Home | History | Annotate | Line # | Download | only in dist
      1 
      2  OpenSSL CHANGES
      3  _______________
      4 
      5  Changes between 1.0.1s and 1.0.1t [3 May 2016]
      6 
      7   *) Prevent padding oracle in AES-NI CBC MAC check
      8 
      9      A MITM attacker can use a padding oracle attack to decrypt traffic
     10      when the connection uses an AES CBC cipher and the server support
     11      AES-NI.
     12 
     13      This issue was introduced as part of the fix for Lucky 13 padding
     14      attack (CVE-2013-0169). The padding check was rewritten to be in
     15      constant time by making sure that always the same bytes are read and
     16      compared against either the MAC or padding bytes. But it no longer
     17      checked that there was enough data to have both the MAC and padding
     18      bytes.
     19 
     20      This issue was reported by Juraj Somorovsky using TLS-Attacker.
     21      (CVE-2016-2107)
     22      [Kurt Roeckx]
     23 
     24   *) Fix EVP_EncodeUpdate overflow
     25 
     26      An overflow can occur in the EVP_EncodeUpdate() function which is used for
     27      Base64 encoding of binary data. If an attacker is able to supply very large
     28      amounts of input data then a length check can overflow resulting in a heap
     29      corruption.
     30 
     31      Internally to OpenSSL the EVP_EncodeUpdate() function is primarly used by
     32      the PEM_write_bio* family of functions. These are mainly used within the
     33      OpenSSL command line applications, so any application which processes data
     34      from an untrusted source and outputs it as a PEM file should be considered
     35      vulnerable to this issue. User applications that call these APIs directly
     36      with large amounts of untrusted data may also be vulnerable.
     37 
     38      This issue was reported by Guido Vranken.
     39      (CVE-2016-2105)
     40      [Matt Caswell]
     41 
     42   *) Fix EVP_EncryptUpdate overflow
     43 
     44      An overflow can occur in the EVP_EncryptUpdate() function. If an attacker
     45      is able to supply very large amounts of input data after a previous call to
     46      EVP_EncryptUpdate() with a partial block then a length check can overflow
     47      resulting in a heap corruption. Following an analysis of all OpenSSL
     48      internal usage of the EVP_EncryptUpdate() function all usage is one of two
     49      forms. The first form is where the EVP_EncryptUpdate() call is known to be
     50      the first called function after an EVP_EncryptInit(), and therefore that
     51      specific call must be safe. The second form is where the length passed to
     52      EVP_EncryptUpdate() can be seen from the code to be some small value and
     53      therefore there is no possibility of an overflow. Since all instances are
     54      one of these two forms, it is believed that there can be no overflows in
     55      internal code due to this problem. It should be noted that
     56      EVP_DecryptUpdate() can call EVP_EncryptUpdate() in certain code paths.
     57      Also EVP_CipherUpdate() is a synonym for EVP_EncryptUpdate(). All instances
     58      of these calls have also been analysed too and it is believed there are no
     59      instances in internal usage where an overflow could occur.
     60 
     61      This issue was reported by Guido Vranken.
     62      (CVE-2016-2106)
     63      [Matt Caswell]
     64 
     65   *) Prevent ASN.1 BIO excessive memory allocation
     66 
     67      When ASN.1 data is read from a BIO using functions such as d2i_CMS_bio()
     68      a short invalid encoding can casuse allocation of large amounts of memory
     69      potentially consuming excessive resources or exhausting memory.
     70 
     71      Any application parsing untrusted data through d2i BIO functions is
     72      affected. The memory based functions such as d2i_X509() are *not* affected.
     73      Since the memory based functions are used by the TLS library, TLS
     74      applications are not affected.
     75 
     76      This issue was reported by Brian Carpenter.
     77      (CVE-2016-2109)
     78      [Stephen Henson]
     79 
     80   *) EBCDIC overread
     81 
     82      ASN1 Strings that are over 1024 bytes can cause an overread in applications
     83      using the X509_NAME_oneline() function on EBCDIC systems. This could result
     84      in arbitrary stack data being returned in the buffer.
     85 
     86      This issue was reported by Guido Vranken.
     87      (CVE-2016-2176)
     88      [Matt Caswell]
     89 
     90   *) Modify behavior of ALPN to invoke callback after SNI/servername
     91      callback, such that updates to the SSL_CTX affect ALPN.
     92      [Todd Short]
     93 
     94   *) Remove LOW from the DEFAULT cipher list.  This removes singles DES from the
     95      default.
     96      [Kurt Roeckx]
     97 
     98   *) Only remove the SSLv2 methods with the no-ssl2-method option. When the
     99      methods are enabled and ssl2 is disabled the methods return NULL.
    100      [Kurt Roeckx]
    101 
    102  Changes between 1.0.1r and 1.0.1s [1 Mar 2016]
    103 
    104   * Disable weak ciphers in SSLv3 and up in default builds of OpenSSL.
    105     Builds that are not configured with "enable-weak-ssl-ciphers" will not
    106     provide any "EXPORT" or "LOW" strength ciphers.
    107     [Viktor Dukhovni]
    108 
    109   * Disable SSLv2 default build, default negotiation and weak ciphers.  SSLv2
    110     is by default disabled at build-time.  Builds that are not configured with
    111     "enable-ssl2" will not support SSLv2.  Even if "enable-ssl2" is used,
    112     users who want to negotiate SSLv2 via the version-flexible SSLv23_method()
    113     will need to explicitly call either of:
    114 
    115         SSL_CTX_clear_options(ctx, SSL_OP_NO_SSLv2);
    116     or
    117         SSL_clear_options(ssl, SSL_OP_NO_SSLv2);
    118 
    119     as appropriate.  Even if either of those is used, or the application
    120     explicitly uses the version-specific SSLv2_method() or its client and
    121     server variants, SSLv2 ciphers vulnerable to exhaustive search key
    122     recovery have been removed.  Specifically, the SSLv2 40-bit EXPORT
    123     ciphers, and SSLv2 56-bit DES are no longer available.
    124     (CVE-2016-0800)
    125     [Viktor Dukhovni]
    126 
    127   *) Fix a double-free in DSA code
    128 
    129      A double free bug was discovered when OpenSSL parses malformed DSA private
    130      keys and could lead to a DoS attack or memory corruption for applications
    131      that receive DSA private keys from untrusted sources.  This scenario is
    132      considered rare.
    133 
    134      This issue was reported to OpenSSL by Adam Langley(Google/BoringSSL) using
    135      libFuzzer.
    136      (CVE-2016-0705)
    137      [Stephen Henson]
    138 
    139   *) Disable SRP fake user seed to address a server memory leak.
    140 
    141      Add a new method SRP_VBASE_get1_by_user that handles the seed properly.
    142 
    143      SRP_VBASE_get_by_user had inconsistent memory management behaviour.
    144      In order to fix an unavoidable memory leak, SRP_VBASE_get_by_user
    145      was changed to ignore the "fake user" SRP seed, even if the seed
    146      is configured.
    147 
    148      Users should use SRP_VBASE_get1_by_user instead. Note that in
    149      SRP_VBASE_get1_by_user, caller must free the returned value. Note
    150      also that even though configuring the SRP seed attempts to hide
    151      invalid usernames by continuing the handshake with fake
    152      credentials, this behaviour is not constant time and no strong
    153      guarantees are made that the handshake is indistinguishable from
    154      that of a valid user.
    155      (CVE-2016-0798)
    156      [Emilia Ksper]
    157 
    158   *) Fix BN_hex2bn/BN_dec2bn NULL pointer deref/heap corruption
    159 
    160      In the BN_hex2bn function the number of hex digits is calculated using an
    161      int value |i|. Later |bn_expand| is called with a value of |i * 4|. For
    162      large values of |i| this can result in |bn_expand| not allocating any
    163      memory because |i * 4| is negative. This can leave the internal BIGNUM data
    164      field as NULL leading to a subsequent NULL ptr deref. For very large values
    165      of |i|, the calculation |i * 4| could be a positive value smaller than |i|.
    166      In this case memory is allocated to the internal BIGNUM data field, but it
    167      is insufficiently sized leading to heap corruption. A similar issue exists
    168      in BN_dec2bn. This could have security consequences if BN_hex2bn/BN_dec2bn
    169      is ever called by user applications with very large untrusted hex/dec data.
    170      This is anticipated to be a rare occurrence.
    171 
    172      All OpenSSL internal usage of these functions use data that is not expected
    173      to be untrusted, e.g. config file data or application command line
    174      arguments. If user developed applications generate config file data based
    175      on untrusted data then it is possible that this could also lead to security
    176      consequences. This is also anticipated to be rare.
