reads.
   2497      The first one involves inputs when 'lzero' is greater than
   2498      'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
   2499      before the beginning of from). The second one involves inputs where
   2500      the 'db' section contains nothing but zeroes (there is a one-byte
   2501      invalid read after the end of 'db').
   2502      [Ivan Nestlerode <inestlerode (a] us.ibm.com>]
   2503 
   2504   *) Partial backport from 0.9.9-dev:
   2505 
   2506      Introduce bn_mul_mont (dedicated Montgomery multiplication
   2507      procedure) as a candidate for BIGNUM assembler implementation.
   2508      While 0.9.9-dev uses assembler for various architectures, only
   2509      x86_64 is available by default here in the 0.9.8 branch, and
   2510      32-bit x86 is available through a compile-time setting.
   2511 
   2512      To try the 32-bit x86 assembler implementation, use Configure
   2513      option "enable-montasm" (which exists only for this backport).
   2514 
   2515      As "enable-montasm" for 32-bit x86 disclaims code stability
   2516      anyway, in this constellation we activate additional code
   2517      backported from 0.9.9-dev for further performance improvements,
   2518      namely BN_from_montgomery_word.  (To enable this otherwise,
   2519      e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
   2520 
   2521      [Andy Polyakov (backport partially by Bodo Moeller)]
   2522 
   2523   *) Add TLS session ticket callback. This allows an application to set
   2524      TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
   2525      values. This is useful for key rollover for example where several key
   2526      sets may exist with different names.
   2527      [Steve Henson]
   2528 
   2529   *) Reverse ENGINE-internal logic for caching default ENGINE handles.
   2530      This was broken until now in 0.9.8 releases, such that the only way
   2531      a registered ENGINE could be used (assuming it initialises
   2532      successfully on the host) was to explicitly set it as the default
   2533      for the relevant algorithms. This is in contradiction with 0.9.7
   2534      behaviour and the documentation. With this fix, when an ENGINE is
   2535      registered into a given algorithm's table of implementations, the
   2536      'uptodate' flag is reset so that auto-discovery will be used next
   2537      time a new context for that algorithm attempts to select an
   2538      implementation.
   2539      [Ian Lister (tweaked by Geoff Thorpe)]
   2540 
   2541   *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
   2542      implemention in the following ways:
   2543 
   2544      Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
   2545      hard coded.
   2546 
   2547      Lack of BER streaming support means one pass streaming processing is
   2548      only supported if data is detached: setting the streaming flag is
   2549      ignored for embedded content.
   2550 
   2551      CMS support is disabled by default and must be explicitly enabled
   2552      with the enable-cms configuration option.
   2553      [Steve Henson]
   2554 
   2555   *) Update the GMP engine glue to do direct copies between BIGNUM and
   2556      mpz_t when openssl and GMP use the same limb size. Otherwise the
   2557      existing "conversion via a text string export" trick is still used.
   2558      [Paul Sheer <paulsheer (a] gmail.com>]
   2559 
   2560   *) Zlib compression BIO. This is a filter BIO which compressed and
   2561      uncompresses any data passed through it.
   2562      [Steve Henson]
   2563 
   2564   *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
   2565      RFC3394 compatible AES key wrapping.
   2566      [Steve Henson]
   2567 
   2568   *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
   2569      sets string data without copying. X509_ALGOR_set0() and
   2570      X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
   2571      data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
   2572      from an X509_ATTRIBUTE structure optionally checking it occurs only
   2573      once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
   2574      data.
   2575      [Steve Henson]
   2576 
   2577   *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
   2578      to get the expected BN_FLG_CONSTTIME behavior.
   2579      [Bodo Moeller (Google)]
   2580   
   2581   *) Netware support:
   2582 
   2583      - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
   2584      - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
   2585      - added some more tests to do_tests.pl
   2586      - fixed RunningProcess usage so that it works with newer LIBC NDKs too
   2587      - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
   2588      - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
   2589        netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
   2590      - various changes to netware.pl to enable gcc-cross builds on Win32
   2591        platform
   2592      - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
   2593      - various changes to fix missing prototype warnings
   2594      - fixed x86nasm.pl to create correct asm files for NASM COFF output
   2595      - added AES, WHIRLPOOL and CPUID assembler code to build files
   2596      - added missing AES assembler make rules to mk1mf.pl
   2597      - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
   2598      [Guenter Knauf <eflash (a] gmx.net>]
   2599 
   2600   *) Implement certificate status request TLS extension defined in RFC3546.
   2601      A client can set the appropriate parameters and receive the encoded
   2602      OCSP response via a callback. A server can query the supplied parameters
   2603      and set the encoded OCSP response in the callback. Add simplified examples
   2604      to s_client and s_server.
   2605      [Steve Henson]
   2606 
   2607  Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
   2608 
   2609   *) Fix various bugs:
   2610      + Binary incompatibility of ssl_ctx_st structure
   2611      + DTLS interoperation with non-compliant servers
   2612      + Don't call get_session_cb() without proposed session
   2613      + Fix ia64 assembler code
   2614      [Andy Polyakov, Steve Henson]
   2615 
   2616  Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
   2617 
   2618   *) DTLS Handshake overhaul. There were longstanding issues with
   2619      OpenSSL DTLS implementation, which were making it impossible for
   2620      RFC 4347 compliant client to communicate with OpenSSL server.
   2621      Unfortunately just fixing these incompatibilities would "cut off"
   2622      pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
   2623      server keeps tolerating non RFC compliant syntax. The opposite is
   2624      not true, 0.9.8f client can not communicate with earlier server.
   2625      This update even addresses CVE-2007-4995.
   2626      [Andy Polyakov]
   2627 
   2628   *) Changes to avoid need for function casts in OpenSSL: some compilers
   2629      (gcc 4.2 and later) reject their use.
   2630      [Kurt Roeckx <kurt (a] roeckx.be>, Peter Hartley <pdh (a] utter.chaos.org.uk>,
   2631       Steve Henson]
   2632   
   2633   *) Add RFC4507 support to OpenSSL. This includes the corrections in
   2634      RFC4507bis. The encrypted ticket format is an encrypted encoded
   2635      SSL_SESSION structure, that way new session features are automatically
   2636      supported.
   2637 
   2638      If a client application caches session in an SSL_SESSION structure
   2639      support is transparent because tickets are now stored in the encoded
   2640      SSL_SESSION.
   2641      
   2642      The SSL_CTX structure automatically generates keys for ticket
   2643      protection in servers so again support should be possible
   2644      with no application modification.
   2645 
   2646      If a client or server wishes to disable RFC4507 support then the option
   2647      SSL_OP_NO_TICKET can be set.
   2648 
   2649      Add a TLS extension debugging callback to allow the contents of any client
   2650      or server extensions to be examined.
   2651 
   2652      This work was sponsored by Google.
   2653      [Steve Henson]
   2654 
   2655   *) Add initial support for TLS extensions, specifically for the server_name
   2656      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
   2657      have new members for a host name.  The SSL data structure has an
   2658      additional member SSL_CTX *initial_ctx so that new sessions can be
   2659      stored in that context to allow for session resumption, even after the
   2660      SSL has been switched to a new SSL_CTX in reaction to a client's
   2661      server_name extension.
   2662 
   2663      New functions (subject to change):
   2664 
   2665          SSL_get_servername()
   2666          SSL_get_servername_type()
   2667          SSL_set_SSL_CTX()
   2668 
   2669      New CTRL codes and macros (subject to change):
   2670 
   2671          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
   2672                                  - SSL_CTX_set_tlsext_servername_callback()
   2673          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
   2674                                       - SSL_CTX_set_tlsext_servername_arg()
   2675          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
   2676 
   2677      openssl s_client has a new '-servername ...' option.
   2678 
   2679      openssl s_server has new options '-servername_host ...', '-cert2 ...',
   2680      '-key2 ...', '-servername_fatal' (subject to change).  This allows
   2681      testing the HostName extension for a specific single host name ('-cert'
   2682      and '-key' remain fallbacks for handshakes without HostName
   2683      negotiation).  If the unrecogninzed_name alert has to be sent, this by
   2684      default is a warning; it becomes fatal with the '-servername_fatal'
   2685      option.
   2686 
   2687      [Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson]
   2688 
   2689   *) Add AES and SSE2 assembly language support to VC++ build.
   2690      [Steve Henson]
   2691 
   2692   *) Mitigate attack on final subtraction in Montgomery reduction.
   2693      [Andy Polyakov]
   2694 
   2695   *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
   2696      (which previously caused an internal error).
   2697      [Bodo Moeller]
   2698 
   2699   *) Squeeze another 10% out of IGE mode when in != out.
   2700      [Ben Laurie]
   2701 
   2702   *) AES IGE mode speedup.
   2703      [Dean Gaudet (Google)]
   2704 
   2705   *) Add the Korean symmetric 128-bit cipher SEED (see
   2706      http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
   2707      add SEED ciphersuites from RFC 4162:
   2708 
   2709         TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
   2710         TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
   2711         TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
   2712         TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
   2713 
   2714      To minimize changes between patchlevels in the OpenSSL 0.9.8
   2715      series, SEED remains excluded from compilation unless OpenSSL
   2716      is configured with 'enable-seed'.
   2717      [KISA, Bodo Moeller]
   2718 
   2719   *) Mitigate branch prediction attacks, which can be practical if a
   2720      single processor is shared, allowing a spy process to extract
   2721      information.  For detailed background information, see
   2722      http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
   2723      J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
   2724      and Necessary Software Countermeasures").  The core of the change
   2725      are new versions BN_div_no_branch() and
   2726      BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
   2727      respectively, which are slower, but avoid the security-relevant
   2728      conditional branches.  These are automatically called by BN_div()
   2729      and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
   2730      of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
   2731      remove a conditional branch.
   2732 
   2733      BN_FLG_CONSTTIME is the new name for the previous
   2734      BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
   2735      modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
   2736      in the exponent causes BN_mod_exp_mont() to use the alternative
   2737      implementation in BN_mod_exp_mont_consttime().)  The old name
   2738      remains as a deprecated alias.
   2739 
   2740      Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
   2741      RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
   2742      constant-time implementations for more than just exponentiation.
   2743      Here too the old name is kept as a deprecated alias.
   2744 
   2745      BN_BLINDING_new() will now use BN_dup() for the modulus so that
   2746      the BN_BLINDING structure gets an independent copy of the
   2747      modulus.  This means that the previous "BIGNUM *m" argument to
   2748      BN_BLINDING_new() and to BN_BLINDING_create_param() now
   2749      essentially becomes "const BIGNUM *m", although we can't actually
   2750      change this in the header file before 0.9.9.  It allows
   2751      RSA_setup_blinding() to use BN_with_flags() on the modulus to
   2752      enable BN_FLG_CONSTTIME.
   2753 
   2754      [Matthew D Wood (Intel Corp)]
   2755 
   2756   *) In the SSL/TLS server implementation, be strict about session ID
   2757      context matching (which matters if an application uses a single
   2758      external cache for different purposes).  Previously,
   2759      out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
   2760      set.  This did ensure strict client verification, but meant that,
   2761      with applications using a single external cache for quite
   2762      different requirements, clients could circumvent ciphersuite
   2763      restrictions for a given session ID context by starting a session
   2764      in a different context.
   2765      [Bodo Moeller]
   2766 
   2767   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
   2768      a ciphersuite string such as "DEFAULT:RSA" cannot enable
   2769      authentication-only ciphersuites.
   2770      [Bodo Moeller]
   2771 
   2772   *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
   2773      not complete and could lead to a possible single byte overflow
   2774      (CVE-2007-5135) [Ben Laurie]
   2775 
   2776  Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
   2777 
   2778   *) Since AES128 and AES256 (and similarly Camellia128 and
   2779      Camellia256) share a single mask bit in the logic of
   2780      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
   2781      kludge to work properly if AES128 is available and AES256 isn't
   2782      (or if Camellia128 is available and Camellia256 isn't).
   2783      [Victor Duchovni]
   2784 
   2785   *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
   2786      (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
   2787      When a point or a seed is encoded in a BIT STRING, we need to
   2788      prevent the removal of trailing zero bits to get the proper DER
   2789      encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
   2790      of a NamedBitList, for which trailing 0 bits need to be removed.)
   2791      [Bodo Moeller]
   2792 
   2793   *) Have SSL/TLS server implementation tolerate "mismatched" record
   2794      protocol version while receiving ClientHello even if the
   2795      ClientHello is fragmented.  (The server can't insist on the
   2796      particular protocol version it has chosen before the ServerHello
   2797      message has informed the client about his choice.)
   2798      [Bodo Moeller]
   2799 
   2800   *) Add RFC 3779 support.
   2801      [Rob Austein for ARIN, Ben Laurie]
   2802 
   2803   *) Load error codes if they are not already present instead of using a
   2804      static variable. This allows them to be cleanly unloaded and reloaded.
   2805      Improve header file function name parsing.
   2806      [Steve Henson]
   2807 
   2808   *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
   2809      or CAPABILITY handshake as required by RFCs.
   2810      [Goetz Babin-Ebell]
   2811 
   2812  Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
   2813 
   2814   *) Introduce limits to prevent malicious keys being able to
   2815      cause a denial of service.  (CVE-2006-2940)
   2816      [Steve Henson, Bodo Moeller]
   2817 
   2818   *) Fix ASN.1 parsing of certain invalid structures that can result
   2819      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
   2820 
   2821   *) Fix buffer overflow in SSL_get_shared_ciphers() function. 
   2822      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
   2823 
   2824   *) Fix SSL client code which could crash if connecting to a
   2825      malicious SSLv2 server.  (CVE-2006-4343)
   2826      [Tavis Ormandy and Will Drewry, Google Security Team]
   2827 
   2828   *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
   2829      match only those.  Before that, "AES256-SHA" would be interpreted
   2830      as a pattern and match "AES128-SHA" too (since AES128-SHA got
   2831      the same strength classification in 0.9.7h) as we currently only
   2832      have a single AES bit in the ciphersuite description bitmap.
   2833      That change, however, also applied to ciphersuite strings such as
   2834      "RC4-MD5" that intentionally matched multiple ciphersuites --
   2835      namely, SSL 2.0 ciphersuites in addition to the more common ones
   2836      from SSL 3.0/TLS 1.0.
   2837 
   2838      So we change the selection algorithm again: Naming an explicit
   2839      ciphersuite selects this one ciphersuite, and any other similar
   2840      ciphersuite (same bitmap) from *other* protocol versions.
   2841      Thus, "RC4-MD5" again will properly select both the SSL 2.0
   2842      ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
   2843 
   2844      Since SSL 2.0 does not have any ciphersuites for which the
   2845      128/256 bit distinction would be relevant, this works for now.
   2846      The proper fix will be to use different bits for AES128 and
   2847      AES256, which would have avoided the problems from the beginning;
   2848      however, bits are scarce, so we can only do this in a new release
   2849      (not just a patchlevel) when we can change the SSL_CIPHER
   2850      definition to split the single 'unsigned long mask' bitmap into
   2851      multiple values to extend the available space.
   2852 
   2853      [Bodo Moeller]
   2854 
   2855  Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
   2856 
   2857   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
   2858      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
   2859 
   2860   *) Add AES IGE and biIGE modes.
   2861      [Ben Laurie]
   2862 
   2863   *) Change the Unix randomness entropy gathering to use poll() when
   2864      possible instead of select(), since the latter has some
   2865      undesirable limitations.
   2866      [Darryl Miles via Richard Levitte and Bodo Moeller]
   2867 
   2868   *) Disable "ECCdraft" ciphersuites more thoroughly.  Now special
   2869      treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
   2870      cannot be implicitly activated as part of, e.g., the "AES" alias.
   2871      However, please upgrade to OpenSSL 0.9.9[-dev] for
   2872      non-experimental use of the ECC ciphersuites to get TLS extension
   2873      support, which is required for curve and point format negotiation
   2874      to avoid potential handshake problems.
   2875      [Bodo Moeller]
   2876 
   2877   *) Disable rogue ciphersuites:
   2878 
   2879       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
   2880       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
   2881       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
   2882 
   2883      The latter two were purportedly from
   2884      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
   2885      appear there.
   2886 
   2887      Also deactivate the remaining ciphersuites from
   2888      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
   2889      unofficial, and the ID has long expired.
   2890      [Bodo Moeller]
   2891 
   2892   *) Fix RSA blinding Heisenbug (problems sometimes occured on
   2893      dual-core machines) and other potential thread-safety issues.
   2894      [Bodo Moeller]
   2895 
   2896   *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
   2897      versions), which is now available for royalty-free use
   2898      (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
   2899      Also, add Camellia TLS ciphersuites from RFC 4132.
   2900 
   2901      To minimize changes between patchlevels in the OpenSSL 0.9.8
   2902      series, Camellia remains excluded from compilation unless OpenSSL
   2903      is configured with 'enable-camellia'.
   2904      [NTT]
   2905 
   2906   *) Disable the padding bug check when compression is in use. The padding
   2907      bug check assumes the first packet is of even length, this is not
   2908      necessarily true if compresssion is enabled and can result in false
   2909      positives causing handshake failure. The actual bug test is ancient
   2910      code so it is hoped that implementations will either have fixed it by
   2911      now or any which still have the bug do not support compression.
   2912      [Steve Henson]
   2913 
   2914  Changes between 0.9.8a and 0.9.8b  [04 May 2006]
   2915 
   2916   *) When applying a cipher rule check to see if string match is an explicit
   2917      cipher suite and only match that one cipher suite if it is.
   2918      [Steve Henson]
   2919 
   2920   *) Link in manifests for VC++ if needed.
   2921      [Austin Ziegler <halostatue (a] gmail.com>]
   2922 
   2923   *) Update support for ECC-based TLS ciphersuites according to
   2924      draft-ietf-tls-ecc-12.txt with proposed changes (but without
   2925      TLS extensions, which are supported starting with the 0.9.9
   2926      branch, not in the OpenSSL 0.9.8 branch).
   2927      [Douglas Stebila]
   2928 
   2929   *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
   2930      opaque EVP_CIPHER_CTX handling.
   2931      [Steve Henson]
   2932 
   2933   *) Fixes and enhancements to zlib compression code. We now only use
   2934      "zlib1.dll" and use the default __cdecl calling convention on Win32
   2935      to conform with the standards mentioned here:
   2936            http://www.zlib.net/DLL_FAQ.txt
   2937      Static zlib linking now works on Windows and the new --with-zlib-include
   2938      --with-zlib-lib options to Configure can be used to supply the location
   2939      of the headers and library. Gracefully handle case where zlib library
   2940      can't be loaded.
   2941      [Steve Henson]
   2942 
   2943   *) Several fixes and enhancements to the OID generation code. The old code
   2944      sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
   2945      handle numbers larger than ULONG_MAX, truncated printing and had a
   2946      non standard OBJ_obj2txt() behaviour.
   2947      [Steve Henson]
   2948 
   2949   *) Add support for building of engines under engine/ as shared libraries
   2950      under VC++ build system.
   2951      [Steve Henson]
   2952 
   2953   *) Corrected the numerous bugs in the Win32 path splitter in DSO.
   2954      Hopefully, we will not see any false combination of paths any more.
   2955      [Richard Levitte]
   2956 
   2957  Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
   2958 
   2959   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
   2960      (part of SSL_OP_ALL).  This option used to disable the
   2961      countermeasure against man-in-the-middle protocol-version
   2962      rollback in the SSL 2.0 server implementation, which is a bad
   2963      idea.  (CVE-2005-2969)
   2964 
   2965      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
   2966      for Information Security, National Institute of Advanced Industrial
   2967      Science and Technology [AIST], Japan)]
   2968 
   2969   *) Add two function to clear and return the verify parameter flags.
   2970      [Steve Henson]
   2971 
   2972   *) Keep cipherlists sorted in the source instead of sorting them at
   2973      runtime, thus removing the need for a lock.
   2974      [Nils Larsch]
   2975 
   2976   *) Avoid some small subgroup attacks in Diffie-Hellman.
   2977      [Nick Mathewson and Ben Laurie]
   2978 
   2979   *) Add functions for well-known primes.
   2980      [Nick Mathewson]
   2981 
   2982   *) Extended Windows CE support.
   2983      [Satoshi Nakamura and Andy Polyakov]
   2984 
   2985   *) Initialize SSL_METHOD structures at compile time instead of during
   2986      runtime, thus removing the need for a lock.
   2987      [Steve Henson]
   2988 
   2989   *) Make PKCS7_decrypt() work even if no certificate is supplied by
   2990      attempting to decrypt each encrypted key in turn. Add support to
   2991      smime utility.
   2992      [Steve Henson]
   2993 
   2994  Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
   2995 
   2996   [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
   2997   OpenSSL 0.9.8.]
   2998 
   2999   *) Add libcrypto.pc and libssl.pc for those who feel they need them.
   3000      [Richard Levitte]
   3001 
   3002   *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
   3003      key into the same file any more.
   3004      [Richard Levitte]
   3005 
   3006   *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
   3007      [Andy Polyakov]
   3008 
   3009   *) Add -utf8 command line and config file option to 'ca'.
   3010      [Stefan <stf (a] udoma.org]
   3011 
   3012   *) Removed the macro des_crypt(), as it seems to conflict with some
   3013      libraries.  Use DES_crypt().
   3014      [Richard Levitte]
   3015 
   3016   *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
   3017      involves renaming the source and generated shared-libs for
   3018      both. The engines will accept the corrected or legacy ids
   3019      ('ncipher' and '4758_cca' respectively) when binding. NB,
   3020      this only applies when building 'shared'.
   3021      [Corinna Vinschen <vinschen (a] redhat.com> and Geoff Thorpe]
   3022 
   3023   *) Add attribute functions to EVP_PKEY structure. Modify
   3024      PKCS12_create() to recognize a CSP name attribute and
   3025      use it. Make -CSP option work again in pkcs12 utility.
   3026      [Steve Henson]
   3027 
   3028   *) Add new functionality to the bn blinding code:
   3029      - automatic re-creation of the BN_BLINDING parameters after
   3030        a fixed number of uses (currently 32)
   3031      - add new function for parameter creation
   3032      - introduce flags to control the update behaviour of the
   3033        BN_BLINDING parameters
   3034      - hide BN_BLINDING structure
   3035      Add a second BN_BLINDING slot to the RSA structure to improve
   3036      performance when a single RSA object is shared among several
   3037      threads.
   3038      [Nils Larsch]
   3039 
   3040   *) Add support for DTLS.
   3041      [Nagendra Modadugu <nagendra (a] cs.stanford.edu> and Ben Laurie]
   3042 
   3043   *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
   3044      to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
   3045      [Walter Goulet]
   3046 
   3047   *) Remove buggy and incompletet DH cert support from
   3048      ssl/ssl_rsa.c and ssl/s3_both.c
   3049      [Nils Larsch]
   3050 
   3051   *) Use SHA-1 instead of MD5 as the default digest algorithm for
   3052      the apps/openssl applications.
   3053      [Nils Larsch]
   3054 
   3055   *) Compile clean with "-Wall -Wmissing-prototypes
   3056      -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
   3057      DEBUG_SAFESTACK must also be set.
   3058      [Ben Laurie]
   3059 
   3060   *) Change ./Configure so that certain algorithms can be disabled by default.
   3061      The new counterpiece to "no-xxx" is "enable-xxx".
   3062 
   3063      The patented RC5 and MDC2 algorithms will now be disabled unless
   3064      "enable-rc5" and "enable-mdc2", respectively, are specified.
   3065 
   3066      (IDEA remains enabled despite being patented.  This is because IDEA
   3067      is frequently required for interoperability, and there is no license
   3068      fee for non-commercial use.  As before, "no-idea" can be used to
   3069      avoid this algorithm.)
   3070 
   3071      [Bodo Moeller]
   3072 
   3073   *) Add processing of proxy certificates (see RFC 3820).  This work was
   3074      sponsored by KTH (The Royal Institute of Technology in Stockholm) and
   3075      EGEE (Enabling Grids for E-science in Europe).
   3076      [Richard Levitte]
   3077 
   3078   *) RC4 performance overhaul on modern architectures/implementations, such
   3079      as Intel P4, IA-64 and AMD64.
   3080      [Andy Polyakov]
   3081 
   3082   *) New utility extract-section.pl. This can be used specify an alternative
   3083      section number in a pod file instead of having to treat each file as
   3084      a separate case in Makefile. This can be done by adding two lines to the
   3085      pod file:
   3086 
   3087      =for comment openssl_section:XXX
   3088 
   3089      The blank line is mandatory.
   3090 
   3091      [Steve Henson]
   3092 
   3093   *) New arguments -certform, -keyform and -pass for s_client and s_server
   3094      to allow alternative format key and certificate files and passphrase
   3095      sources.
   3096      [Steve Henson]
   3097 
   3098   *) New structure X509_VERIFY_PARAM which combines current verify parameters,
   3099      update associated structures and add various utility functions.
   3100 
   3101      Add new policy related verify parameters, include policy checking in 
   3102      standard verify code. Enhance 'smime' application with extra parameters
   3103      to support policy checking and print out.
   3104      [Steve Henson]
   3105 
   3106   *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
   3107      Nehemiah processors. These extensions support AES encryption in hardware
   3108      as well as RNG (though RNG support is currently disabled).
   3109      [Michal Ludvig <michal (a] logix.cz>, with help from Andy Polyakov]
   3110 
   3111   *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
   3112      [Geoff Thorpe]
   3113 
   3114   *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
   3115      [Andy Polyakov and a number of other people]
   3116 
   3117   *) Improved PowerPC platform support. Most notably BIGNUM assembler
   3118      implementation contributed by IBM.
   3119      [Suresh Chari, Peter Waltenberg, Andy Polyakov]
   3120 
   3121   *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
   3122      exponent rather than 'unsigned long'. There is a corresponding change to
   3123      the new 'rsa_keygen' element of the RSA_METHOD structure.
   3124      [Jelte Jansen, Geoff Thorpe]
   3125 
   3126   *) Functionality for creating the initial serial number file is now
   3127      moved from CA.pl to the 'ca' utility with a new option -create_serial.
   3128 
   3129      (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
   3130      number file to 1, which is bound to cause problems.  To avoid
   3131      the problems while respecting compatibility between different 0.9.7
   3132      patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
   3133      CA.pl for serial number initialization.  With the new release 0.9.8,
   3134      we can fix the problem directly in the 'ca' utility.)
   3135      [Steve Henson]
   3136 
   3137   *) Reduced header interdepencies by declaring more opaque objects in
   3138      ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
   3139      give fewer recursive includes, which could break lazy source code - so
   3140      this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
   3141      developers should define this symbol when building and using openssl to
   3142      ensure they track the recommended behaviour, interfaces, [etc], but
   3143      backwards-compatible behaviour prevails when this isn't defined.
   3144      [Geoff Thorpe]
   3145 
   3146   *) New function X509_POLICY_NODE_print() which prints out policy nodes.
   3147      [Steve Henson]
   3148 
   3149   *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
   3150      This will generate a random key of the appropriate length based on the 
   3151      cipher context. The EVP_CIPHER can provide its own random key generation
   3152      routine to support keys of a specific form. This is used in the des and 
   3153      3des routines to generate a key of the correct parity. Update S/MIME
   3154      code to use new functions and hence generate correct parity DES keys.
   3155      Add EVP_CHECK_DES_KEY #define to return an error if the key is not 
   3156      valid (weak or incorrect parity).
   3157      [Steve Henson]
   3158 
   3159   *) Add a local set of CRLs that can be used by X509_verify_cert() as well
   3160      as looking them up. This is useful when the verified structure may contain
   3161      CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
   3162      present unless the new PKCS7_NO_CRL flag is asserted.
   3163      [Steve Henson]
   3164 
   3165   *) Extend ASN1 oid configuration module. It now additionally accepts the
   3166      syntax:
   3167 
   3168      shortName = some long name, 1.2.3.4
   3169      [Steve Henson]
   3170 
   3171   *) Reimplemented the BN_CTX implementation. There is now no more static
   3172      limitation on the number of variables it can handle nor the depth of the
   3173      "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
   3174      information can now expand as required, and rather than having a single
   3175      static array of bignums, BN_CTX now uses a linked-list of such arrays
   3176      allowing it to expand on demand whilst maintaining the usefulness of
   3177      BN_CTX's "bundling".
   3178      [Geoff Thorpe]
   3179 
   3180   *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
   3181      to allow all RSA operations to function using a single BN_CTX.
   3182      [Geoff Thorpe]
   3183 
   3184   *) Preliminary support for certificate policy evaluation and checking. This
   3185      is initially intended to pass the tests outlined in "Conformance Testing
   3186      of Relying Party Client Certificate Path Processing Logic" v1.07.
   3187      [Steve Henson]
   3188 
   3189   *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
   3190      remained unused and not that useful. A variety of other little bignum
   3191      tweaks and fixes have also been made continuing on from the audit (see
   3192      below).
   3193      [Geoff Thorpe]
   3194 
   3195   *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
   3196      associated ASN1, EVP and SSL functions and old ASN1 macros.
   3197      [Richard Levitte]
   3198 
   3199   *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
   3200      and this should never fail. So the return value from the use of
   3201      BN_set_word() (which can fail due to needless expansion) is now deprecated;
   3202      if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
   3203      [Geoff Thorpe]
   3204 
   3205   *) BN_CTX_get() should return zero-valued bignums, providing the same
   3206      initialised value as BN_new().
   3207      [Geoff Thorpe, suggested by Ulf Mller]
   3208 
   3209   *) Support for inhibitAnyPolicy certificate extension.
   3210      [Steve Henson]
   3211 
   3212   *) An audit of the BIGNUM code is underway, for which debugging code is
   3213      enabled when BN_DEBUG is defined. This makes stricter enforcements on what
   3214      is considered valid when processing BIGNUMs, and causes execution to
   3215      assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
   3216      further steps are taken to deliberately pollute unused data in BIGNUM
   3217      structures to try and expose faulty code further on. For now, openssl will
   3218      (in its default mode of operation) continue to tolerate the inconsistent
   3219      forms that it has tolerated in the past, but authors and packagers should
   3220      consider trying openssl and their own applications when compiled with
   3221      these debugging symbols defined. It will help highlight potential bugs in
   3222      their own code, and will improve the test coverage for OpenSSL itself. At
   3223      some point, these tighter rules will become openssl's default to improve
   3224      maintainability, though the assert()s and other overheads will remain only
   3225      in debugging configurations. See bn.h for more details.
   3226      [Geoff Thorpe, Nils Larsch, Ulf Mller]
   3227 
   3228   *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
   3229      that can only be obtained through BN_CTX_new() (which implicitly
   3230      initialises it). The presence of this function only made it possible
   3231      to overwrite an existing structure (and cause memory leaks).
   3232      [Geoff Thorpe]
   3233 
   3234   *) Because of the callback-based approach for implementing LHASH as a
   3235      template type, lh_insert() adds opaque objects to hash-tables and
   3236      lh_doall() or lh_doall_arg() are typically used with a destructor callback
   3237      to clean up those corresponding objects before destroying the hash table
   3238      (and losing the object pointers). So some over-zealous constifications in
   3239      LHASH have been relaxed so that lh_insert() does not take (nor store) the
   3240      objects as "const" and the lh_doall[_arg] callback wrappers are not
   3241      prototyped to have "const" restrictions on the object pointers they are
   3242      given (and so aren't required to cast them away any more).
   3243      [Geoff Thorpe]
   3244 
   3245   *) The tmdiff.h API was so ugly and minimal that our own timing utility
   3246      (speed) prefers to use its own implementation. The two implementations
   3247      haven't been consolidated as yet (volunteers?) but the tmdiff API has had
   3248      its object type properly exposed (MS_TM) instead of casting to/from "char
   3249      *". This may still change yet if someone realises MS_TM and "ms_time_***"
   3250      aren't necessarily the greatest nomenclatures - but this is what was used
   3251      internally to the implementation so I've used that for now.
   3252      [Geoff Thorpe]
   3253 
   3254   *) Ensure that deprecated functions do not get compiled when
   3255      OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
   3256      the self-tests were still using deprecated key-generation functions so
   3257      these have been updated also.
   3258      [Geoff Thorpe]
   3259 
   3260   *) Reorganise PKCS#7 code to separate the digest location functionality
   3261      into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
   3262      New function PKCS7_set_digest() to set the digest type for PKCS#7
   3263      digestedData type. Add additional code to correctly generate the
   3264      digestedData type and add support for this type in PKCS7 initialization
   3265      functions.
   3266      [Steve Henson]
   3267 
   3268   *) New function PKCS7_set0_type_other() this initializes a PKCS7 
   3269      structure of type "other".
   3270      [Steve Henson]
   3271 
   3272   *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
   3273      sure the loop does correctly stop and breaking ("division by zero")
   3274      modulus operations are not performed. The (pre-generated) prime
   3275      table crypto/bn/bn_prime.h was already correct, but it could not be
   3276      re-generated on some platforms because of the "division by zero"
   3277      situation in the script.
   3278      [Ralf S. Engelschall]
   3279 
   3280   *) Update support for ECC-based TLS ciphersuites according to
   3281      draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
   3282      SHA-1 now is only used for "small" curves (where the
   3283      representation of a field element takes up to 24 bytes); for
   3284      larger curves, the field element resulting from ECDH is directly
   3285      used as premaster secret.
   3286      [Douglas Stebila (Sun Microsystems Laboratories)]
   3287 
   3288   *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
   3289      curve secp160r1 to the tests.
   3290      [Douglas Stebila (Sun Microsystems Laboratories)]
   3291 
   3292   *) Add the possibility to load symbols globally with DSO.
   3293      [Gtz Babin-Ebell <babin-ebell (a] trustcenter.de> via Richard Levitte]
   3294 
   3295   *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
   3296      control of the error stack.
   3297      [Richard Levitte]
   3298 
   3299   *) Add support for STORE in ENGINE.