    177 
    178      This issue was reported to OpenSSL by Guido Vranken.
    179      (CVE-2016-0797)
    180      [Matt Caswell]
    181 
    182   *) Fix memory issues in BIO_*printf functions
    183 
    184      The internal |fmtstr| function used in processing a "%s" format string in
    185      the BIO_*printf functions could overflow while calculating the length of a
    186      string and cause an OOB read when printing very long strings.
    187 
    188      Additionally the internal |doapr_outch| function can attempt to write to an
    189      OOB memory location (at an offset from the NULL pointer) in the event of a
    190      memory allocation failure. In 1.0.2 and below this could be caused where
    191      the size of a buffer to be allocated is greater than INT_MAX. E.g. this
    192      could be in processing a very long "%s" format string. Memory leaks can
    193      also occur.
    194 
    195      The first issue may mask the second issue dependent on compiler behaviour.
    196      These problems could enable attacks where large amounts of untrusted data
    197      is passed to the BIO_*printf functions. If applications use these functions
    198      in this way then they could be vulnerable. OpenSSL itself uses these
    199      functions when printing out human-readable dumps of ASN.1 data. Therefore
    200      applications that print this data could be vulnerable if the data is from
    201      untrusted sources. OpenSSL command line applications could also be
    202      vulnerable where they print out ASN.1 data, or if untrusted data is passed
    203      as command line arguments.
    204 
    205      Libssl is not considered directly vulnerable. Additionally certificates etc
    206      received via remote connections via libssl are also unlikely to be able to
    207      trigger these issues because of message size limits enforced within libssl.
    208 
    209      This issue was reported to OpenSSL Guido Vranken.
    210      (CVE-2016-0799)
    211      [Matt Caswell]
    212 
    213   *) Side channel attack on modular exponentiation
    214 
    215      A side-channel attack was found which makes use of cache-bank conflicts on
    216      the Intel Sandy-Bridge microarchitecture which could lead to the recovery
    217      of RSA keys.  The ability to exploit this issue is limited as it relies on
    218      an attacker who has control of code in a thread running on the same
    219      hyper-threaded core as the victim thread which is performing decryptions.
    220 
    221      This issue was reported to OpenSSL by Yuval Yarom, The University of
    222      Adelaide and NICTA, Daniel Genkin, Technion and Tel Aviv University, and
    223      Nadia Heninger, University of Pennsylvania with more information at
    224      http://cachebleed.info.
    225      (CVE-2016-0702)
    226      [Andy Polyakov]
    227 
    228   *) Change the req app to generate a 2048-bit RSA/DSA key by default,
    229      if no keysize is specified with default_bits. This fixes an
    230      omission in an earlier change that changed all RSA/DSA key generation
    231      apps to use 2048 bits by default.
    232      [Emilia Ksper]
    233 
    234  Changes between 1.0.1q and 1.0.1r [28 Jan 2016]
    235 
    236   *) Protection for DH small subgroup attacks
    237 
    238      As a precautionary measure the SSL_OP_SINGLE_DH_USE option has been
    239      switched on by default and cannot be disabled. This could have some
    240      performance impact.
    241      [Matt Caswell]
    242 
    243   *) SSLv2 doesn't block disabled ciphers
    244 
    245      A malicious client can negotiate SSLv2 ciphers that have been disabled on
    246      the server and complete SSLv2 handshakes even if all SSLv2 ciphers have
    247      been disabled, provided that the SSLv2 protocol was not also disabled via
    248      SSL_OP_NO_SSLv2.
    249 
    250      This issue was reported to OpenSSL on 26th December 2015 by Nimrod Aviram
    251      and Sebastian Schinzel.
    252      (CVE-2015-3197)
    253      [Viktor Dukhovni]
    254 
    255   *) Reject DH handshakes with parameters shorter than 1024 bits.
    256      [Kurt Roeckx]
    257 
    258  Changes between 1.0.1p and 1.0.1q [3 Dec 2015]
    259 
    260   *) Certificate verify crash with missing PSS parameter
    261 
    262      The signature verification routines will crash with a NULL pointer
    263      dereference if presented with an ASN.1 signature using the RSA PSS
    264      algorithm and absent mask generation function parameter. Since these
    265      routines are used to verify certificate signature algorithms this can be
    266      used to crash any certificate verification operation and exploited in a
    267      DoS attack. Any application which performs certificate verification is
    268      vulnerable including OpenSSL clients and servers which enable client
    269      authentication.
    270 
    271      This issue was reported to OpenSSL by Loc Jonas Etienne (Qnective AG).
    272      (CVE-2015-3194)
    273      [Stephen Henson]
    274 
    275   *) X509_ATTRIBUTE memory leak
    276 
    277      When presented with a malformed X509_ATTRIBUTE structure OpenSSL will leak
    278      memory. This structure is used by the PKCS#7 and CMS routines so any
    279      application which reads PKCS#7 or CMS data from untrusted sources is
    280      affected. SSL/TLS is not affected.
    281 
    282      This issue was reported to OpenSSL by Adam Langley (Google/BoringSSL) using
    283      libFuzzer.
    284      (CVE-2015-3195)
    285      [Stephen Henson]
    286 
    287   *) Rewrite EVP_DecodeUpdate (base64 decoding) to fix several bugs.
    288      This changes the decoding behaviour for some invalid messages,
    289      though the change is mostly in the more lenient direction, and
    290      legacy behaviour is preserved as much as possible.
    291      [Emilia Ksper]
    292 
    293   *) In DSA_generate_parameters_ex, if the provided seed is too short,
    294      use a random seed, as already documented.
    295      [Rich Salz and Ismo Puustinen <ismo.puustinen (a] intel.com>]
    296 
    297  Changes between 1.0.1o and 1.0.1p [9 Jul 2015]
    298 
    299   *) Alternate chains certificate forgery
    300 
    301      During certificate verfification, OpenSSL will attempt to find an
    302      alternative certificate chain if the first attempt to build such a chain
    303      fails. An error in the implementation of this logic can mean that an
    304      attacker could cause certain checks on untrusted certificates to be
    305      bypassed, such as the CA flag, enabling them to use a valid leaf
    306      certificate to act as a CA and "issue" an invalid certificate.
    307 
    308      This issue was reported to OpenSSL by Adam Langley/David Benjamin
    309      (Google/BoringSSL).
    310      (CVE-2015-1793)
    311      [Matt Caswell]
    312 
    313   *) Race condition handling PSK identify hint
    314 
    315      If PSK identity hints are received by a multi-threaded client then
    316      the values are wrongly updated in the parent SSL_CTX structure. This can
    317      result in a race condition potentially leading to a double free of the
    318      identify hint data.
    319      (CVE-2015-3196)
    320      [Stephen Henson]
    321 
    322  Changes between 1.0.1n and 1.0.1o [12 Jun 2015]
    323   *) Fix HMAC ABI incompatibility. The previous version introduced an ABI
    324      incompatibility in the handling of HMAC. The previous ABI has now been
    325      restored.
    326 
    327  Changes between 1.0.1m and 1.0.1n [11 Jun 2015]
    328 
    329   *) Malformed ECParameters causes infinite loop
    330 
    331      When processing an ECParameters structure OpenSSL enters an infinite loop
    332      if the curve specified is over a specially malformed binary polynomial
    333      field.
    334 
    335      This can be used to perform denial of service against any
    336      system which processes public keys, certificate requests or
    337      certificates.  This includes TLS clients and TLS servers with
    338      client authentication enabled.
    339 
    340      This issue was reported to OpenSSL by Joseph Barr-Pixton.
    341      (CVE-2015-1788)
    342      [Andy Polyakov]
    343 
    344   *) Exploitable out-of-bounds read in X509_cmp_time
    345 
    346      X509_cmp_time does not properly check the length of the ASN1_TIME
    347      string and can read a few bytes out of bounds. In addition,
    348      X509_cmp_time accepts an arbitrary number of fractional seconds in the
    349      time string.
    350 
    351      An attacker can use this to craft malformed certificates and CRLs of
    352      various sizes and potentially cause a segmentation fault, resulting in
    353      a DoS on applications that verify certificates or CRLs. TLS clients
    354      that verify CRLs are affected. TLS clients and servers with client
    355      authentication enabled may be affected if they use custom verification
    356      callbacks.
    357 
    358      This issue was reported to OpenSSL by Robert Swiecki (Google), and
    359      independently by Hanno Bck.
    360      (CVE-2015-1789)
    361      [Emilia Ksper]
    362 
    363   *) PKCS7 crash with missing EnvelopedContent
    364 
    365      The PKCS#7 parsing code does not handle missing inner EncryptedContent
    366      correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs
    367      with missing content and trigger a NULL pointer dereference on parsing.
    368 
    369      Applications that decrypt PKCS#7 data or otherwise parse PKCS#7
    370      structures from untrusted sources are affected. OpenSSL clients and
    371      servers are not affected.
    372 
    373      This issue was reported to OpenSSL by Michal Zalewski (Google).
    374      (CVE-2015-1790)
    375      [Emilia Ksper]
    376 
    377   *) CMS verify infinite loop with unknown hash function
    378 
    379      When verifying a signedData message the CMS code can enter an infinite loop
    380      if presented with an unknown hash function OID. This can be used to perform
    381      denial of service against any system which verifies signedData messages using
    382      the CMS code.
    383      This issue was reported to OpenSSL by Johannes Bauer.
    384      (CVE-2015-1792)
    385      [Stephen Henson]
    386 
    387   *) Race condition handling NewSessionTicket
    388 
    389      If a NewSessionTicket is received by a multi-threaded client when attempting to
    390      reuse a previous ticket then a race condition can occur potentially leading to
    391      a double free of the ticket data.
    392      (CVE-2015-1791)
    393      [Matt Caswell]
    394 
    395   *) Reject DH handshakes with parameters shorter than 768 bits.
    396      [Kurt Roeckx and Emilia Kasper]
    397 
    398   *) dhparam: generate 2048-bit parameters by default.
    399      [Kurt Roeckx and Emilia Kasper]
    400 
    401  Changes between 1.0.1l and 1.0.1m [19 Mar 2015]
    402 
    403   *) Segmentation fault in ASN1_TYPE_cmp fix
    404 
    405      The function ASN1_TYPE_cmp will crash with an invalid read if an attempt is
    406      made to compare ASN.1 boolean types. Since ASN1_TYPE_cmp is used to check
    407      certificate signature algorithm consistency this can be used to crash any
    408      certificate verification operation and exploited in a DoS attack. Any
    409      application which performs certificate verification is vulnerable including
    410      OpenSSL clients and servers which enable client authentication.
    411      (CVE-2015-0286)
    412      [Stephen Henson]
    413 
    414   *) ASN.1 structure reuse memory corruption fix
    415 
    416      Reusing a structure in ASN.1 parsing may allow an attacker to cause
    417      memory corruption via an invalid write. Such reuse is and has been
    418      strongly discouraged and is believed to be rare.
    419 
    420      Applications that parse structures containing CHOICE or ANY DEFINED BY
    421      components may be affected. Certificate parsing (d2i_X509 and related
    422      functions) are however not affected. OpenSSL clients and servers are
    423      not affected.
    424      (CVE-2015-0287)
    425      [Stephen Henson]
    426 
    427   *) PKCS7 NULL pointer dereferences fix
    428 
    429      The PKCS#7 parsing code does not handle missing outer ContentInfo
    430      correctly. An attacker can craft malformed ASN.1-encoded PKCS#7 blobs with
    431      missing content and trigger a NULL pointer dereference on parsing.
    432 
    433      Applications that verify PKCS#7 signatures, decrypt PKCS#7 data or
    434      otherwise parse PKCS#7 structures from untrusted sources are
    435      affected. OpenSSL clients and servers are not affected.
    436 
    437      This issue was reported to OpenSSL by Michal Zalewski (Google).
    438      (CVE-2015-0289)
    439      [Emilia Ksper]
    440 
    441   *) DoS via reachable assert in SSLv2 servers fix
    442 
    443      A malicious client can trigger an OPENSSL_assert (i.e., an abort) in
    444      servers that both support SSLv2 and enable export cipher suites by sending
    445      a specially crafted SSLv2 CLIENT-MASTER-KEY message.
    446 
    447      This issue was discovered by Sean Burford (Google) and Emilia Ksper
    448      (OpenSSL development team).
    449      (CVE-2015-0293)
    450      [Emilia Ksper]
    451 
    452   *) Use After Free following d2i_ECPrivatekey error fix
    453 
    454      A malformed EC private key file consumed via the d2i_ECPrivateKey function
    455      could cause a use after free condition. This, in turn, could cause a double
    456      free in several private key parsing functions (such as d2i_PrivateKey
    457      or EVP_PKCS82PKEY) and could lead to a DoS attack or memory corruption
    458      for applications that receive EC private keys from untrusted
    459      sources. This scenario is considered rare.
    460 
    461      This issue was discovered by the BoringSSL project and fixed in their
    462      commit 517073cd4b.
    463      (CVE-2015-0209)
    464      [Matt Caswell]
    465 
    466   *) X509_to_X509_REQ NULL pointer deref fix
    467 
    468      The function X509_to_X509_REQ will crash with a NULL pointer dereference if
    469      the certificate key is invalid. This function is rarely used in practice.
    470 
    471      This issue was discovered by Brian Carpenter.
    472      (CVE-2015-0288)
    473      [Stephen Henson]
    474 
    475   *) Removed the export ciphers from the DEFAULT ciphers
    476      [Kurt Roeckx]
    477 
    478  Changes between 1.0.1k and 1.0.1l [15 Jan 2015]
    479 
    480   *) Build fixes for the Windows and OpenVMS platforms
    481      [Matt Caswell and Richard Levitte]
    482 
    483  Changes between 1.0.1j and 1.0.1k [8 Jan 2015]
    484 
    485   *) Fix DTLS segmentation fault in dtls1_get_record. A carefully crafted DTLS
    486      message can cause a segmentation fault in OpenSSL due to a NULL pointer
    487      dereference. This could lead to a Denial Of Service attack. Thanks to
    488      Markus Stenberg of Cisco Systems, Inc. for reporting this issue.
    489      (CVE-2014-3571)
    490      [Steve Henson]
    491 
    492   *) Fix DTLS memory leak in dtls1_buffer_record. A memory leak can occur in the
    493      dtls1_buffer_record function under certain conditions. In particular this
    494      could occur if an attacker sent repeated DTLS records with the same
    495      sequence number but for the next epoch. The memory leak could be exploited
    496      by an attacker in a Denial of Service attack through memory exhaustion.
    497      Thanks to Chris Mueller for reporting this issue.
    498      (CVE-2015-0206)
    499      [Matt Caswell]
    500 
    501   *) Fix issue where no-ssl3 configuration sets method to NULL. When openssl is
    502      built with the no-ssl3 option and a SSL v3 ClientHello is received the ssl
    503      method would be set to NULL which could later result in a NULL pointer
    504      dereference. Thanks to Frank Schmirler for reporting this issue.
    505      (CVE-2014-3569)
    506      [Kurt Roeckx]
    507 
    508   *) Abort handshake if server key exchange message is omitted for ephemeral
    509      ECDH ciphersuites.
    510 
    511      Thanks to Karthikeyan Bhargavan of the PROSECCO team at INRIA for
    512      reporting this issue.
    513      (CVE-2014-3572)
    514      [Steve Henson]
    515 
    516   *) Remove non-export ephemeral RSA code on client and server. This code
    517      violated the TLS standard by allowing the use of temporary RSA keys in
    518      non-export ciphersuites and could be used by a server to effectively
    519      downgrade the RSA key length used to a value smaller than the server
    520      certificate. Thanks for Karthikeyan Bhargavan of the PROSECCO team at
    521      INRIA or reporting this issue.
    522      (CVE-2015-0204)
    523      [Steve Henson]
    524 
    525   *) Fixed issue where DH client certificates are accepted without verification.
    526      An OpenSSL server will accept a DH certificate for client authentication
    527      without the certificate verify message. This effectively allows a client to
    528      authenticate without the use of a private key. This only affects servers
    529      which trust a client certificate authority which issues certificates
    530      containing DH keys: these are extremely rare and hardly ever encountered.
    531      Thanks for Karthikeyan Bhargavan of the PROSECCO team at INRIA or reporting
    532      this issue.
    533      (CVE-2015-0205)
    534      [Steve Henson]
    535 
    536   *) Ensure that the session ID context of an SSL is updated when its
    537      SSL_CTX is updated via SSL_set_SSL_CTX.
    538 
    539      The session ID context is typically set from the parent SSL_CTX,
    540      and can vary with the CTX.
    541      [Adam Langley]
    542 
    543   *) Fix various certificate fingerprint issues.
    544 
    545      By using non-DER or invalid encodings outside the signed portion of a
    546      certificate the fingerprint can be changed without breaking the signature.
    547      Although no details of the signed portion of the certificate can be changed
    548      this can cause problems with some applications: e.g. those using the
    549      certificate fingerprint for blacklists.
    550 
    551      1. Reject signatures with non zero unused bits.
    552 
    553      If the BIT STRING containing the signature has non zero unused bits reject
    554      the signature. All current signature algorithms require zero unused bits.
    555 
    556      2. Check certificate algorithm consistency.
    557 
    558      Check the AlgorithmIdentifier inside TBS matches the one in the
    559      certificate signature. NB: this will result in signature failure
    560      errors for some broken certificates.
    561 
    562      Thanks to Konrad Kraszewski from Google for reporting this issue.
    563 
    564      3. Check DSA/ECDSA signatures use DER.