   3300      [Richard Levitte]
   3301 
   3302   *) Add the STORE type.  The intention is to provide a common interface
   3303      to certificate and key stores, be they simple file-based stores, or
   3304      HSM-type store, or LDAP stores, or...
   3305      NOTE: The code is currently UNTESTED and isn't really used anywhere.
   3306      [Richard Levitte]
   3307 
   3308   *) Add a generic structure called OPENSSL_ITEM.  This can be used to
   3309      pass a list of arguments to any function as well as provide a way
   3310      for a function to pass data back to the caller.
   3311      [Richard Levitte]
   3312 
   3313   *) Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
   3314      works like BUF_strdup() but can be used to duplicate a portion of
   3315      a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
   3316      a memory area.
   3317      [Richard Levitte]
   3318 
   3319   *) Add the function sk_find_ex() which works like sk_find(), but will
   3320      return an index to an element even if an exact match couldn't be
   3321      found.  The index is guaranteed to point at the element where the
   3322      searched-for key would be inserted to preserve sorting order.
   3323      [Richard Levitte]
   3324 
   3325   *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
   3326      takes an extra flags argument for optional functionality.  Currently,
   3327      the following flags are defined:
   3328 
   3329 	OBJ_BSEARCH_VALUE_ON_NOMATCH
   3330 	This one gets OBJ_bsearch_ex() to return a pointer to the first
   3331 	element where the comparing function returns a negative or zero
   3332 	number.
   3333 
   3334 	OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
   3335 	This one gets OBJ_bsearch_ex() to return a pointer to the first
   3336 	element where the comparing function returns zero.  This is useful
   3337 	if there are more than one element where the comparing function
   3338 	returns zero.
   3339      [Richard Levitte]
   3340 
   3341   *) Make it possible to create self-signed certificates with 'openssl ca'
   3342      in such a way that the self-signed certificate becomes part of the
   3343      CA database and uses the same mechanisms for serial number generation
   3344      as all other certificate signing.  The new flag '-selfsign' enables
   3345      this functionality.  Adapt CA.sh and CA.pl.in.
   3346      [Richard Levitte]
   3347 
   3348   *) Add functionality to check the public key of a certificate request
   3349      against a given private.  This is useful to check that a certificate
   3350      request can be signed by that key (self-signing).
   3351      [Richard Levitte]
   3352 
   3353   *) Make it possible to have multiple active certificates with the same
   3354      subject in the CA index file.  This is done only if the keyword
   3355      'unique_subject' is set to 'no' in the main CA section (default
   3356      if 'CA_default') of the configuration file.  The value is saved
   3357      with the database itself in a separate index attribute file,
   3358      named like the index file with '.attr' appended to the name.
   3359      [Richard Levitte]
   3360 
   3361   *) Generate muti valued AVAs using '+' notation in config files for
   3362      req and dirName.
   3363      [Steve Henson]
   3364 
   3365   *) Support for nameConstraints certificate extension.
   3366      [Steve Henson]
   3367 
   3368   *) Support for policyConstraints certificate extension.
   3369      [Steve Henson]
   3370 
   3371   *) Support for policyMappings certificate extension.
   3372      [Steve Henson]
   3373 
   3374   *) Make sure the default DSA_METHOD implementation only uses its
   3375      dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
   3376      and change its own handlers to be NULL so as to remove unnecessary
   3377      indirection. This lets alternative implementations fallback to the
   3378      default implementation more easily.
   3379      [Geoff Thorpe]
   3380 
   3381   *) Support for directoryName in GeneralName related extensions
   3382      in config files.
   3383      [Steve Henson]
   3384 
   3385   *) Make it possible to link applications using Makefile.shared.
   3386      Make that possible even when linking against static libraries!
   3387      [Richard Levitte]
   3388 
   3389   *) Support for single pass processing for S/MIME signing. This now
   3390      means that S/MIME signing can be done from a pipe, in addition
   3391      cleartext signing (multipart/signed type) is effectively streaming
   3392      and the signed data does not need to be all held in memory.
   3393 
   3394      This is done with a new flag PKCS7_STREAM. When this flag is set
   3395      PKCS7_sign() only initializes the PKCS7 structure and the actual signing
   3396      is done after the data is output (and digests calculated) in
   3397      SMIME_write_PKCS7().
   3398      [Steve Henson]
   3399 
   3400   *) Add full support for -rpath/-R, both in shared libraries and
   3401      applications, at least on the platforms where it's known how
   3402      to do it.
   3403      [Richard Levitte]
   3404 
   3405   *) In crypto/ec/ec_mult.c, implement fast point multiplication with
   3406      precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
   3407      will now compute a table of multiples of the generator that
   3408      makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
   3409      faster (notably in the case of a single point multiplication,
   3410      scalar * generator).
   3411      [Nils Larsch, Bodo Moeller]
   3412 
   3413   *) IPv6 support for certificate extensions. The various extensions
   3414      which use the IP:a.b.c.d can now take IPv6 addresses using the
   3415      formats of RFC1884 2.2 . IPv6 addresses are now also displayed
   3416      correctly.
   3417      [Steve Henson]
   3418 
   3419   *) Added an ENGINE that implements RSA by performing private key
   3420      exponentiations with the GMP library. The conversions to and from
   3421      GMP's mpz_t format aren't optimised nor are any montgomery forms
   3422      cached, and on x86 it appears OpenSSL's own performance has caught up.
   3423      However there are likely to be other architectures where GMP could
   3424      provide a boost. This ENGINE is not built in by default, but it can be
   3425      specified at Configure time and should be accompanied by the necessary
   3426      linker additions, eg;
   3427          ./config -DOPENSSL_USE_GMP -lgmp
   3428      [Geoff Thorpe]
   3429 
   3430   *) "openssl engine" will not display ENGINE/DSO load failure errors when
   3431      testing availability of engines with "-t" - the old behaviour is
   3432      produced by increasing the feature's verbosity with "-tt".
   3433      [Geoff Thorpe]
   3434 
   3435   *) ECDSA routines: under certain error conditions uninitialized BN objects
   3436      could be freed. Solution: make sure initialization is performed early
   3437      enough. (Reported and fix supplied by Nils Larsch <nla (a] trustcenter.de>
   3438      via PR#459)
   3439      [Lutz Jaenicke]
   3440 
   3441   *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
   3442      and DH_METHOD (eg. by ENGINE implementations) to override the normal
   3443      software implementations. For DSA and DH, parameter generation can
   3444      also be overriden by providing the appropriate method callbacks.
   3445      [Geoff Thorpe]
   3446 
   3447   *) Change the "progress" mechanism used in key-generation and
   3448      primality testing to functions that take a new BN_GENCB pointer in
   3449      place of callback/argument pairs. The new API functions have "_ex"
   3450      postfixes and the older functions are reimplemented as wrappers for
   3451      the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
   3452      declarations of the old functions to help (graceful) attempts to
   3453      migrate to the new functions. Also, the new key-generation API
   3454      functions operate on a caller-supplied key-structure and return
   3455      success/failure rather than returning a key or NULL - this is to
   3456      help make "keygen" another member function of RSA_METHOD etc.
   3457 
   3458      Example for using the new callback interface:
   3459 
   3460           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
   3461           void *my_arg = ...;
   3462           BN_GENCB my_cb;
   3463 
   3464           BN_GENCB_set(&my_cb, my_callback, my_arg);
   3465 
   3466           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
   3467           /* For the meaning of a, b in calls to my_callback(), see the
   3468            * documentation of the function that calls the callback.
   3469            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
   3470            * my_callback should return 1 if it wants BN_is_prime_ex()
   3471            * to continue, or 0 to stop.
   3472            */
   3473 
   3474      [Geoff Thorpe]
   3475 
   3476   *) Change the ZLIB compression method to be stateful, and make it
   3477      available to TLS with the number defined in 
   3478      draft-ietf-tls-compression-04.txt.
   3479      [Richard Levitte]
   3480 
   3481   *) Add the ASN.1 structures and functions for CertificatePair, which
   3482      is defined as follows (according to X.509_4thEditionDraftV6.pdf):
   3483 
   3484      CertificatePair ::= SEQUENCE {
   3485         forward		[0]	Certificate OPTIONAL,
   3486         reverse		[1]	Certificate OPTIONAL,
   3487         -- at least one of the pair shall be present -- }
   3488 
   3489      Also implement the PEM functions to read and write certificate
   3490      pairs, and defined the PEM tag as "CERTIFICATE PAIR".
   3491 
   3492      This needed to be defined, mostly for the sake of the LDAP
   3493      attribute crossCertificatePair, but may prove useful elsewhere as
   3494      well.
   3495      [Richard Levitte]
   3496 
   3497   *) Make it possible to inhibit symlinking of shared libraries in
   3498      Makefile.shared, for Cygwin's sake.
   3499      [Richard Levitte]
   3500 
   3501   *) Extend the BIGNUM API by creating a function 
   3502           void BN_set_negative(BIGNUM *a, int neg);
   3503      and a macro that behave like
   3504           int  BN_is_negative(const BIGNUM *a);
   3505 
   3506      to avoid the need to access 'a->neg' directly in applications.
   3507      [Nils Larsch]
   3508 
   3509   *) Implement fast modular reduction for pseudo-Mersenne primes
   3510      used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
   3511      EC_GROUP_new_curve_GFp() will now automatically use this
   3512      if applicable.
   3513      [Nils Larsch <nla (a] trustcenter.de>]
   3514 
   3515   *) Add new lock type (CRYPTO_LOCK_BN).
   3516      [Bodo Moeller]
   3517 
   3518   *) Change the ENGINE framework to automatically load engines
   3519      dynamically from specific directories unless they could be
   3520      found to already be built in or loaded.  Move all the
   3521      current engines except for the cryptodev one to a new
   3522      directory engines/.
   3523      The engines in engines/ are built as shared libraries if
   3524      the "shared" options was given to ./Configure or ./config.
   3525      Otherwise, they are inserted in libcrypto.a.
   3526      /usr/local/ssl/engines is the default directory for dynamic
   3527      engines, but that can be overriden at configure time through
   3528      the usual use of --prefix and/or --openssldir, and at run
   3529      time with the environment variable OPENSSL_ENGINES.
   3530      [Geoff Thorpe and Richard Levitte]
   3531 
   3532   *) Add Makefile.shared, a helper makefile to build shared
   3533      libraries.  Addapt Makefile.org.
   3534      [Richard Levitte]
   3535 
   3536   *) Add version info to Win32 DLLs.
   3537      [Peter 'Luna' Runestig" <peter (a] runestig.com>]
   3538 
   3539   *) Add new 'medium level' PKCS#12 API. Certificates and keys
   3540      can be added using this API to created arbitrary PKCS#12
   3541      files while avoiding the low level API.
   3542 
   3543      New options to PKCS12_create(), key or cert can be NULL and
   3544      will then be omitted from the output file. The encryption
   3545      algorithm NIDs can be set to -1 for no encryption, the mac
   3546      iteration count can be set to 0 to omit the mac.
   3547 
   3548      Enhance pkcs12 utility by making the -nokeys and -nocerts
   3549      options work when creating a PKCS#12 file. New option -nomac
   3550      to omit the mac, NONE can be set for an encryption algorithm.
   3551      New code is modified to use the enhanced PKCS12_create()
   3552      instead of the low level API.
   3553      [Steve Henson]
   3554 
   3555   *) Extend ASN1 encoder to support indefinite length constructed
   3556      encoding. This can output sequences tags and octet strings in
   3557      this form. Modify pk7_asn1.c to support indefinite length
   3558      encoding. This is experimental and needs additional code to
   3559      be useful, such as an ASN1 bio and some enhanced streaming
   3560      PKCS#7 code.
   3561 
   3562      Extend template encode functionality so that tagging is passed
   3563      down to the template encoder.
   3564      [Steve Henson]
   3565 
   3566   *) Let 'openssl req' fail if an argument to '-newkey' is not
   3567      recognized instead of using RSA as a default.
   3568      [Bodo Moeller]
   3569 
   3570   *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
   3571      As these are not official, they are not included in "ALL";
   3572      the "ECCdraft" ciphersuite group alias can be used to select them.
   3573      [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
   3574 
   3575   *) Add ECDH engine support.
   3576      [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
   3577 
   3578   *) Add ECDH in new directory crypto/ecdh/.
   3579      [Douglas Stebila (Sun Microsystems Laboratories)]
   3580 
   3581   *) Let BN_rand_range() abort with an error after 100 iterations
   3582      without success (which indicates a broken PRNG).
   3583      [Bodo Moeller]
   3584 
   3585   *) Change BN_mod_sqrt() so that it verifies that the input value
   3586      is really the square of the return value.  (Previously,
   3587      BN_mod_sqrt would show GIGO behaviour.)
   3588      [Bodo Moeller]
   3589 
   3590   *) Add named elliptic curves over binary fields from X9.62, SECG,
   3591      and WAP/WTLS; add OIDs that were still missing.
   3592 
   3593      [Sheueling Chang Shantz and Douglas Stebila
   3594      (Sun Microsystems Laboratories)]
   3595 
   3596   *) Extend the EC library for elliptic curves over binary fields
   3597      (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
   3598      New EC_METHOD:
   3599 
   3600           EC_GF2m_simple_method
   3601 
   3602      New API functions:
   3603 
   3604           EC_GROUP_new_curve_GF2m
   3605           EC_GROUP_set_curve_GF2m
   3606           EC_GROUP_get_curve_GF2m
   3607           EC_POINT_set_affine_coordinates_GF2m
   3608           EC_POINT_get_affine_coordinates_GF2m
   3609           EC_POINT_set_compressed_coordinates_GF2m
   3610 
   3611      Point compression for binary fields is disabled by default for
   3612      patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
   3613      enable it).
   3614 
   3615      As binary polynomials are represented as BIGNUMs, various members
   3616      of the EC_GROUP and EC_POINT data structures can be shared
   3617      between the implementations for prime fields and binary fields;
   3618      the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
   3619      are essentially identical to their ..._GFp counterparts.
   3620      (For simplicity, the '..._GFp' prefix has been dropped from
   3621      various internal method names.)
   3622 
   3623      An internal 'field_div' method (similar to 'field_mul' and
   3624      'field_sqr') has been added; this is used only for binary fields.
   3625 
   3626      [Sheueling Chang Shantz and Douglas Stebila
   3627      (Sun Microsystems Laboratories)]
   3628 
   3629   *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
   3630      through methods ('mul', 'precompute_mult').
   3631 
   3632      The generic implementations (now internally called 'ec_wNAF_mul'
   3633      and 'ec_wNAF_precomputed_mult') remain the default if these
   3634      methods are undefined.
   3635 
   3636      [Sheueling Chang Shantz and Douglas Stebila
   3637      (Sun Microsystems Laboratories)]
   3638 
   3639   *) New function EC_GROUP_get_degree, which is defined through
   3640      EC_METHOD.  For curves over prime fields, this returns the bit
   3641      length of the modulus.
   3642 
   3643      [Sheueling Chang Shantz and Douglas Stebila
   3644      (Sun Microsystems Laboratories)]
   3645 
   3646   *) New functions EC_GROUP_dup, EC_POINT_dup.
   3647      (These simply call ..._new  and ..._copy).
   3648 
   3649      [Sheueling Chang Shantz and Douglas Stebila
   3650      (Sun Microsystems Laboratories)]
   3651 
   3652   *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
   3653      Polynomials are represented as BIGNUMs (where the sign bit is not
   3654      used) in the following functions [macros]:  
   3655 
   3656           BN_GF2m_add
   3657           BN_GF2m_sub             [= BN_GF2m_add]
   3658           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
   3659           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
   3660           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
   3661           BN_GF2m_mod_inv
   3662           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
   3663           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
   3664           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
   3665           BN_GF2m_cmp             [= BN_ucmp]
   3666 
   3667      (Note that only the 'mod' functions are actually for fields GF(2^m).
   3668      BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
   3669 
   3670      For some functions, an the irreducible polynomial defining a
   3671      field can be given as an 'unsigned int[]' with strictly
   3672      decreasing elements giving the indices of those bits that are set;
   3673      i.e., p[] represents the polynomial
   3674           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
   3675      where
   3676           p[0] > p[1] > ... > p[k] = 0.
   3677      This applies to the following functions:
   3678 
   3679           BN_GF2m_mod_arr
   3680           BN_GF2m_mod_mul_arr
   3681           BN_GF2m_mod_sqr_arr
   3682           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
   3683           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
   3684           BN_GF2m_mod_exp_arr
   3685           BN_GF2m_mod_sqrt_arr
   3686           BN_GF2m_mod_solve_quad_arr
   3687           BN_GF2m_poly2arr
   3688           BN_GF2m_arr2poly
   3689 
   3690      Conversion can be performed by the following functions:
   3691 
   3692           BN_GF2m_poly2arr
   3693           BN_GF2m_arr2poly
   3694 
   3695      bntest.c has additional tests for binary polynomial arithmetic.
   3696 
   3697      Two implementations for BN_GF2m_mod_div() are available.
   3698      The default algorithm simply uses BN_GF2m_mod_inv() and
   3699      BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
   3700      if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
   3701      copyright notice in crypto/bn/bn_gf2m.c before enabling it).
   3702 
   3703      [Sheueling Chang Shantz and Douglas Stebila
   3704      (Sun Microsystems Laboratories)]
   3705 
   3706   *) Add new error code 'ERR_R_DISABLED' that can be used when some
   3707      functionality is disabled at compile-time.
   3708      [Douglas Stebila <douglas.stebila (a] sun.com>]
   3709 
   3710   *) Change default behaviour of 'openssl asn1parse' so that more
   3711      information is visible when viewing, e.g., a certificate:
   3712 
   3713      Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
   3714      mode the content of non-printable OCTET STRINGs is output in a
   3715      style similar to INTEGERs, but with '[HEX DUMP]' prepended to
   3716      avoid the appearance of a printable string.
   3717      [Nils Larsch <nla (a] trustcenter.de>]
   3718 
   3719   *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
   3720      functions
   3721           EC_GROUP_set_asn1_flag()
   3722           EC_GROUP_get_asn1_flag()
   3723           EC_GROUP_set_point_conversion_form()
   3724           EC_GROUP_get_point_conversion_form()
   3725      These control ASN1 encoding details:
   3726      - Curves (i.e., groups) are encoded explicitly unless asn1_flag
   3727        has been set to OPENSSL_EC_NAMED_CURVE.
   3728      - Points are encoded in uncompressed form by default; options for
   3729        asn1_for are as for point2oct, namely
   3730           POINT_CONVERSION_COMPRESSED
   3731           POINT_CONVERSION_UNCOMPRESSED
   3732           POINT_CONVERSION_HYBRID
   3733 
   3734      Also add 'seed' and 'seed_len' members to EC_GROUP with access
   3735      functions
   3736           EC_GROUP_set_seed()
   3737           EC_GROUP_get0_seed()
   3738           EC_GROUP_get_seed_len()
   3739      This is used only for ASN1 purposes (so far).
   3740      [Nils Larsch <nla (a] trustcenter.de>]
   3741 
   3742   *) Add 'field_type' member to EC_METHOD, which holds the NID
   3743      of the appropriate field type OID.  The new function
   3744      EC_METHOD_get_field_type() returns this value.
   3745      [Nils Larsch <nla (a] trustcenter.de>]
   3746 
   3747   *) Add functions 
   3748           EC_POINT_point2bn()
   3749           EC_POINT_bn2point()
   3750           EC_POINT_point2hex()
   3751           EC_POINT_hex2point()
   3752      providing useful interfaces to EC_POINT_point2oct() and
   3753      EC_POINT_oct2point().
   3754      [Nils Larsch <nla (a] trustcenter.de>]
   3755 
   3756   *) Change internals of the EC library so that the functions
   3757           EC_GROUP_set_generator()
   3758           EC_GROUP_get_generator()
   3759           EC_GROUP_get_order()
   3760           EC_GROUP_get_cofactor()
   3761      are implemented directly in crypto/ec/ec_lib.c and not dispatched
   3762      to methods, which would lead to unnecessary code duplication when
   3763      adding different types of curves.
   3764      [Nils Larsch <nla (a] trustcenter.de> with input by Bodo Moeller]
   3765 
   3766   *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
   3767      arithmetic, and such that modified wNAFs are generated
   3768      (which avoid length expansion in many cases).
   3769      [Bodo Moeller]
   3770 
   3771   *) Add a function EC_GROUP_check_discriminant() (defined via
   3772      EC_METHOD) that verifies that the curve discriminant is non-zero.
   3773 
   3774      Add a function EC_GROUP_check() that makes some sanity tests
   3775      on a EC_GROUP, its generator and order.  This includes
   3776      EC_GROUP_check_discriminant().
   3777      [Nils Larsch <nla (a] trustcenter.de>]
   3778 
   3779   *) Add ECDSA in new directory crypto/ecdsa/.
   3780 
   3781      Add applications 'openssl ecparam' and 'openssl ecdsa'
   3782      (these are based on 'openssl dsaparam' and 'openssl dsa').
   3783 
   3784      ECDSA support is also included in various other files across the
   3785      library.  Most notably,
   3786      - 'openssl req' now has a '-newkey ecdsa:file' option;
   3787      - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
   3788      - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
   3789        d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
   3790        them suitable for ECDSA where domain parameters must be
   3791        extracted before the specific public key;
   3792      - ECDSA engine support has been added.
   3793      [Nils Larsch <nla (a] trustcenter.de>]
   3794 
   3795   *) Include some named elliptic curves, and add OIDs from X9.62,
   3796      SECG, and WAP/WTLS.  Each curve can be obtained from the new
   3797      function
   3798           EC_GROUP_new_by_curve_name(),
   3799      and the list of available named curves can be obtained with
   3800           EC_get_builtin_curves().
   3801      Also add a 'curve_name' member to EC_GROUP objects, which can be
   3802      accessed via
   3803          EC_GROUP_set_curve_name()
   3804          EC_GROUP_get_curve_name()
   3805      [Nils Larsch <larsch (a] trustcenter.de, Bodo Moeller]
   3806  
   3807   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
   3808      was actually never needed) and in BN_mul().  The removal in BN_mul()
   3809      required a small change in bn_mul_part_recursive() and the addition
   3810      of the functions bn_cmp_part_words(), bn_sub_part_words() and
   3811      bn_add_part_words(), which do the same thing as bn_cmp_words(),
   3812      bn_sub_words() and bn_add_words() except they take arrays with
   3813      differing sizes.
   3814      [Richard Levitte]
   3815 
   3816  Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
   3817 
   3818   *) Cleanse PEM buffers before freeing them since they may contain 
   3819      sensitive data.
   3820      [Benjamin Bennett <ben (a] psc.edu>]
   3821 
   3822   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
   3823      a ciphersuite string such as "DEFAULT:RSA" cannot enable
   3824      authentication-only ciphersuites.
   3825      [Bodo Moeller]
   3826 
   3827   *) Since AES128 and AES256 share a single mask bit in the logic of
   3828      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
   3829      kludge to work properly if AES128 is available and AES256 isn't.
   3830      [Victor Duchovni]
   3831 
   3832   *) Expand security boundary to match 1.1.1 module.
   3833      [Steve Henson]
   3834 
   3835   *) Remove redundant features: hash file source, editing of test vectors
   3836      modify fipsld to use external fips_premain.c signature.
   3837      [Steve Henson]
   3838 
   3839   *) New perl script mkfipsscr.pl to create shell scripts or batch files to
   3840      run algorithm test programs.
   3841      [Steve Henson]
   3842 
   3843   *) Make algorithm test programs more tolerant of whitespace.
   3844      [Steve Henson]
   3845 
   3846   *) Have SSL/TLS server implementation tolerate "mismatched" record
   3847      protocol version while receiving ClientHello even if the
   3848      ClientHello is fragmented.  (The server can't insist on the
   3849      particular protocol version it has chosen before the ServerHello
   3850      message has informed the client about his choice.)
   3851      [Bodo Moeller]
   3852 
   3853   *) Load error codes if they are not already present instead of using a
   3854      static variable. This allows them to be cleanly unloaded and reloaded.
   3855      [Steve Henson]
   3856 
   3857  Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
   3858 
   3859   *) Introduce limits to prevent malicious keys being able to
   3860      cause a denial of service.  (CVE-2006-2940)
   3861      [Steve Henson, Bodo Moeller]
   3862 
   3863   *) Fix ASN.1 parsing of certain invalid structures that can result
   3864      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
   3865 
   3866   *) Fix buffer overflow in SSL_get_shared_ciphers() function. 
   3867      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
   3868 
   3869   *) Fix SSL client code which could crash if connecting to a
   3870      malicious SSLv2 server.  (CVE-2006-4343)
   3871      [Tavis Ormandy and Will Drewry, Google Security Team]
   3872 
   3873   *) Change ciphersuite string processing so that an explicit
   3874      ciphersuite selects this one ciphersuite (so that "AES256-SHA"
   3875      will no longer include "AES128-SHA"), and any other similar
   3876      ciphersuite (same bitmap) from *other* protocol versions (so that
   3877      "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
   3878      SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
   3879      changes from 0.9.8b and 0.9.8d.
   3880      [Bodo Moeller]
   3881 
   3882  Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
   3883 
   3884   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
   3885      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
   3886 
   3887   *) Change the Unix randomness entropy gathering to use poll() when
   3888      possible instead of select(), since the latter has some
   3889      undesirable limitations.
   3890      [Darryl Miles via Richard Levitte and Bodo Moeller]
   3891 
   3892   *) Disable rogue ciphersuites:
   3893 
   3894       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
   3895       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
   3896       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
   3897 
   3898      The latter two were purportedly from
   3899      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
   3900      appear there.
   3901 
   3902      Also deactive the remaining ciphersuites from
   3903      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
   3904      unofficial, and the ID has long expired.
   3905      [Bodo Moeller]
   3906 
   3907   *) Fix RSA blinding Heisenbug (problems sometimes occured on
   3908      dual-core machines) and other potential thread-safety issues.
   3909      [Bodo Moeller]
   3910 
   3911  Changes between 0.9.7i and 0.9.7j  [04 May 2006]
   3912 
   3913   *) Adapt fipsld and the build system to link against the validated FIPS
   3914      module in FIPS mode.
   3915      [Steve Henson]
   3916 
   3917   *) Fixes for VC++ 2005 build under Windows.
   3918      [Steve Henson]
   3919 
   3920   *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 
   3921      from a Windows bash shell such as MSYS. It is autodetected from the
   3922      "config" script when run from a VC++ environment. Modify standard VC++
   3923      build to use fipscanister.o from the GNU make build. 
   3924      [Steve Henson]
   3925 
   3926  Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
   3927 
   3928   *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
   3929      The value now differs depending on if you build for FIPS or not.
   3930      BEWARE!  A program linked with a shared FIPSed libcrypto can't be
   3931      safely run with a non-FIPSed libcrypto, as it may crash because of
   3932      the difference induced by this change.
   3933      [Andy Polyakov]
   3934 
   3935  Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
   3936 
   3937   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
   3938      (part of SSL_OP_ALL).  This option used to disable the
   3939      countermeasure against man-in-the-middle protocol-version
   3940      rollback in the SSL 2.0 server implementation, which is a bad
   3941      idea.  (CVE-2005-2969)
   3942 
   3943      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
   3944      for Information Security, National Institute of Advanced Industrial
   3945      Science and Technology [AIST], Japan)]
   3946 
   3947   *) Minimal support for X9.31 signatures and PSS padding modes. This is
   3948      mainly for FIPS compliance and not fully integrated at this stage.
   3949      [Steve Henson]
   3950 
   3951   *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
   3952      the exponentiation using a fixed-length exponent.  (Otherwise,
   3953      the information leaked through timing could expose the secret key
   3954      after many signatures; cf. Bleichenbacher's attack on DSA with
   3955      biased k.)
   3956      [Bodo Moeller]
   3957 
   3958   *) Make a new fixed-window mod_exp implementation the default for
   3959      RSA, DSA, and DH private-key operations so that the sequence of
   3960      squares and multiplies and the memory access pattern are
   3961      independent of the particular secret key.  This will mitigate
   3962      cache-timing and potential related attacks.
   3963 
   3964      BN_mod_exp_mont_consttime() is the new exponentiation implementation,
   3965      and this is automatically used by BN_mod_exp_mont() if the new flag
   3966      BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
   3967      will use this BN flag for private exponents unless the flag
   3968      RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
   3969      DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
   3970 
   3971      [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
   3972 
   3973   *) Change the client implementation for SSLv23_method() and
   3974      SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
   3975      Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
   3976      (Previously, the SSL 2.0 backwards compatible Client Hello
   3977      message format would be used even with SSL_OP_NO_SSLv2.)
   3978      [Bodo Moeller]
   3979 
   3980   *) Add support for smime-type MIME parameter in S/MIME messages which some
   3981      clients need.
   3982      [Steve Henson]
   3983 
   3984   *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
   3985      a threadsafe manner. Modify rsa code to use new function and add calls
   3986      to dsa and dh code (which had race conditions before).
   3987      [Steve Henson]
   3988 
   3989   *) Include the fixed error library code in the C error file definitions
   3990      instead of fixing them up at runtime. This keeps the error code
   3991      structures constant.
   3992      [Steve Henson]
   3993 
   3994  Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
   3995 
   3996   [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
   3997   OpenSSL 0.9.8.]
   3998 
   3999   *) Fixes for newer kerberos headers. NB: the casts are needed because
   4000      the 'length' field is signed on one version and unsigned on another
   4001      with no (?) obvious way to tell the difference, without these VC++
   4002      complains. Also the "definition" of FAR (blank) is no longer included
   4003      nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
   4004      some needed definitions.
   4005      [Steve Henson]
   4006 
   4007   *) Undo Cygwin change.
   4008      [Ulf Mller]
   4009 
   4010   *) Added support for proxy certificates according to RFC 3820.
   4011      Because they may be a security thread to unaware applications,
   4012      they must be explicitely allowed in run-time.  See
   4013      docs/HOWTO/proxy_certificates.txt for further information.
   4014      [Richard Levitte]
   4015 
   4016  Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
   4017 
   4018   *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
   4019      server and client random values. Previously
   4020      (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
   4021      less random data when sizeof(time_t) > 4 (some 64 bit platforms).
   4022 
   4023      This change has negligible security impact because:
   4024 
   4025      1. Server and client random values still have 24 bytes of pseudo random
   4026         data.
   4027 
   4028      2. Server and client random values are sent in the clear in the initial
   4029         handshake.
   4030 
   4031      3. The master secret is derived using the premaster secret (48 bytes in
   4032         size for static RSA ciphersuites) as well as client server and random
   4033         values.
   4034 
   4035      The OpenSSL team would like to thank the UK NISCC for bringing this issue
   4036      to our attention. 
   4037 
   4038      [Stephen Henson, reported by UK NISCC]
   4039 
   4040   *) Use Windows randomness collection on Cygwin.
   4041      [Ulf Mller]
   4042 
   4043   *) Fix hang in EGD/PRNGD query when communication socket is closed
   4044      prematurely by EGD/PRNGD.
   4045      [Darren Tucker <dtucker (a] zip.com.au> via Lutz Jnicke, resolves #1014]
   4046 
   4047   *) Prompt for pass phrases when appropriate for PKCS12 input format.
   4048      [Steve Henson]
   4049 
   4050   *) Back-port of selected performance improvements from development
   4051      branch, as well as improved support for PowerPC platforms.
   4052      [Andy Polyakov]
   4053 
   4054   *) Add lots of checks for memory allocation failure, error codes to indicate
   4055      failure and freeing up memory if a failure occurs.
   4056      [Nauticus Networks SSL Team <openssl (a] nauticusnet.com>, Steve Henson]
   4057 
   4058   *) Add new -passin argument to dgst.
   4059      [Steve Henson]
   4060 
   4061   *) Perform some character comparisons of different types in X509_NAME_cmp:
   4062      this is needed for some certificates that reencode DNs into UTF8Strings
   4063      (in violation of RFC3280) and can't or wont issue name rollover
   4064      certificates.
   4065      [Steve Henson]
   4066 
   4067   *) Make an explicit check during certificate validation to see that
   4068      the CA setting in each certificate on the chain is correct.  As a
   4069      side effect always do the following basic checks on extensions,
   4070      not just when there's an associated purpose to the check:
   4071 
   4072       - if there is an unhandled critical extension (unless the user
   4073         has chosen to ignore this fault)
   4074       - if the path length has been exceeded (if one is set at all)
   4075       - that certain extensions fit the associated purpose (if one has
   4076         been given)
   4077      [Richard Levitte]
   4078 
   4079  Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
   4080 
   4081   *) Avoid a race condition when CRLs are checked in a multi threaded 
   4082      environment. This would happen due to the reordering of the revoked
   4083      entries during signature checking and serial number lookup. Now the
   4084      encoding is cached and the serial number sort performed under a lock.
   4085      Add new STACK function sk_is_sorted().
   4086      [Steve Henson]
   4087 
   4088   *) Add Delta CRL to the extension code.
   4089      [Steve Henson]
   4090 
   4091   *) Various fixes to s3_pkt.c so alerts are sent properly.
   4092      [David Holmes <d.holmes (a] f5.com>]
   4093 
   4094   *) Reduce the chances of duplicate issuer name and serial numbers (in
   4095      violation of RFC3280) using the OpenSSL certificate creation utilities.
   4096      This is done by creating a random 64 bit value for the initial serial
   4097      number when a serial number file is created or when a self signed
   4098      certificate is created using 'openssl req -x509'. The initial serial
   4099      number file is created using 'openssl x509 -next_serial' in CA.pl
   4100      rather than being initialized to 1.
   4101      [Steve Henson]
   4102 
   4103  Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
   4104 
   4105   *) Fix null-pointer assignment in do_change_cipher_spec() revealed           
   4106      by using the Codenomicon TLS Test Tool (CVE-2004-0079)                    
   4107      [Joe Orton, Steve Henson]   
   4108 
   4109   *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
   4110      (CVE-2004-0112)
   4111      [Joe Orton, Steve Henson]   
   4112 
   4113   *) Make it possible to have multiple active certificates with the same
   4114      subject in the CA index file.  This is done only if the keyword
   4115      'unique_subject' is set to 'no' in the main CA section (default
   4116      if 'CA_default') of the configuration file.  The value is saved
   4117      with the database itself in a separate index attribute file,
   4118      named like the index file with '.attr' appended to the name.