    565 
    566      Reencode DSA/ECDSA signatures and compare with the original received
    567      signature. Return an error if there is a mismatch.
    568 
    569      This will reject various cases including garbage after signature
    570      (thanks to Antti Karjalainen and Tuomo Untinen from the Codenomicon CROSS
    571      program for discovering this case) and use of BER or invalid ASN.1 INTEGERs
    572      (negative or with leading zeroes).
    573 
    574      Further analysis was conducted and fixes were developed by Stephen Henson
    575      of the OpenSSL core team.
    576 
    577      (CVE-2014-8275)
    578      [Steve Henson]
    579 
    580    *) Correct Bignum squaring. Bignum squaring (BN_sqr) may produce incorrect
    581       results on some platforms, including x86_64. This bug occurs at random
    582       with a very low probability, and is not known to be exploitable in any
    583       way, though its exact impact is difficult to determine. Thanks to Pieter
    584       Wuille (Blockstream) who reported this issue and also suggested an initial
    585       fix. Further analysis was conducted by the OpenSSL development team and
    586       Adam Langley of Google. The final fix was developed by Andy Polyakov of
    587       the OpenSSL core team.
    588       (CVE-2014-3570)
    589       [Andy Polyakov]
    590 
    591    *) Do not resume sessions on the server if the negotiated protocol
    592       version does not match the session's version. Resuming with a different
    593       version, while not strictly forbidden by the RFC, is of questionable
    594       sanity and breaks all known clients.
    595       [David Benjamin, Emilia Ksper]
    596 
    597    *) Tighten handling of the ChangeCipherSpec (CCS) message: reject
    598       early CCS messages during renegotiation. (Note that because
    599       renegotiation is encrypted, this early CCS was not exploitable.)
    600       [Emilia Ksper]
    601 
    602    *) Tighten client-side session ticket handling during renegotiation:
    603       ensure that the client only accepts a session ticket if the server sends
    604       the extension anew in the ServerHello. Previously, a TLS client would
    605       reuse the old extension state and thus accept a session ticket if one was
    606       announced in the initial ServerHello.
    607 
    608       Similarly, ensure that the client requires a session ticket if one
    609       was advertised in the ServerHello. Previously, a TLS client would
    610       ignore a missing NewSessionTicket message.
    611       [Emilia Ksper]
    612 
    613  Changes between 1.0.1i and 1.0.1j [15 Oct 2014]
    614 
    615   *) SRTP Memory Leak.
    616 
    617      A flaw in the DTLS SRTP extension parsing code allows an attacker, who
    618      sends a carefully crafted handshake message, to cause OpenSSL to fail
    619      to free up to 64k of memory causing a memory leak. This could be
    620      exploited in a Denial Of Service attack. This issue affects OpenSSL
    621      1.0.1 server implementations for both SSL/TLS and DTLS regardless of
    622      whether SRTP is used or configured. Implementations of OpenSSL that
    623      have been compiled with OPENSSL_NO_SRTP defined are not affected.
    624 
    625      The fix was developed by the OpenSSL team.
    626      (CVE-2014-3513)
    627      [OpenSSL team]
    628 
    629   *) Session Ticket Memory Leak.
    630 
    631      When an OpenSSL SSL/TLS/DTLS server receives a session ticket the
    632      integrity of that ticket is first verified. In the event of a session
    633      ticket integrity check failing, OpenSSL will fail to free memory
    634      causing a memory leak. By sending a large number of invalid session
    635      tickets an attacker could exploit this issue in a Denial Of Service
    636      attack.
    637      (CVE-2014-3567)
    638      [Steve Henson]
    639 
    640   *) Build option no-ssl3 is incomplete.
    641 
    642      When OpenSSL is configured with "no-ssl3" as a build option, servers
    643      could accept and complete a SSL 3.0 handshake, and clients could be
    644      configured to send them.
    645      (CVE-2014-3568)
    646      [Akamai and the OpenSSL team]
    647 
    648   *) Add support for TLS_FALLBACK_SCSV.
    649      Client applications doing fallback retries should call
    650      SSL_set_mode(s, SSL_MODE_SEND_FALLBACK_SCSV).
    651      (CVE-2014-3566)
    652      [Adam Langley, Bodo Moeller]
    653 
    654   *) Add additional DigestInfo checks.
    655  
    656      Reencode DigestInto in DER and check against the original when
    657      verifying RSA signature: this will reject any improperly encoded
    658      DigestInfo structures.
    659 
    660      Note: this is a precautionary measure and no attacks are currently known.
    661 
    662      [Steve Henson]
    663 
    664  Changes between 1.0.1h and 1.0.1i [6 Aug 2014]
    665 
    666   *) Fix SRP buffer overrun vulnerability. Invalid parameters passed to the
    667      SRP code can be overrun an internal buffer. Add sanity check that
    668      g, A, B < N to SRP code.
    669 
    670      Thanks to Sean Devlin and Watson Ladd of Cryptography Services, NCC
    671      Group for discovering this issue.
    672      (CVE-2014-3512)
    673      [Steve Henson]
    674 
    675   *) A flaw in the OpenSSL SSL/TLS server code causes the server to negotiate
    676      TLS 1.0 instead of higher protocol versions when the ClientHello message
    677      is badly fragmented. This allows a man-in-the-middle attacker to force a
    678      downgrade to TLS 1.0 even if both the server and the client support a
    679      higher protocol version, by modifying the client's TLS records.
    680 
    681      Thanks to David Benjamin and Adam Langley (Google) for discovering and
    682      researching this issue.
    683      (CVE-2014-3511)
    684      [David Benjamin]
    685 
    686   *) OpenSSL DTLS clients enabling anonymous (EC)DH ciphersuites are subject
    687      to a denial of service attack. A malicious server can crash the client
    688      with a null pointer dereference (read) by specifying an anonymous (EC)DH
    689      ciphersuite and sending carefully crafted handshake messages.
    690 
    691      Thanks to Felix Grbert (Google) for discovering and researching this
    692      issue.
    693      (CVE-2014-3510)
    694      [Emilia Ksper]
    695 
    696   *) By sending carefully crafted DTLS packets an attacker could cause openssl
    697      to leak memory. This can be exploited through a Denial of Service attack.
    698      Thanks to Adam Langley for discovering and researching this issue.
    699      (CVE-2014-3507)
    700      [Adam Langley]
    701 
    702   *) An attacker can force openssl to consume large amounts of memory whilst
    703      processing DTLS handshake messages. This can be exploited through a
    704      Denial of Service attack.
    705      Thanks to Adam Langley for discovering and researching this issue.
    706      (CVE-2014-3506)
    707      [Adam Langley]
    708 
    709   *) An attacker can force an error condition which causes openssl to crash
    710      whilst processing DTLS packets due to memory being freed twice. This
    711      can be exploited through a Denial of Service attack.
    712      Thanks to Adam Langley and Wan-Teh Chang for discovering and researching
    713      this issue.
    714      (CVE-2014-3505)
    715      [Adam Langley]
    716 
    717   *) If a multithreaded client connects to a malicious server using a resumed
    718      session and the server sends an ec point format extension it could write
    719      up to 255 bytes to freed memory.
    720 
    721      Thanks to Gabor Tyukasz (LogMeIn Inc) for discovering and researching this
    722      issue.
    723      (CVE-2014-3509)
    724      [Gabor Tyukasz]
    725 
    726   *) A malicious server can crash an OpenSSL client with a null pointer
    727      dereference (read) by specifying an SRP ciphersuite even though it was not
    728      properly negotiated with the client. This can be exploited through a
    729      Denial of Service attack.
    730 
    731      Thanks to Joonas Kuorilehto and Riku Hietamki (Codenomicon) for
    732      discovering and researching this issue.
    733      (CVE-2014-5139)
    734      [Steve Henson]
    735 
    736   *) A flaw in OBJ_obj2txt may cause pretty printing functions such as
    737      X509_name_oneline, X509_name_print_ex et al. to leak some information
    738      from the stack. Applications may be affected if they echo pretty printing
    739      output to the attacker.
    740 
    741      Thanks to Ivan Fratric (Google) for discovering this issue.
    742      (CVE-2014-3508)
    743      [Emilia Ksper, and Steve Henson]
    744 
    745   *) Fix ec_GFp_simple_points_make_affine (thus, EC_POINTs_mul etc.)
    746      for corner cases. (Certain input points at infinity could lead to
    747      bogus results, with non-infinity inputs mapped to infinity too.)
    748      [Bodo Moeller]
    749 
    750  Changes between 1.0.1g and 1.0.1h [5 Jun 2014]
    751 
    752   *) Fix for SSL/TLS MITM flaw. An attacker using a carefully crafted
    753      handshake can force the use of weak keying material in OpenSSL
    754      SSL/TLS clients and servers.
    755 
    756      Thanks to KIKUCHI Masashi (Lepidum Co. Ltd.) for discovering and
    757      researching this issue. (CVE-2014-0224)
    758      [KIKUCHI Masashi, Steve Henson]
    759 
    760   *) Fix DTLS recursion flaw. By sending an invalid DTLS handshake to an
    761      OpenSSL DTLS client the code can be made to recurse eventually crashing
    762      in a DoS attack.
    763 
    764      Thanks to Imre Rad (Search-Lab Ltd.) for discovering this issue.
    765      (CVE-2014-0221)
    766      [Imre Rad, Steve Henson]
    767 
    768   *) Fix DTLS invalid fragment vulnerability. A buffer overrun attack can
    769      be triggered by sending invalid DTLS fragments to an OpenSSL DTLS
    770      client or server. This is potentially exploitable to run arbitrary
    771      code on a vulnerable client or server.
    772 
    773      Thanks to Jri Aedla for reporting this issue. (CVE-2014-0195)
    774      [Jri Aedla, Steve Henson]
    775 
    776   *) Fix bug in TLS code where clients enable anonymous ECDH ciphersuites
    777      are subject to a denial of service attack.
    778 
    779      Thanks to Felix Grbert and Ivan Fratric at Google for discovering
    780      this issue. (CVE-2014-3470)
    781      [Felix Grbert, Ivan Fratric, Steve Henson]
    782 
    783   *) Harmonize version and its documentation. -f flag is used to display
    784      compilation flags.
    785      [mancha <mancha1 (a] zoho.com>]
    786 
    787   *) Fix eckey_priv_encode so it immediately returns an error upon a failure
    788      in i2d_ECPrivateKey.
    789      [mancha <mancha1 (a] zoho.com>]
    790 
    791   *) Fix some double frees. These are not thought to be exploitable.
    792      [mancha <mancha1 (a] zoho.com>]
    793 
    794  Changes between 1.0.1f and 1.0.1g [7 Apr 2014]
    795 
    796   *) A missing bounds check in the handling of the TLS heartbeat extension
    797      can be used to reveal up to 64k of memory to a connected client or
    798      server.
    799 
    800      Thanks for Neel Mehta of Google Security for discovering this bug and to
    801      Adam Langley <agl (a] chromium.org> and Bodo Moeller <bmoeller (a] acm.org> for
    802      preparing the fix (CVE-2014-0160)
    803      [Adam Langley, Bodo Moeller]
    804 
    805   *) Fix for the attack described in the paper "Recovering OpenSSL
    806      ECDSA Nonces Using the FLUSH+RELOAD Cache Side-channel Attack"
    807      by Yuval Yarom and Naomi Benger. Details can be obtained from:
    808      http://eprint.iacr.org/2014/140
    809 
    810      Thanks to Yuval Yarom and Naomi Benger for discovering this
    811      flaw and to Yuval Yarom for supplying a fix (CVE-2014-0076)
    812      [Yuval Yarom and Naomi Benger]
    813 
    814   *) TLS pad extension: draft-agl-tls-padding-03
    815 
    816      Workaround for the "TLS hang bug" (see FAQ and PR#2771): if the
    817      TLS client Hello record length value would otherwise be > 255 and
    818      less that 512 pad with a dummy extension containing zeroes so it
    819      is at least 512 bytes long.
    820 
    821      [Adam Langley, Steve Henson]
    822 
    823  Changes between 1.0.1e and 1.0.1f [6 Jan 2014]
    824 
    825   *) Fix for TLS record tampering bug. A carefully crafted invalid 
    826      handshake could crash OpenSSL with a NULL pointer exception.
    827      Thanks to Anton Johansson for reporting this issues.
    828      (CVE-2013-4353)
    829 
    830   *) Keep original DTLS digest and encryption contexts in retransmission
    831      structures so we can use the previous session parameters if they need
    832      to be resent. (CVE-2013-6450)
    833      [Steve Henson]
    834 
    835   *) Add option SSL_OP_SAFARI_ECDHE_ECDSA_BUG (part of SSL_OP_ALL) which
    836      avoids preferring ECDHE-ECDSA ciphers when the client appears to be
    837      Safari on OS X.  Safari on OS X 10.8..10.8.3 advertises support for
    838      several ECDHE-ECDSA ciphers, but fails to negotiate them.  The bug
    839      is fixed in OS X 10.8.4, but Apple have ruled out both hot fixing
    840      10.8..10.8.3 and forcing users to upgrade to 10.8.4 or newer.
    841      [Rob Stradling, Adam Langley]
    842 
    843  Changes between 1.0.1d and 1.0.1e [11 Feb 2013]
    844 
    845   *) Correct fix for CVE-2013-0169. The original didn't work on AES-NI
    846      supporting platforms or when small records were transferred.
    847      [Andy Polyakov, Steve Henson]
    848 
    849  Changes between 1.0.1c and 1.0.1d [5 Feb 2013]
    850 
    851   *) Make the decoding of SSLv3, TLS and DTLS CBC records constant time.
    852 
    853      This addresses the flaw in CBC record processing discovered by 
    854      Nadhem Alfardan and Kenny Paterson. Details of this attack can be found
    855      at: http://www.isg.rhul.ac.uk/tls/     
    856 
    857      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
    858      Security Group at Royal Holloway, University of London
    859      (www.isg.rhul.ac.uk) for discovering this flaw and Adam Langley and
    860      Emilia Ksper for the initial patch.
    861      (CVE-2013-0169)
    862      [Emilia Ksper, Adam Langley, Ben Laurie, Andy Polyakov, Steve Henson]
    863 
    864   *) Fix flaw in AESNI handling of TLS 1.2 and 1.1 records for CBC mode
    865      ciphersuites which can be exploited in a denial of service attack.
    866      Thanks go to and to Adam Langley <agl (a] chromium.org> for discovering
    867      and detecting this bug and to Wolfgang Ettlinger
    868      <wolfgang.ettlinger (a] gmail.com> for independently discovering this issue.
    869      (CVE-2012-2686)
    870      [Adam Langley]
    871 
    872   *) Return an error when checking OCSP signatures when key is NULL.
    873      This fixes a DoS attack. (CVE-2013-0166)
    874      [Steve Henson]
    875 
    876   *) Make openssl verify return errors.
    877      [Chris Palmer <palmer (a] google.com> and Ben Laurie]
    878 
    879   *) Call OCSP Stapling callback after ciphersuite has been chosen, so
    880      the right response is stapled. Also change SSL_get_certificate()
    881      so it returns the certificate actually sent.
    882      See http://rt.openssl.org/Ticket/Display.html?id=2836.
    883      [Rob Stradling <rob.stradling (a] comodo.com>]
    884 
    885   *) Fix possible deadlock when decoding public keys.
    886      [Steve Henson]
    887 
    888   *) Don't use TLS 1.0 record version number in initial client hello
    889      if renegotiating.
    890      [Steve Henson]
    891 
    892  Changes between 1.0.1b and 1.0.1c [10 May 2012]
    893 
    894   *) Sanity check record length before skipping explicit IV in TLS
    895      1.2, 1.1 and DTLS to fix DoS attack.
    896 
    897      Thanks to Codenomicon for discovering this issue using Fuzz-o-Matic
    898      fuzzing as a service testing platform.
    899      (CVE-2012-2333)
    900      [Steve Henson]
    901 
    902   *) Initialise tkeylen properly when encrypting CMS messages.
    903      Thanks to Solar Designer of Openwall for reporting this issue.
    904      [Steve Henson]
    905 
    906   *) In FIPS mode don't try to use composite ciphers as they are not
    907      approved.
    908      [Steve Henson]
    909 
    910  Changes between 1.0.1a and 1.0.1b [26 Apr 2012]
    911 
    912   *) OpenSSL 1.0.0 sets SSL_OP_ALL to 0x80000FFFL and OpenSSL 1.0.1 and
    913      1.0.1a set SSL_OP_NO_TLSv1_1 to 0x00000400L which would unfortunately
    914      mean any application compiled against OpenSSL 1.0.0 headers setting
    915      SSL_OP_ALL would also set SSL_OP_NO_TLSv1_1, unintentionally disablng
    916      TLS 1.1 also. Fix this by changing the value of SSL_OP_NO_TLSv1_1 to
    917      0x10000000L Any application which was previously compiled against
    918      OpenSSL 1.0.1 or 1.0.1a headers and which cares about SSL_OP_NO_TLSv1_1
    919      will need to be recompiled as a result. Letting be results in
    920      inability to disable specifically TLS 1.1 and in client context,
    921      in unlike event, limit maximum offered version to TLS 1.0 [see below].