   4119      [Richard Levitte]
   4120 
   4121   *) X509 verify fixes. Disable broken certificate workarounds when 
   4122      X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
   4123      keyUsage extension present. Don't accept CRLs with unhandled critical
   4124      extensions: since verify currently doesn't process CRL extensions this
   4125      rejects a CRL with *any* critical extensions. Add new verify error codes
   4126      for these cases.
   4127      [Steve Henson]
   4128 
   4129   *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
   4130      A clarification of RFC2560 will require the use of OCTET STRINGs and 
   4131      some implementations cannot handle the current raw format. Since OpenSSL
   4132      copies and compares OCSP nonces as opaque blobs without any attempt at
   4133      parsing them this should not create any compatibility issues.
   4134      [Steve Henson]
   4135 
   4136   *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
   4137      calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
   4138      this HMAC (and other) operations are several times slower than OpenSSL
   4139      < 0.9.7.
   4140      [Steve Henson]
   4141 
   4142   *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
   4143      [Peter Sylvester <Peter.Sylvester (a] EdelWeb.fr>]
   4144 
   4145   *) Use the correct content when signing type "other".
   4146      [Steve Henson]
   4147 
   4148  Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
   4149 
   4150   *) Fix various bugs revealed by running the NISCC test suite:
   4151 
   4152      Stop out of bounds reads in the ASN1 code when presented with
   4153      invalid tags (CVE-2003-0543 and CVE-2003-0544).
   4154      
   4155      Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
   4156 
   4157      If verify callback ignores invalid public key errors don't try to check
   4158      certificate signature with the NULL public key.
   4159 
   4160      [Steve Henson]
   4161 
   4162   *) New -ignore_err option in ocsp application to stop the server
   4163      exiting on the first error in a request.
   4164      [Steve Henson]
   4165 
   4166   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
   4167      if the server requested one: as stated in TLS 1.0 and SSL 3.0
   4168      specifications.
   4169      [Steve Henson]
   4170 
   4171   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
   4172      extra data after the compression methods not only for TLS 1.0
   4173      but also for SSL 3.0 (as required by the specification).
   4174      [Bodo Moeller; problem pointed out by Matthias Loepfe]
   4175 
   4176   *) Change X509_certificate_type() to mark the key as exported/exportable
   4177      when it's 512 *bits* long, not 512 bytes.
   4178      [Richard Levitte]
   4179 
   4180   *) Change AES_cbc_encrypt() so it outputs exact multiple of
   4181      blocks during encryption.
   4182      [Richard Levitte]
   4183 
   4184   *) Various fixes to base64 BIO and non blocking I/O. On write 
   4185      flushes were not handled properly if the BIO retried. On read
   4186      data was not being buffered properly and had various logic bugs.
   4187      This also affects blocking I/O when the data being decoded is a
   4188      certain size.
   4189      [Steve Henson]
   4190 
   4191   *) Various S/MIME bugfixes and compatibility changes:
   4192      output correct application/pkcs7 MIME type if
   4193      PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
   4194      Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
   4195      of files as .eml work). Correctly handle very long lines in MIME
   4196      parser.
   4197      [Steve Henson]
   4198 
   4199  Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
   4200 
   4201   *) Countermeasure against the Klima-Pokorny-Rosa extension of
   4202      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
   4203      a protocol version number mismatch like a decryption error
   4204      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
   4205      [Bodo Moeller]
   4206 
   4207   *) Turn on RSA blinding by default in the default implementation
   4208      to avoid a timing attack. Applications that don't want it can call
   4209      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
   4210      They would be ill-advised to do so in most cases.
   4211      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
   4212 
   4213   *) Change RSA blinding code so that it works when the PRNG is not
   4214      seeded (in this case, the secret RSA exponent is abused as
   4215      an unpredictable seed -- if it is not unpredictable, there
   4216      is no point in blinding anyway).  Make RSA blinding thread-safe
   4217      by remembering the creator's thread ID in rsa->blinding and
   4218      having all other threads use local one-time blinding factors
   4219      (this requires more computation than sharing rsa->blinding, but
   4220      avoids excessive locking; and if an RSA object is not shared
   4221      between threads, blinding will still be very fast).
   4222      [Bodo Moeller]
   4223 
   4224   *) Fixed a typo bug that would cause ENGINE_set_default() to set an
   4225      ENGINE as defaults for all supported algorithms irrespective of
   4226      the 'flags' parameter. 'flags' is now honoured, so applications
   4227      should make sure they are passing it correctly.
   4228      [Geoff Thorpe]
   4229 
   4230   *) Target "mingw" now allows native Windows code to be generated in
   4231      the Cygwin environment as well as with the MinGW compiler.
   4232      [Ulf Moeller] 
   4233 
   4234  Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
   4235 
   4236   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
   4237      via timing by performing a MAC computation even if incorrrect
   4238      block cipher padding has been found.  This is a countermeasure
   4239      against active attacks where the attacker has to distinguish
   4240      between bad padding and a MAC verification error. (CVE-2003-0078)
   4241 
   4242      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
   4243      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
   4244      Martin Vuagnoux (EPFL, Ilion)]
   4245 
   4246   *) Make the no-err option work as intended.  The intention with no-err
   4247      is not to have the whole error stack handling routines removed from
   4248      libcrypto, it's only intended to remove all the function name and
   4249      reason texts, thereby removing some of the footprint that may not
   4250      be interesting if those errors aren't displayed anyway.
   4251 
   4252      NOTE: it's still possible for any application or module to have it's
   4253      own set of error texts inserted.  The routines are there, just not
   4254      used by default when no-err is given.
   4255      [Richard Levitte]
   4256 
   4257   *) Add support for FreeBSD on IA64.
   4258      [dirk.meyer (a] dinoex.sub.org via Richard Levitte, resolves #454]
   4259 
   4260   *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
   4261      Kerberos function mit_des_cbc_cksum().  Before this change,
   4262      the value returned by DES_cbc_cksum() was like the one from
   4263      mit_des_cbc_cksum(), except the bytes were swapped.
   4264      [Kevin Greaney <Kevin.Greaney (a] hp.com> and Richard Levitte]
   4265 
   4266   *) Allow an application to disable the automatic SSL chain building.
   4267      Before this a rather primitive chain build was always performed in
   4268      ssl3_output_cert_chain(): an application had no way to send the 
   4269      correct chain if the automatic operation produced an incorrect result.
   4270 
   4271      Now the chain builder is disabled if either:
   4272 
   4273      1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
   4274 
   4275      2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
   4276 
   4277      The reasoning behind this is that an application would not want the
   4278      auto chain building to take place if extra chain certificates are
   4279      present and it might also want a means of sending no additional
   4280      certificates (for example the chain has two certificates and the
   4281      root is omitted).
   4282      [Steve Henson]
   4283 
   4284   *) Add the possibility to build without the ENGINE framework.
   4285      [Steven Reddie <smr (a] essemer.com.au> via Richard Levitte]
   4286 
   4287   *) Under Win32 gmtime() can return NULL: check return value in
   4288      OPENSSL_gmtime(). Add error code for case where gmtime() fails.
   4289      [Steve Henson]
   4290 
   4291   *) DSA routines: under certain error conditions uninitialized BN objects
   4292      could be freed. Solution: make sure initialization is performed early
   4293      enough. (Reported and fix supplied by Ivan D Nestlerode <nestler (a] MIT.EDU>,
   4294      Nils Larsch <nla (a] trustcenter.de> via PR#459)
   4295      [Lutz Jaenicke]
   4296 
   4297   *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
   4298      checked on reconnect on the client side, therefore session resumption
   4299      could still fail with a "ssl session id is different" error. This
   4300      behaviour is masked when SSL_OP_ALL is used due to
   4301      SSL_OP_MICROSOFT_SESS_ID_BUG being set.
   4302      Behaviour observed by Crispin Flowerday <crispin (a] flowerday.cx> as
   4303      followup to PR #377.
   4304      [Lutz Jaenicke]
   4305 
   4306   *) IA-32 assembler support enhancements: unified ELF targets, support
   4307      for SCO/Caldera platforms, fix for Cygwin shared build.
   4308      [Andy Polyakov]
   4309 
   4310   *) Add support for FreeBSD on sparc64.  As a consequence, support for
   4311      FreeBSD on non-x86 processors is separate from x86 processors on
   4312      the config script, much like the NetBSD support.
   4313      [Richard Levitte & Kris Kennaway <kris (a] obsecurity.org>]
   4314 
   4315  Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
   4316 
   4317   [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
   4318   OpenSSL 0.9.7.]
   4319 
   4320   *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
   4321      code (06) was taken as the first octet of the session ID and the last
   4322      octet was ignored consequently. As a result SSLv2 client side session
   4323      caching could not have worked due to the session ID mismatch between
   4324      client and server.
   4325      Behaviour observed by Crispin Flowerday <crispin (a] flowerday.cx> as
   4326      PR #377.
   4327      [Lutz Jaenicke]
   4328 
   4329   *) Change the declaration of needed Kerberos libraries to use EX_LIBS
   4330      instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
   4331      removed entirely.
   4332      [Richard Levitte]
   4333 
   4334   *) The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
   4335      seems that in spite of existing for more than a year, many application
   4336      author have done nothing to provide the necessary callbacks, which
   4337      means that this particular engine will not work properly anywhere.
   4338      This is a very unfortunate situation which forces us, in the name
   4339      of usability, to give the hw_ncipher.c a static lock, which is part
   4340      of libcrypto.
   4341      NOTE: This is for the 0.9.7 series ONLY.  This hack will never
   4342      appear in 0.9.8 or later.  We EXPECT application authors to have
   4343      dealt properly with this when 0.9.8 is released (unless we actually
   4344      make such changes in the libcrypto locking code that changes will
   4345      have to be made anyway).
   4346      [Richard Levitte]
   4347 
   4348   *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
   4349      octets have been read, EOF or an error occurs. Without this change
   4350      some truncated ASN1 structures will not produce an error.
   4351      [Steve Henson]
   4352 
   4353   *) Disable Heimdal support, since it hasn't been fully implemented.
   4354      Still give the possibility to force the use of Heimdal, but with
   4355      warnings and a request that patches get sent to openssl-dev.
   4356      [Richard Levitte]
   4357 
   4358   *) Add the VC-CE target, introduce the WINCE sysname, and add
   4359      INSTALL.WCE and appropriate conditionals to make it build.
   4360      [Steven Reddie <smr (a] essemer.com.au> via Richard Levitte]
   4361 
   4362   *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
   4363      cygssl-x.y.z.dll, where x, y and z are the major, minor and
   4364      edit numbers of the version.
   4365      [Corinna Vinschen <vinschen (a] redhat.com> and Richard Levitte]
   4366 
   4367   *) Introduce safe string copy and catenation functions
   4368      (BUF_strlcpy() and BUF_strlcat()).
   4369      [Ben Laurie (CHATS) and Richard Levitte]
   4370 
   4371   *) Avoid using fixed-size buffers for one-line DNs.
   4372      [Ben Laurie (CHATS)]
   4373 
   4374   *) Add BUF_MEM_grow_clean() to avoid information leakage when
   4375      resizing buffers containing secrets, and use where appropriate.
   4376      [Ben Laurie (CHATS)]
   4377 
   4378   *) Avoid using fixed size buffers for configuration file location.
   4379      [Ben Laurie (CHATS)]
   4380 
   4381   *) Avoid filename truncation for various CA files.
   4382      [Ben Laurie (CHATS)]
   4383 
   4384   *) Use sizeof in preference to magic numbers.
   4385      [Ben Laurie (CHATS)]
   4386 
   4387   *) Avoid filename truncation in cert requests.
   4388      [Ben Laurie (CHATS)]
   4389 
   4390   *) Add assertions to check for (supposedly impossible) buffer
   4391      overflows.
   4392      [Ben Laurie (CHATS)]
   4393 
   4394   *) Don't cache truncated DNS entries in the local cache (this could
   4395      potentially lead to a spoofing attack).
   4396      [Ben Laurie (CHATS)]
   4397 
   4398   *) Fix various buffers to be large enough for hex/decimal
   4399      representations in a platform independent manner.
   4400      [Ben Laurie (CHATS)]
   4401 
   4402   *) Add CRYPTO_realloc_clean() to avoid information leakage when
   4403      resizing buffers containing secrets, and use where appropriate.
   4404      [Ben Laurie (CHATS)]
   4405 
   4406   *) Add BIO_indent() to avoid much slightly worrying code to do
   4407      indents.
   4408      [Ben Laurie (CHATS)]
   4409 
   4410   *) Convert sprintf()/BIO_puts() to BIO_printf().
   4411      [Ben Laurie (CHATS)]
   4412 
   4413   *) buffer_gets() could terminate with the buffer only half
   4414      full. Fixed.
   4415      [Ben Laurie (CHATS)]
   4416 
   4417   *) Add assertions to prevent user-supplied crypto functions from
   4418      overflowing internal buffers by having large block sizes, etc.
   4419      [Ben Laurie (CHATS)]
   4420 
   4421   *) New OPENSSL_assert() macro (similar to assert(), but enabled
   4422      unconditionally).
   4423      [Ben Laurie (CHATS)]
   4424 
   4425   *) Eliminate unused copy of key in RC4.
   4426      [Ben Laurie (CHATS)]
   4427 
   4428   *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
   4429      [Ben Laurie (CHATS)]
   4430 
   4431   *) Fix off-by-one error in EGD path.
   4432      [Ben Laurie (CHATS)]
   4433 
   4434   *) If RANDFILE path is too long, ignore instead of truncating.
   4435      [Ben Laurie (CHATS)]
   4436 
   4437   *) Eliminate unused and incorrectly sized X.509 structure
   4438      CBCParameter.
   4439      [Ben Laurie (CHATS)]
   4440 
   4441   *) Eliminate unused and dangerous function knumber().
   4442      [Ben Laurie (CHATS)]
   4443 
   4444   *) Eliminate unused and dangerous structure, KSSL_ERR.
   4445      [Ben Laurie (CHATS)]
   4446 
   4447   *) Protect against overlong session ID context length in an encoded
   4448      session object. Since these are local, this does not appear to be
   4449      exploitable.
   4450      [Ben Laurie (CHATS)]
   4451 
   4452   *) Change from security patch (see 0.9.6e below) that did not affect
   4453      the 0.9.6 release series:
   4454 
   4455      Remote buffer overflow in SSL3 protocol - an attacker could
   4456      supply an oversized master key in Kerberos-enabled versions.
   4457      (CVE-2002-0657)
   4458      [Ben Laurie (CHATS)]
   4459 
   4460   *) Change the SSL kerb5 codes to match RFC 2712.
   4461      [Richard Levitte]
   4462 
   4463   *) Make -nameopt work fully for req and add -reqopt switch.
   4464      [Michael Bell <michael.bell (a] rz.hu-berlin.de>, Steve Henson]
   4465 
   4466   *) The "block size" for block ciphers in CFB and OFB mode should be 1.
   4467      [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve (a] opera.com>]
   4468 
   4469   *) Make sure tests can be performed even if the corresponding algorithms
   4470      have been removed entirely.  This was also the last step to make
   4471      OpenSSL compilable with DJGPP under all reasonable conditions.
   4472      [Richard Levitte, Doug Kaufman <dkaufman (a] rahul.net>]
   4473 
   4474   *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
   4475      to allow version independent disabling of normally unselected ciphers,
   4476      which may be activated as a side-effect of selecting a single cipher.
   4477 
   4478      (E.g., cipher list string "RSA" enables ciphersuites that are left
   4479      out of "ALL" because they do not provide symmetric encryption.
   4480      "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
   4481      [Lutz Jaenicke, Bodo Moeller]
   4482 
   4483   *) Add appropriate support for separate platform-dependent build
   4484      directories.  The recommended way to make a platform-dependent
   4485      build directory is the following (tested on Linux), maybe with
   4486      some local tweaks:
   4487 
   4488 	# Place yourself outside of the OpenSSL source tree.  In
   4489 	# this example, the environment variable OPENSSL_SOURCE
   4490 	# is assumed to contain the absolute OpenSSL source directory.
   4491 	mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
   4492 	cd objtree/"`uname -s`-`uname -r`-`uname -m`"
   4493 	(cd $OPENSSL_SOURCE; find . -type f) | while read F; do
   4494 		mkdir -p `dirname $F`
   4495 		ln -s $OPENSSL_SOURCE/$F $F
   4496 	done
   4497 
   4498      To be absolutely sure not to disturb the source tree, a "make clean"
   4499      is a good thing.  If it isn't successfull, don't worry about it,
   4500      it probably means the source directory is very clean.
   4501      [Richard Levitte]
   4502 
   4503   *) Make sure any ENGINE control commands make local copies of string
   4504      pointers passed to them whenever necessary. Otherwise it is possible
   4505      the caller may have overwritten (or deallocated) the original string
   4506      data when a later ENGINE operation tries to use the stored values.
   4507      [Gtz Babin-Ebell <babinebell (a] trustcenter.de>]
   4508 
   4509   *) Improve diagnostics in file reading and command-line digests.
   4510      [Ben Laurie aided and abetted by Solar Designer <solar (a] openwall.com>]
   4511 
   4512   *) Add AES modes CFB and OFB to the object database.  Correct an
   4513      error in AES-CFB decryption.
   4514      [Richard Levitte]
   4515 
   4516   *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 
   4517      allows existing EVP_CIPHER_CTX structures to be reused after
   4518      calling EVP_*Final(). This behaviour is used by encryption
   4519      BIOs and some applications. This has the side effect that
   4520      applications must explicitly clean up cipher contexts with
   4521      EVP_CIPHER_CTX_cleanup() or they will leak memory.
   4522      [Steve Henson]
   4523 
   4524   *) Check the values of dna and dnb in bn_mul_recursive before calling
   4525      bn_mul_comba (a non zero value means the a or b arrays do not contain
   4526      n2 elements) and fallback to bn_mul_normal if either is not zero.
   4527      [Steve Henson]
   4528 
   4529   *) Fix escaping of non-ASCII characters when using the -subj option
   4530      of the "openssl req" command line tool. (Robert Joop <joop (a] fokus.gmd.de>)
   4531      [Lutz Jaenicke]
   4532 
   4533   *) Make object definitions compliant to LDAP (RFC2256): SN is the short
   4534      form for "surname", serialNumber has no short form.
   4535      Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
   4536      therefore remove "mail" short name for "internet 7".
   4537      The OID for unique identifiers in X509 certificates is
   4538      x500UniqueIdentifier, not uniqueIdentifier.
   4539      Some more OID additions. (Michael Bell <michael.bell (a] rz.hu-berlin.de>)
   4540      [Lutz Jaenicke]
   4541 
   4542   *) Add an "init" command to the ENGINE config module and auto initialize
   4543      ENGINEs. Without any "init" command the ENGINE will be initialized 
   4544      after all ctrl commands have been executed on it. If init=1 the 
   4545      ENGINE is initailized at that point (ctrls before that point are run
   4546      on the uninitialized ENGINE and after on the initialized one). If
   4547      init=0 then the ENGINE will not be iniatialized at all.
   4548      [Steve Henson]
   4549 
   4550   *) Fix the 'app_verify_callback' interface so that the user-defined
   4551      argument is actually passed to the callback: In the
   4552      SSL_CTX_set_cert_verify_callback() prototype, the callback
   4553      declaration has been changed from
   4554           int (*cb)()
   4555      into
   4556           int (*cb)(X509_STORE_CTX *,void *);
   4557      in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
   4558           i=s->ctx->app_verify_callback(&ctx)
   4559      has been changed into
   4560           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
   4561 
   4562      To update applications using SSL_CTX_set_cert_verify_callback(),
   4563      a dummy argument can be added to their callback functions.
   4564      [D. K. Smetters <smetters (a] parc.xerox.com>]
   4565 
   4566   *) Added the '4758cca' ENGINE to support IBM 4758 cards.
   4567      [Maurice Gittens <maurice (a] gittens.nl>, touchups by Geoff Thorpe]
   4568 
   4569   *) Add and OPENSSL_LOAD_CONF define which will cause
   4570      OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
   4571      This allows older applications to transparently support certain
   4572      OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
   4573      Two new functions OPENSSL_add_all_algorithms_noconf() which will never
   4574      load the config file and OPENSSL_add_all_algorithms_conf() which will
   4575      always load it have also been added.
   4576      [Steve Henson]
   4577 
   4578   *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
   4579      Adjust NIDs and EVP layer.
   4580      [Stephen Sprunk <stephen (a] sprunk.org> and Richard Levitte]
   4581 
   4582   *) Config modules support in openssl utility.
   4583 
   4584      Most commands now load modules from the config file,
   4585      though in a few (such as version) this isn't done 
   4586      because it couldn't be used for anything.
   4587 
   4588      In the case of ca and req the config file used is
   4589      the same as the utility itself: that is the -config
   4590      command line option can be used to specify an
   4591      alternative file.
   4592      [Steve Henson]
   4593 
   4594   *) Move default behaviour from OPENSSL_config(). If appname is NULL
   4595      use "openssl_conf" if filename is NULL use default openssl config file.
   4596      [Steve Henson]
   4597 
   4598   *) Add an argument to OPENSSL_config() to allow the use of an alternative
   4599      config section name. Add a new flag to tolerate a missing config file
   4600      and move code to CONF_modules_load_file().
   4601      [Steve Henson]
   4602 
   4603   *) Support for crypto accelerator cards from Accelerated Encryption
   4604      Processing, www.aep.ie.  (Use engine 'aep')
   4605      The support was copied from 0.9.6c [engine] and adapted/corrected
   4606      to work with the new engine framework.
   4607      [AEP Inc. and Richard Levitte]
   4608 
   4609   *) Support for SureWare crypto accelerator cards from Baltimore
   4610      Technologies.  (Use engine 'sureware')
   4611      The support was copied from 0.9.6c [engine] and adapted
   4612      to work with the new engine framework.
   4613      [Richard Levitte]
   4614 
   4615   *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
   4616      make the newer ENGINE framework commands for the CHIL engine work.
   4617      [Toomas Kiisk <vix (a] cyber.ee> and Richard Levitte]
   4618 
   4619   *) Make it possible to produce shared libraries on ReliantUNIX.
   4620      [Robert Dahlem <Robert.Dahlem (a] ffm2.siemens.de> via Richard Levitte]
   4621 
   4622   *) Add the configuration target debug-linux-ppro.
   4623      Make 'openssl rsa' use the general key loading routines
   4624      implemented in apps.c, and make those routines able to
   4625      handle the key format FORMAT_NETSCAPE and the variant
   4626      FORMAT_IISSGC.
   4627      [Toomas Kiisk <vix (a] cyber.ee> via Richard Levitte]
   4628 
   4629  *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
   4630      [Toomas Kiisk <vix (a] cyber.ee> via Richard Levitte]
   4631 
   4632   *) Add -keyform to rsautl, and document -engine.
   4633      [Richard Levitte, inspired by Toomas Kiisk <vix (a] cyber.ee>]
   4634 
   4635   *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
   4636      BIO_R_NO_SUCH_FILE error code rather than the generic
   4637      ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
   4638      [Ben Laurie]
   4639 
   4640   *) Add new functions
   4641           ERR_peek_last_error
   4642           ERR_peek_last_error_line
   4643           ERR_peek_last_error_line_data.
   4644      These are similar to
   4645           ERR_peek_error
   4646           ERR_peek_error_line
   4647           ERR_peek_error_line_data,
   4648      but report on the latest error recorded rather than the first one
   4649      still in the error queue.
   4650      [Ben Laurie, Bodo Moeller]
   4651         
   4652   *) default_algorithms option in ENGINE config module. This allows things
   4653      like:
   4654      default_algorithms = ALL
   4655      default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
   4656      [Steve Henson]
   4657 
   4658   *) Prelminary ENGINE config module.
   4659      [Steve Henson]
   4660 
   4661   *) New experimental application configuration code.
   4662      [Steve Henson]
   4663 
   4664   *) Change the AES code to follow the same name structure as all other
   4665      symmetric ciphers, and behave the same way.  Move everything to
   4666      the directory crypto/aes, thereby obsoleting crypto/rijndael.
   4667      [Stephen Sprunk <stephen (a] sprunk.org> and Richard Levitte]
   4668 
   4669   *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
   4670      [Ben Laurie and Theo de Raadt]
   4671 
   4672   *) Add option to output public keys in req command.
   4673      [Massimiliano Pala madwolf (a] openca.org]
   4674 
   4675   *) Use wNAFs in EC_POINTs_mul() for improved efficiency
   4676      (up to about 10% better than before for P-192 and P-224).
   4677      [Bodo Moeller]
   4678 
   4679   *) New functions/macros
   4680 
   4681           SSL_CTX_set_msg_callback(ctx, cb)
   4682           SSL_CTX_set_msg_callback_arg(ctx, arg)
   4683           SSL_set_msg_callback(ssl, cb)
   4684           SSL_set_msg_callback_arg(ssl, arg)
   4685 
   4686      to request calling a callback function
   4687 
   4688           void cb(int write_p, int version, int content_type,
   4689                   const void *buf, size_t len, SSL *ssl, void *arg)
   4690 
   4691      whenever a protocol message has been completely received
   4692      (write_p == 0) or sent (write_p == 1).  Here 'version' is the
   4693      protocol version  according to which the SSL library interprets
   4694      the current protocol message (SSL2_VERSION, SSL3_VERSION, or
   4695      TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
   4696      the content type as defined in the SSL 3.0/TLS 1.0 protocol
   4697      specification (change_cipher_spec(20), alert(21), handshake(22)).
   4698      'buf' and 'len' point to the actual message, 'ssl' to the
   4699      SSL object, and 'arg' is the application-defined value set by
   4700      SSL[_CTX]_set_msg_callback_arg().
   4701 
   4702      'openssl s_client' and 'openssl s_server' have new '-msg' options
   4703      to enable a callback that displays all protocol messages.
   4704      [Bodo Moeller]
   4705 
   4706   *) Change the shared library support so shared libraries are built as
   4707      soon as the corresponding static library is finished, and thereby get
   4708      openssl and the test programs linked against the shared library.
   4709      This still only happens when the keyword "shard" has been given to
   4710      the configuration scripts.
   4711 
   4712      NOTE: shared library support is still an experimental thing, and
   4713      backward binary compatibility is still not guaranteed.
   4714      ["Maciej W. Rozycki" <macro (a] ds2.pg.gda.pl> and Richard Levitte]
   4715 
   4716   *) Add support for Subject Information Access extension.
   4717      [Peter Sylvester <Peter.Sylvester (a] EdelWeb.fr>]
   4718 
   4719   *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
   4720      additional bytes when new memory had to be allocated, not just
   4721      when reusing an existing buffer.
   4722      [Bodo Moeller]
   4723 
   4724   *) New command line and configuration option 'utf8' for the req command.
   4725      This allows field values to be specified as UTF8 strings.
   4726      [Steve Henson]
   4727 
   4728   *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
   4729      runs for the former and machine-readable output for the latter.
   4730      [Ben Laurie]
   4731 
   4732   *) Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
   4733      of the e-mail address in the DN (i.e., it will go into a certificate
   4734      extension only).  The new configuration file option 'email_in_dn = no'
   4735      has the same effect.
   4736      [Massimiliano Pala madwolf (a] openca.org]
   4737 
   4738   *) Change all functions with names starting with des_ to be starting
   4739      with DES_ instead.  Add wrappers that are compatible with libdes,
   4740      but are named _ossl_old_des_*.  Finally, add macros that map the
   4741      des_* symbols to the corresponding _ossl_old_des_* if libdes
   4742      compatibility is desired.  If OpenSSL 0.9.6c compatibility is
   4743      desired, the des_* symbols will be mapped to DES_*, with one
   4744      exception.
   4745 
   4746      Since we provide two compatibility mappings, the user needs to
   4747      define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
   4748      compatibility is desired.  The default (i.e., when that macro
   4749      isn't defined) is OpenSSL 0.9.6c compatibility.
   4750 
   4751      There are also macros that enable and disable the support of old
   4752      des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
   4753      and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
   4754      are defined, the default will apply: to support the old des routines.
   4755 
   4756      In either case, one must include openssl/des.h to get the correct
   4757      definitions.  Do not try to just include openssl/des_old.h, that
   4758      won't work.
   4759 
   4760      NOTE: This is a major break of an old API into a new one.  Software
   4761      authors are encouraged to switch to the DES_ style functions.  Some
   4762      time in the future, des_old.h and the libdes compatibility functions
   4763      will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
   4764      default), and then completely removed.
   4765      [Richard Levitte]
   4766 
   4767   *) Test for certificates which contain unsupported critical extensions.
   4768      If such a certificate is found during a verify operation it is 
   4769      rejected by default: this behaviour can be overridden by either
   4770      handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
   4771      by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
   4772      X509_supported_extension() has also been added which returns 1 if a
   4773      particular extension is supported.
   4774      [Steve Henson]
   4775 
   4776   *) Modify the behaviour of EVP cipher functions in similar way to digests
   4777      to retain compatibility with existing code.
   4778      [Steve Henson]
   4779 
   4780   *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
   4781      compatibility with existing code. In particular the 'ctx' parameter does
   4782      not have to be to be initialized before the call to EVP_DigestInit() and
   4783      it is tidied up after a call to EVP_DigestFinal(). New function
   4784      EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
   4785      EVP_MD_CTX_copy() changed to not require the destination to be
   4786      initialized valid and new function EVP_MD_CTX_copy_ex() added which
   4787      requires the destination to be valid.
   4788 
   4789      Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
   4790      EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
   4791      [Steve Henson]
   4792 
   4793   *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
   4794      so that complete 'Handshake' protocol structures are kept in memory
   4795      instead of overwriting 'msg_type' and 'length' with 'body' data.
   4796      [Bodo Moeller]
   4797 
   4798   *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
   4799      [Massimo Santin via Richard Levitte]
   4800 
   4801   *) Major restructuring to the underlying ENGINE code. This includes
   4802      reduction of linker bloat, separation of pure "ENGINE" manipulation
   4803      (initialisation, etc) from functionality dealing with implementations
   4804      of specific crypto iterfaces. This change also introduces integrated
   4805      support for symmetric ciphers and digest implementations - so ENGINEs
   4806      can now accelerate these by providing EVP_CIPHER and EVP_MD
   4807      implementations of their own. This is detailed in crypto/engine/README
   4808      as it couldn't be adequately described here. However, there are a few
   4809      API changes worth noting - some RSA, DSA, DH, and RAND functions that
   4810      were changed in the original introduction of ENGINE code have now
   4811      reverted back - the hooking from this code to ENGINE is now a good
   4812      deal more passive and at run-time, operations deal directly with
   4813      RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
   4814      dereferencing through an ENGINE pointer any more. Also, the ENGINE
   4815      functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
   4816      they were not being used by the framework as there is no concept of a
   4817      BIGNUM_METHOD and they could not be generalised to the new
   4818      'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
   4819      ENGINE_cpy() has been removed as it cannot be consistently defined in
   4820      the new code.
   4821      [Geoff Thorpe]
   4822 
   4823   *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
   4824      [Steve Henson]
   4825 
   4826   *) Change mkdef.pl to sort symbols that get the same entry number,
   4827      and make sure the automatically generated functions ERR_load_*
   4828      become part of libeay.num as well.
   4829      [Richard Levitte]
   4830 
   4831   *) New function SSL_renegotiate_pending().  This returns true once
   4832      renegotiation has been requested (either SSL_renegotiate() call
   4833      or HelloRequest/ClientHello receveived from the peer) and becomes
   4834      false once a handshake has been completed.
   4835      (For servers, SSL_renegotiate() followed by SSL_do_handshake()
   4836      sends a HelloRequest, but does not ensure that a handshake takes
   4837      place.  SSL_renegotiate_pending() is useful for checking if the
   4838      client has followed the request.)
   4839      [Bodo Moeller]
   4840 
   4841   *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
   4842      By default, clients may request session resumption even during
   4843      renegotiation (if session ID contexts permit); with this option,
   4844      session resumption is possible only in the first handshake.
   4845 
   4846      SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
   4847      more bits available for options that should not be part of
   4848      SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
   4849      [Bodo Moeller]
   4850 
   4851   *) Add some demos for certificate and certificate request creation.
   4852      [Steve Henson]
   4853 
   4854   *) Make maximum certificate chain size accepted from the peer application
   4855      settable (SSL*_get/set_max_cert_list()), as proposed by
   4856      "Douglas E. Engert" <deengert (a] anl.gov>.
   4857      [Lutz Jaenicke]
   4858 
   4859   *) Add support for shared libraries for Unixware-7
   4860      (Boyd Lynn Gerber <gerberb (a] zenez.com>).
   4861      [Lutz Jaenicke]
   4862 
   4863   *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
   4864      be done prior to destruction. Use this to unload error strings from
   4865      ENGINEs that load their own error strings. NB: This adds two new API
   4866      functions to "get" and "set" this destroy handler in an ENGINE.
   4867      [Geoff Thorpe]
   4868 
   4869   *) Alter all existing ENGINE implementations (except "openssl" and
   4870      "openbsd") to dynamically instantiate their own error strings. This
   4871      makes them more flexible to be built both as statically-linked ENGINEs
   4872      and self-contained shared-libraries loadable via the "dynamic" ENGINE.
   4873      Also, add stub code to each that makes building them as self-contained
   4874      shared-libraries easier (see README.ENGINE).