    922      [Steve Henson]
    923 
    924   *) In order to ensure interoperabilty SSL_OP_NO_protocolX does not
    925      disable just protocol X, but all protocols above X *if* there are
    926      protocols *below* X still enabled. In more practical terms it means
    927      that if application wants to disable TLS1.0 in favor of TLS1.1 and
    928      above, it's not sufficient to pass SSL_OP_NO_TLSv1, one has to pass
    929      SSL_OP_NO_TLSv1|SSL_OP_NO_SSLv3|SSL_OP_NO_SSLv2. This applies to
    930      client side.
    931      [Andy Polyakov]
    932 
    933  Changes between 1.0.1 and 1.0.1a [19 Apr 2012]
    934 
    935   *) Check for potentially exploitable overflows in asn1_d2i_read_bio
    936      BUF_mem_grow and BUF_mem_grow_clean. Refuse attempts to shrink buffer
    937      in CRYPTO_realloc_clean.
    938 
    939      Thanks to Tavis Ormandy, Google Security Team, for discovering this
    940      issue and to Adam Langley <agl (a] chromium.org> for fixing it.
    941      (CVE-2012-2110)
    942      [Adam Langley (Google), Tavis Ormandy, Google Security Team]
    943 
    944   *) Don't allow TLS 1.2 SHA-256 ciphersuites in TLS 1.0, 1.1 connections.
    945      [Adam Langley]
    946 
    947   *) Workarounds for some broken servers that "hang" if a client hello
    948      record length exceeds 255 bytes.
    949 
    950      1. Do not use record version number > TLS 1.0 in initial client
    951         hello: some (but not all) hanging servers will now work.
    952      2. If we set OPENSSL_MAX_TLS1_2_CIPHER_LENGTH this will truncate
    953 	the number of ciphers sent in the client hello. This should be
    954         set to an even number, such as 50, for example by passing:
    955         -DOPENSSL_MAX_TLS1_2_CIPHER_LENGTH=50 to config or Configure.
    956         Most broken servers should now work.
    957      3. If all else fails setting OPENSSL_NO_TLS1_2_CLIENT will disable
    958 	TLS 1.2 client support entirely.
    959      [Steve Henson]
    960 
    961   *) Fix SEGV in Vector Permutation AES module observed in OpenSSH.
    962      [Andy Polyakov]
    963 
    964  Changes between 1.0.0h and 1.0.1  [14 Mar 2012]
    965 
    966   *) Add compatibility with old MDC2 signatures which use an ASN1 OCTET
    967      STRING form instead of a DigestInfo.
    968      [Steve Henson]
    969 
    970   *) The format used for MDC2 RSA signatures is inconsistent between EVP
    971      and the RSA_sign/RSA_verify functions. This was made more apparent when
    972      OpenSSL used RSA_sign/RSA_verify for some RSA signatures in particular
    973      those which went through EVP_PKEY_METHOD in 1.0.0 and later. Detect 
    974      the correct format in RSA_verify so both forms transparently work.
    975      [Steve Henson]
    976 
    977   *) Some servers which support TLS 1.0 can choke if we initially indicate
    978      support for TLS 1.2 and later renegotiate using TLS 1.0 in the RSA
    979      encrypted premaster secret. As a workaround use the maximum pemitted
    980      client version in client hello, this should keep such servers happy
    981      and still work with previous versions of OpenSSL.
    982      [Steve Henson]
    983 
    984   *) Add support for TLS/DTLS heartbeats.
    985      [Robin Seggelmann <seggelmann (a] fh-muenster.de>]
    986 
    987   *) Add support for SCTP.
    988      [Robin Seggelmann <seggelmann (a] fh-muenster.de>]
    989 
    990   *) Improved PRNG seeding for VOS.
    991      [Paul Green <Paul.Green (a] stratus.com>]
    992 
    993   *) Extensive assembler packs updates, most notably:
    994 
    995 	- x86[_64]:     AES-NI, PCLMULQDQ, RDRAND support;
    996 	- x86[_64]:     SSSE3 support (SHA1, vector-permutation AES);
    997 	- x86_64:       bit-sliced AES implementation;
    998 	- ARM:          NEON support, contemporary platforms optimizations;
    999 	- s390x:        z196 support;
   1000 	- *:            GHASH and GF(2^m) multiplication implementations;
   1001 
   1002      [Andy Polyakov]
   1003 
   1004   *) Make TLS-SRP code conformant with RFC 5054 API cleanup
   1005      (removal of unnecessary code)
   1006      [Peter Sylvester <peter.sylvester (a] edelweb.fr>]
   1007 
   1008   *) Add TLS key material exporter from RFC 5705.
   1009      [Eric Rescorla]
   1010 
   1011   *) Add DTLS-SRTP negotiation from RFC 5764.
   1012      [Eric Rescorla]
   1013 
   1014   *) Add Next Protocol Negotiation,
   1015      http://tools.ietf.org/html/draft-agl-tls-nextprotoneg-00. Can be
   1016      disabled with a no-npn flag to config or Configure. Code donated
   1017      by Google.
   1018      [Adam Langley <agl (a] google.com> and Ben Laurie]
   1019 
   1020   *) Add optional 64-bit optimized implementations of elliptic curves NIST-P224,
   1021      NIST-P256, NIST-P521, with constant-time single point multiplication on
   1022      typical inputs. Compiler support for the nonstandard type __uint128_t is
   1023      required to use this (present in gcc 4.4 and later, for 64-bit builds).
   1024      Code made available under Apache License version 2.0.
   1025 
   1026      Specify "enable-ec_nistp_64_gcc_128" on the Configure (or config) command
   1027      line to include this in your build of OpenSSL, and run "make depend" (or
   1028      "make update"). This enables the following EC_METHODs:
   1029 
   1030          EC_GFp_nistp224_method()
   1031          EC_GFp_nistp256_method()
   1032          EC_GFp_nistp521_method()
   1033 
   1034      EC_GROUP_new_by_curve_name() will automatically use these (while
   1035      EC_GROUP_new_curve_GFp() currently prefers the more flexible
   1036      implementations).
   1037      [Emilia Ksper, Adam Langley, Bodo Moeller (Google)]
   1038 
   1039   *) Use type ossl_ssize_t instad of ssize_t which isn't available on
   1040      all platforms. Move ssize_t definition from e_os.h to the public
   1041      header file e_os2.h as it now appears in public header file cms.h
   1042      [Steve Henson]
   1043 
   1044   *) New -sigopt option to the ca, req and x509 utilities. Additional
   1045      signature parameters can be passed using this option and in
   1046      particular PSS. 
   1047      [Steve Henson]
   1048 
   1049   *) Add RSA PSS signing function. This will generate and set the
   1050      appropriate AlgorithmIdentifiers for PSS based on those in the
   1051      corresponding EVP_MD_CTX structure. No application support yet.
   1052      [Steve Henson]
   1053 
   1054   *) Support for companion algorithm specific ASN1 signing routines.
   1055      New function ASN1_item_sign_ctx() signs a pre-initialised
   1056      EVP_MD_CTX structure and sets AlgorithmIdentifiers based on
   1057      the appropriate parameters.
   1058      [Steve Henson]
   1059 
   1060   *) Add new algorithm specific ASN1 verification initialisation function
   1061      to EVP_PKEY_ASN1_METHOD: this is not in EVP_PKEY_METHOD since the ASN1
   1062      handling will be the same no matter what EVP_PKEY_METHOD is used.
   1063      Add a PSS handler to support verification of PSS signatures: checked
   1064      against a number of sample certificates.
   1065      [Steve Henson]
   1066 
   1067   *) Add signature printing for PSS. Add PSS OIDs.
   1068      [Steve Henson, Martin Kaiser <lists (a] kaiser.cx>]
   1069 
   1070   *) Add algorithm specific signature printing. An individual ASN1 method
   1071      can now print out signatures instead of the standard hex dump. 
   1072 
   1073      More complex signatures (e.g. PSS) can print out more meaningful
   1074      information. Include DSA version that prints out the signature
   1075      parameters r, s.
   1076      [Steve Henson]
   1077 
   1078   *) Password based recipient info support for CMS library: implementing
   1079      RFC3211.
   1080      [Steve Henson]
   1081 
   1082   *) Split password based encryption into PBES2 and PBKDF2 functions. This
   1083      neatly separates the code into cipher and PBE sections and is required
   1084      for some algorithms that split PBES2 into separate pieces (such as
   1085      password based CMS).
   1086      [Steve Henson]
   1087 
   1088   *) Session-handling fixes:
   1089      - Fix handling of connections that are resuming with a session ID,
   1090        but also support Session Tickets.
   1091      - Fix a bug that suppressed issuing of a new ticket if the client
   1092        presented a ticket with an expired session.
   1093      - Try to set the ticket lifetime hint to something reasonable.
   1094      - Make tickets shorter by excluding irrelevant information.
   1095      - On the client side, don't ignore renewed tickets.
   1096      [Adam Langley, Bodo Moeller (Google)]
   1097 
   1098   *) Fix PSK session representation.
   1099      [Bodo Moeller]
   1100 
   1101   *) Add RC4-MD5 and AESNI-SHA1 "stitched" implementations.
   1102 
   1103      This work was sponsored by Intel.
   1104      [Andy Polyakov]
   1105 
   1106   *) Add GCM support to TLS library. Some custom code is needed to split
   1107      the IV between the fixed (from PRF) and explicit (from TLS record)
   1108      portions. This adds all GCM ciphersuites supported by RFC5288 and 
   1109      RFC5289. Generalise some AES* cipherstrings to inlclude GCM and
   1110      add a special AESGCM string for GCM only.
   1111      [Steve Henson]
   1112 
   1113   *) Expand range of ctrls for AES GCM. Permit setting invocation
   1114      field on decrypt and retrieval of invocation field only on encrypt.
   1115      [Steve Henson]
   1116 
   1117   *) Add HMAC ECC ciphersuites from RFC5289. Include SHA384 PRF support.
   1118      As required by RFC5289 these ciphersuites cannot be used if for
   1119      versions of TLS earlier than 1.2.
   1120      [Steve Henson]
   1121 
   1122   *) For FIPS capable OpenSSL interpret a NULL default public key method
   1123      as unset and return the appopriate default but do *not* set the default.
   1124      This means we can return the appopriate method in applications that
   1125      swicth between FIPS and non-FIPS modes.
   1126      [Steve Henson]
   1127 
   1128   *) Redirect HMAC and CMAC operations to FIPS module in FIPS mode. If an
   1129      ENGINE is used then we cannot handle that in the FIPS module so we
   1130      keep original code iff non-FIPS operations are allowed.
   1131      [Steve Henson]
   1132 
   1133   *) Add -attime option to openssl utilities.
   1134      [Peter Eckersley <pde (a] eff.org>, Ben Laurie and Steve Henson]
   1135 
   1136   *) Redirect DSA and DH operations to FIPS module in FIPS mode.
   1137      [Steve Henson]
   1138 
   1139   *) Redirect ECDSA and ECDH operations to FIPS module in FIPS mode. Also use
   1140      FIPS EC methods unconditionally for now.
   1141      [Steve Henson]
   1142 
   1143   *) New build option no-ec2m to disable characteristic 2 code.
   1144      [Steve Henson]
   1145 
   1146   *) Backport libcrypto audit of return value checking from 1.1.0-dev; not
   1147      all cases can be covered as some introduce binary incompatibilities.
   1148      [Steve Henson]
   1149 
   1150   *) Redirect RSA operations to FIPS module including keygen,
   1151      encrypt, decrypt, sign and verify. Block use of non FIPS RSA methods.
   1152      [Steve Henson]
   1153 
   1154   *) Add similar low level API blocking to ciphers.
   1155      [Steve Henson]
   1156 
   1157   *) Low level digest APIs are not approved in FIPS mode: any attempt
   1158      to use these will cause a fatal error. Applications that *really* want
   1159      to use them can use the private_* version instead.
   1160      [Steve Henson]
   1161 
   1162   *) Redirect cipher operations to FIPS module for FIPS builds. 
   1163      [Steve Henson]
   1164 
   1165   *) Redirect digest operations to FIPS module for FIPS builds. 
   1166      [Steve Henson]
   1167 
   1168   *) Update build system to add "fips" flag which will link in fipscanister.o
   1169      for static and shared library builds embedding a signature if needed.
   1170      [Steve Henson]
   1171 
   1172   *) Output TLS supported curves in preference order instead of numerical
   1173      order. This is currently hardcoded for the highest order curves first.
   1174      This should be configurable so applications can judge speed vs strength.
   1175      [Steve Henson]
   1176 
   1177   *) Add TLS v1.2 server support for client authentication. 
   1178      [Steve Henson]
   1179 
   1180   *) Add support for FIPS mode in ssl library: disable SSLv3, non-FIPS ciphers
   1181      and enable MD5.
   1182      [Steve Henson]
   1183 
   1184   *) Functions FIPS_mode_set() and FIPS_mode() which call the underlying
   1185      FIPS modules versions.
   1186      [Steve Henson]
   1187 
   1188   *) Add TLS v1.2 client side support for client authentication. Keep cache
   1189      of handshake records longer as we don't know the hash algorithm to use
   1190      until after the certificate request message is received.
   1191      [Steve Henson]
   1192 
   1193   *) Initial TLS v1.2 client support. Add a default signature algorithms
   1194      extension including all the algorithms we support. Parse new signature
   1195      format in client key exchange. Relax some ECC signing restrictions for
   1196      TLS v1.2 as indicated in RFC5246.
   1197      [Steve Henson]
   1198 
   1199   *) Add server support for TLS v1.2 signature algorithms extension. Switch
   1200      to new signature format when needed using client digest preference.
   1201      All server ciphersuites should now work correctly in TLS v1.2. No client
   1202      support yet and no support for client certificates.
   1203      [Steve Henson]
   1204 
   1205   *) Initial TLS v1.2 support. Add new SHA256 digest to ssl code, switch
   1206      to SHA256 for PRF when using TLS v1.2 and later. Add new SHA256 based
   1207      ciphersuites. At present only RSA key exchange ciphersuites work with
   1208      TLS v1.2. Add new option for TLS v1.2 replacing the old and obsolete
   1209      SSL_OP_PKCS1_CHECK flags with SSL_OP_NO_TLSv1_2. New TLSv1.2 methods
   1210      and version checking.
   1211      [Steve Henson]
   1212 
   1213   *) New option OPENSSL_NO_SSL_INTERN. If an application can be compiled
   1214      with this defined it will not be affected by any changes to ssl internal
   1215      structures. Add several utility functions to allow openssl application
   1216      to work with OPENSSL_NO_SSL_INTERN defined.
   1217      [Steve Henson]
   1218 
   1219   *) Add SRP support.
   1220      [Tom Wu <tjw (a] cs.stanford.edu> and Ben Laurie]
   1221 
   1222   *) Add functions to copy EVP_PKEY_METHOD and retrieve flags and id.
   1223      [Steve Henson]
   1224 
   1225   *) Permit abbreviated handshakes when renegotiating using the function
   1226      SSL_renegotiate_abbreviated().
   1227      [Robin Seggelmann <seggelmann (a] fh-muenster.de>]
   1228 
   1229   *) Add call to ENGINE_register_all_complete() to
   1230      ENGINE_load_builtin_engines(), so some implementations get used
   1231      automatically instead of needing explicit application support.
   1232      [Steve Henson]
   1233 
   1234   *) Add support for TLS key exporter as described in RFC5705.
   1235      [Robin Seggelmann <seggelmann (a] fh-muenster.de>, Steve Henson]
   1236 
   1237   *) Initial TLSv1.1 support. Since TLSv1.1 is very similar to TLS v1.0 only
   1238      a few changes are required:
   1239 
   1240        Add SSL_OP_NO_TLSv1_1 flag.
   1241        Add TLSv1_1 methods.
   1242        Update version checking logic to handle version 1.1.
   1243        Add explicit IV handling (ported from DTLS code).
   1244        Add command line options to s_client/s_server.
   1245      [Steve Henson]
   1246 
   1247  Changes between 1.0.0g and 1.0.0h [12 Mar 2012]
   1248 
   1249   *) Fix MMA (Bleichenbacher's attack on PKCS #1 v1.5 RSA padding) weakness
   1250      in CMS and PKCS7 code. When RSA decryption fails use a random key for
   1251      content decryption and always return the same error. Note: this attack
   1252      needs on average 2^20 messages so it only affects automated senders. The
   1253      old behaviour can be reenabled in the CMS code by setting the
   1254      CMS_DEBUG_DECRYPT flag: this is useful for debugging and testing where
   1255      an MMA defence is not necessary.
   1256      Thanks to Ivan Nestlerode <inestlerode (a] us.ibm.com> for discovering
   1257      this issue. (CVE-2012-0884)
   1258      [Steve Henson]
   1259 
   1260   *) Fix CVE-2011-4619: make sure we really are receiving a 
   1261      client hello before rejecting multiple SGC restarts. Thanks to
   1262      Ivan Nestlerode <inestlerode (a] us.ibm.com> for discovering this bug.
   1263      [Steve Henson]
   1264 
   1265  Changes between 1.0.0f and 1.0.0g [18 Jan 2012]
   1266 
   1267   *) Fix for DTLS DoS issue introduced by fix for CVE-2011-4109.