   4875      [Geoff Thorpe]
   4876 
   4877   *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
   4878      implementations into applications that are completely implemented in
   4879      self-contained shared-libraries. The "dynamic" ENGINE exposes control
   4880      commands that can be used to configure what shared-library to load and
   4881      to control aspects of the way it is handled. Also, made an update to
   4882      the README.ENGINE file that brings its information up-to-date and
   4883      provides some information and instructions on the "dynamic" ENGINE
   4884      (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
   4885      [Geoff Thorpe]
   4886 
   4887   *) Make it possible to unload ranges of ERR strings with a new
   4888      "ERR_unload_strings" function.
   4889      [Geoff Thorpe]
   4890 
   4891   *) Add a copy() function to EVP_MD.
   4892      [Ben Laurie]
   4893 
   4894   *) Make EVP_MD routines take a context pointer instead of just the
   4895      md_data void pointer.
   4896      [Ben Laurie]
   4897 
   4898   *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
   4899      that the digest can only process a single chunk of data
   4900      (typically because it is provided by a piece of
   4901      hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
   4902      is only going to provide a single chunk of data, and hence the
   4903      framework needn't accumulate the data for oneshot drivers.
   4904      [Ben Laurie]
   4905 
   4906   *) As with "ERR", make it possible to replace the underlying "ex_data"
   4907      functions. This change also alters the storage and management of global
   4908      ex_data state - it's now all inside ex_data.c and all "class" code (eg.
   4909      RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
   4910      index counters. The API functions that use this state have been changed
   4911      to take a "class_index" rather than pointers to the class's local STACK
   4912      and counter, and there is now an API function to dynamically create new
   4913      classes. This centralisation allows us to (a) plug a lot of the
   4914      thread-safety problems that existed, and (b) makes it possible to clean
   4915      up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
   4916      such data would previously have always leaked in application code and
   4917      workarounds were in place to make the memory debugging turn a blind eye
   4918      to it. Application code that doesn't use this new function will still
   4919      leak as before, but their memory debugging output will announce it now
   4920      rather than letting it slide.
   4921 
   4922      Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
   4923      induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
   4924      has a return value to indicate success or failure.
   4925      [Geoff Thorpe]
   4926 
   4927   *) Make it possible to replace the underlying "ERR" functions such that the
   4928      global state (2 LHASH tables and 2 locks) is only used by the "default"
   4929      implementation. This change also adds two functions to "get" and "set"
   4930      the implementation prior to it being automatically set the first time
   4931      any other ERR function takes place. Ie. an application can call "get",
   4932      pass the return value to a module it has just loaded, and that module
   4933      can call its own "set" function using that value. This means the
   4934      module's "ERR" operations will use (and modify) the error state in the
   4935      application and not in its own statically linked copy of OpenSSL code.
   4936      [Geoff Thorpe]
   4937 
   4938   *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
   4939      reference counts. This performs normal REF_PRINT/REF_CHECK macros on
   4940      the operation, and provides a more encapsulated way for external code
   4941      (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
   4942      to use these functions rather than manually incrementing the counts.
   4943 
   4944      Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
   4945      [Geoff Thorpe]
   4946 
   4947   *) Add EVP test program.
   4948      [Ben Laurie]
   4949 
   4950   *) Add symmetric cipher support to ENGINE. Expect the API to change!
   4951      [Ben Laurie]
   4952 
   4953   *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
   4954      X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
   4955      X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
   4956      These allow a CRL to be built without having to access X509_CRL fields
   4957      directly. Modify 'ca' application to use new functions.
   4958      [Steve Henson]
   4959 
   4960   *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
   4961      bug workarounds. Rollback attack detection is a security feature.
   4962      The problem will only arise on OpenSSL servers when TLSv1 is not
   4963      available (sslv3_server_method() or SSL_OP_NO_TLSv1).
   4964      Software authors not wanting to support TLSv1 will have special reasons
   4965      for their choice and can explicitly enable this option.
   4966      [Bodo Moeller, Lutz Jaenicke]
   4967 
   4968   *) Rationalise EVP so it can be extended: don't include a union of
   4969      cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
   4970      (similar to those existing for EVP_CIPHER_CTX).
   4971      Usage example:
   4972 
   4973          EVP_MD_CTX md;
   4974 
   4975          EVP_MD_CTX_init(&md);             /* new function call */
   4976          EVP_DigestInit(&md, EVP_sha1());
   4977          EVP_DigestUpdate(&md, in, len);
   4978          EVP_DigestFinal(&md, out, NULL);
   4979          EVP_MD_CTX_cleanup(&md);          /* new function call */
   4980 
   4981      [Ben Laurie]
   4982 
   4983   *) Make DES key schedule conform to the usual scheme, as well as
   4984      correcting its structure. This means that calls to DES functions
   4985      now have to pass a pointer to a des_key_schedule instead of a
   4986      plain des_key_schedule (which was actually always a pointer
   4987      anyway): E.g.,
   4988 
   4989          des_key_schedule ks;
   4990 
   4991 	 des_set_key_checked(..., &ks);
   4992 	 des_ncbc_encrypt(..., &ks, ...);
   4993 
   4994      (Note that a later change renames 'des_...' into 'DES_...'.)
   4995      [Ben Laurie]
   4996 
   4997   *) Initial reduction of linker bloat: the use of some functions, such as
   4998      PEM causes large amounts of unused functions to be linked in due to
   4999      poor organisation. For example pem_all.c contains every PEM function
   5000      which has a knock on effect of linking in large amounts of (unused)
   5001      ASN1 code. Grouping together similar functions and splitting unrelated
   5002      functions prevents this.
   5003      [Steve Henson]
   5004 
   5005   *) Cleanup of EVP macros.
   5006      [Ben Laurie]
   5007 
   5008   *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
   5009      correct _ecb suffix.
   5010      [Ben Laurie]
   5011 
   5012   *) Add initial OCSP responder support to ocsp application. The
   5013      revocation information is handled using the text based index
   5014      use by the ca application. The responder can either handle
   5015      requests generated internally, supplied in files (for example
   5016      via a CGI script) or using an internal minimal server.
   5017      [Steve Henson]
   5018 
   5019   *) Add configuration choices to get zlib compression for TLS.
   5020      [Richard Levitte]
   5021 
   5022   *) Changes to Kerberos SSL for RFC 2712 compliance:
   5023      1.  Implemented real KerberosWrapper, instead of just using
   5024          KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw (a] sxw.org.uk>]
   5025      2.  Implemented optional authenticator field of KerberosWrapper.
   5026 
   5027      Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
   5028      and authenticator structs; see crypto/krb5/.
   5029 
   5030      Generalized Kerberos calls to support multiple Kerberos libraries.
   5031      [Vern Staats <staatsvr (a] asc.hpc.mil>,
   5032       Jeffrey Altman <jaltman (a] columbia.edu>
   5033       via Richard Levitte]
   5034 
   5035   *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
   5036      already does with RSA. testdsa.h now has 'priv_key/pub_key'
   5037      values for each of the key sizes rather than having just
   5038      parameters (and 'speed' generating keys each time).
   5039      [Geoff Thorpe]
   5040 
   5041   *) Speed up EVP routines.
   5042      Before:
   5043 encrypt
   5044 type              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
   5045 des-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
   5046 des-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
   5047 des-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
   5048 decrypt
   5049 des-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
   5050 des-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
   5051 des-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
   5052      After:
   5053 encrypt
   5054 des-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
   5055 decrypt
   5056 des-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
   5057      [Ben Laurie]
   5058 
   5059   *) Added the OS2-EMX target.
   5060      ["Brian Havard" <brianh (a] kheldar.apana.org.au> and Richard Levitte]
   5061 
   5062   *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
   5063      to support NCONF routines in extension code. New function CONF_set_nconf()
   5064      to allow functions which take an NCONF to also handle the old LHASH
   5065      structure: this means that the old CONF compatible routines can be
   5066      retained (in particular wrt extensions) without having to duplicate the
   5067      code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
   5068      [Steve Henson]
   5069 
   5070   *) Enhance the general user interface with mechanisms for inner control
   5071      and with possibilities to have yes/no kind of prompts.
   5072      [Richard Levitte]
   5073 
   5074   *) Change all calls to low level digest routines in the library and
   5075      applications to use EVP. Add missing calls to HMAC_cleanup() and
   5076      don't assume HMAC_CTX can be copied using memcpy().
   5077      [Verdon Walker <VWalker (a] novell.com>, Steve Henson]
   5078 
   5079   *) Add the possibility to control engines through control names but with
   5080      arbitrary arguments instead of just a string.
   5081      Change the key loaders to take a UI_METHOD instead of a callback
   5082      function pointer.  NOTE: this breaks binary compatibility with earlier
   5083      versions of OpenSSL [engine].
   5084      Adapt the nCipher code for these new conditions and add a card insertion
   5085      callback.
   5086      [Richard Levitte]
   5087 
   5088   *) Enhance the general user interface with mechanisms to better support
   5089      dialog box interfaces, application-defined prompts, the possibility
   5090      to use defaults (for example default passwords from somewhere else)
   5091      and interrupts/cancellations.
   5092      [Richard Levitte]
   5093 
   5094   *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
   5095      attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
   5096      [Steve Henson]
   5097 
   5098   *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
   5099      tidy up some unnecessarily weird code in 'sk_new()').
   5100      [Geoff, reported by Diego Tartara <dtartara (a] novamens.com>]
   5101 
   5102   *) Change the key loading routines for ENGINEs to use the same kind
   5103      callback (pem_password_cb) as all other routines that need this
   5104      kind of callback.
   5105      [Richard Levitte]
   5106 
   5107   *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
   5108      256 bit (=32 byte) keys. Of course seeding with more entropy bytes
   5109      than this minimum value is recommended.
   5110      [Lutz Jaenicke]
   5111 
   5112   *) New random seeder for OpenVMS, using the system process statistics
   5113      that are easily reachable.
   5114      [Richard Levitte]
   5115 
   5116   *) Windows apparently can't transparently handle global
   5117      variables defined in DLLs. Initialisations such as:
   5118 
   5119         const ASN1_ITEM *it = &ASN1_INTEGER_it;
   5120 
   5121      wont compile. This is used by the any applications that need to
   5122      declare their own ASN1 modules. This was fixed by adding the option
   5123      EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
   5124      needed for static libraries under Win32.
   5125      [Steve Henson]
   5126 
   5127   *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
   5128      setting of purpose and trust fields. New X509_STORE trust and
   5129      purpose functions and tidy up setting in other SSL functions.
   5130      [Steve Henson]
   5131 
   5132   *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
   5133      structure. These are inherited by X509_STORE_CTX when it is 
   5134      initialised. This allows various defaults to be set in the
   5135      X509_STORE structure (such as flags for CRL checking and custom
   5136      purpose or trust settings) for functions which only use X509_STORE_CTX
   5137      internally such as S/MIME.
   5138 
   5139      Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
   5140      trust settings if they are not set in X509_STORE. This allows X509_STORE
   5141      purposes and trust (in S/MIME for example) to override any set by default.
   5142 
   5143      Add command line options for CRL checking to smime, s_client and s_server
   5144      applications.
   5145      [Steve Henson]
   5146 
   5147   *) Initial CRL based revocation checking. If the CRL checking flag(s)
   5148      are set then the CRL is looked up in the X509_STORE structure and
   5149      its validity and signature checked, then if the certificate is found
   5150      in the CRL the verify fails with a revoked error.
   5151 
   5152      Various new CRL related callbacks added to X509_STORE_CTX structure.
   5153 
   5154      Command line options added to 'verify' application to support this.
   5155 
   5156      This needs some additional work, such as being able to handle multiple
   5157      CRLs with different times, extension based lookup (rather than just
   5158      by subject name) and ultimately more complete V2 CRL extension
   5159      handling.
   5160      [Steve Henson]
   5161 
   5162   *) Add a general user interface API (crypto/ui/).  This is designed
   5163      to replace things like des_read_password and friends (backward
   5164      compatibility functions using this new API are provided).
   5165      The purpose is to remove prompting functions from the DES code
   5166      section as well as provide for prompting through dialog boxes in
   5167      a window system and the like.
   5168      [Richard Levitte]
   5169 
   5170   *) Add "ex_data" support to ENGINE so implementations can add state at a
   5171      per-structure level rather than having to store it globally.
   5172      [Geoff]
   5173 
   5174   *) Make it possible for ENGINE structures to be copied when retrieved by
   5175      ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
   5176      This causes the "original" ENGINE structure to act like a template,
   5177      analogous to the RSA vs. RSA_METHOD type of separation. Because of this
   5178      operational state can be localised to each ENGINE structure, despite the
   5179      fact they all share the same "methods". New ENGINE structures returned in
   5180      this case have no functional references and the return value is the single
   5181      structural reference. This matches the single structural reference returned
   5182      by ENGINE_by_id() normally, when it is incremented on the pre-existing
   5183      ENGINE structure.
   5184      [Geoff]
   5185 
   5186   *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
   5187      needs to match any other type at all we need to manually clear the
   5188      tag cache.
   5189      [Steve Henson]
   5190 
   5191   *) Changes to the "openssl engine" utility to include;
   5192      - verbosity levels ('-v', '-vv', and '-vvv') that provide information
   5193        about an ENGINE's available control commands.
   5194      - executing control commands from command line arguments using the
   5195        '-pre' and '-post' switches. '-post' is only used if '-t' is
   5196        specified and the ENGINE is successfully initialised. The syntax for
   5197        the individual commands are colon-separated, for example;
   5198 	 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
   5199      [Geoff]
   5200 
   5201   *) New dynamic control command support for ENGINEs. ENGINEs can now
   5202      declare their own commands (numbers), names (strings), descriptions,
   5203      and input types for run-time discovery by calling applications. A
   5204      subset of these commands are implicitly classed as "executable"
   5205      depending on their input type, and only these can be invoked through
   5206      the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
   5207      can be based on user input, config files, etc). The distinction is
   5208      that "executable" commands cannot return anything other than a boolean
   5209      result and can only support numeric or string input, whereas some
   5210      discoverable commands may only be for direct use through
   5211      ENGINE_ctrl(), eg. supporting the exchange of binary data, function
   5212      pointers, or other custom uses. The "executable" commands are to
   5213      support parameterisations of ENGINE behaviour that can be
   5214      unambiguously defined by ENGINEs and used consistently across any
   5215      OpenSSL-based application. Commands have been added to all the
   5216      existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
   5217      control over shared-library paths without source code alterations.
   5218      [Geoff]
   5219 
   5220   *) Changed all ENGINE implementations to dynamically allocate their
   5221      ENGINEs rather than declaring them statically. Apart from this being
   5222      necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
   5223      this also allows the implementations to compile without using the
   5224      internal engine_int.h header.
   5225      [Geoff]
   5226 
   5227   *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
   5228      'const' value. Any code that should be able to modify a RAND_METHOD
   5229      should already have non-const pointers to it (ie. they should only
   5230      modify their own ones).
   5231      [Geoff]
   5232 
   5233   *) Made a variety of little tweaks to the ENGINE code.
   5234      - "atalla" and "ubsec" string definitions were moved from header files
   5235        to C code. "nuron" string definitions were placed in variables
   5236        rather than hard-coded - allowing parameterisation of these values
   5237        later on via ctrl() commands.
   5238      - Removed unused "#if 0"'d code.
   5239      - Fixed engine list iteration code so it uses ENGINE_free() to release
   5240        structural references.
   5241      - Constified the RAND_METHOD element of ENGINE structures.
   5242      - Constified various get/set functions as appropriate and added
   5243        missing functions (including a catch-all ENGINE_cpy that duplicates
   5244        all ENGINE values onto a new ENGINE except reference counts/state).
   5245      - Removed NULL parameter checks in get/set functions. Setting a method
   5246        or function to NULL is a way of cancelling out a previously set
   5247        value.  Passing a NULL ENGINE parameter is just plain stupid anyway
   5248        and doesn't justify the extra error symbols and code.
   5249      - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
   5250        flags from engine_int.h to engine.h.
   5251      - Changed prototypes for ENGINE handler functions (init(), finish(),
   5252        ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
   5253      [Geoff]
   5254 
   5255   *) Implement binary inversion algorithm for BN_mod_inverse in addition
   5256      to the algorithm using long division.  The binary algorithm can be
   5257      used only if the modulus is odd.  On 32-bit systems, it is faster
   5258      only for relatively small moduli (roughly 20-30% for 128-bit moduli,
   5259      roughly 5-15% for 256-bit moduli), so we use it only for moduli
   5260      up to 450 bits.  In 64-bit environments, the binary algorithm
   5261      appears to be advantageous for much longer moduli; here we use it
   5262      for moduli up to 2048 bits.
   5263      [Bodo Moeller]
   5264 
   5265   *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
   5266      could not support the combine flag in choice fields.
   5267      [Steve Henson]
   5268 
   5269   *) Add a 'copy_extensions' option to the 'ca' utility. This copies
   5270      extensions from a certificate request to the certificate.
   5271      [Steve Henson]
   5272 
   5273   *) Allow multiple 'certopt' and 'nameopt' options to be separated
   5274      by commas. Add 'namopt' and 'certopt' options to the 'ca' config
   5275      file: this allows the display of the certificate about to be
   5276      signed to be customised, to allow certain fields to be included
   5277      or excluded and extension details. The old system didn't display
   5278      multicharacter strings properly, omitted fields not in the policy
   5279      and couldn't display additional details such as extensions.
   5280      [Steve Henson]
   5281 
   5282   *) Function EC_POINTs_mul for multiple scalar multiplication
   5283      of an arbitrary number of elliptic curve points
   5284           \sum scalars[i]*points[i],
   5285      optionally including the generator defined for the EC_GROUP:
   5286           scalar*generator +  \sum scalars[i]*points[i].
   5287 
   5288      EC_POINT_mul is a simple wrapper function for the typical case
   5289      that the point list has just one item (besides the optional
   5290      generator).
   5291      [Bodo Moeller]
   5292 
   5293   *) First EC_METHODs for curves over GF(p):
   5294 
   5295      EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
   5296      operations and provides various method functions that can also
   5297      operate with faster implementations of modular arithmetic.     
   5298 
   5299      EC_GFp_mont_method() reuses most functions that are part of
   5300      EC_GFp_simple_method, but uses Montgomery arithmetic.
   5301 
   5302      [Bodo Moeller; point addition and point doubling
   5303      implementation directly derived from source code provided by
   5304      Lenka Fibikova <fibikova (a] exp-math.uni-essen.de>]
   5305 
   5306   *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
   5307      crypto/ec/ec_lib.c):
   5308 
   5309      Curves are EC_GROUP objects (with an optional group generator)
   5310      based on EC_METHODs that are built into the library.
   5311 
   5312      Points are EC_POINT objects based on EC_GROUP objects.
   5313 
   5314      Most of the framework would be able to handle curves over arbitrary
   5315      finite fields, but as there are no obvious types for fields other
   5316      than GF(p), some functions are limited to that for now.
   5317      [Bodo Moeller]
   5318 
   5319   *) Add the -HTTP option to s_server.  It is similar to -WWW, but requires
   5320      that the file contains a complete HTTP response.
   5321      [Richard Levitte]
   5322 
   5323   *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
   5324      change the def and num file printf format specifier from "%-40sXXX"
   5325      to "%-39s XXX". The latter will always guarantee a space after the
   5326      field while the former will cause them to run together if the field
   5327      is 40 of more characters long.
   5328      [Steve Henson]
   5329 
   5330   *) Constify the cipher and digest 'method' functions and structures
   5331      and modify related functions to take constant EVP_MD and EVP_CIPHER
   5332      pointers.
   5333      [Steve Henson]
   5334 
   5335   *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
   5336      in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
   5337      [Bodo Moeller]
   5338 
   5339   *) Modify EVP_Digest*() routines so they now return values. Although the
   5340      internal software routines can never fail additional hardware versions
   5341      might.
   5342      [Steve Henson]
   5343 
   5344   *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
   5345 
   5346      Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
   5347      (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
   5348 
   5349      ASN1 error codes
   5350           ERR_R_NESTED_ASN1_ERROR
   5351           ...
   5352           ERR_R_MISSING_ASN1_EOS
   5353      were 4 .. 9, conflicting with
   5354           ERR_LIB_RSA (= ERR_R_RSA_LIB)
   5355           ...
   5356           ERR_LIB_PEM (= ERR_R_PEM_LIB).
   5357      They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
   5358 
   5359      Add new error code 'ERR_R_INTERNAL_ERROR'.
   5360      [Bodo Moeller]
   5361 
   5362   *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
   5363      suffices.
   5364      [Bodo Moeller]
   5365 
   5366   *) New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
   5367      sets the subject name for a new request or supersedes the
   5368      subject name in a given request. Formats that can be parsed are
   5369           'CN=Some Name, OU=myOU, C=IT'
   5370      and
   5371           'CN=Some Name/OU=myOU/C=IT'.
   5372 
   5373      Add options '-batch' and '-verbose' to 'openssl req'.
   5374      [Massimiliano Pala <madwolf (a] hackmasters.net>]
   5375 
   5376   *) Introduce the possibility to access global variables through
   5377      functions on platform were that's the best way to handle exporting
   5378      global variables in shared libraries.  To enable this functionality,
   5379      one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
   5380      "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
   5381      is normally done by Configure or something similar).
   5382 
   5383      To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
   5384      in the source file (foo.c) like this:
   5385 
   5386 	OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
   5387 	OPENSSL_IMPLEMENT_GLOBAL(double,bar);
   5388 
   5389      To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
   5390      and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
   5391 
   5392 	OPENSSL_DECLARE_GLOBAL(int,foo);
   5393 	#define foo OPENSSL_GLOBAL_REF(foo)
   5394 	OPENSSL_DECLARE_GLOBAL(double,bar);
   5395 	#define bar OPENSSL_GLOBAL_REF(bar)
   5396 
   5397      The #defines are very important, and therefore so is including the
   5398      header file everywhere where the defined globals are used.
   5399 
   5400      The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
   5401      of ASN.1 items, but that structure is a bit different.
   5402 
   5403      The largest change is in util/mkdef.pl which has been enhanced with
   5404      better and easier to understand logic to choose which symbols should
   5405      go into the Windows .def files as well as a number of fixes and code
   5406      cleanup (among others, algorithm keywords are now sorted
   5407      lexicographically to avoid constant rewrites).
   5408      [Richard Levitte]
   5409 
   5410   *) In BN_div() keep a copy of the sign of 'num' before writing the
   5411      result to 'rm' because if rm==num the value will be overwritten
   5412      and produce the wrong result if 'num' is negative: this caused
   5413      problems with BN_mod() and BN_nnmod().
   5414      [Steve Henson]
   5415 
   5416   *) Function OCSP_request_verify(). This checks the signature on an
   5417      OCSP request and verifies the signer certificate. The signer
   5418      certificate is just checked for a generic purpose and OCSP request
   5419      trust settings.
   5420      [Steve Henson]
   5421 
   5422   *) Add OCSP_check_validity() function to check the validity of OCSP
   5423      responses. OCSP responses are prepared in real time and may only
   5424      be a few seconds old. Simply checking that the current time lies
   5425      between thisUpdate and nextUpdate max reject otherwise valid responses
   5426      caused by either OCSP responder or client clock inaccuracy. Instead
   5427      we allow thisUpdate and nextUpdate to fall within a certain period of
   5428      the current time. The age of the response can also optionally be
   5429      checked. Two new options -validity_period and -status_age added to
   5430      ocsp utility.
   5431      [Steve Henson]
   5432 
   5433   *) If signature or public key algorithm is unrecognized print out its
   5434      OID rather that just UNKNOWN.
   5435      [Steve Henson]
   5436 
   5437   *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
   5438      OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
   5439      ID to be generated from the issuer certificate alone which can then be
   5440      passed to OCSP_id_issuer_cmp().
   5441      [Steve Henson]
   5442 
   5443   *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
   5444      ASN1 modules to export functions returning ASN1_ITEM pointers
   5445      instead of the ASN1_ITEM structures themselves. This adds several
   5446      new macros which allow the underlying ASN1 function/structure to
   5447      be accessed transparently. As a result code should not use ASN1_ITEM
   5448      references directly (such as &X509_it) but instead use the relevant
   5449      macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
   5450      use of the new ASN1 code on platforms where exporting structures
   5451      is problematical (for example in shared libraries) but exporting
   5452      functions returning pointers to structures is not.
   5453      [Steve Henson]
   5454 
   5455   *) Add support for overriding the generation of SSL/TLS session IDs.
   5456      These callbacks can be registered either in an SSL_CTX or per SSL.
   5457      The purpose of this is to allow applications to control, if they wish,
   5458      the arbitrary values chosen for use as session IDs, particularly as it
   5459      can be useful for session caching in multiple-server environments. A
   5460      command-line switch for testing this (and any client code that wishes
   5461      to use such a feature) has been added to "s_server".
   5462      [Geoff Thorpe, Lutz Jaenicke]
   5463 
   5464   *) Modify mkdef.pl to recognise and parse preprocessor conditionals
   5465      of the form '#if defined(...) || defined(...) || ...' and
   5466      '#if !defined(...) && !defined(...) && ...'.  This also avoids
   5467      the growing number of special cases it was previously handling.
   5468      [Richard Levitte]
   5469 
   5470   *) Make all configuration macros available for application by making
   5471      sure they are available in opensslconf.h, by giving them names starting
   5472      with "OPENSSL_" to avoid conflicts with other packages and by making
   5473      sure e_os2.h will cover all platform-specific cases together with
   5474      opensslconf.h.
   5475      Additionally, it is now possible to define configuration/platform-
   5476      specific names (called "system identities").  In the C code, these
   5477      are prefixed with "OPENSSL_SYSNAME_".  e_os2.h will create another
   5478      macro with the name beginning with "OPENSSL_SYS_", which is determined
   5479      from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
   5480      what is available.
   5481      [Richard Levitte]
   5482 
   5483   *) New option -set_serial to 'req' and 'x509' this allows the serial
   5484      number to use to be specified on the command line. Previously self
   5485      signed certificates were hard coded with serial number 0 and the 
   5486      CA options of 'x509' had to use a serial number in a file which was
   5487      auto incremented.
   5488      [Steve Henson]
   5489 
   5490   *) New options to 'ca' utility to support V2 CRL entry extensions.
   5491      Currently CRL reason, invalidity date and hold instruction are
   5492      supported. Add new CRL extensions to V3 code and some new objects.
   5493      [Steve Henson]
   5494 
   5495   *) New function EVP_CIPHER_CTX_set_padding() this is used to
   5496      disable standard block padding (aka PKCS#5 padding) in the EVP
   5497      API, which was previously mandatory. This means that the data is
   5498      not padded in any way and so the total length much be a multiple
   5499      of the block size, otherwise an error occurs.
   5500      [Steve Henson]
   5501 
   5502   *) Initial (incomplete) OCSP SSL support.
   5503      [Steve Henson]
   5504 
   5505   *) New function OCSP_parse_url(). This splits up a URL into its host,
   5506      port and path components: primarily to parse OCSP URLs. New -url
   5507      option to ocsp utility.
   5508      [Steve Henson]
   5509 
   5510   *) New nonce behavior. The return value of OCSP_check_nonce() now 
   5511      reflects the various checks performed. Applications can decide
   5512      whether to tolerate certain situations such as an absent nonce
   5513      in a response when one was present in a request: the ocsp application
   5514      just prints out a warning. New function OCSP_add1_basic_nonce()
   5515      this is to allow responders to include a nonce in a response even if
   5516      the request is nonce-less.
   5517      [Steve Henson]
   5518 
   5519   *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
   5520      skipped when using openssl x509 multiple times on a single input file,
   5521      e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
   5522      [Bodo Moeller]
   5523 
   5524   *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
   5525      set string type: to handle setting ASN1_TIME structures. Fix ca
   5526      utility to correctly initialize revocation date of CRLs.
   5527      [Steve Henson]
   5528 
   5529   *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
   5530      the clients preferred ciphersuites and rather use its own preferences.
   5531      Should help to work around M$ SGC (Server Gated Cryptography) bug in
   5532      Internet Explorer by ensuring unchanged hash method during stepup.
   5533      (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
   5534      [Lutz Jaenicke]
   5535 
   5536   *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
   5537      to aes and add a new 'exist' option to print out symbols that don't
   5538      appear to exist.
   5539      [Steve Henson]
   5540 
   5541   *) Additional options to ocsp utility to allow flags to be set and
   5542      additional certificates supplied.
   5543      [Steve Henson]
   5544 
   5545   *) Add the option -VAfile to 'openssl ocsp', so the user can give the
   5546      OCSP client a number of certificate to only verify the response
   5547      signature against.
   5548      [Richard Levitte]
   5549 
   5550   *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
   5551      handle the new API. Currently only ECB, CBC modes supported. Add new
   5552      AES OIDs.
   5553 
   5554      Add TLS AES ciphersuites as described in RFC3268, "Advanced
   5555      Encryption Standard (AES) Ciphersuites for Transport Layer
   5556      Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
   5557      not enabled by default and were not part of the "ALL" ciphersuite
   5558      alias because they were not yet official; they could be
   5559      explicitly requested by specifying the "AESdraft" ciphersuite
   5560      group alias.  In the final release of OpenSSL 0.9.7, the group
   5561      alias is called "AES" and is part of "ALL".)
   5562      [Ben Laurie, Steve  Henson, Bodo Moeller]
   5563 
   5564   *) New function OCSP_copy_nonce() to copy nonce value (if present) from
   5565      request to response.
   5566      [Steve Henson]
   5567 
   5568   *) Functions for OCSP responders. OCSP_request_onereq_count(),
   5569      OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
   5570      extract information from a certificate request. OCSP_response_create()
   5571      creates a response and optionally adds a basic response structure.
   5572      OCSP_basic_add1_status() adds a complete single response to a basic
   5573      response and returns the OCSP_SINGLERESP structure just added (to allow
   5574      extensions to be included for example). OCSP_basic_add1_cert() adds a
   5575      certificate to a basic response and OCSP_basic_sign() signs a basic
   5576      response with various flags. New helper functions ASN1_TIME_check()
   5577      (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
   5578      (converts ASN1_TIME to GeneralizedTime).
   5579      [Steve Henson]
   5580 
   5581   *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
   5582      in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
   5583      structure from a certificate. X509_pubkey_digest() digests the public_key
   5584      contents: this is used in various key identifiers. 
   5585      [Steve Henson]
   5586 
   5587   *) Make sk_sort() tolerate a NULL argument.
   5588      [Steve Henson reported by Massimiliano Pala <madwolf (a] comune.modena.it>]
   5589 
   5590   *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
   5591      passed by the function are trusted implicitly. If any of them signed the
   5592      response then it is assumed to be valid and is not verified.
   5593      [Steve Henson]
   5594 
   5595   *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
   5596      to data. This was previously part of the PKCS7 ASN1 code. This
   5597      was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
   5598      [Steve Henson, reported by Kenneth R. Robinette
   5599 				<support (a] securenetterm.com>]
   5600 
   5601   *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
   5602      routines: without these tracing memory leaks is very painful.
   5603      Fix leaks in PKCS12 and PKCS7 routines.
   5604      [Steve Henson]
   5605 
   5606   *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
   5607      Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
   5608      effectively meant GeneralizedTime would never be used. Now it
   5609      is initialised to -1 but X509_time_adj() now has to check the value
   5610      and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
   5611      V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
   5612      [Steve Henson, reported by Kenneth R. Robinette
   5613 				<support (a] securenetterm.com>]
   5614 
   5615   *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
   5616      result in a zero length in the ASN1_INTEGER structure which was
   5617      not consistent with the structure when d2i_ASN1_INTEGER() was used
   5618      and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
   5619      to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
   5620      where it did not print out a minus for negative ASN1_INTEGER.
   5621      [Steve Henson]
   5622 
   5623   *) Add summary printout to ocsp utility. The various functions which
   5624      convert status values to strings have been renamed to:
   5625      OCSP_response_status_str(), OCSP_cert_status_str() and
   5626      OCSP_crl_reason_str() and are no longer static. New options
   5627      to verify nonce values and to disable verification. OCSP response
   5628      printout format cleaned up.
   5629      [Steve Henson]
   5630 
   5631   *) Add additional OCSP certificate checks. These are those specified
   5632      in RFC2560. This consists of two separate checks: the CA of the
   5633      certificate being checked must either be the OCSP signer certificate
   5634      or the issuer of the OCSP signer certificate. In the latter case the
   5635      OCSP signer certificate must contain the OCSP signing extended key
   5636      usage. This check is performed by attempting to match the OCSP
   5637      signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
   5638      in the OCSP_CERTID structures of the response.
   5639      [Steve Henson]
   5640 
   5641   *) Initial OCSP certificate verification added to OCSP_basic_verify()
   5642      and related routines. This uses the standard OpenSSL certificate
   5643      verify routines to perform initial checks (just CA validity) and
   5644      to obtain the certificate chain. Then additional checks will be
   5645      performed on the chain. Currently the root CA is checked to see
   5646      if it is explicitly trusted for OCSP signing. This is used to set
   5647      a root CA as a global signing root: that is any certificate that
   5648      chains to that CA is an acceptable OCSP signing certificate.
   5649      [Steve Henson]
   5650 
   5651   *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
   5652      extensions from a separate configuration file.
   5653      As when reading extensions from the main configuration file,
   5654      the '-extensions ...' option may be used for specifying the
   5655      section to use.
   5656      [Massimiliano Pala <madwolf (a] comune.modena.it>]
   5657 
   5658   *) New OCSP utility. Allows OCSP requests to be generated or
   5659      read. The request can be sent to a responder and the output
   5660      parsed, outputed or printed in text form. Not complete yet:
   5661      still needs to check the OCSP response validity.
   5662      [Steve Henson]
   5663 
   5664   *) New subcommands for 'openssl ca':
   5665      'openssl ca -status <serial>' prints the status of the cert with
   5666      the given serial number (according to the index file).
   5667      'openssl ca -updatedb' updates the expiry status of certificates
   5668      in the index file.
   5669      [Massimiliano Pala <madwolf (a] comune.modena.it>]
   5670 
   5671   *) New '-newreq-nodes' command option to CA.pl.  This is like
   5672      '-newreq', but calls 'openssl req' with the '-nodes' option
   5673      so that the resulting key is not encrypted.
   5674      [Damien Miller <djm (a] mindrot.org>]
   5675 
   5676   *) New configuration for the GNU Hurd.
   5677      [Jonathan Bartlett <johnnyb (a] wolfram.com> via Richard Levitte]
   5678 
   5679   *) Initial code to implement OCSP basic response verify. This
   5680      is currently incomplete. Currently just finds the signer's
   5681      certificate and verifies the signature on the response.
   5682      [Steve Henson]
   5683 
   5684   *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
   5685      value of OPENSSLDIR.  This is available via the new '-d' option
   5686      to 'openssl version', and is also included in 'openssl version -a'.
   5687      [Bodo Moeller]
   5688 
   5689   *) Allowing defining memory allocation callbacks that will be given
   5690      file name and line number information in additional arguments
   5691      (a const char* and an int).  The basic functionality remains, as
   5692      well as the original possibility to just replace malloc(),
   5693      realloc() and free() by functions that do not know about these
   5694      additional arguments.  To register and find out the current
   5695      settings for extended allocation functions, the following
   5696      functions are provided:
   5697 
   5698 	CRYPTO_set_mem_ex_functions
   5699 	CRYPTO_set_locked_mem_ex_functions
   5700 	CRYPTO_get_mem_ex_functions
   5701 	CRYPTO_get_locked_mem_ex_functions
   5702 
   5703      These work the same way as CRYPTO_set_mem_functions and friends.
   5704      CRYPTO_get_[locked_]mem_functions now writes 0 where such an
   5705      extended allocation function is enabled.
   5706      Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
   5707      a conventional allocation function is enabled.
   5708      [Richard Levitte, Bodo Moeller]
   5709 
   5710   *) Finish off removing the remaining LHASH function pointer casts.
   5711      There should no longer be any prototype-casting required when using
   5712      the LHASH abstraction, and any casts that remain are "bugs". See
   5713      the callback types and macros at the head of lhash.h for details
   5714      (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
   5715      [Geoff Thorpe]
   5716 
   5717   *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
   5718      If /dev/[u]random devices are not available or do not return enough
   5719      entropy, EGD style sockets (served by EGD or PRNGD) will automatically
   5720      be queried.
   5721      The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
   5722      /etc/entropy will be queried once each in this sequence, quering stops
   5723      when enough entropy was collected without querying more sockets.
   5724      [Lutz Jaenicke]
   5725 
   5726   *) Change the Unix RAND_poll() variant to be able to poll several
   5727      random devices, as specified by DEVRANDOM, until a sufficient amount
   5728      of data has been collected.   We spend at most 10 ms on each file
   5729      (select timeout) and read in non-blocking mode.  DEVRANDOM now
   5730      defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
   5731      (previously it was just the string "/dev/urandom"), so on typical
   5732      platforms the 10 ms delay will never occur.
   5733      Also separate out the Unix variant to its own file, rand_unix.c.
   5734      For VMS, there's a currently-empty rand_vms.c.
   5735      [Richard Levitte]
   5736 
   5737   *) Move OCSP client related routines to ocsp_cl.c. These
   5738      provide utility functions which an application needing
   5739      to issue a request to an OCSP responder and analyse the
   5740      response will typically need: as opposed to those which an
   5741      OCSP responder itself would need which will be added later.
   5742 
   5743      OCSP_request_sign() signs an OCSP request with an API similar
   5744      to PKCS7_sign(). OCSP_response_status() returns status of OCSP
   5745      response. OCSP_response_get1_basic() extracts basic response
   5746      from response. OCSP_resp_find_status(): finds and extracts status
   5747      information from an OCSP_CERTID structure (which will be created
   5748      when the request structure is built). These are built from lower
   5749      level functions which work on OCSP_SINGLERESP structures but
   5750      wont normally be used unless the application wishes to examine
   5751      extensions in the OCSP response for example.
   5752 
   5753      Replace nonce routines with a pair of functions.
   5754      OCSP_request_add1_nonce() adds a nonce value and optionally
   5755      generates a random value. OCSP_check_nonce() checks the
   5756      validity of the nonce in an OCSP response.
   5757      [Steve Henson]
   5758 
   5759   *) Change function OCSP_request_add() to OCSP_request_add0_id().
   5760      This doesn't copy the supplied OCSP_CERTID and avoids the
   5761      need to free up the newly created id. Change return type
   5762      to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
   5763      This can then be used to add extensions to the request.
   5764      Deleted OCSP_request_new(), since most of its functionality
   5765      is now in OCSP_REQUEST_new() (and the case insensitive name
   5766      clash) apart from the ability to set the request name which
   5767      will be added elsewhere.
   5768      [Steve Henson]
   5769 
   5770   *) Update OCSP API. Remove obsolete extensions argument from
   5771      various functions. Extensions are now handled using the new
   5772      OCSP extension code. New simple OCSP HTTP function which 
   5773      can be used to send requests and parse the response.
   5774      [Steve Henson]
   5775 
   5776   *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
   5777      ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
   5778      uses the special reorder version of SET OF to sort the attributes
   5779      and reorder them to match the encoded order. This resolves a long
   5780      standing problem: a verify on a PKCS7 structure just after signing
   5781      it used to fail because the attribute order did not match the
   5782      encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
   5783      it uses the received order. This is necessary to tolerate some broken
   5784      software that does not order SET OF. This is handled by encoding
   5785      as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
   5786      to produce the required SET OF.
   5787      [Steve Henson]
   5788 
   5789   *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
   5790      OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
   5791      files to get correct declarations of the ASN.1 item variables.
   5792      [Richard Levitte]
   5793 
   5794   *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
   5795      PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
   5796      asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
   5797      NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
   5798      New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
   5799      ASN1_ITEM and no wrapper functions.
   5800      [Steve Henson]
   5801 
   5802   *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
   5803      replace the old function pointer based I/O routines. Change most of
   5804      the *_d2i_bio() and *_d2i_fp() functions to use these.
   5805      [Steve Henson]
   5806 
   5807   *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
   5808      lines, recognice more "algorithms" that can be deselected, and make
   5809      it complain about algorithm deselection that isn't recognised.
   5810      [Richard Levitte]
   5811 
   5812   *) New ASN1 functions to handle dup, sign, verify, digest, pack and
   5813      unpack operations in terms of ASN1_ITEM. Modify existing wrappers
   5814      to use new functions. Add NO_ASN1_OLD which can be set to remove
   5815      some old style ASN1 functions: this can be used to determine if old
   5816      code will still work when these eventually go away.
   5817      [Steve Henson]
   5818 
   5819   *) New extension functions for OCSP structures, these follow the
   5820      same conventions as certificates and CRLs.
   5821      [Steve Henson]
   5822 
   5823   *) New function X509V3_add1_i2d(). This automatically encodes and
   5824      adds an extension. Its behaviour can be customised with various
   5825      flags to append, replace or delete. Various wrappers added for
   5826      certifcates and CRLs.
   5827      [Steve Henson]
   5828 
   5829   *) Fix to avoid calling the underlying ASN1 print routine when
   5830      an extension cannot be parsed. Correct a typo in the
   5831      OCSP_SERVICELOC extension. Tidy up print OCSP format.
   5832      [Steve Henson]
   5833 
   5834   *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
   5835      entries for variables.
   5836      [Steve Henson]
   5837 
   5838   *) Add functionality to apps/openssl.c for detecting locking
   5839      problems: As the program is single-threaded, all we have
   5840      to do is register a locking callback using an array for
   5841      storing which locks are currently held by the program.
   5842      [Bodo Moeller]
   5843 
   5844   *) Use a lock around the call to CRYPTO_get_ex_new_index() in
   5845      SSL_get_ex_data_X509_STORE_idx(), which is used in
   5846      ssl_verify_cert_chain() and thus can be called at any time
   5847      during TLS/SSL handshakes so that thread-safety is essential.
   5848      Unfortunately, the ex_data design is not at all suited
   5849      for multi-threaded use, so it probably should be abolished.
   5850      [Bodo Moeller]
   5851 
   5852   *) Added Broadcom "ubsec" ENGINE to OpenSSL.
   5853      [Broadcom, tweaked and integrated by Geoff Thorpe]
   5854 
   5855   *) Move common extension printing code to new function
   5856      X509V3_print_extensions(). Reorganise OCSP print routines and
   5857      implement some needed OCSP ASN1 functions. Add OCSP extensions.
   5858      [Steve Henson]
   5859 
   5860   *) New function X509_signature_print() to remove duplication in some
   5861      print routines.
   5862      [Steve Henson]
   5863 
   5864   *) Add a special meaning when SET OF and SEQUENCE OF flags are both
   5865      set (this was treated exactly the same as SET OF previously). This
   5866      is used to reorder the STACK representing the structure to match the
   5867      encoding. This will be used to get round a problem where a PKCS7
   5868      structure which was signed could not be verified because the STACK
   5869      order did not reflect the encoded order.
   5870      [Steve Henson]
   5871 
   5872   *) Reimplement the OCSP ASN1 module using the new code.
   5873      [Steve Henson]
   5874 
   5875   *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
   5876      for its ASN1 operations. The old style function pointers still exist
   5877      for now but they will eventually go away.
   5878      [Steve Henson]
   5879 
   5880   *) Merge in replacement ASN1 code from the ASN1 branch. This almost
   5881      completely replaces the old ASN1 functionality with a table driven
   5882      encoder and decoder which interprets an ASN1_ITEM structure describing
   5883      the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
   5884      largely maintained. Almost all of the old asn1_mac.h macro based ASN1
   5885      has also been converted to the new form.
   5886      [Steve Henson]
   5887 
   5888   *) Change BN_mod_exp_recp so that negative moduli are tolerated
   5889      (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
   5890      so that BN_mod_exp_mont and BN_mod_exp_mont_word work
   5891      for negative moduli.
   5892      [Bodo Moeller]
   5893 
   5894   *) Fix BN_uadd and BN_usub: Always return non-negative results instead
   5895      of not touching the result's sign bit.
   5896      [Bodo Moeller]
   5897 
   5898   *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
   5899      set.
   5900      [Bodo Moeller]
   5901 
   5902   *) Changed the LHASH code to use prototypes for callbacks, and created
   5903      macros to declare and implement thin (optionally static) functions
   5904      that provide type-safety and avoid function pointer casting for the
   5905      type-specific callbacks.
   5906      [Geoff Thorpe]
   5907 
   5908   *) Added Kerberos Cipher Suites to be used with TLS, as written in
   5909      RFC 2712.
   5910      [Veers Staats <staatsvr (a] asc.hpc.mil>,
   5911       Jeffrey Altman <jaltman (a] columbia.edu>, via Richard Levitte]
   5912 
   5913   *) Reformat the FAQ so the different questions and answers can be divided
   5914      in sections depending on the subject.
   5915      [Richard Levitte]
   5916 
   5917   *) Have the zlib compression code load ZLIB.DLL dynamically under
   5918      Windows.
   5919      [Richard Levitte]
   5920 
   5921   *) New function BN_mod_sqrt for computing square roots modulo a prime
   5922      (using the probabilistic Tonelli-Shanks algorithm unless
   5923      p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
   5924      be handled deterministically).
   5925      [Lenka Fibikova <fibikova (a] exp-math.uni-essen.de>, Bodo Moeller]
   5926 
   5927   *) Make BN_mod_inverse faster by explicitly handling small quotients
   5928      in the Euclid loop. (Speed gain about 20% for small moduli [256 or
   5929      512 bits], about 30% for larger ones [1024 or 2048 bits].)
   5930      [Bodo Moeller]
   5931 
   5932   *) New function BN_kronecker.
   5933      [Bodo Moeller]
   5934 
   5935   *) Fix BN_gcd so that it works on negative inputs; the result is
   5936      positive unless both parameters are zero.
   5937      Previously something reasonably close to an infinite loop was
   5938      possible because numbers could be growing instead of shrinking
   5939      in the implementation of Euclid's algorithm.
   5940      [Bodo Moeller]
   5941 
   5942   *) Fix BN_is_word() and BN_is_one() macros to take into account the
   5943      sign of the number in question.
   5944 
   5945      Fix BN_is_word(a,w) to work correctly for w == 0.
   5946 
   5947      The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
   5948      because its test if the absolute value of 'a' equals 'w'.
   5949      Note that BN_abs_is_word does *not* handle w == 0 reliably;
   5950      it exists mostly for use in the implementations of BN_is_zero(),
   5951      BN_is_one(), and BN_is_word().
   5952      [Bodo Moeller]
   5953 
   5954   *) New function BN_swap.
   5955      [Bodo Moeller]
   5956 
   5957   *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
   5958      the exponentiation functions are more likely to produce reasonable
   5959      results on negative inputs.
   5960      [Bodo Moeller]
   5961 
   5962   *) Change BN_mod_mul so that the result is always non-negative.
   5963      Previously, it could be negative if one of the factors was negative;
   5964      I don't think anyone really wanted that behaviour.
   5965      [Bodo Moeller]
   5966 
   5967   *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
   5968      (except for exponentiation, which stays in crypto/bn/bn_exp.c,
   5969      and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
   5970      and add new functions:
   5971 
   5972           BN_nnmod
   5973           BN_mod_sqr
   5974           BN_mod_add
   5975           BN_mod_add_quick
   5976           BN_mod_sub
   5977           BN_mod_sub_quick
   5978           BN_mod_lshift1
   5979           BN_mod_lshift1_quick
   5980           BN_mod_lshift
   5981           BN_mod_lshift_quick
   5982 
   5983      These functions always generate non-negative results.
   5984 
   5985      BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder  r
   5986      such that  |m| < r < 0,  BN_nnmod will output  rem + |m|  instead).
   5987 
   5988      BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
   5989      BN_mod_XXX(r, a, [b,] m, ctx), but requires that  a  [and  b]
   5990      be reduced modulo  m.
   5991      [Lenka Fibikova <fibikova (a] exp-math.uni-essen.de>, Bodo Moeller]
   5992 
   5993 #if 0
   5994      The following entry accidentily appeared in the CHANGES file
   5995      distributed with OpenSSL 0.9.7.  The modifications described in
   5996      it do *not* apply to OpenSSL 0.9.7.
   5997 
   5998   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
   5999      was actually never needed) and in BN_mul().  The removal in BN_mul()
   6000      required a small change in bn_mul_part_recursive() and the addition
   6001      of the functions bn_cmp_part_words(), bn_sub_part_words() and
   6002      bn_add_part_words(), which do the same thing as bn_cmp_words(),
   6003      bn_sub_words() and bn_add_words() except they take arrays with
   6004      differing sizes.
   6005      [Richard Levitte]
   6006 #endif
   6007 
   6008   *) In 'openssl passwd', verify passwords read from the terminal
   6009      unless the '-salt' option is used (which usually means that
   6010      verification would just waste user's time since the resulting
   6011      hash is going to be compared with some given password hash)
   6012      or the new '-noverify' option is used.
   6013 
   6014      This is an incompatible change, but it does not affect
   6015      non-interactive use of 'openssl passwd' (passwords on the command
   6016      line, '-stdin' option, '-in ...' option) and thus should not
   6017      cause any problems.
   6018      [Bodo Moeller]
   6019 
   6020   *) Remove all references to RSAref, since there's no more need for it.
   6021      [Richard Levitte]
   6022 
   6023   *) Make DSO load along a path given through an environment variable
   6024      (SHLIB_PATH) with shl_load().
   6025      [Richard Levitte]
   6026 
   6027   *) Constify the ENGINE code as a result of BIGNUM constification.
   6028      Also constify the RSA code and most things related to it.  In a
   6029      few places, most notable in the depth of the ASN.1 code, ugly
   6030      casts back to non-const were required (to be solved at a later
   6031      time)
   6032      [Richard Levitte]
   6033 
   6034   *) Make it so the openssl application has all engines loaded by default.
   6035      [Richard Levitte]
   6036 
   6037   *) Constify the BIGNUM routines a little more.
   6038      [Richard Levitte]
   6039 
   6040   *) Add the following functions:
   6041 
   6042 	ENGINE_load_cswift()
   6043 	ENGINE_load_chil()
   6044 	ENGINE_load_atalla()
   6045 	ENGINE_load_nuron()
   6046 	ENGINE_load_builtin_engines()
   6047 
   6048      That way, an application can itself choose if external engines that
   6049      are built-in in OpenSSL shall ever be used or not.  The benefit is
   6050      that applications won't have to be linked with libdl or other dso
   6051      libraries unless it's really needed.
   6052 
   6053      Changed 'openssl engine' to load all engines on demand.
   6054      Changed the engine header files to avoid the duplication of some
   6055      declarations (they differed!).
   6056      [Richard Levitte]
   6057 
   6058   *) 'openssl engine' can now list capabilities.
   6059      [Richard Levitte]
   6060 
   6061   *) Better error reporting in 'openssl engine'.
   6062      [Richard Levitte]
   6063 
   6064   *) Never call load_dh_param(NULL) in s_server.
   6065      [Bodo Moeller]
   6066 
   6067   *) Add engine application.  It can currently list engines by name and
   6068      identity, and test if they are actually available.
   6069      [Richard Levitte]
   6070 
   6071   *) Improve RPM specification file by forcing symbolic linking and making
   6072      sure the installed documentation is also owned by root.root.
   6073      [Damien Miller <djm (a] mindrot.org>]
   6074 
   6075   *) Give the OpenSSL applications more possibilities to make use of
   6076      keys (public as well as private) handled by engines.
   6077      [Richard Levitte]
   6078 
   6079   *) Add OCSP code that comes from CertCo.
   6080      [Richard Levitte]
   6081 
   6082   *) Add VMS support for the Rijndael code.
   6083      [Richard Levitte]
   6084 
   6085   *) Added untested support for Nuron crypto accelerator.
   6086      [Ben Laurie]
   6087 
   6088   *) Add support for external cryptographic devices.  This code was
   6089      previously distributed separately as the "engine" branch.
   6090      [Geoff Thorpe, Richard Levitte]
   6091 
   6092   *) Rework the filename-translation in the DSO code. It is now possible to
   6093      have far greater control over how a "name" is turned into a filename
   6094      depending on the operating environment and any oddities about the
   6095      different shared library filenames on each system.
   6096      [Geoff Thorpe]
   6097 
   6098   *) Support threads on FreeBSD-elf in Configure.
   6099      [Richard Levitte]
   6100 
   6101   *) Fix for SHA1 assembly problem with MASM: it produces
   6102      warnings about corrupt line number information when assembling
   6103      with debugging information. This is caused by the overlapping
   6104      of two sections.
   6105      [Bernd Matthes <mainbug (a] celocom.de>, Steve Henson]
   6106 
   6107   *) NCONF changes.
   6108      NCONF_get_number() has no error checking at all.  As a replacement,
   6109      NCONF_get_number_e() is defined (_e for "error checking") and is
   6110      promoted strongly.  The old NCONF_get_number is kept around for
   6111      binary backward compatibility.
   6112      Make it possible for methods to load from something other than a BIO,
   6113      by providing a function pointer that is given a name instead of a BIO.
   6114      For example, this could be used to load configuration data from an
   6115      LDAP server.
   6116      [Richard Levitte]
   6117 
   6118   *) Fix for non blocking accept BIOs. Added new I/O special reason
   6119      BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
   6120      with non blocking I/O was not possible because no retry code was
   6121      implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
   6122      this case.
   6123      [Steve Henson]
   6124 
   6125   *) Added the beginnings of Rijndael support.
   6126      [Ben Laurie]
   6127 
   6128   *) Fix for bug in DirectoryString mask setting. Add support for
   6129      X509_NAME_print_ex() in 'req' and X509_print_ex() function
   6130      to allow certificate printing to more controllable, additional
   6131      'certopt' option to 'x509' to allow new printing options to be
   6132      set.
   6133      [Steve Henson]
   6134 
   6135   *) Clean old EAY MD5 hack from e_os.h.
   6136      [Richard Levitte]
   6137 
   6138  Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
   6139 
   6140   *) Fix null-pointer assignment in do_change_cipher_spec() revealed
   6141      by using the Codenomicon TLS Test Tool (CVE-2004-0079)
   6142      [Joe Orton, Steve Henson]
   6143 
   6144  Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
   6145 
   6146   *) Fix additional bug revealed by the NISCC test suite:
   6147 
   6148      Stop bug triggering large recursion when presented with
   6149      certain ASN.1 tags (CVE-2003-0851)
   6150      [Steve Henson]
   6151 
   6152  Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
   6153 
   6154   *) Fix various bugs revealed by running the NISCC test suite:
   6155 
   6156      Stop out of bounds reads in the ASN1 code when presented with
   6157      invalid tags (CVE-2003-0543 and CVE-2003-0544).
   6158      
   6159      If verify callback ignores invalid public key errors don't try to check
   6160      certificate signature with the NULL public key.
   6161 
   6162      [Steve Henson]
   6163 
   6164   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
   6165      if the server requested one: as stated in TLS 1.0 and SSL 3.0
   6166      specifications.
   6167      [Steve Henson]
   6168 
   6169   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
   6170      extra data after the compression methods not only for TLS 1.0
   6171      but also for SSL 3.0 (as required by the specification).
   6172      [Bodo Moeller; problem pointed out by Matthias Loepfe]
   6173 
   6174   *) Change X509_certificate_type() to mark the key as exported/exportable
   6175      when it's 512 *bits* long, not 512 bytes.
   6176      [Richard Levitte]
   6177 
   6178  Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
   6179 
   6180   *) Countermeasure against the Klima-Pokorny-Rosa extension of
   6181      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
   6182      a protocol version number mismatch like a decryption error
   6183      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
   6184      [Bodo Moeller]
   6185 
   6186   *) Turn on RSA blinding by default in the default implementation
   6187      to avoid a timing attack. Applications that don't want it can call
   6188      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
   6189      They would be ill-advised to do so in most cases.
   6190      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
   6191 
   6192   *) Change RSA blinding code so that it works when the PRNG is not
   6193      seeded (in this case, the secret RSA exponent is abused as
   6194      an unpredictable seed -- if it is not unpredictable, there
   6195      is no point in blinding anyway).  Make RSA blinding thread-safe
   6196      by remembering the creator's thread ID in rsa->blinding and
   6197      having all other threads use local one-time blinding factors
   6198      (this requires more computation than sharing rsa->blinding, but
   6199      avoids excessive locking; and if an RSA object is not shared
   6200      between threads, blinding will still be very fast).
   6201      [Bodo Moeller]
   6202 
   6203  Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
   6204 
   6205   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
   6206      via timing by performing a MAC computation even if incorrrect
   6207      block cipher padding has been found.  This is a countermeasure
   6208      against active attacks where the attacker has to distinguish
   6209      between bad padding and a MAC verification error. (CVE-2003-0078)
   6210 
   6211      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
   6212      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
   6213      Martin Vuagnoux (EPFL, Ilion)]
   6214 
   6215  Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
   6216 
   6217   *) New function OPENSSL_cleanse(), which is used to cleanse a section of
   6218      memory from it's contents.  This is done with a counter that will
   6219      place alternating values in each byte.  This can be used to solve
   6220      two issues: 1) the removal of calls to memset() by highly optimizing
   6221      compilers, and 2) cleansing with other values than 0, since those can
   6222      be read through on certain media, for example a swap space on disk.
   6223      [Geoff Thorpe]
   6224 
   6225   *) Bugfix: client side session caching did not work with external caching,
   6226      because the session->cipher setting was not restored when reloading
   6227      from the external cache. This problem was masked, when
   6228      SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
   6229      (Found by Steve Haslam <steve (a] araqnid.ddts.net>.)
   6230      [Lutz Jaenicke]
   6231 
   6232   *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
   6233      length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
   6234      [Zeev Lieber <zeev-l (a] yahoo.com>]
   6235 
   6236   *) Undo an undocumented change introduced in 0.9.6e which caused
   6237      repeated calls to OpenSSL_add_all_ciphers() and 
   6238      OpenSSL_add_all_digests() to be ignored, even after calling
   6239      EVP_cleanup().
   6240      [Richard Levitte]
   6241 
   6242   *) Change the default configuration reader to deal with last line not
   6243      being properly terminated.
   6244      [Richard Levitte]
   6245 
   6246   *) Change X509_NAME_cmp() so it applies the special rules on handling
   6247      DN values that are of type PrintableString, as well as RDNs of type
   6248      emailAddress where the value has the type ia5String.
   6249      [stefank (a] valicert.com via Richard Levitte]
   6250 
   6251   *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
   6252      the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
   6253      doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
   6254      the bitwise-OR of the two for use by the majority of applications
   6255      wanting this behaviour, and update the docs. The documented
   6256      behaviour and actual behaviour were inconsistent and had been
   6257      changing anyway, so this is more a bug-fix than a behavioural
   6258      change.
   6259      [Geoff Thorpe, diagnosed by Nadav Har'El]
   6260 
   6261   *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
   6262      (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
   6263      [Bodo Moeller]
   6264 
   6265   *) Fix initialization code race conditions in
   6266         SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
   6267         SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
   6268         SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
   6269         TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
   6270         ssl2_get_cipher_by_char(),
   6271         ssl3_get_cipher_by_char().
   6272      [Patrick McCormick <patrick (a] tellme.com>, Bodo Moeller]
   6273 
   6274   *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
   6275      the cached sessions are flushed, as the remove_cb() might use ex_data
   6276      contents. Bug found by Sam Varshavchik <mrsam (a] courier-mta.com>
   6277      (see [openssl.org #212]).
   6278      [Geoff Thorpe, Lutz Jaenicke]
   6279 
   6280   *) Fix typo in OBJ_txt2obj which incorrectly passed the content
   6281      length, instead of the encoding length to d2i_ASN1_OBJECT.
   6282      [Steve Henson]
   6283 
   6284  Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
   6285 
   6286   *) [In 0.9.6g-engine release:]
   6287      Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
   6288      [Lynn Gazis <lgazis (a] rainbow.com>]
   6289 
   6290  Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
   6291 
   6292   *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
   6293      and get fix the header length calculation.
   6294      [Florian Weimer <Weimer (a] CERT.Uni-Stuttgart.DE>,
   6295 	Alon Kantor <alonk (a] checkpoint.com> (and others),
   6296 	Steve Henson]
   6297 
   6298   *) Use proper error handling instead of 'assertions' in buffer
   6299      overflow checks added in 0.9.6e.  This prevents DoS (the
   6300      assertions could call abort()).
   6301      [Arne Ansper <arne (a] ats.cyber.ee>, Bodo Moeller]
   6302 
   6303  Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
   6304 
   6305   *) Add various sanity checks to asn1_get_length() to reject
   6306      the ASN1 length bytes if they exceed sizeof(long), will appear
   6307      negative or the content length exceeds the length of the
   6308      supplied buffer.
   6309      [Steve Henson, Adi Stav <stav (a] mercury.co.il>, James Yonan <jim (a] ntlp.com>]
   6310 
   6311   *) Fix cipher selection routines: ciphers without encryption had no flags
   6312      for the cipher strength set and where therefore not handled correctly
   6313      by the selection routines (PR #130).
   6314      [Lutz Jaenicke]
   6315 
   6316   *) Fix EVP_dsa_sha macro.
   6317      [Nils Larsch]
   6318 
   6319   *) New option
   6320           SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
   6321      for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
   6322      that was added in OpenSSL 0.9.6d.
   6323 
   6324      As the countermeasure turned out to be incompatible with some
   6325      broken SSL implementations, the new option is part of SSL_OP_ALL.
   6326      SSL_OP_ALL is usually employed when compatibility with weird SSL
   6327      implementations is desired (e.g. '-bugs' option to 's_client' and
   6328      's_server'), so the new option is automatically set in many
   6329      applications.
   6330      [Bodo Moeller]
   6331 
   6332   *) Changes in security patch:
   6333 
   6334      Changes marked "(CHATS)" were sponsored by the Defense Advanced
   6335      Research Projects Agency (DARPA) and Air Force Research Laboratory,
   6336      Air Force Materiel Command, USAF, under agreement number
   6337      F30602-01-2-0537.
   6338 
   6339   *) Add various sanity checks to asn1_get_length() to reject
   6340      the ASN1 length bytes if they exceed sizeof(long), will appear
   6341      negative or the content length exceeds the length of the
   6342      supplied buffer. (CVE-2002-0659)
   6343      [Steve Henson, Adi Stav <stav (a] mercury.co.il>, James Yonan <jim (a] ntlp.com>]
   6344 
   6345   *) Assertions for various potential buffer overflows, not known to
   6346      happen in practice.
   6347      [Ben Laurie (CHATS)]
   6348 
   6349   *) Various temporary buffers to hold ASCII versions of integers were
   6350      too small for 64 bit platforms. (CVE-2002-0655)
   6351      [Matthew Byng-Maddick <mbm (a] aldigital.co.uk> and Ben Laurie (CHATS)>
   6352 
   6353   *) Remote buffer overflow in SSL3 protocol - an attacker could
   6354      supply an oversized session ID to a client. (CVE-2002-0656)
   6355      [Ben Laurie (CHATS)]
   6356 
   6357   *) Remote buffer overflow in SSL2 protocol - an attacker could
   6358      supply an oversized client master key. (CVE-2002-0656)
   6359      [Ben Laurie (CHATS)]
   6360 
   6361  Changes between 0.9.6c and 0.9.6d  [9 May 2002]
   6362 
   6363   *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
   6364      encoded as NULL) with id-dsa-with-sha1.
   6365      [Nils Larsch <nla (a] trustcenter.de>; problem pointed out by Bodo Moeller]
   6366 
   6367   *) Check various X509_...() return values in apps/req.c.
   6368      [Nils Larsch <nla (a] trustcenter.de>]
   6369 
   6370   *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
   6371      an end-of-file condition would erronously be flagged, when the CRLF
   6372      was just at the end of a processed block. The bug was discovered when
   6373      processing data through a buffering memory BIO handing the data to a
   6374      BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
   6375      <ptsekov (a] syntrex.com> and Nedelcho Stanev.
   6376      [Lutz Jaenicke]
   6377 
   6378   *) Implement a countermeasure against a vulnerability recently found
   6379      in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
   6380      before application data chunks to avoid the use of known IVs
   6381      with data potentially chosen by the attacker.
   6382      [Bodo Moeller]
   6383 
   6384   *) Fix length checks in ssl3_get_client_hello().
   6385      [Bodo Moeller]
   6386 
   6387   *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
   6388      to prevent ssl3_read_internal() from incorrectly assuming that
   6389      ssl3_read_bytes() found application data while handshake
   6390      processing was enabled when in fact s->s3->in_read_app_data was
   6391      merely automatically cleared during the initial handshake.
   6392      [Bodo Moeller; problem pointed out by Arne Ansper <arne (a] ats.cyber.ee>]
   6393 
   6394   *) Fix object definitions for Private and Enterprise: they were not
   6395      recognized in their shortname (=lowercase) representation. Extend
   6396      obj_dat.pl to issue an error when using undefined keywords instead
   6397      of silently ignoring the problem (Svenning Sorensen
   6398      <sss (a] sss.dnsalias.net>).
   6399      [Lutz Jaenicke]
   6400 
   6401   *) Fix DH_generate_parameters() so that it works for 'non-standard'
   6402      generators, i.e. generators other than 2 and 5.  (Previously, the
   6403      code did not properly initialise the 'add' and 'rem' values to
   6404      BN_generate_prime().)
   6405 
   6406      In the new general case, we do not insist that 'generator' is
   6407      actually a primitive root: This requirement is rather pointless;
   6408      a generator of the order-q subgroup is just as good, if not
   6409      better.
   6410      [Bodo Moeller]
   6411  
   6412   *) Map new X509 verification errors to alerts. Discovered and submitted by
   6413      Tom Wu <tom (a] arcot.com>.
   6414      [Lutz Jaenicke]
   6415 
   6416   *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
   6417      returning non-zero before the data has been completely received
   6418      when using non-blocking I/O.
   6419      [Bodo Moeller; problem pointed out by John Hughes]
   6420 
   6421   *) Some of the ciphers missed the strength entry (SSL_LOW etc).
   6422      [Ben Laurie, Lutz Jaenicke]
   6423 
   6424   *) Fix bug in SSL_clear(): bad sessions were not removed (found by
   6425      Yoram Zahavi <YoramZ (a] gilian.com>).
   6426      [Lutz Jaenicke]
   6427 
   6428   *) Add information about CygWin 1.3 and on, and preserve proper
   6429      configuration for the versions before that.
   6430      [Corinna Vinschen <vinschen (a] redhat.com> and Richard Levitte]
   6431 
   6432   *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
   6433      check whether we deal with a copy of a session and do not delete from
   6434      the cache in this case. Problem reported by "Izhar Shoshani Levi"
   6435      <izhar (a] checkpoint.com>.
   6436      [Lutz Jaenicke]
   6437 
   6438   *) Do not store session data into the internal session cache, if it
   6439      is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
   6440      flag is set). Proposed by Aslam <aslam (a] funk.com>.
   6441      [Lutz Jaenicke]
   6442 
   6443   *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
   6444      value is 0.
   6445      [Richard Levitte]
   6446 
   6447   *) [In 0.9.6d-engine release:]
   6448      Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
   6449      [Toomas Kiisk <vix (a] cyber.ee> via Richard Levitte]
   6450 
   6451   *) Add the configuration target linux-s390x.
   6452      [Neale Ferguson <Neale.Ferguson (a] SoftwareAG-USA.com> via Richard Levitte]
   6453 
   6454   *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
   6455      ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
   6456      variable as an indication that a ClientHello message has been
   6457      received.  As the flag value will be lost between multiple
   6458      invocations of ssl3_accept when using non-blocking I/O, the
   6459      function may not be aware that a handshake has actually taken
   6460      place, thus preventing a new session from being added to the
   6461      session cache.
   6462 
   6463      To avoid this problem, we now set s->new_session to 2 instead of
   6464      using a local variable.
   6465      [Lutz Jaenicke, Bodo Moeller]
   6466 
   6467   *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
   6468      if the SSL_R_LENGTH_MISMATCH error is detected.
   6469      [Geoff Thorpe, Bodo Moeller]
   6470 
   6471   *) New 'shared_ldflag' column in Configure platform table.
   6472      [Richard Levitte]
   6473 
   6474   *) Fix EVP_CIPHER_mode macro.
   6475      ["Dan S. Camper" <dan (a] bti.net>]
   6476 
   6477   *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
   6478      type, we must throw them away by setting rr->length to 0.
   6479      [D P Chang <dpc (a] qualys.com>]
   6480 
   6481  Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
   6482 
   6483   *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
   6484      <Dominikus.Scherkl (a] biodata.com>.  (The previous implementation
   6485      worked incorrectly for those cases where  range = 10..._2  and
   6486      3*range  is two bits longer than  range.)
   6487      [Bodo Moeller]
   6488 
   6489   *) Only add signing time to PKCS7 structures if it is not already
   6490      present.
   6491      [Steve Henson]
   6492 
   6493   *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
   6494      OBJ_ld_ce should be OBJ_id_ce.
   6495      Also some ip-pda OIDs in crypto/objects/objects.txt were
   6496      incorrect (cf. RFC 3039).
   6497      [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
   6498 
   6499   *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
   6500      returns early because it has nothing to do.
   6501      [Andy Schneider <andy.schneider (a] bjss.co.uk>]
   6502 
   6503   *) [In 0.9.6c-engine release:]
   6504      Fix mutex callback return values in crypto/engine/hw_ncipher.c.
   6505      [Andy Schneider <andy.schneider (a] bjss.co.uk>]
   6506 
   6507   *) [In 0.9.6c-engine release:]
   6508      Add support for Cryptographic Appliance's keyserver technology.
   6509      (Use engine 'keyclient')
   6510      [Cryptographic Appliances and Geoff Thorpe]
   6511 
   6512   *) Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
   6513      is called via tools/c89.sh because arguments have to be
   6514      rearranged (all '-L' options must appear before the first object
   6515      modules).
   6516      [Richard Shapiro <rshapiro (a] abinitio.com>]
   6517 
   6518   *) [In 0.9.6c-engine release:]
   6519      Add support for Broadcom crypto accelerator cards, backported
   6520      from 0.9.7.
   6521      [Broadcom, Nalin Dahyabhai <nalin (a] redhat.com>, Mark Cox]
   6522 
   6523   *) [In 0.9.6c-engine release:]
   6524      Add support for SureWare crypto accelerator cards from 
   6525      Baltimore Technologies.  (Use engine 'sureware')
   6526      [Baltimore Technologies and Mark Cox]
   6527 
   6528   *) [In 0.9.6c-engine release:]
   6529      Add support for crypto accelerator cards from Accelerated
   6530      Encryption Processing, www.aep.ie.  (Use engine 'aep')
   6531      [AEP Inc. and Mark Cox]
   6532 
   6533   *) Add a configuration entry for gcc on UnixWare.
   6534      [Gary Benson <gbenson (a] redhat.com>]
   6535 
   6536   *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
   6537      messages are stored in a single piece (fixed-length part and
   6538      variable-length part combined) and fix various bugs found on the way.
   6539      [Bodo Moeller]
   6540 
   6541   *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
   6542      instead.  BIO_gethostbyname() does not know what timeouts are
   6543      appropriate, so entries would stay in cache even when they have
   6544      become invalid.
   6545      [Bodo Moeller; problem pointed out by Rich Salz <rsalz (a] zolera.com>
   6546 
   6547   *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
   6548      faced with a pathologically small ClientHello fragment that does
   6549      not contain client_version: Instead of aborting with an error,
   6550      simply choose the highest available protocol version (i.e.,
   6551      TLS 1.0 unless it is disabled).  In practice, ClientHello
   6552      messages are never sent like this, but this change gives us
   6553      strictly correct behaviour at least for TLS.
   6554      [Bodo Moeller]
   6555 
   6556   *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
   6557      never resets s->method to s->ctx->method when called from within
   6558      one of the SSL handshake functions.
   6559      [Bodo Moeller; problem pointed out by Niko Baric]
   6560 
   6561   *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
   6562      (sent using the client's version number) if client_version is
   6563      smaller than the protocol version in use.  Also change
   6564      ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
   6565      the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
   6566      the client will at least see that alert.
   6567      [Bodo Moeller]
   6568 
   6569   *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
   6570      correctly.
   6571      [Bodo Moeller]
   6572 
   6573   *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
   6574      client receives HelloRequest while in a handshake.
   6575      [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider (a] bjss.co.uk>]
   6576 
   6577   *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
   6578      should end in 'break', not 'goto end' which circuments various
   6579      cleanups done in state SSL_ST_OK.   But session related stuff
   6580      must be disabled for SSL_ST_OK in the case that we just sent a
   6581      HelloRequest.
   6582 
   6583      Also avoid some overhead by not calling ssl_init_wbio_buffer()
   6584      before just sending a HelloRequest.
   6585      [Bodo Moeller, Eric Rescorla <ekr (a] rtfm.com>]
   6586 
   6587   *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
   6588      reveal whether illegal block cipher padding was found or a MAC
   6589      verification error occured.  (Neither SSLerr() codes nor alerts
   6590      are directly visible to potential attackers, but the information
   6591      may leak via logfiles.)
   6592 
   6593      Similar changes are not required for the SSL 2.0 implementation
   6594      because the number of padding bytes is sent in clear for SSL 2.0,
   6595      and the extra bytes are just ignored.  However ssl/s2_pkt.c
   6596      failed to verify that the purported number of padding bytes is in
   6597      the legal range.
   6598      [Bodo Moeller]
   6599 
   6600   *) Add OpenUNIX-8 support including shared libraries
   6601      (Boyd Lynn Gerber <gerberb (a] zenez.com>).
   6602      [Lutz Jaenicke]
   6603 
   6604   *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
   6605      'wristwatch attack' using huge encoding parameters (cf.
   6606      James H. Manger's CRYPTO 2001 paper).  Note that the
   6607      RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
   6608      encoding parameters and hence was not vulnerable.
   6609      [Bodo Moeller]
   6610 
   6611   *) BN_sqr() bug fix.
   6612      [Ulf Mller, reported by Jim Ellis <jim.ellis (a] cavium.com>]
   6613 
   6614   *) Rabin-Miller test analyses assume uniformly distributed witnesses,
   6615      so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
   6616      followed by modular reduction.
   6617      [Bodo Moeller; pointed out by Adam Young <AYoung1 (a] NCSUS.JNJ.COM>]
   6618 
   6619   *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
   6620      equivalent based on BN_pseudo_rand() instead of BN_rand().
   6621      [Bodo Moeller]
   6622 
   6623   *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
   6624      This function was broken, as the check for a new client hello message
   6625      to handle SGC did not allow these large messages.
   6626      (Tracked down by "Douglas E. Engert" <deengert (a] anl.gov>.)
   6627      [Lutz Jaenicke]
   6628 
   6629   *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
   6630      [Lutz Jaenicke]
   6631 
   6632   *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
   6633      for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton (a] netopia.com>).
   6634      [Lutz Jaenicke]
   6635 
   6636   *) Rework the configuration and shared library support for Tru64 Unix.
   6637      The configuration part makes use of modern compiler features and
   6638      still retains old compiler behavior for those that run older versions
   6639      of the OS.  The shared library support part includes a variant that
   6640      uses the RPATH feature, and is available through the special
   6641      configuration target "alpha-cc-rpath", which will never be selected
   6642      automatically.
   6643      [Tim Mooney <mooney (a] dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
   6644 
   6645   *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
   6646      with the same message size as in ssl3_get_certificate_request().
   6647      Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
   6648      messages might inadvertently be reject as too long.
   6649      [Petr Lampa <lampa (a] fee.vutbr.cz>]
   6650 
   6651   *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
   6652      [Andy Polyakov]
   6653 
   6654   *) Modified SSL library such that the verify_callback that has been set
   6655      specificly for an SSL object with SSL_set_verify() is actually being
   6656      used. Before the change, a verify_callback set with this function was
   6657      ignored and the verify_callback() set in the SSL_CTX at the time of
   6658      the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
   6659      to allow the necessary settings.
   6660      [Lutz Jaenicke]
   6661 
   6662   *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
   6663      explicitly to NULL, as at least on Solaris 8 this seems not always to be
   6664      done automatically (in contradiction to the requirements of the C
   6665      standard). This made problems when used from OpenSSH.
   6666      [Lutz Jaenicke]
   6667 
   6668   *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
   6669      dh->length and always used
   6670 
   6671           BN_rand_range(priv_key, dh->p).
   6672 
   6673      BN_rand_range() is not necessary for Diffie-Hellman, and this
   6674      specific range makes Diffie-Hellman unnecessarily inefficient if
   6675      dh->length (recommended exponent length) is much smaller than the
   6676      length of dh->p.  We could use BN_rand_range() if the order of
   6677      the subgroup was stored in the DH structure, but we only have
   6678      dh->length.
   6679 
   6680      So switch back to
   6681 
   6682           BN_rand(priv_key, l, ...)
   6683 
   6684      where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
   6685      otherwise.
   6686      [Bodo Moeller]
   6687 
   6688   *) In
   6689 
   6690           RSA_eay_public_encrypt
   6691           RSA_eay_private_decrypt
   6692           RSA_eay_private_encrypt (signing)
   6693           RSA_eay_public_decrypt (signature verification)
   6694 
   6695      (default implementations for RSA_public_encrypt,
   6696      RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
   6697      always reject numbers >= n.
   6698      [Bodo Moeller]
   6699 
   6700   *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
   6701      to synchronize access to 'locking_thread'.  This is necessary on
   6702      systems where access to 'locking_thread' (an 'unsigned long'
   6703      variable) is not atomic.
   6704      [Bodo Moeller]
   6705 
   6706   *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
   6707      *before* setting the 'crypto_lock_rand' flag.  The previous code had
   6708      a race condition if 0 is a valid thread ID.
   6709      [Travis Vitek <vitek (a] roguewave.com>]
   6710 
   6711   *) Add support for shared libraries under Irix.
   6712      [Albert Chin-A-Young <china (a] thewrittenword.com>]
   6713 
   6714   *) Add configuration option to build on Linux on both big-endian and
   6715      little-endian MIPS.
   6716      [Ralf Baechle <ralf (a] uni-koblenz.de>]
   6717 
   6718   *) Add the possibility to create shared libraries on HP-UX.
   6719      [Richard Levitte]
   6720 
   6721  Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
   6722 
   6723   *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
   6724      to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
   6725      Markku-Juhani O. Saarinen <markku-juhani.saarinen (a] nokia.com>:
   6726      PRNG state recovery was possible based on the output of
   6727      one PRNG request appropriately sized to gain knowledge on
   6728      'md' followed by enough consecutive 1-byte PRNG requests
   6729      to traverse all of 'state'.
   6730 
   6731      1. When updating 'md_local' (the current thread's copy of 'md')
   6732         during PRNG output generation, hash all of the previous
   6733         'md_local' value, not just the half used for PRNG output.
   6734 
   6735      2. Make the number of bytes from 'state' included into the hash
   6736         independent from the number of PRNG bytes requested.
   6737 
   6738      The first measure alone would be sufficient to avoid
   6739      Markku-Juhani's attack.  (Actually it had never occurred
   6740      to me that the half of 'md_local' used for chaining was the
   6741      half from which PRNG output bytes were taken -- I had always
   6742      assumed that the secret half would be used.)  The second
   6743      measure makes sure that additional data from 'state' is never
   6744      mixed into 'md_local' in small portions; this heuristically
   6745      further strengthens the PRNG.
   6746      [Bodo Moeller]
   6747 
   6748   *) Fix crypto/bn/asm/mips3.s.
   6749      [Andy Polyakov]
   6750 
   6751   *) When only the key is given to "enc", the IV is undefined. Print out
   6752      an error message in this case.
   6753      [Lutz Jaenicke]
   6754 
   6755   *) Handle special case when X509_NAME is empty in X509 printing routines.
   6756      [Steve Henson]
   6757 
   6758   *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
   6759      positive and less than q.
   6760      [Bodo Moeller]
   6761 
   6762   *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
   6763      used: it isn't thread safe and the add_lock_callback should handle
   6764      that itself.
   6765      [Paul Rose <Paul.Rose (a] bridge.com>]
   6766 
   6767   *) Verify that incoming data obeys the block size in
   6768      ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
   6769      [Bodo Moeller]
   6770 
   6771   *) Fix OAEP check.
   6772      [Ulf Mller, Bodo Mller]
   6773 
   6774   *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
   6775      RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
   6776      when fixing the server behaviour for backwards-compatible 'client
   6777      hello' messages.  (Note that the attack is impractical against
   6778      SSL 3.0 and TLS 1.0 anyway because length and version checking
   6779      means that the probability of guessing a valid ciphertext is
   6780      around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
   6781      paper.)
   6782 
   6783      Before 0.9.5, the countermeasure (hide the error by generating a
   6784      random 'decryption result') did not work properly because
   6785      ERR_clear_error() was missing, meaning that SSL_get_error() would
   6786      detect the supposedly ignored error.
   6787 
   6788      Both problems are now fixed.
   6789      [Bodo Moeller]
   6790 
   6791   *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
   6792      (previously it was 1024).
   6793      [Bodo Moeller]
   6794 
   6795   *) Fix for compatibility mode trust settings: ignore trust settings
   6796      unless some valid trust or reject settings are present.
   6797      [Steve Henson]
   6798 
   6799   *) Fix for blowfish EVP: its a variable length cipher.
   6800      [Steve Henson]
   6801 
   6802   *) Fix various bugs related to DSA S/MIME verification. Handle missing
   6803      parameters in DSA public key structures and return an error in the
   6804      DSA routines if parameters are absent.
   6805      [Steve Henson]
   6806 
   6807   *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
   6808      in the current directory if neither $RANDFILE nor $HOME was set.
   6809      RAND_file_name() in 0.9.6a returned NULL in this case.  This has
   6810      caused some confusion to Windows users who haven't defined $HOME.
   6811      Thus RAND_file_name() is changed again: e_os.h can define a
   6812      DEFAULT_HOME, which will be used if $HOME is not set.
   6813      For Windows, we use "C:"; on other platforms, we still require
   6814      environment variables.
   6815 
   6816   *) Move 'if (!initialized) RAND_poll()' into regions protected by
   6817      CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
   6818      having multiple threads call RAND_poll() concurrently.
   6819      [Bodo Moeller]
   6820 
   6821   *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
   6822      combination of a flag and a thread ID variable.
   6823      Otherwise while one thread is in ssleay_rand_bytes (which sets the
   6824      flag), *other* threads can enter ssleay_add_bytes without obeying
   6825      the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
   6826      that they do not hold after the first thread unsets add_do_not_lock).
   6827      [Bodo Moeller]
   6828 
   6829   *) Change bctest again: '-x' expressions are not available in all
   6830      versions of 'test'.
   6831      [Bodo Moeller]
   6832 
   6833  Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
   6834 
   6835   *) Fix a couple of memory leaks in PKCS7_dataDecode()
   6836      [Steve Henson, reported by Heyun Zheng <hzheng (a] atdsprint.com>]
   6837 
   6838   *) Change Configure and Makefiles to provide EXE_EXT, which will contain
   6839      the default extension for executables, if any.  Also, make the perl
   6840      scripts that use symlink() to test if it really exists and use "cp"
   6841      if it doesn't.  All this made OpenSSL compilable and installable in
   6842      CygWin.
   6843      [Richard Levitte]
   6844 
   6845   *) Fix for asn1_GetSequence() for indefinite length constructed data.
   6846      If SEQUENCE is length is indefinite just set c->slen to the total
   6847      amount of data available.
   6848      [Steve Henson, reported by shige (a] FreeBSD.org]
   6849      [This change does not apply to 0.9.7.]
   6850 
   6851   *) Change bctest to avoid here-documents inside command substitution
   6852      (workaround for FreeBSD /bin/sh bug).
   6853      For compatibility with Ultrix, avoid shell functions (introduced
   6854      in the bctest version that searches along $PATH).
   6855      [Bodo Moeller]
   6856 
   6857   *) Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
   6858      with des_encrypt() defined on some operating systems, like Solaris
   6859      and UnixWare.
   6860      [Richard Levitte]
   6861 
   6862   *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
   6863      On the Importance of Eliminating Errors in Cryptographic
   6864      Computations, J. Cryptology 14 (2001) 2, 101-119,
   6865      http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
   6866      [Ulf Moeller]
   6867   
   6868   *) MIPS assembler BIGNUM division bug fix. 
   6869      [Andy Polyakov]
   6870 
   6871   *) Disabled incorrect Alpha assembler code.
   6872      [Richard Levitte]
   6873 
   6874   *) Fix PKCS#7 decode routines so they correctly update the length
   6875      after reading an EOC for the EXPLICIT tag.
   6876      [Steve Henson]
   6877      [This change does not apply to 0.9.7.]
   6878 
   6879   *) Fix bug in PKCS#12 key generation routines. This was triggered
   6880      if a 3DES key was generated with a 0 initial byte. Include
   6881      PKCS12_BROKEN_KEYGEN compilation option to retain the old
   6882      (but broken) behaviour.
   6883      [Steve Henson]
   6884 
   6885   *) Enhance bctest to search for a working bc along $PATH and print
   6886      it when found.
   6887      [Tim Rice <tim (a] multitalents.net> via Richard Levitte]
   6888 
   6889   *) Fix memory leaks in err.c: free err_data string if necessary;
   6890      don't write to the wrong index in ERR_set_error_data.
   6891      [Bodo Moeller]
   6892 
   6893   *) Implement ssl23_peek (analogous to ssl23_read), which previously
   6894      did not exist.
   6895      [Bodo Moeller]
   6896 
   6897   *) Replace rdtsc with _emit statements for VC++ version 5.
   6898      [Jeremy Cooper <jeremy (a] baymoo.org>]
   6899 
   6900   *) Make it possible to reuse SSLv2 sessions.
   6901      [Richard Levitte]
   6902 
   6903   *) In copy_email() check for >= 0 as a return value for
   6904      X509_NAME_get_index_by_NID() since 0 is a valid index.
   6905      [Steve Henson reported by Massimiliano Pala <madwolf (a] opensca.org>]
   6906 
   6907   *) Avoid coredump with unsupported or invalid public keys by checking if
   6908      X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
   6909      PKCS7_verify() fails with non detached data.
   6910      [Steve Henson]
   6911 
   6912   *) Don't use getenv in library functions when run as setuid/setgid.
   6913      New function OPENSSL_issetugid().
   6914      [Ulf Moeller]
   6915 
   6916   *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
   6917      due to incorrect handling of multi-threading:
   6918 
   6919      1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
   6920 
   6921      2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
   6922 
   6923      3. Count how many times MemCheck_off() has been called so that
   6924         nested use can be treated correctly.  This also avoids 
   6925         inband-signalling in the previous code (which relied on the
   6926         assumption that thread ID 0 is impossible).
   6927      [Bodo Moeller]
   6928 
   6929   *) Add "-rand" option also to s_client and s_server.
   6930      [Lutz Jaenicke]
   6931 
   6932   *) Fix CPU detection on Irix 6.x.
   6933      [Kurt Hockenbury <khockenb (a] stevens-tech.edu> and
   6934       "Bruce W. Forsberg" <bruce.forsberg (a] baesystems.com>]
   6935 
   6936   *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
   6937      was empty.
   6938      [Steve Henson]
   6939      [This change does not apply to 0.9.7.]
   6940 
   6941   *) Use the cached encoding of an X509_NAME structure rather than
   6942      copying it. This is apparently the reason for the libsafe "errors"
   6943      but the code is actually correct.
   6944      [Steve Henson]
   6945 
   6946   *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
   6947      Bleichenbacher's DSA attack.
   6948      Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
   6949      to be set and top=0 forces the highest bit to be set; top=-1 is new
   6950      and leaves the highest bit random.
   6951      [Ulf Moeller, Bodo Moeller]
   6952 
   6953   *) In the NCONF_...-based implementations for CONF_... queries
   6954      (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
   6955      a temporary CONF structure with the data component set to NULL
   6956      (which gives segmentation faults in lh_retrieve).
   6957      Instead, use NULL for the CONF pointer in CONF_get_string and
   6958      CONF_get_number (which may use environment variables) and directly
   6959      return NULL from CONF_get_section.
   6960      [Bodo Moeller]
   6961 
   6962   *) Fix potential buffer overrun for EBCDIC.
   6963      [Ulf Moeller]
   6964 
   6965   *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
   6966      keyUsage if basicConstraints absent for a CA.
   6967      [Steve Henson]
   6968 
   6969   *) Make SMIME_write_PKCS7() write mail header values with a format that
   6970      is more generally accepted (no spaces before the semicolon), since
   6971      some programs can't parse those values properly otherwise.  Also make
   6972      sure BIO's that break lines after each write do not create invalid
   6973      headers.
   6974      [Richard Levitte]
   6975 
   6976   *) Make the CRL encoding routines work with empty SEQUENCE OF. The
   6977      macros previously used would not encode an empty SEQUENCE OF
   6978      and break the signature.
   6979      [Steve Henson]
   6980      [This change does not apply to 0.9.7.]
   6981 
   6982   *) Zero the premaster secret after deriving the master secret in
   6983      DH ciphersuites.
   6984      [Steve Henson]
   6985 
   6986   *) Add some EVP_add_digest_alias registrations (as found in
   6987      OpenSSL_add_all_digests()) to SSL_library_init()
   6988      aka OpenSSL_add_ssl_algorithms().  This provides improved
   6989      compatibility with peers using X.509 certificates
   6990      with unconventional AlgorithmIdentifier OIDs.
   6991      [Bodo Moeller]
   6992 
   6993   *) Fix for Irix with NO_ASM.
   6994      ["Bruce W. Forsberg" <bruce.forsberg (a] baesystems.com>]
   6995 
   6996   *) ./config script fixes.
   6997      [Ulf Moeller, Richard Levitte]
   6998 
   6999   *) Fix 'openssl passwd -1'.
   7000      [Bodo Moeller]
   7001 
   7002   *) Change PKCS12_key_gen_asc() so it can cope with non null
   7003      terminated strings whose length is passed in the passlen
   7004      parameter, for example from PEM callbacks. This was done
   7005      by adding an extra length parameter to asc2uni().
   7006      [Steve Henson, reported by <oddissey (a] samsung.co.kr>]
   7007 
   7008   *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
   7009      call failed, free the DSA structure.
   7010      [Bodo Moeller]
   7011 
   7012   *) Fix to uni2asc() to cope with zero length Unicode strings.
   7013      These are present in some PKCS#12 files.
   7014      [Steve Henson]
   7015 
   7016   *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
   7017      Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
   7018      when writing a 32767 byte record.
   7019      [Bodo Moeller; problem reported by Eric Day <eday (a] concentric.net>]
   7020 
   7021   *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
   7022      obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
   7023 
   7024      (RSA objects have a reference count access to which is protected
   7025      by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
   7026      so they are meant to be shared between threads.)
   7027      [Bodo Moeller, Geoff Thorpe; original patch submitted by
   7028      "Reddie, Steven" <Steven.Reddie (a] ca.com>]
   7029 
   7030   *) Fix a deadlock in CRYPTO_mem_leaks().
   7031      [Bodo Moeller]
   7032 
   7033   *) Use better test patterns in bntest.
   7034      [Ulf Mller]
   7035 
   7036   *) rand_win.c fix for Borland C.
   7037      [Ulf Mller]
   7038  
   7039   *) BN_rshift bugfix for n == 0.
   7040      [Bodo Moeller]
   7041 
   7042   *) Add a 'bctest' script that checks for some known 'bc' bugs
   7043      so that 'make test' does not abort just because 'bc' is broken.
   7044      [Bodo Moeller]
   7045 
   7046   *) Store verify_result within SSL_SESSION also for client side to
   7047      avoid potential security hole. (Re-used sessions on the client side
   7048      always resulted in verify_result==X509_V_OK, not using the original
   7049      result of the server certificate verification.)
   7050      [Lutz Jaenicke]
   7051 
   7052   *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
   7053      SSL3_RT_APPLICATION_DATA, return 0.
   7054      Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
   7055      [Bodo Moeller]
   7056 
   7057   *) Fix SSL_peek:
   7058      Both ssl2_peek and ssl3_peek, which were totally broken in earlier
   7059      releases, have been re-implemented by renaming the previous
   7060      implementations of ssl2_read and ssl3_read to ssl2_read_internal
   7061      and ssl3_read_internal, respectively, and adding 'peek' parameters
   7062      to them.  The new ssl[23]_{read,peek} functions are calls to
   7063      ssl[23]_read_internal with the 'peek' flag set appropriately.
   7064      A 'peek' parameter has also been added to ssl3_read_bytes, which
   7065      does the actual work for ssl3_read_internal.
   7066      [Bodo Moeller]
   7067 
   7068   *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
   7069      the method-specific "init()" handler. Also clean up ex_data after
   7070      calling the method-specific "finish()" handler. Previously, this was
   7071      happening the other way round.
   7072      [Geoff Thorpe]
   7073 
   7074   *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
   7075      The previous value, 12, was not always sufficient for BN_mod_exp().
   7076      [Bodo Moeller]
   7077 
   7078   *) Make sure that shared libraries get the internal name engine with
   7079      the full version number and not just 0.  This should mark the
   7080      shared libraries as not backward compatible.  Of course, this should
   7081      be changed again when we can guarantee backward binary compatibility.
   7082      [Richard Levitte]
   7083 
   7084   *) Fix typo in get_cert_by_subject() in by_dir.c
   7085      [Jean-Marc Desperrier <jean-marc.desperrier (a] certplus.com>]
   7086 
   7087   *) Rework the system to generate shared libraries:
   7088 
   7089      - Make note of the expected extension for the shared libraries and
   7090        if there is a need for symbolic links from for example libcrypto.so.0
   7091        to libcrypto.so.0.9.7.  There is extended info in Configure for
   7092        that.
   7093 
   7094      - Make as few rebuilds of the shared libraries as possible.
   7095 
   7096      - Still avoid linking the OpenSSL programs with the shared libraries.
   7097 
   7098      - When installing, install the shared libraries separately from the
   7099        static ones.
   7100      [Richard Levitte]
   7101 
   7102   *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
   7103 
   7104      Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
   7105      and not in SSL_clear because the latter is also used by the
   7106      accept/connect functions; previously, the settings made by
   7107      SSL_set_read_ahead would be lost during the handshake.
   7108      [Bodo Moeller; problems reported by Anders Gertz <gertz (a] epact.se>]     
   7109 
   7110   *) Correct util/mkdef.pl to be selective about disabled algorithms.
   7111      Previously, it would create entries for disableed algorithms no
   7112      matter what.
   7113      [Richard Levitte]
   7114 
   7115   *) Added several new manual pages for SSL_* function.
   7116      [Lutz Jaenicke]
   7117 
   7118  Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
   7119 
   7120   *) In ssl23_get_client_hello, generate an error message when faced
   7121      with an initial SSL 3.0/TLS record that is too small to contain the
   7122      first two bytes of the ClientHello message, i.e. client_version.
   7123      (Note that this is a pathologic case that probably has never happened
   7124      in real life.)  The previous approach was to use the version number
   7125      from the record header as a substitute; but our protocol choice
   7126      should not depend on that one because it is not authenticated
   7127      by the Finished messages.
   7128      [Bodo Moeller]
   7129 
   7130   *) More robust randomness gathering functions for Windows.
   7131      [Jeffrey Altman <jaltman (a] columbia.edu>]
   7132 
   7133   *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
   7134      not set then we don't setup the error code for issuer check errors
   7135      to avoid possibly overwriting other errors which the callback does
   7136      handle. If an application does set the flag then we assume it knows
   7137      what it is doing and can handle the new informational codes
   7138      appropriately.
   7139      [Steve Henson]
   7140 
   7141   *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
   7142      a general "ANY" type, as such it should be able to decode anything
   7143      including tagged types. However it didn't check the class so it would
   7144      wrongly interpret tagged types in the same way as their universal
   7145      counterpart and unknown types were just rejected. Changed so that the
   7146      tagged and unknown types are handled in the same way as a SEQUENCE:
   7147      that is the encoding is stored intact. There is also a new type
   7148      "V_ASN1_OTHER" which is used when the class is not universal, in this
   7149      case we have no idea what the actual type is so we just lump them all
   7150      together.
   7151      [Steve Henson]
   7152 
   7153   *) On VMS, stdout may very well lead to a file that is written to
   7154      in a record-oriented fashion.  That means that every write() will
   7155      write a separate record, which will be read separately by the
   7156      programs trying to read from it.  This can be very confusing.
   7157 
   7158      The solution is to put a BIO filter in the way that will buffer
   7159      text until a linefeed is reached, and then write everything a
   7160      line at a time, so every record written will be an actual line,
   7161      not chunks of lines and not (usually doesn't happen, but I've
   7162      seen it once) several lines in one record.  BIO_f_linebuffer() is
   7163      the answer.
   7164 
   7165      Currently, it's a VMS-only method, because that's where it has
   7166      been tested well enough.
   7167      [Richard Levitte]
   7168 
   7169   *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
   7170      it can return incorrect results.
   7171      (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
   7172      but it was in 0.9.6-beta[12].)
   7173      [Bodo Moeller]
   7174 
   7175   *) Disable the check for content being present when verifying detached
   7176      signatures in pk7_smime.c. Some versions of Netscape (wrongly)
   7177      include zero length content when signing messages.
   7178      [Steve Henson]
   7179 
   7180   *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
   7181      BIO_ctrl (for BIO pairs).
   7182      [Bodo Mller]
   7183 
   7184   *) Add DSO method for VMS.
   7185      [Richard Levitte]
   7186 
   7187   *) Bug fix: Montgomery multiplication could produce results with the
   7188      wrong sign.
   7189      [Ulf Mller]
   7190 
   7191   *) Add RPM specification openssl.spec and modify it to build three
   7192      packages.  The default package contains applications, application
   7193      documentation and run-time libraries.  The devel package contains
   7194      include files, static libraries and function documentation.  The
   7195      doc package contains the contents of the doc directory.  The original
   7196      openssl.spec was provided by Damien Miller <djm (a] mindrot.org>.
   7197      [Richard Levitte]
   7198      
   7199   *) Add a large number of documentation files for many SSL routines.
   7200      [Lutz Jaenicke <Lutz.Jaenicke (a] aet.TU-Cottbus.DE>]
   7201 
   7202   *) Add a configuration entry for Sony News 4.
   7203      [NAKAJI Hiroyuki <nakaji (a] tutrp.tut.ac.jp>]
   7204 
   7205   *) Don't set the two most significant bits to one when generating a
   7206      random number < q in the DSA library.
   7207      [Ulf Mller]
   7208 
   7209   *) New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
   7210      behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
   7211      the underlying transport is blocking) if a handshake took place.
   7212      (The default behaviour is needed by applications such as s_client
   7213      and s_server that use select() to determine when to use SSL_read;
   7214      but for applications that know in advance when to expect data, it
   7215      just makes things more complicated.)
   7216      [Bodo Moeller]
   7217 
   7218   *) Add RAND_egd_bytes(), which gives control over the number of bytes read
   7219      from EGD.
   7220      [Ben Laurie]
   7221 
   7222   *) Add a few more EBCDIC conditionals that make `req' and `x509'
   7223      work better on such systems.
   7224      [Martin Kraemer <Martin.Kraemer (a] MchP.Siemens.De>]
   7225 
   7226   *) Add two demo programs for PKCS12_parse() and PKCS12_create().
   7227      Update PKCS12_parse() so it copies the friendlyName and the
   7228      keyid to the certificates aux info.
   7229      [Steve Henson]
   7230 
   7231   *) Fix bug in PKCS7_verify() which caused an infinite loop
   7232      if there was more than one signature.
   7233      [Sven Uszpelkat <su (a] celocom.de>]
   7234 
   7235   *) Major change in util/mkdef.pl to include extra information
   7236      about each symbol, as well as presentig variables as well
   7237      as functions.  This change means that there's n more need
   7238      to rebuild the .num files when some algorithms are excluded.
   7239      [Richard Levitte]
   7240 
   7241   *) Allow the verify time to be set by an application,
   7242      rather than always using the current time.
   7243      [Steve Henson]
   7244   
   7245   *) Phase 2 verify code reorganisation. The certificate
   7246      verify code now looks up an issuer certificate by a
   7247      number of criteria: subject name, authority key id
   7248      and key usage. It also verifies self signed certificates
   7249      by the same criteria. The main comparison function is
   7250      X509_check_issued() which performs these checks.
   7251  
   7252      Lot of changes were necessary in order to support this
   7253      without completely rewriting the lookup code.
   7254  
   7255      Authority and subject key identifier are now cached.
   7256  
   7257      The LHASH 'certs' is X509_STORE has now been replaced
   7258      by a STACK_OF(X509_OBJECT). This is mainly because an
   7259      LHASH can't store or retrieve multiple objects with
   7260      the same hash value.
   7261 
   7262      As a result various functions (which were all internal
   7263      use only) have changed to handle the new X509_STORE
   7264      structure. This will break anything that messed round
   7265      with X509_STORE internally.
   7266  
   7267      The functions X509_STORE_add_cert() now checks for an
   7268      exact match, rather than just subject name.
   7269  
   7270      The X509_STORE API doesn't directly support the retrieval
   7271      of multiple certificates matching a given criteria, however
   7272      this can be worked round by performing a lookup first
   7273      (which will fill the cache with candidate certificates)
   7274      and then examining the cache for matches. This is probably
   7275      the best we can do without throwing out X509_LOOKUP
   7276      entirely (maybe later...).
   7277  
   7278      The X509_VERIFY_CTX structure has been enhanced considerably.
   7279  
   7280      All certificate lookup operations now go via a get_issuer()
   7281      callback. Although this currently uses an X509_STORE it
   7282      can be replaced by custom lookups. This is a simple way
   7283      to bypass the X509_STORE hackery necessary to make this
   7284      work and makes it possible to use more efficient techniques
   7285      in future. A very simple version which uses a simple
   7286      STACK for its trusted certificate store is also provided
   7287      using X509_STORE_CTX_trusted_stack().
   7288  
   7289      The verify_cb() and verify() callbacks now have equivalents
   7290      in the X509_STORE_CTX structure.
   7291  
   7292      X509_STORE_CTX also has a 'flags' field which can be used
   7293      to customise the verify behaviour.
   7294      [Steve Henson]
   7295  
   7296   *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 
   7297      excludes S/MIME capabilities.
   7298      [Steve Henson]
   7299 
   7300   *) When a certificate request is read in keep a copy of the
   7301      original encoding of the signed data and use it when outputing
   7302      again. Signatures then use the original encoding rather than
   7303      a decoded, encoded version which may cause problems if the
   7304      request is improperly encoded.
   7305      [Steve Henson]
   7306 
   7307   *) For consistency with other BIO_puts implementations, call
   7308      buffer_write(b, ...) directly in buffer_puts instead of calling
   7309      BIO_write(b, ...).
   7310 
   7311      In BIO_puts, increment b->num_write as in BIO_write.
   7312      [Peter.Sylvester (a] EdelWeb.fr]
   7313 
   7314   *) Fix BN_mul_word for the case where the word is 0. (We have to use
   7315      BN_zero, we may not return a BIGNUM with an array consisting of
   7316      words set to zero.)
   7317      [Bodo Moeller]
   7318 
   7319   *) Avoid calling abort() from within the library when problems are
   7320      detected, except if preprocessor symbols have been defined
   7321      (such as REF_CHECK, BN_DEBUG etc.).
   7322      [Bodo Moeller]
   7323 
   7324   *) New openssl application 'rsautl'. This utility can be
   7325      used for low level RSA operations. DER public key
   7326      BIO/fp routines also added.
   7327      [Steve Henson]
   7328 
   7329   *) New Configure entry and patches for compiling on QNX 4.
   7330      [Andreas Schneider <andreas (a] ds3.etech.fh-hamburg.de>]
   7331 
   7332   *) A demo state-machine implementation was sponsored by
   7333      Nuron (http://www.nuron.com/) and is now available in
   7334      demos/state_machine.
   7335      [Ben Laurie]
   7336 
   7337   *) New options added to the 'dgst' utility for signature
   7338      generation and verification.
   7339      [Steve Henson]
   7340 
   7341   *) Unrecognized PKCS#7 content types are now handled via a
   7342      catch all ASN1_TYPE structure. This allows unsupported
   7343      types to be stored as a "blob" and an application can
   7344      encode and decode it manually.
   7345      [Steve Henson]
   7346 
   7347   *) Fix various signed/unsigned issues to make a_strex.c
   7348      compile under VC++.
   7349      [Oscar Jacobsson <oscar.jacobsson (a] celocom.com>]
   7350 
   7351   *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
   7352      length if passed a buffer. ASN1_INTEGER_to_BN failed
   7353      if passed a NULL BN and its argument was negative.
   7354      [Steve Henson, pointed out by Sven Heiberg <sven (a] tartu.cyber.ee>]
   7355 
   7356   *) Modification to PKCS#7 encoding routines to output definite
   7357      length encoding. Since currently the whole structures are in
   7358      memory there's not real point in using indefinite length 
   7359      constructed encoding. However if OpenSSL is compiled with
   7360      the flag PKCS7_INDEFINITE_ENCODING the old form is used.
   7361      [Steve Henson]
   7362 
   7363   *) Added BIO_vprintf() and BIO_vsnprintf().
   7364      [Richard Levitte]
   7365 
   7366   *) Added more prefixes to parse for in the the strings written
   7367      through a logging bio, to cover all the levels that are available
   7368      through syslog.  The prefixes are now:
   7369 
   7370 	PANIC, EMERG, EMR	=>	LOG_EMERG
   7371 	ALERT, ALR		=>	LOG_ALERT
   7372 	CRIT, CRI		=>	LOG_CRIT
   7373 	ERROR, ERR		=>	LOG_ERR
   7374 	WARNING, WARN, WAR	=>	LOG_WARNING
   7375 	NOTICE, NOTE, NOT	=>	LOG_NOTICE
   7376 	INFO, INF		=>	LOG_INFO
   7377 	DEBUG, DBG		=>	LOG_DEBUG
   7378 
   7379      and as before, if none of those prefixes are present at the
   7380      beginning of the string, LOG_ERR is chosen.
   7381 
   7382      On Win32, the LOG_* levels are mapped according to this:
   7383 
   7384 	LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR	=> EVENTLOG_ERROR_TYPE
   7385 	LOG_WARNING				=> EVENTLOG_WARNING_TYPE
   7386 	LOG_NOTICE, LOG_INFO, LOG_DEBUG		=> EVENTLOG_INFORMATION_TYPE
   7387 
   7388      [Richard Levitte]
   7389 
   7390   *) Made it possible to reconfigure with just the configuration
   7391      argument "reconf" or "reconfigure".  The command line arguments
   7392      are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
   7393      and are retrieved from there when reconfiguring.
   7394      [Richard Levitte]
   7395 
   7396   *) MD4 implemented.
   7397      [Assar Westerlund <assar (a] sics.se>, Richard Levitte]
   7398 
   7399   *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
   7400      [Richard Levitte]
   7401 
   7402   *) The obj_dat.pl script was messing up the sorting of object
   7403      names. The reason was that it compared the quoted version
   7404      of strings as a result "OCSP" > "OCSP Signing" because
   7405      " > SPACE. Changed script to store unquoted versions of
   7406      names and add quotes on output. It was also omitting some
   7407      names from the lookup table if they were given a default
   7408      value (that is if SN is missing it is given the same
   7409      value as LN and vice versa), these are now added on the
   7410      grounds that if an object has a name we should be able to
   7411      look it up. Finally added warning output when duplicate
   7412      short or long names are found.
   7413      [Steve Henson]
   7414 
   7415   *) Changes needed for Tandem NSK.
   7416      [Scott Uroff <scott (a] xypro.com>]
   7417 
   7418   *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
   7419      RSA_padding_check_SSLv23(), special padding was never detected
   7420      and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
   7421      version rollback attacks was not effective.
   7422 
   7423      In s23_clnt.c, don't use special rollback-attack detection padding
   7424      (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
   7425      client; similarly, in s23_srvr.c, don't do the rollback check if
   7426      SSL 2.0 is the only protocol enabled in the server.
   7427      [Bodo Moeller]
   7428 
   7429   *) Make it possible to get hexdumps of unprintable data with 'openssl
   7430      asn1parse'.  By implication, the functions ASN1_parse_dump() and
   7431      BIO_dump_indent() are added.
   7432      [Richard Levitte]
   7433 
   7434   *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
   7435      these print out strings and name structures based on various
   7436      flags including RFC2253 support and proper handling of
   7437      multibyte characters. Added options to the 'x509' utility 
   7438      to allow the various flags to be set.
   7439      [Steve Henson]
   7440 
   7441   *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
   7442      Also change the functions X509_cmp_current_time() and
   7443      X509_gmtime_adj() work with an ASN1_TIME structure,
   7444      this will enable certificates using GeneralizedTime in validity
   7445      dates to be checked.
   7446      [Steve Henson]
   7447 
   7448   *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
   7449      negative public key encodings) on by default,
   7450      NO_NEG_PUBKEY_BUG can be set to disable it.
   7451      [Steve Henson]
   7452 
   7453   *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
   7454      content octets. An i2c_ASN1_OBJECT is unnecessary because
   7455      the encoding can be trivially obtained from the structure.
   7456      [Steve Henson]
   7457 
   7458   *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
   7459      not read locks (CRYPTO_r_[un]lock).
   7460      [Bodo Moeller]
   7461 
   7462   *) A first attempt at creating official support for shared
   7463      libraries through configuration.  I've kept it so the
   7464      default is static libraries only, and the OpenSSL programs
   7465      are always statically linked for now, but there are
   7466      preparations for dynamic linking in place.
   7467      This has been tested on Linux and Tru64.
   7468      [Richard Levitte]
   7469 
   7470   *) Randomness polling function for Win9x, as described in:
   7471      Peter Gutmann, Software Generation of Practically Strong
   7472      Random Numbers.
   7473      [Ulf Mller]
   7474 
   7475   *) Fix so PRNG is seeded in req if using an already existing
   7476      DSA key.
   7477      [Steve Henson]
   7478 
   7479   *) New options to smime application. -inform and -outform
   7480      allow alternative formats for the S/MIME message including
   7481      PEM and DER. The -content option allows the content to be
   7482      specified separately. This should allow things like Netscape
   7483      form signing output easier to verify.
   7484      [Steve Henson]
   7485 
   7486   *) Fix the ASN1 encoding of tags using the 'long form'.
   7487      [Steve Henson]
   7488 
   7489   *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
   7490      STRING types. These convert content octets to and from the
   7491      underlying type. The actual tag and length octets are
   7492      already assumed to have been read in and checked. These
   7493      are needed because all other string types have virtually
   7494      identical handling apart from the tag. By having versions
   7495      of the ASN1 functions that just operate on content octets
   7496      IMPLICIT tagging can be handled properly. It also allows
   7497      the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
   7498      and ASN1_INTEGER are identical apart from the tag.
   7499      [Steve Henson]
   7500 
   7501   *) Change the handling of OID objects as follows:
   7502 
   7503      - New object identifiers are inserted in objects.txt, following
   7504        the syntax given in objects.README.
   7505      - objects.pl is used to process obj_mac.num and create a new
   7506        obj_mac.h.
   7507      - obj_dat.pl is used to create a new obj_dat.h, using the data in
   7508        obj_mac.h.
   7509 
   7510      This is currently kind of a hack, and the perl code in objects.pl
   7511      isn't very elegant, but it works as I intended.  The simplest way
   7512      to check that it worked correctly is to look in obj_dat.h and
   7513      check the array nid_objs and make sure the objects haven't moved
   7514      around (this is important!).  Additions are OK, as well as
   7515      consistent name changes. 
   7516      [Richard Levitte]
   7517 
   7518   *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
   7519      [Bodo Moeller]
   7520 
   7521   *) Addition of the command line parameter '-rand file' to 'openssl req'.
   7522      The given file adds to whatever has already been seeded into the
   7523      random pool through the RANDFILE configuration file option or
   7524      environment variable, or the default random state file.
   7525      [Richard Levitte]
   7526 
   7527   *) mkstack.pl now sorts each macro group into lexical order.
   7528      Previously the output order depended on the order the files
   7529      appeared in the directory, resulting in needless rewriting
   7530      of safestack.h .
   7531      [Steve Henson]
   7532 
   7533   *) Patches to make OpenSSL compile under Win32 again. Mostly
   7534      work arounds for the VC++ problem that it treats func() as
   7535      func(void). Also stripped out the parts of mkdef.pl that
   7536      added extra typesafe functions: these no longer exist.
   7537      [Steve Henson]
   7538 
   7539   *) Reorganisation of the stack code. The macros are now all 
   7540      collected in safestack.h . Each macro is defined in terms of
   7541      a "stack macro" of the form SKM_<name>(type, a, b). The 
   7542      DEBUG_SAFESTACK is now handled in terms of function casts,
   7543      this has the advantage of retaining type safety without the
   7544      use of additional functions. If DEBUG_SAFESTACK is not defined
   7545      then the non typesafe macros are used instead. Also modified the
   7546      mkstack.pl script to handle the new form. Needs testing to see
   7547      if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
   7548      the default if no major problems. Similar behaviour for ASN1_SET_OF
   7549      and PKCS12_STACK_OF.
   7550      [Steve Henson]
   7551 
   7552   *) When some versions of IIS use the 'NET' form of private key the
   7553      key derivation algorithm is different. Normally MD5(password) is
   7554      used as a 128 bit RC4 key. In the modified case
   7555      MD5(MD5(password) + "SGCKEYSALT")  is used insted. Added some
   7556      new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
   7557      as the old Netscape_RSA functions except they have an additional
   7558      'sgckey' parameter which uses the modified algorithm. Also added
   7559      an -sgckey command line option to the rsa utility. Thanks to 
   7560      Adrian Peck <bertie (a] ncipher.com> for posting details of the modified
   7561      algorithm to openssl-dev.
   7562      [Steve Henson]
   7563 
   7564   *) The evp_local.h macros were using 'c.##kname' which resulted in
   7565      invalid expansion on some systems (SCO 5.0.5 for example).
   7566      Corrected to 'c.kname'.
   7567      [Phillip Porch <root (a] theporch.com>]
   7568 
   7569   *) New X509_get1_email() and X509_REQ_get1_email() functions that return
   7570      a STACK of email addresses from a certificate or request, these look
   7571      in the subject name and the subject alternative name extensions and 
   7572      omit any duplicate addresses.
   7573      [Steve Henson]
   7574 
   7575   *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
   7576      This makes DSA verification about 2 % faster.
   7577      [Bodo Moeller]
   7578 
   7579   *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
   7580      (meaning that now 2^5 values will be precomputed, which is only 4 KB
   7581      plus overhead for 1024 bit moduli).
   7582      This makes exponentiations about 0.5 % faster for 1024 bit
   7583      exponents (as measured by "openssl speed rsa2048").
   7584      [Bodo Moeller]
   7585 
   7586   *) Rename memory handling macros to avoid conflicts with other
   7587      software:
   7588           Malloc         =>  OPENSSL_malloc
   7589           Malloc_locked  =>  OPENSSL_malloc_locked
   7590           Realloc        =>  OPENSSL_realloc
   7591           Free           =>  OPENSSL_free
   7592      [Richard Levitte]
   7593 
   7594   *) New function BN_mod_exp_mont_word for small bases (roughly 15%
   7595      faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
   7596      [Bodo Moeller]
   7597 
   7598   *) CygWin32 support.
   7599      [John Jarvie <jjarvie (a] newsguy.com>]
   7600 
   7601   *) The type-safe stack code has been rejigged. It is now only compiled
   7602      in when OpenSSL is configured with the DEBUG_SAFESTACK option and
   7603      by default all type-specific stack functions are "#define"d back to
   7604      standard stack functions. This results in more streamlined output
   7605      but retains the type-safety checking possibilities of the original
   7606      approach.
   7607      [Geoff Thorpe]
   7608 
   7609   *) The STACK code has been cleaned up, and certain type declarations
   7610      that didn't make a lot of sense have been brought in line. This has
   7611      also involved a cleanup of sorts in safestack.h to more correctly
   7612      map type-safe stack functions onto their plain stack counterparts.
   7613      This work has also resulted in a variety of "const"ifications of
   7614      lots of the code, especially "_cmp" operations which should normally
   7615      be prototyped with "const" parameters anyway.
   7616      [Geoff Thorpe]
   7617 
   7618   *) When generating bytes for the first time in md_rand.c, 'stir the pool'
   7619      by seeding with STATE_SIZE dummy bytes (with zero entropy count).
   7620      (The PRNG state consists of two parts, the large pool 'state' and 'md',
   7621      where all of 'md' is used each time the PRNG is used, but 'state'
   7622      is used only indexed by a cyclic counter. As entropy may not be
   7623      well distributed from the beginning, 'md' is important as a
   7624      chaining variable. However, the output function chains only half
   7625      of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
   7626      all of 'md', and seeding with STATE_SIZE dummy bytes will result
   7627      in all of 'state' being rewritten, with the new values depending
   7628      on virtually all of 'md'.  This overcomes the 80 bit limitation.)
   7629      [Bodo Moeller]
   7630 
   7631   *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
   7632      the handshake is continued after ssl_verify_cert_chain();
   7633      otherwise, if SSL_VERIFY_NONE is set, remaining error codes
   7634      can lead to 'unexplainable' connection aborts later.
   7635      [Bodo Moeller; problem tracked down by Lutz Jaenicke]
   7636 
   7637   *) Major EVP API cipher revision.
   7638      Add hooks for extra EVP features. This allows various cipher
   7639      parameters to be set in the EVP interface. Support added for variable
   7640      key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
   7641      setting of RC2 and RC5 parameters.
   7642 
   7643      Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
   7644      ciphers.
   7645 
   7646      Remove lots of duplicated code from the EVP library. For example *every*
   7647      cipher init() function handles the 'iv' in the same way according to the
   7648      cipher mode. They also all do nothing if the 'key' parameter is NULL and
   7649      for CFB and OFB modes they zero ctx->num.
   7650 
   7651      New functionality allows removal of S/MIME code RC2 hack.
   7652 
   7653      Most of the routines have the same form and so can be declared in terms
   7654      of macros.
   7655 
   7656      By shifting this to the top level EVP_CipherInit() it can be removed from
   7657      all individual ciphers. If the cipher wants to handle IVs or keys
   7658      differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
   7659      flags.
   7660 
   7661      Change lots of functions like EVP_EncryptUpdate() to now return a
   7662      value: although software versions of the algorithms cannot fail
   7663      any installed hardware versions can.
   7664      [Steve Henson]
   7665 
   7666   *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
   7667      this option is set, tolerate broken clients that send the negotiated
   7668      protocol version number instead of the requested protocol version
   7669      number.
   7670      [Bodo Moeller]
   7671 
   7672   *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
   7673      i.e. non-zero for export ciphersuites, zero otherwise.
   7674      Previous versions had this flag inverted, inconsistent with
   7675      rsa_tmp_cb (..._TMP_RSA_CB).
   7676      [Bodo Moeller; problem reported by Amit Chopra]
   7677 
   7678   *) Add missing DSA library text string. Work around for some IIS
   7679      key files with invalid SEQUENCE encoding.
   7680      [Steve Henson]
   7681 
   7682   *) Add a document (doc/standards.txt) that list all kinds of standards
   7683      and so on that are implemented in OpenSSL.
   7684      [Richard Levitte]
   7685 
   7686   *) Enhance c_rehash script. Old version would mishandle certificates
   7687      with the same subject name hash and wouldn't handle CRLs at all.
   7688      Added -fingerprint option to crl utility, to support new c_rehash
   7689      features.
   7690      [Steve Henson]
   7691 
   7692   *) Eliminate non-ANSI declarations in crypto.h and stack.h.
   7693      [Ulf Mller]
   7694 
   7695   *) Fix for SSL server purpose checking. Server checking was
   7696      rejecting certificates which had extended key usage present
   7697      but no ssl client purpose.
   7698      [Steve Henson, reported by Rene Grosser <grosser (a] hisolutions.com>]
   7699 
   7700   *) Make PKCS#12 code work with no password. The PKCS#12 spec
   7701      is a little unclear about how a blank password is handled.
   7702      Since the password in encoded as a BMPString with terminating
   7703      double NULL a zero length password would end up as just the
   7704      double NULL. However no password at all is different and is
   7705      handled differently in the PKCS#12 key generation code. NS
   7706      treats a blank password as zero length. MSIE treats it as no
   7707      password on export: but it will try both on import. We now do
   7708      the same: PKCS12_parse() tries zero length and no password if
   7709      the password is set to "" or NULL (NULL is now a valid password:
   7710      it wasn't before) as does the pkcs12 application.
   7711      [Steve Henson]
   7712 
   7713   *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
   7714      perror when PEM_read_bio_X509_REQ fails, the error message must
   7715      be obtained from the error queue.
   7716      [Bodo Moeller]
   7717 
   7718   *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
   7719      it in ERR_remove_state if appropriate, and change ERR_get_state
   7720      accordingly to avoid race conditions (this is necessary because
   7721      thread_hash is no longer constant once set).
   7722      [Bodo Moeller]
   7723 
   7724   *) Bugfix for linux-elf makefile.one.
   7725      [Ulf Mller]
   7726 
   7727   *) RSA_get_default_method() will now cause a default
   7728      RSA_METHOD to be chosen if one doesn't exist already.
   7729      Previously this was only set during a call to RSA_new()
   7730      or RSA_new_method(NULL) meaning it was possible for
   7731      RSA_get_default_method() to return NULL.
   7732      [Geoff Thorpe]
   7733 
   7734   *) Added native name translation to the existing DSO code
   7735      that will convert (if the flag to do so is set) filenames
   7736      that are sufficiently small and have no path information
   7737      into a canonical native form. Eg. "blah" converted to
   7738      "libblah.so" or "blah.dll" etc.
   7739      [Geoff Thorpe]
   7740 
   7741   *) New function ERR_error_string_n(e, buf, len) which is like
   7742      ERR_error_string(e, buf), but writes at most 'len' bytes
   7743      including the 0 terminator.  For ERR_error_string_n, 'buf'
   7744      may not be NULL.
   7745      [Damien Miller <djm (a] mindrot.org>, Bodo Moeller]
   7746 
   7747   *) CONF library reworked to become more general.  A new CONF
   7748      configuration file reader "class" is implemented as well as a
   7749      new functions (NCONF_*, for "New CONF") to handle it.  The now
   7750      old CONF_* functions are still there, but are reimplemented to
   7751      work in terms of the new functions.  Also, a set of functions
   7752      to handle the internal storage of the configuration data is
   7753      provided to make it easier to write new configuration file
   7754      reader "classes" (I can definitely see something reading a
   7755      configuration file in XML format, for example), called _CONF_*,
   7756      or "the configuration storage API"...
   7757 
   7758      The new configuration file reading functions are:
   7759 
   7760         NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
   7761         NCONF_get_section, NCONF_get_string, NCONF_get_numbre
   7762 
   7763         NCONF_default, NCONF_WIN32
   7764 
   7765         NCONF_dump_fp, NCONF_dump_bio
   7766 
   7767      NCONF_default and NCONF_WIN32 are method (or "class") choosers,
   7768      NCONF_new creates a new CONF object.  This works in the same way
   7769      as other interfaces in OpenSSL, like the BIO interface.
   7770      NCONF_dump_* dump the internal storage of the configuration file,
   7771      which is useful for debugging.  All other functions take the same
   7772      arguments as the old CONF_* functions wth the exception of the
   7773      first that must be a `CONF *' instead of a `LHASH *'.
   7774 
   7775      To make it easer to use the new classes with the old CONF_* functions,
   7776      the function CONF_set_default_method is provided.
   7777      [Richard Levitte]
   7778 
   7779   *) Add '-tls1' option to 'openssl ciphers', which was already
   7780      mentioned in the documentation but had not been implemented.
   7781      (This option is not yet really useful because even the additional
   7782      experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
   7783      [Bodo Moeller]
   7784 
   7785   *) Initial DSO code added into libcrypto for letting OpenSSL (and
   7786      OpenSSL-based applications) load shared libraries and bind to
   7787      them in a portable way.
   7788      [Geoff Thorpe, with contributions from Richard Levitte]
   7789 
   7790  Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
   7791 
   7792   *) Make sure _lrotl and _lrotr are only used with MSVC.
   7793 
   7794   *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
   7795      (the default implementation of RAND_status).
   7796 
   7797   *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
   7798      to '-clrext' (= clear extensions), as intended and documented.
   7799      [Bodo Moeller; inconsistency pointed out by Michael Attili
   7800      <attili (a] amaxo.com>]
   7801 
   7802   *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
   7803      was larger than the MD block size.      
   7804      [Steve Henson, pointed out by Yost William <YostW (a] tce.com>]
   7805 
   7806   *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
   7807      fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
   7808      using the passed key: if the passed key was a private key the result
   7809      of X509_print(), for example, would be to print out all the private key
   7810      components.
   7811      [Steve Henson]
   7812 
   7813   *) des_quad_cksum() byte order bug fix.
   7814      [Ulf Mller, using the problem description in krb4-0.9.7, where
   7815       the solution is attributed to Derrick J Brashear <shadow (a] DEMENTIA.ORG>]
   7816 
   7817   *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
   7818      discouraged.
   7819      [Steve Henson, pointed out by Brian Korver <briank (a] cs.stanford.edu>]
   7820 
   7821   *) For easily testing in shell scripts whether some command
   7822      'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
   7823      returns with exit code 0 iff no command of the given name is available.
   7824      'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
   7825      the output goes to stdout and nothing is printed to stderr.
   7826      Additional arguments are always ignored.
   7827 
   7828      Since for each cipher there is a command of the same name,
   7829      the 'no-cipher' compilation switches can be tested this way.
   7830 
   7831      ('openssl no-XXX' is not able to detect pseudo-commands such
   7832      as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
   7833      [Bodo Moeller]
   7834 
   7835   *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
   7836      [Bodo Moeller]
   7837 
   7838   *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
   7839      is set; it will be thrown away anyway because each handshake creates
   7840      its own key.
   7841      ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
   7842      to parameters -- in previous versions (since OpenSSL 0.9.3) the
   7843      'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
   7844      you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
   7845      [Bodo Moeller]
   7846 
   7847   *) New s_client option -ign_eof: EOF at stdin is ignored, and
   7848      'Q' and 'R' lose their special meanings (quit/renegotiate).
   7849      This is part of what -quiet does; unlike -quiet, -ign_eof
   7850      does not suppress any output.
   7851      [Richard Levitte]
   7852 
   7853   *) Add compatibility options to the purpose and trust code. The
   7854      purpose X509_PURPOSE_ANY is "any purpose" which automatically
   7855      accepts a certificate or CA, this was the previous behaviour,
   7856      with all the associated security issues.
   7857 
   7858      X509_TRUST_COMPAT is the old trust behaviour: only and
   7859      automatically trust self signed roots in certificate store. A
   7860      new trust setting X509_TRUST_DEFAULT is used to specify that
   7861      a purpose has no associated trust setting and it should instead
   7862      use the value in the default purpose.
   7863      [Steve Henson]
   7864 
   7865   *) Fix the PKCS#8 DSA private key code so it decodes keys again
   7866      and fix a memory leak.
   7867      [Steve Henson]
   7868 
   7869   *) In util/mkerr.pl (which implements 'make errors'), preserve
   7870      reason strings from the previous version of the .c file, as
   7871      the default to have only downcase letters (and digits) in
   7872      automatically generated reasons codes is not always appropriate.
   7873      [Bodo Moeller]
   7874 
   7875   *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
   7876      using strerror.  Previously, ERR_reason_error_string() returned
   7877      library names as reason strings for SYSerr; but SYSerr is a special
   7878      case where small numbers are errno values, not library numbers.
   7879      [Bodo Moeller]
   7880 
   7881   *) Add '-dsaparam' option to 'openssl dhparam' application.  This
   7882      converts DSA parameters into DH parameters. (When creating parameters,
   7883      DSA_generate_parameters is used.)
   7884      [Bodo Moeller]
   7885 
   7886   *) Include 'length' (recommended exponent length) in C code generated
   7887      by 'openssl dhparam -C'.
   7888      [Bodo Moeller]
   7889 
   7890   *) The second argument to set_label in perlasm was already being used
   7891      so couldn't be used as a "file scope" flag. Moved to third argument
   7892      which was free.
   7893      [Steve Henson]
   7894 
   7895   *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
   7896      instead of RAND_bytes for encryption IVs and salts.
   7897      [Bodo Moeller]
   7898 
   7899   *) Include RAND_status() into RAND_METHOD instead of implementing
   7900      it only for md_rand.c  Otherwise replacing the PRNG by calling
   7901      RAND_set_rand_method would be impossible.
   7902      [Bodo Moeller]
   7903 
   7904   *) Don't let DSA_generate_key() enter an infinite loop if the random
   7905      number generation fails.
   7906      [Bodo Moeller]
   7907 
   7908   *) New 'rand' application for creating pseudo-random output.
   7909      [Bodo Moeller]
   7910 
   7911   *) Added configuration support for Linux/IA64
   7912      [Rolf Haberrecker <rolf (a] suse.de>]
   7913 
   7914   *) Assembler module support for Mingw32.
   7915      [Ulf Mller]
   7916 
   7917   *) Shared library support for HPUX (in shlib/).
   7918      [Lutz Jaenicke <Lutz.Jaenicke (a] aet.TU-Cottbus.DE> and Anonymous]
   7919 
   7920   *) Shared library support for Solaris gcc.
   7921      [Lutz Behnke <behnke (a] trustcenter.de>]
   7922 
   7923  Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
   7924 
   7925   *) PKCS7_encrypt() was adding text MIME headers twice because they
   7926      were added manually and by SMIME_crlf_copy().
   7927      [Steve Henson]
   7928 
   7929   *) In bntest.c don't call BN_rand with zero bits argument.
   7930      [Steve Henson, pointed out by Andrew W. Gray <agray (a] iconsinc.com>]
   7931 
   7932   *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
   7933      case was implemented. This caused BN_div_recp() to fail occasionally.
   7934      [Ulf Mller]
   7935 
   7936   *) Add an optional second argument to the set_label() in the perl
   7937      assembly language builder. If this argument exists and is set
   7938      to 1 it signals that the assembler should use a symbol whose 
   7939      scope is the entire file, not just the current function. This
   7940      is needed with MASM which uses the format label:: for this scope.
   7941      [Steve Henson, pointed out by Peter Runestig <peter (a] runestig.com>]
   7942 
   7943   *) Change the ASN1 types so they are typedefs by default. Before
   7944      almost all types were #define'd to ASN1_STRING which was causing
   7945      STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
   7946      for example.
   7947      [Steve Henson]
   7948 
   7949   *) Change names of new functions to the new get1/get0 naming
   7950      convention: After 'get1', the caller owns a reference count
   7951      and has to call ..._free; 'get0' returns a pointer to some
   7952      data structure without incrementing reference counters.
   7953      (Some of the existing 'get' functions increment a reference
   7954      counter, some don't.)
   7955      Similarly, 'set1' and 'add1' functions increase reference
   7956      counters or duplicate objects.
   7957      [Steve Henson]
   7958 
   7959   *) Allow for the possibility of temp RSA key generation failure:
   7960      the code used to assume it always worked and crashed on failure.
   7961      [Steve Henson]
   7962 
   7963   *) Fix potential buffer overrun problem in BIO_printf().
   7964      [Ulf Mller, using public domain code by Patrick Powell; problem
   7965       pointed out by David Sacerdote <das33 (a] cornell.edu>]
   7966 
   7967   *) Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
   7968      RAND_egd() and RAND_status().  In the command line application,
   7969      the EGD socket can be specified like a seed file using RANDFILE
   7970      or -rand.
   7971      [Ulf Mller]
   7972 
   7973   *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
   7974      Some CAs (e.g. Verisign) distribute certificates in this form.
   7975      [Steve Henson]
   7976 
   7977   *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
   7978      list to exclude them. This means that no special compilation option
   7979      is needed to use anonymous DH: it just needs to be included in the
   7980      cipher list.
   7981      [Steve Henson]
   7982 
   7983   *) Change the EVP_MD_CTX_type macro so its meaning consistent with
   7984      EVP_MD_type. The old functionality is available in a new macro called
   7985      EVP_MD_md(). Change code that uses it and update docs.
   7986      [Steve Henson]
   7987 
   7988   *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
   7989      where the 'void *' argument is replaced by a function pointer argument.
   7990      Previously 'void *' was abused to point to functions, which works on
   7991      many platforms, but is not correct.  As these functions are usually
   7992      called by macros defined in OpenSSL header files, most source code
   7993      should work without changes.
   7994      [Richard Levitte]
   7995 
   7996   *) <openssl/opensslconf.h> (which is created by Configure) now contains
   7997      sections with information on -D... compiler switches used for
   7998      compiling the library so that applications can see them.  To enable
   7999      one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
   8000      must be defined.  E.g.,
   8001         #define OPENSSL_ALGORITHM_DEFINES
   8002         #include <openssl/opensslconf.h>
   8003      defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
   8004      [Richard Levitte, Ulf and Bodo Mller]
   8005 
   8006   *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
   8007      record layer.
   8008      [Bodo Moeller]
   8009 
   8010   *) Change the 'other' type in certificate aux info to a STACK_OF
   8011      X509_ALGOR. Although not an AlgorithmIdentifier as such it has
   8012      the required ASN1 format: arbitrary types determined by an OID.
   8013      [Steve Henson]
   8014 
   8015   *) Add some PEM_write_X509_REQ_NEW() functions and a command line
   8016      argument to 'req'. This is not because the function is newer or
   8017      better than others it just uses the work 'NEW' in the certificate
   8018      request header lines. Some software needs this.
   8019      [Steve Henson]
   8020 
   8021   *) Reorganise password command line arguments: now passwords can be
   8022      obtained from various sources. Delete the PEM_cb function and make
   8023      it the default behaviour: i.e. if the callback is NULL and the
   8024      usrdata argument is not NULL interpret it as a null terminated pass
   8025      phrase. If usrdata and the callback are NULL then the pass phrase
   8026      is prompted for as usual.
   8027      [Steve Henson]
   8028 
   8029   *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
   8030      the support is automatically enabled. The resulting binaries will
   8031      autodetect the card and use it if present.
   8032      [Ben Laurie and Compaq Inc.]
   8033 
   8034   *) Work around for Netscape hang bug. This sends certificate request
   8035      and server done in one record. Since this is perfectly legal in the
   8036      SSL/TLS protocol it isn't a "bug" option and is on by default. See
   8037      the bugs/SSLv3 entry for more info.
   8038      [Steve Henson]
   8039 
   8040   *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
   8041      [Andy Polyakov]
   8042 
   8043   *) Add -rand argument to smime and pkcs12 applications and read/write
   8044      of seed file.
   8045      [Steve Henson]
   8046 
   8047   *) New 'passwd' tool for crypt(3) and apr1 password hashes.
   8048      [Bodo Moeller]
   8049 
   8050   *) Add command line password options to the remaining applications.
   8051      [Steve Henson]
   8052 
   8053   *) Bug fix for BN_div_recp() for numerators with an even number of
   8054      bits.
   8055      [Ulf Mller]
   8056 
   8057   *) More tests in bntest.c, and changed test_bn output.
   8058      [Ulf Mller]
   8059 
   8060   *) ./config recognizes MacOS X now.
   8061      [Andy Polyakov]
   8062 
   8063   *) Bug fix for BN_div() when the first words of num and divsor are
   8064      equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
   8065      [Ulf Mller]
   8066 
   8067   *) Add support for various broken PKCS#8 formats, and command line
   8068      options to produce them.
   8069      [Steve Henson]
   8070 
   8071   *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
   8072      get temporary BIGNUMs from a BN_CTX.
   8073      [Ulf Mller]
   8074 
   8075   *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
   8076      for p == 0.
   8077      [Ulf Mller]
   8078 
   8079   *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
   8080      include a #define from the old name to the new. The original intent
   8081      was that statically linked binaries could for example just call
   8082      SSLeay_add_all_ciphers() to just add ciphers to the table and not
   8083      link with digests. This never worked becayse SSLeay_add_all_digests()
   8084      and SSLeay_add_all_ciphers() were in the same source file so calling
   8085      one would link with the other. They are now in separate source files.
   8086      [Steve Henson]
   8087 
   8088   *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
   8089      [Steve Henson]
   8090 
   8091   *) Use a less unusual form of the Miller-Rabin primality test (it used
   8092      a binary algorithm for exponentiation integrated into the Miller-Rabin
   8093      loop, our standard modexp algorithms are faster).
   8094      [Bodo Moeller]
   8095 
   8096   *) Support for the EBCDIC character set completed.
   8097      [Martin Kraemer <Martin.Kraemer (a] Mch.SNI.De>]
   8098 
   8099   *) Source code cleanups: use const where appropriate, eliminate casts,
   8100      use void * instead of char * in lhash.
   8101      [Ulf Mller] 
   8102 
   8103   *) Bugfix: ssl3_send_server_key_exchange was not restartable
   8104      (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
   8105      this the server could overwrite ephemeral keys that the client
   8106      has already seen).
   8107      [Bodo Moeller]
   8108 
   8109   *) Turn DSA_is_prime into a macro that calls BN_is_prime,
   8110      using 50 iterations of the Rabin-Miller test.
   8111 
   8112      DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
   8113      iterations of the Rabin-Miller test as required by the appendix
   8114      to FIPS PUB 186[-1]) instead of DSA_is_prime.
   8115      As BN_is_prime_fasttest includes trial division, DSA parameter
   8116      generation becomes much faster.
   8117 
   8118      This implies a change for the callback functions in DSA_is_prime
   8119      and DSA_generate_parameters: The callback function is called once
   8120      for each positive witness in the Rabin-Miller test, not just
   8121      occasionally in the inner loop; and the parameters to the
   8122      callback function now provide an iteration count for the outer
   8123      loop rather than for the current invocation of the inner loop.
   8124      DSA_generate_parameters additionally can call the callback
   8125      function with an 'iteration count' of -1, meaning that a
   8126      candidate has passed the trial division test (when q is generated 
   8127      from an application-provided seed, trial division is skipped).
   8128      [Bodo Moeller]
   8129 
   8130   *) New function BN_is_prime_fasttest that optionally does trial
   8131      division before starting the Rabin-Miller test and has
   8132      an additional BN_CTX * argument (whereas BN_is_prime always
   8133      has to allocate at least one BN_CTX).
   8134      'callback(1, -1, cb_arg)' is called when a number has passed the
   8135      trial division stage.
   8136      [Bodo Moeller]
   8137 
   8138   *) Fix for bug in CRL encoding. The validity dates weren't being handled
   8139      as ASN1_TIME.
   8140      [Steve Henson]
   8141 
   8142   *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
   8143      [Steve Henson]
   8144 
   8145   *) New function BN_pseudo_rand().
   8146      [Ulf Mller]
   8147 
   8148   *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
   8149      bignum version of BN_from_montgomery() with the working code from
   8150      SSLeay 0.9.0 (the word based version is faster anyway), and clean up
   8151      the comments.
   8152      [Ulf Mller]
   8153 
   8154   *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
   8155      made it impossible to use the same SSL_SESSION data structure in
   8156      SSL2 clients in multiple threads.
   8157      [Bodo Moeller]
   8158 
   8159   *) The return value of RAND_load_file() no longer counts bytes obtained
   8160      by stat().  RAND_load_file(..., -1) is new and uses the complete file
   8161      to seed the PRNG (previously an explicit byte count was required).
   8162      [Ulf Mller, Bodo Mller]
   8163 
   8164   *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
   8165      used (char *) instead of (void *) and had casts all over the place.
   8166      [Steve Henson]
   8167 
   8168   *) Make BN_generate_prime() return NULL on error if ret!=NULL.
   8169      [Ulf Mller]
   8170 
   8171   *) Retain source code compatibility for BN_prime_checks macro:
   8172      BN_is_prime(..., BN_prime_checks, ...) now uses
   8173      BN_prime_checks_for_size to determine the appropriate number of
   8174      Rabin-Miller iterations.
   8175      [Ulf Mller]
   8176 
   8177   *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
   8178      DH_CHECK_P_NOT_SAFE_PRIME.
   8179      (Check if this is true? OpenPGP calls them "strong".)
   8180      [Ulf Mller]
   8181 
   8182   *) Merge the functionality of "dh" and "gendh" programs into a new program
   8183      "dhparam". The old programs are retained for now but will handle DH keys
   8184      (instead of parameters) in future.
   8185      [Steve Henson]
   8186 
   8187   *) Make the ciphers, s_server and s_client programs check the return values
   8188      when a new cipher list is set.
   8189      [Steve Henson]
   8190 
   8191   *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
   8192      ciphers. Before when the 56bit ciphers were enabled the sorting was
   8193      wrong.
   8194 
   8195      The syntax for the cipher sorting has been extended to support sorting by
   8196      cipher-strength (using the strength_bits hard coded in the tables).
   8197      The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
   8198 
   8199      Fix a bug in the cipher-command parser: when supplying a cipher command
   8200      string with an "undefined" symbol (neither command nor alphanumeric
   8201      [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
   8202      an error is flagged.
   8203 
   8204      Due to the strength-sorting extension, the code of the
   8205      ssl_create_cipher_list() function was completely rearranged. I hope that
   8206      the readability was also increased :-)
   8207      [Lutz Jaenicke <Lutz.Jaenicke (a] aet.TU-Cottbus.DE>]
   8208 
   8209   *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
   8210      for the first serial number and places 2 in the serial number file. This
   8211      avoids problems when the root CA is created with serial number zero and
   8212      the first user certificate has the same issuer name and serial number
   8213      as the root CA.
   8214      [Steve Henson]
   8215 
   8216   *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
   8217      the new code. Add documentation for this stuff.
   8218      [Steve Henson]
   8219 
   8220   *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
   8221      X509_*() to X509at_*() on the grounds that they don't handle X509
   8222      structures and behave in an analagous way to the X509v3 functions:
   8223      they shouldn't be called directly but wrapper functions should be used
   8224      instead.
   8225 
   8226      So we also now have some wrapper functions that call the X509at functions
   8227      when passed certificate requests. (TO DO: similar things can be done with
   8228      PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
   8229      things. Some of these need some d2i or i2d and print functionality
   8230      because they handle more complex structures.)
   8231      [Steve Henson]
   8232 
   8233   *) Add missing #ifndefs that caused missing symbols when building libssl
   8234      as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
   8235      NO_RSA in ssl/s2*.c. 
   8236      [Kris Kennaway <kris (a] hub.freebsd.org>, modified by Ulf Mller]
   8237 
   8238   *) Precautions against using the PRNG uninitialized: RAND_bytes() now
   8239      has a return value which indicates the quality of the random data
   8240      (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
   8241      error queue. New function RAND_pseudo_bytes() generates output that is
   8242      guaranteed to be unique but not unpredictable. RAND_add is like
   8243      RAND_seed, but takes an extra argument for an entropy estimate
   8244      (RAND_seed always assumes full entropy).
   8245      [Ulf Mller]
   8246 
   8247   *) Do more iterations of Rabin-Miller probable prime test (specifically,
   8248      3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
   8249      instead of only 2 for all lengths; see BN_prime_checks_for_size definition
   8250      in crypto/bn/bn_prime.c for the complete table).  This guarantees a
   8251      false-positive rate of at most 2^-80 for random input.
   8252      [Bodo Moeller]
   8253 
   8254   *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
   8255      [Bodo Moeller]
   8256 
   8257   *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
   8258      in the 0.9.5 release), this returns the chain
   8259      from an X509_CTX structure with a dup of the stack and all
   8260      the X509 reference counts upped: so the stack will exist
   8261      after X509_CTX_cleanup() has been called. Modify pkcs12.c
   8262      to use this.
   8263 
   8264      Also make SSL_SESSION_print() print out the verify return
   8265      code.
   8266      [Steve Henson]
   8267 
   8268   *) Add manpage for the pkcs12 command. Also change the default
   8269      behaviour so MAC iteration counts are used unless the new
   8270      -nomaciter option is used. This improves file security and
   8271      only older versions of MSIE (4.0 for example) need it.
   8272      [Steve Henson]
   8273 
   8274   *) Honor the no-xxx Configure options when creating .DEF files.
   8275      [Ulf Mller]
   8276 
   8277   *) Add PKCS#10 attributes to field table: challengePassword, 
   8278      unstructuredName and unstructuredAddress. These are taken from
   8279      draft PKCS#9 v2.0 but are compatible with v1.2 provided no 
   8280      international characters are used.
   8281 
   8282      More changes to X509_ATTRIBUTE code: allow the setting of types
   8283      based on strings. Remove the 'loc' parameter when adding
   8284      attributes because these will be a SET OF encoding which is sorted
   8285      in ASN1 order.
   8286      [Steve Henson]
   8287 
   8288   *) Initial changes to the 'req' utility to allow request generation
   8289      automation. This will allow an application to just generate a template
   8290      file containing all the field values and have req construct the
   8291      request.
   8292 
   8293      Initial support for X509_ATTRIBUTE handling. Stacks of these are
   8294      used all over the place including certificate requests and PKCS#7
   8295      structures. They are currently handled manually where necessary with
   8296      some primitive wrappers for PKCS#7. The new functions behave in a
   8297      manner analogous to the X509 extension functions: they allow
   8298      attributes to be looked up by NID and added.
   8299 
   8300      Later something similar to the X509V3 code would be desirable to
   8301      automatically handle the encoding, decoding and printing of the
   8302      more complex types. The string types like challengePassword can
   8303      be handled by the string table functions.
   8304 
   8305      Also modified the multi byte string table handling. Now there is
   8306      a 'global mask' which masks out certain types. The table itself
   8307      can use the flag STABLE_NO_MASK to ignore the mask setting: this
   8308      is useful when for example there is only one permissible type
   8309      (as in countryName) and using the mask might result in no valid
   8310      types at all.
   8311      [Steve Henson]
   8312 
   8313   *) Clean up 'Finished' handling, and add functions SSL_get_finished and
   8314      SSL_get_peer_finished to allow applications to obtain the latest
   8315      Finished messages sent to the peer or expected from the peer,
   8316      respectively.  (SSL_get_peer_finished is usually the Finished message
   8317      actually received from the peer, otherwise the protocol will be aborted.)
   8318 
   8319      As the Finished message are message digests of the complete handshake
   8320      (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
   8321      be used for external authentication procedures when the authentication
   8322      provided by SSL/TLS is not desired or is not enough.
   8323      [Bodo Moeller]
   8324 
   8325   *) Enhanced support for Alpha Linux is added. Now ./config checks if
   8326      the host supports BWX extension and if Compaq C is present on the
   8327      $PATH. Just exploiting of the BWX extension results in 20-30%
   8328      performance kick for some algorithms, e.g. DES and RC4 to mention
   8329      a couple. Compaq C in turn generates ~20% faster code for MD5 and
   8330      SHA1.
   8331      [Andy Polyakov]
   8332 
   8333   *) Add support for MS "fast SGC". This is arguably a violation of the
   8334      SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
   8335      weak crypto and after checking the certificate is SGC a second one
   8336      with strong crypto. MS SGC stops the first handshake after receiving
   8337      the server certificate message and sends a second client hello. Since
   8338      a server will typically do all the time consuming operations before
   8339      expecting any further messages from the client (server key exchange
   8340      is the most expensive) there is little difference between the two.
   8341 
   8342      To get OpenSSL to support MS SGC we have to permit a second client
   8343      hello message after we have sent server done. In addition we have to
   8344      reset the MAC if we do get this second client hello.
   8345      [Steve Henson]
   8346 
   8347   *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
   8348      if a DER encoded private key is RSA or DSA traditional format. Changed
   8349      d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
   8350      format DER encoded private key. Newer code should use PKCS#8 format which
   8351      has the key type encoded in the ASN1 structure. Added DER private key
   8352      support to pkcs8 application.
   8353      [Steve Henson]
   8354 
   8355   *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
   8356      ciphersuites has been selected (as required by the SSL 3/TLS 1
   8357      specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
   8358      is set, we interpret this as a request to violate the specification
   8359      (the worst that can happen is a handshake failure, and 'correct'
   8360      behaviour would result in a handshake failure anyway).
   8361      [Bodo Moeller]
   8362 
   8363   *) In SSL_CTX_add_session, take into account that there might be multiple
   8364      SSL_SESSION structures with the same session ID (e.g. when two threads
   8365      concurrently obtain them from an external cache).
   8366      The internal cache can handle only one SSL_SESSION with a given ID,
   8367      so if there's a conflict, we now throw out the old one to achieve
   8368      consistency.
   8369      [Bodo Moeller]
   8370 
   8371   *) Add OIDs for idea and blowfish in CBC mode. This will allow both
   8372      to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
   8373      some routines that use cipher OIDs: some ciphers do not have OIDs
   8374      defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
   8375      example.
   8376      [Steve Henson]
   8377 
   8378   *) Simplify the trust setting structure and code. Now we just have
   8379      two sequences of OIDs for trusted and rejected settings. These will
   8380      typically have values the same as the extended key usage extension
   8381      and any application specific purposes.
   8382 
   8383      The trust checking code now has a default behaviour: it will just
   8384      check for an object with the same NID as the passed id. Functions can
   8385      be provided to override either the default behaviour or the behaviour
   8386      for a given id. SSL client, server and email already have functions
   8387      in place for compatibility: they check the NID and also return "trusted"
   8388      if the certificate is self signed.
   8389      [Steve Henson]
   8390 
   8391   *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
   8392      traditional format into an EVP_PKEY structure.
   8393      [Steve Henson]
   8394 
   8395   *) Add a password callback function PEM_cb() which either prompts for
   8396      a password if usr_data is NULL or otherwise assumes it is a null
   8397      terminated password. Allow passwords to be passed on command line
   8398      environment or config files in a few more utilities.
   8399      [Steve Henson]
   8400 
   8401   *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
   8402      keys. Add some short names for PKCS#8 PBE algorithms and allow them
   8403      to be specified on the command line for the pkcs8 and pkcs12 utilities.
   8404      Update documentation.
   8405      [Steve Henson]
   8406 
   8407   *) Support for ASN1 "NULL" type. This could be handled before by using
   8408      ASN1_TYPE but there wasn't any function that would try to read a NULL
   8409      and produce an error if it couldn't. For compatibility we also have
   8410      ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
   8411      don't allocate anything because they don't need to.
   8412      [Steve Henson]
   8413 
   8414   *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
   8415      for details.
   8416      [Andy Polyakov, Roy Woods <roy (a] centicsystems.ca>]
   8417 
   8418   *) Rebuild of the memory allocation routines used by OpenSSL code and
   8419      possibly others as well.  The purpose is to make an interface that
   8420      provide hooks so anyone can build a separate set of allocation and
   8421      deallocation routines to be used by OpenSSL, for example memory
   8422      pool implementations, or something else, which was previously hard
   8423      since Malloc(), Realloc() and Free() were defined as macros having
   8424      the values malloc, realloc and free, respectively (except for Win32
   8425      compilations).  The same is provided for memory debugging code.
   8426      OpenSSL already comes with functionality to find memory leaks, but
   8427      this gives people a chance to debug other memory problems.
   8428 
   8429      With these changes, a new set of functions and macros have appeared:
   8430 
   8431        CRYPTO_set_mem_debug_functions()	        [F]
   8432        CRYPTO_get_mem_debug_functions()         [F]
   8433        CRYPTO_dbg_set_options()	                [F]
   8434        CRYPTO_dbg_get_options()                 [F]
   8435        CRYPTO_malloc_debug_init()               [M]
   8436 
   8437      The memory debug functions are NULL by default, unless the library
   8438      is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
   8439      wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
   8440      gives the standard debugging functions that come with OpenSSL) or
   8441      CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
   8442      provided by the library user) must be used.  When the standard
   8443      debugging functions are used, CRYPTO_dbg_set_options can be used to
   8444      request additional information:
   8445      CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
   8446      the CRYPTO_MDEBUG_xxx macro when compiling the library.   
   8447 
   8448      Also, things like CRYPTO_set_mem_functions will always give the
   8449      expected result (the new set of functions is used for allocation
   8450      and deallocation) at all times, regardless of platform and compiler
   8451      options.
   8452 
   8453      To finish it up, some functions that were never use in any other
   8454      way than through macros have a new API and new semantic:
   8455 
   8456        CRYPTO_dbg_malloc()
   8457        CRYPTO_dbg_realloc()
   8458        CRYPTO_dbg_free()
   8459 
   8460      All macros of value have retained their old syntax.
   8461      [Richard Levitte and Bodo Moeller]
   8462 
   8463   *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
   8464      ordering of SMIMECapabilities wasn't in "strength order" and there
   8465      was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
   8466      algorithm.
   8467      [Steve Henson]
   8468 
   8469   *) Some ASN1 types with illegal zero length encoding (INTEGER,
   8470      ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
   8471      [Frans Heymans <fheymans (a] isaserver.be>, modified by Steve Henson]
   8472 
   8473   *) Merge in my S/MIME library for OpenSSL. This provides a simple
   8474      S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
   8475      functionality to handle multipart/signed properly) and a utility
   8476      called 'smime' to call all this stuff. This is based on code I
   8477      originally wrote for Celo who have kindly allowed it to be
   8478      included in OpenSSL.
   8479      [Steve Henson]
   8480 
   8481   *) Add variants des_set_key_checked and des_set_key_unchecked of
   8482      des_set_key (aka des_key_sched).  Global variable des_check_key
   8483      decides which of these is called by des_set_key; this way
   8484      des_check_key behaves as it always did, but applications and
   8485      the library itself, which was buggy for des_check_key == 1,
   8486      have a cleaner way to pick the version they need.
   8487      [Bodo Moeller]
   8488 
   8489   *) New function PKCS12_newpass() which changes the password of a
   8490      PKCS12 structure.
   8491      [Steve Henson]
   8492 
   8493   *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
   8494      dynamic mix. In both cases the ids can be used as an index into the
   8495      table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
   8496      functions so they accept a list of the field values and the
   8497      application doesn't need to directly manipulate the X509_TRUST
   8498      structure.
   8499      [Steve Henson]
   8500 
   8501   *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
   8502      need initialising.
   8503      [Steve Henson]
   8504 
   8505   *) Modify the way the V3 extension code looks up extensions. This now
   8506      works in a similar way to the object code: we have some "standard"
   8507      extensions in a static table which is searched with OBJ_bsearch()
   8508      and the application can add dynamic ones if needed. The file
   8509      crypto/x509v3/ext_dat.h now has the info: this file needs to be
   8510      updated whenever a new extension is added to the core code and kept
   8511      in ext_nid order. There is a simple program 'tabtest.c' which checks
   8512      this. New extensions are not added too often so this file can readily
   8513      be maintained manually.
   8514 
   8515      There are two big advantages in doing things this way. The extensions
   8516      can be looked up immediately and no longer need to be "added" using
   8517      X509V3_add_standard_extensions(): this function now does nothing.
   8518      [Side note: I get *lots* of email saying the extension code doesn't
   8519       work because people forget to call this function]
   8520      Also no dynamic allocation is done unless new extensions are added:
   8521      so if we don't add custom extensions there is no need to call
   8522      X509V3_EXT_cleanup().
   8523      [Steve Henson]
   8524 
   8525   *) Modify enc utility's salting as follows: make salting the default. Add a
   8526      magic header, so unsalted files fail gracefully instead of just decrypting
   8527      to garbage. This is because not salting is a big security hole, so people
   8528      should be discouraged from doing it.
   8529      [Ben Laurie]
   8530 
   8531   *) Fixes and enhancements to the 'x509' utility. It allowed a message
   8532      digest to be passed on the command line but it only used this
   8533      parameter when signing a certificate. Modified so all relevant
   8534      operations are affected by the digest parameter including the
   8535      -fingerprint and -x509toreq options. Also -x509toreq choked if a
   8536      DSA key was used because it didn't fix the digest.
   8537      [Steve Henson]
   8538 
   8539   *) Initial certificate chain verify code. Currently tests the untrusted
   8540      certificates for consistency with the verify purpose (which is set
   8541      when the X509_STORE_CTX structure is set up) and checks the pathlength.
   8542 
   8543      There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
   8544      this is because it will reject chains with invalid extensions whereas
   8545      every previous version of OpenSSL and SSLeay made no checks at all.
   8546 
   8547      Trust code: checks the root CA for the relevant trust settings. Trust
   8548      settings have an initial value consistent with the verify purpose: e.g.
   8549      if the verify purpose is for SSL client use it expects the CA to be
   8550      trusted for SSL client use. However the default value can be changed to
   8551      permit custom trust settings: one example of this would be to only trust
   8552      certificates from a specific "secure" set of CAs.
   8553 
   8554      Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
   8555      which should be used for version portability: especially since the
   8556      verify structure is likely to change more often now.
   8557 
   8558      SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
   8559      to set them. If not set then assume SSL clients will verify SSL servers
   8560      and vice versa.
   8561 
   8562      Two new options to the verify program: -untrusted allows a set of
   8563      untrusted certificates to be passed in and -purpose which sets the
   8564      intended purpose of the certificate. If a purpose is set then the
   8565      new chain verify code is used to check extension consistency.
   8566      [Steve Henson]
   8567 
   8568   *) Support for the authority information access extension.
   8569      [Steve Henson]
   8570 
   8571   *) Modify RSA and DSA PEM read routines to transparently handle
   8572      PKCS#8 format private keys. New *_PUBKEY_* functions that handle
   8573      public keys in a format compatible with certificate
   8574      SubjectPublicKeyInfo structures. Unfortunately there were already
   8575      functions called *_PublicKey_* which used various odd formats so
   8576      these are retained for compatibility: however the DSA variants were
   8577      never in a public release so they have been deleted. Changed dsa/rsa
   8578      utilities to handle the new format: note no releases ever handled public
   8579      keys so we should be OK.
   8580 
   8581      The primary motivation for this change is to avoid the same fiasco
   8582      that dogs private keys: there are several incompatible private key
   8583      formats some of which are standard and some OpenSSL specific and
   8584      require various evil hacks to allow partial transparent handling and
   8585      even then it doesn't work with DER formats. Given the option anything
   8586      other than PKCS#8 should be dumped: but the other formats have to
   8587      stay in the name of compatibility.
   8588 
   8589      With public keys and the benefit of hindsight one standard format 
   8590      is used which works with EVP_PKEY, RSA or DSA structures: though
   8591      it clearly returns an error if you try to read the wrong kind of key.
   8592 
   8593      Added a -pubkey option to the 'x509' utility to output the public key.
   8594      Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
   8595      (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
   8596      EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
   8597      that do the same as the EVP_PKEY_assign_*() except they up the
   8598      reference count of the added key (they don't "swallow" the
   8599      supplied key).
   8600      [Steve Henson]
   8601 
   8602   *) Fixes to crypto/x509/by_file.c the code to read in certificates and
   8603      CRLs would fail if the file contained no certificates or no CRLs:
   8604      added a new function to read in both types and return the number
   8605      read: this means that if none are read it will be an error. The
   8606      DER versions of the certificate and CRL reader would always fail
   8607      because it isn't possible to mix certificates and CRLs in DER format
   8608      without choking one or the other routine. Changed this to just read
   8609      a certificate: this is the best we can do. Also modified the code
   8610      in apps/verify.c to take notice of return codes: it was previously
   8611      attempting to read in certificates from NULL pointers and ignoring
   8612      any errors: this is one reason why the cert and CRL reader seemed
   8613      to work. It doesn't check return codes from the default certificate
   8614      routines: these may well fail if the certificates aren't installed.
   8615      [Steve Henson]
   8616 
   8617   *) Code to support otherName option in GeneralName.
   8618      [Steve Henson]
   8619 
   8620   *) First update to verify code. Change the verify utility
   8621      so it warns if it is passed a self signed certificate:
   8622      for consistency with the normal behaviour. X509_verify
   8623      has been modified to it will now verify a self signed
   8624      certificate if *exactly* the same certificate appears
   8625      in the store: it was previously impossible to trust a
   8626      single self signed certificate. This means that:
   8627      openssl verify ss.pem
   8628      now gives a warning about a self signed certificate but
   8629      openssl verify -CAfile ss.pem ss.pem
   8630      is OK.
   8631      [Steve Henson]
   8632 
   8633   *) For servers, store verify_result in SSL_SESSION data structure
   8634      (and add it to external session representation).
   8635      This is needed when client certificate verifications fails,
   8636      but an application-provided verification callback (set by
   8637      SSL_CTX_set_cert_verify_callback) allows accepting the session
   8638      anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
   8639      but returns 1): When the session is reused, we have to set
   8640      ssl->verify_result to the appropriate error code to avoid
   8641      security holes.
   8642      [Bodo Moeller, problem pointed out by Lutz Jaenicke]
   8643 
   8644   *) Fix a bug in the new PKCS#7 code: it didn't consider the
   8645      case in PKCS7_dataInit() where the signed PKCS7 structure
   8646      didn't contain any existing data because it was being created.
   8647      [Po-Cheng Chen <pocheng (a] nst.com.tw>, slightly modified by Steve Henson]
   8648 
   8649   *) Add a salt to the key derivation routines in enc.c. This
   8650      forms the first 8 bytes of the encrypted file. Also add a
   8651      -S option to allow a salt to be input on the command line.
   8652      [Steve Henson]
   8653 
   8654   *) New function X509_cmp(). Oddly enough there wasn't a function
   8655      to compare two certificates. We do this by working out the SHA1
   8656      hash and comparing that. X509_cmp() will be needed by the trust
   8657      code.
   8658      [Steve Henson]
   8659 
   8660   *) SSL_get1_session() is like SSL_get_session(), but increments
   8661      the reference count in the SSL_SESSION returned.
   8662      [Geoff Thorpe <geoff (a] eu.c2.net>]
   8663 
   8664   *) Fix for 'req': it was adding a null to request attributes.
   8665      Also change the X509_LOOKUP and X509_INFO code to handle
   8666      certificate auxiliary information.
   8667      [Steve Henson]
   8668 
   8669   *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
   8670      the 'enc' command.
   8671      [Steve Henson]
   8672 
   8673   *) Add the possibility to add extra information to the memory leak
   8674      detecting output, to form tracebacks, showing from where each
   8675      allocation was originated: CRYPTO_push_info("constant string") adds
   8676      the string plus current file name and line number to a per-thread
   8677      stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
   8678      is like calling CYRPTO_pop_info() until the stack is empty.
   8679      Also updated memory leak detection code to be multi-thread-safe.
   8680      [Richard Levitte]
   8681 
   8682   *) Add options -text and -noout to pkcs7 utility and delete the
   8683      encryption options which never did anything. Update docs.
   8684      [Steve Henson]
   8685 
   8686   *) Add options to some of the utilities to allow the pass phrase
   8687      to be included on either the command line (not recommended on
   8688      OSes like Unix) or read from the environment. Update the
   8689      manpages and fix a few bugs.
   8690      [Steve Henson]
   8691 
   8692   *) Add a few manpages for some of the openssl commands.
   8693      [Steve Henson]
   8694 
   8695   *) Fix the -revoke option in ca. It was freeing up memory twice,
   8696      leaking and not finding already revoked certificates.
   8697      [Steve Henson]
   8698 
   8699   *) Extensive changes to support certificate auxiliary information.
   8700      This involves the use of X509_CERT_AUX structure and X509_AUX
   8701      functions. An X509_AUX function such as PEM_read_X509_AUX()
   8702      can still read in a certificate file in the usual way but it
   8703      will also read in any additional "auxiliary information". By
   8704      doing things this way a fair degree of compatibility can be
   8705      retained: existing certificates can have this information added
   8706      using the new 'x509' options. 
   8707 
   8708      Current auxiliary information includes an "alias" and some trust
   8709      settings. The trust settings will ultimately be used in enhanced
   8710      certificate chain verification routines: currently a certificate
   8711      can only be trusted if it is self signed and then it is trusted
   8712      for all purposes.
   8713      [Steve Henson]
   8714 
   8715   *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
   8716      The problem was that one of the replacement routines had not been working
   8717      since SSLeay releases.  For now the offending routine has been replaced
   8718      with non-optimised assembler.  Even so, this now gives around 95%
   8719      performance improvement for 1024 bit RSA signs.
   8720      [Mark Cox]
   8721 
   8722   *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 
   8723      handling. Most clients have the effective key size in bits equal to
   8724      the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
   8725      A few however don't do this and instead use the size of the decrypted key
   8726      to determine the RC2 key length and the AlgorithmIdentifier to determine
   8727      the effective key length. In this case the effective key length can still
   8728      be 40 bits but the key length can be 168 bits for example. This is fixed
   8729      by manually forcing an RC2 key into the EVP_PKEY structure because the
   8730      EVP code can't currently handle unusual RC2 key sizes: it always assumes
   8731      the key length and effective key length are equal.
   8732      [Steve Henson]
   8733 
   8734   *) Add a bunch of functions that should simplify the creation of 
   8735      X509_NAME structures. Now you should be able to do:
   8736      X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
   8737      and have it automatically work out the correct field type and fill in
   8738      the structures. The more adventurous can try:
   8739      X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
   8740      and it will (hopefully) work out the correct multibyte encoding.
   8741      [Steve Henson]
   8742 
   8743   *) Change the 'req' utility to use the new field handling and multibyte
   8744      copy routines. Before the DN field creation was handled in an ad hoc
   8745      way in req, ca, and x509 which was rather broken and didn't support
   8746      BMPStrings or UTF8Strings. Since some software doesn't implement
   8747      BMPStrings or UTF8Strings yet, they can be enabled using the config file
   8748      using the dirstring_type option. See the new comment in the default
   8749      openssl.cnf for more info.
   8750      [Steve Henson]
   8751 
   8752   *) Make crypto/rand/md_rand.c more robust:
   8753      - Assure unique random numbers after fork().
   8754      - Make sure that concurrent threads access the global counter and
   8755        md serializably so that we never lose entropy in them
   8756        or use exactly the same state in multiple threads.
   8757        Access to the large state is not always serializable because
   8758        the additional locking could be a performance killer, and
   8759        md should be large enough anyway.
   8760      [Bodo Moeller]
   8761 
   8762   *) New file apps/app_rand.c with commonly needed functionality
   8763      for handling the random seed file.
   8764 
   8765      Use the random seed file in some applications that previously did not:
   8766           ca,
   8767           dsaparam -genkey (which also ignored its '-rand' option), 
   8768           s_client,
   8769           s_server,
   8770           x509 (when signing).
   8771      Except on systems with /dev/urandom, it is crucial to have a random
   8772      seed file at least for key creation, DSA signing, and for DH exchanges;
   8773      for RSA signatures we could do without one.
   8774 
   8775      gendh and gendsa (unlike genrsa) used to read only the first byte
   8776      of each file listed in the '-rand' option.  The function as previously
   8777      found in genrsa is now in app_rand.c and is used by all programs
   8778      that support '-rand'.
   8779      [Bodo Moeller]
   8780 
   8781   *) In RAND_write_file, use mode 0600 for creating files;
   8782      don't just chmod when it may be too late.
   8783      [Bodo Moeller]
   8784 
   8785   *) Report an error from X509_STORE_load_locations
   8786      when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
   8787      [Bill Perry]
   8788 
   8789   *) New function ASN1_mbstring_copy() this copies a string in either
   8790      ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
   8791      into an ASN1_STRING type. A mask of permissible types is passed
   8792      and it chooses the "minimal" type to use or an error if not type
   8793      is suitable.
   8794      [Steve Henson]
   8795 
   8796   *) Add function equivalents to the various macros in asn1.h. The old
   8797      macros are retained with an M_ prefix. Code inside the library can
   8798      use the M_ macros. External code (including the openssl utility)
   8799      should *NOT* in order to be "shared library friendly".
   8800      [Steve Henson]
   8801 
   8802   *) Add various functions that can check a certificate's extensions
   8803      to see if it usable for various purposes such as SSL client,
   8804      server or S/MIME and CAs of these types. This is currently 
   8805      VERY EXPERIMENTAL but will ultimately be used for certificate chain
   8806      verification. Also added a -purpose flag to x509 utility to
   8807      print out all the purposes.
   8808      [Steve Henson]
   8809 
   8810   *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
   8811      functions.
   8812      [Steve Henson]
   8813 
   8814   *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
   8815      for, obtain and decode and extension and obtain its critical flag.
   8816      This allows all the necessary extension code to be handled in a
   8817      single function call.
   8818      [Steve Henson]
   8819 
   8820   *) RC4 tune-up featuring 30-40% performance improvement on most RISC
   8821      platforms. See crypto/rc4/rc4_enc.c for further details.
   8822      [Andy Polyakov]
   8823 
   8824   *) New -noout option to asn1parse. This causes no output to be produced
   8825      its main use is when combined with -strparse and -out to extract data
   8826      from a file (which may not be in ASN.1 format).
   8827      [Steve Henson]
   8828 
   8829   *) Fix for pkcs12 program. It was hashing an inval