   1268      Thanks to Antonio Martin, Enterprise Secure Access Research and
   1269      Development, Cisco Systems, Inc. for discovering this bug and
   1270      preparing a fix. (CVE-2012-0050)
   1271      [Antonio Martin]
   1272 
   1273  Changes between 1.0.0e and 1.0.0f [4 Jan 2012]
   1274 
   1275   *) Nadhem Alfardan and Kenny Paterson have discovered an extension
   1276      of the Vaudenay padding oracle attack on CBC mode encryption
   1277      which enables an efficient plaintext recovery attack against
   1278      the OpenSSL implementation of DTLS. Their attack exploits timing
   1279      differences arising during decryption processing. A research
   1280      paper describing this attack can be found at:
   1281                   http://www.isg.rhul.ac.uk/~kp/dtls.pdf
   1282      Thanks go to Nadhem Alfardan and Kenny Paterson of the Information
   1283      Security Group at Royal Holloway, University of London
   1284      (www.isg.rhul.ac.uk) for discovering this flaw and to Robin Seggelmann
   1285      <seggelmann (a] fh-muenster.de> and Michael Tuexen <tuexen (a] fh-muenster.de>
   1286      for preparing the fix. (CVE-2011-4108)
   1287      [Robin Seggelmann, Michael Tuexen]
   1288 
   1289   *) Clear bytes used for block padding of SSL 3.0 records.
   1290      (CVE-2011-4576)
   1291      [Adam Langley (Google)]
   1292 
   1293   *) Only allow one SGC handshake restart for SSL/TLS. Thanks to George
   1294      Kadianakis <desnacked (a] gmail.com> for discovering this issue and
   1295      Adam Langley for preparing the fix. (CVE-2011-4619)
   1296      [Adam Langley (Google)]
   1297 
   1298   *) Check parameters are not NULL in GOST ENGINE. (CVE-2012-0027)
   1299      [Andrey Kulikov <amdeich (a] gmail.com>]
   1300 
   1301   *) Prevent malformed RFC3779 data triggering an assertion failure.
   1302      Thanks to Andrew Chi, BBN Technologies, for discovering the flaw
   1303      and Rob Austein <sra (a] hactrn.net> for fixing it. (CVE-2011-4577)
   1304      [Rob Austein <sra (a] hactrn.net>]
   1305 
   1306   *) Improved PRNG seeding for VOS.
   1307      [Paul Green <Paul.Green (a] stratus.com>]
   1308 
   1309   *) Fix ssl_ciph.c set-up race.
   1310      [Adam Langley (Google)]
   1311 
   1312   *) Fix spurious failures in ecdsatest.c.
   1313      [Emilia Ksper (Google)]
   1314 
   1315   *) Fix the BIO_f_buffer() implementation (which was mixing different
   1316      interpretations of the '..._len' fields).
   1317      [Adam Langley (Google)]
   1318 
   1319   *) Fix handling of BN_BLINDING: now BN_BLINDING_invert_ex (rather than
   1320      BN_BLINDING_invert_ex) calls BN_BLINDING_update, ensuring that concurrent
   1321      threads won't reuse the same blinding coefficients.
   1322 
   1323      This also avoids the need to obtain the CRYPTO_LOCK_RSA_BLINDING
   1324      lock to call BN_BLINDING_invert_ex, and avoids one use of
   1325      BN_BLINDING_update for each BN_BLINDING structure (previously,
   1326      the last update always remained unused).
   1327      [Emilia Ksper (Google)]
   1328 
   1329   *) In ssl3_clear, preserve s3->init_extra along with s3->rbuf.
   1330      [Bob Buckholz (Google)]
   1331 
   1332  Changes between 1.0.0d and 1.0.0e [6 Sep 2011]
   1333 
   1334   *) Fix bug where CRLs with nextUpdate in the past are sometimes accepted
   1335      by initialising X509_STORE_CTX properly. (CVE-2011-3207)
   1336      [Kaspar Brand <ossl (a] velox.ch>]
   1337 
   1338   *) Fix SSL memory handling for (EC)DH ciphersuites, in particular
   1339      for multi-threaded use of ECDH. (CVE-2011-3210)
   1340      [Adam Langley (Google)]
   1341 
   1342   *) Fix x509_name_ex_d2i memory leak on bad inputs.
   1343      [Bodo Moeller]
   1344 
   1345   *) Remove hard coded ecdsaWithSHA1 signature tests in ssl code and check
   1346      signature public key algorithm by using OID xref utilities instead.
   1347      Before this you could only use some ECC ciphersuites with SHA1 only.
   1348      [Steve Henson]
   1349 
   1350   *) Add protection against ECDSA timing attacks as mentioned in the paper
   1351      by Billy Bob Brumley and Nicola Tuveri, see:
   1352 
   1353 	http://eprint.iacr.org/2011/232.pdf
   1354 
   1355      [Billy Bob Brumley and Nicola Tuveri]
   1356 
   1357  Changes between 1.0.0c and 1.0.0d [8 Feb 2011]
   1358 
   1359   *) Fix parsing of OCSP stapling ClientHello extension. CVE-2011-0014
   1360      [Neel Mehta, Adam Langley, Bodo Moeller (Google)]
   1361 
   1362   *) Fix bug in string printing code: if *any* escaping is enabled we must
   1363      escape the escape character (backslash) or the resulting string is
   1364      ambiguous.
   1365      [Steve Henson]
   1366 
   1367  Changes between 1.0.0b and 1.0.0c  [2 Dec 2010]
   1368 
   1369   *) Disable code workaround for ancient and obsolete Netscape browsers
   1370      and servers: an attacker can use it in a ciphersuite downgrade attack.
   1371      Thanks to Martin Rex for discovering this bug. CVE-2010-4180
   1372      [Steve Henson]
   1373 
   1374   *) Fixed J-PAKE implementation error, originally discovered by
   1375      Sebastien Martini, further info and confirmation from Stefan
   1376      Arentz and Feng Hao. Note that this fix is a security fix. CVE-2010-4252
   1377      [Ben Laurie]
   1378 
   1379  Changes between 1.0.0a and 1.0.0b  [16 Nov 2010]
   1380 
   1381   *) Fix extension code to avoid race conditions which can result in a buffer
   1382      overrun vulnerability: resumed sessions must not be modified as they can
   1383      be shared by multiple threads. CVE-2010-3864
   1384      [Steve Henson]
   1385 
   1386   *) Fix WIN32 build system to correctly link an ENGINE directory into
   1387      a DLL. 
   1388      [Steve Henson]
   1389 
   1390  Changes between 1.0.0 and 1.0.0a  [01 Jun 2010]
   1391 
   1392   *) Check return value of int_rsa_verify in pkey_rsa_verifyrecover 
   1393      (CVE-2010-1633)
   1394      [Steve Henson, Peter-Michael Hager <hager (a] dortmund.net>]
   1395 
   1396  Changes between 0.9.8n and 1.0.0  [29 Mar 2010]
   1397 
   1398   *) Add "missing" function EVP_CIPHER_CTX_copy(). This copies a cipher
   1399      context. The operation can be customised via the ctrl mechanism in
   1400      case ENGINEs want to include additional functionality.
   1401      [Steve Henson]
   1402 
   1403   *) Tolerate yet another broken PKCS#8 key format: private key value negative.
   1404      [Steve Henson]
   1405 
   1406   *) Add new -subject_hash_old and -issuer_hash_old options to x509 utility to
   1407      output hashes compatible with older versions of OpenSSL.
   1408      [Willy Weisz <weisz (a] vcpc.univie.ac.at>]
   1409 
   1410   *) Fix compression algorithm handling: if resuming a session use the
   1411      compression algorithm of the resumed session instead of determining
   1412      it from client hello again. Don't allow server to change algorithm.
   1413      [Steve Henson]
   1414 
   1415   *) Add load_crls() function to apps tidying load_certs() too. Add option
   1416      to verify utility to allow additional CRLs to be included.
   1417      [Steve Henson]
   1418 
   1419   *) Update OCSP request code to permit adding custom headers to the request:
   1420      some responders need this.
   1421      [Steve Henson]
   1422 
   1423   *) The function EVP_PKEY_sign() returns <=0 on error: check return code
   1424      correctly.
   1425      [Julia Lawall <julia (a] diku.dk>]
   1426 
   1427   *) Update verify callback code in apps/s_cb.c and apps/verify.c, it
   1428      needlessly dereferenced structures, used obsolete functions and
   1429      didn't handle all updated verify codes correctly.
   1430      [Steve Henson]
   1431 
   1432   *) Disable MD2 in the default configuration.
   1433      [Steve Henson]
   1434 
   1435   *) In BIO_pop() and BIO_push() use the ctrl argument (which was NULL) to
   1436      indicate the initial BIO being pushed or popped. This makes it possible
   1437      to determine whether the BIO is the one explicitly called or as a result
   1438      of the ctrl being passed down the chain. Fix BIO_pop() and SSL BIOs so
   1439      it handles reference counts correctly and doesn't zero out the I/O bio
   1440      when it is not being explicitly popped. WARNING: applications which
   1441      included workarounds for the old buggy behaviour will need to be modified
   1442      or they could free up already freed BIOs.
   1443      [Steve Henson]
   1444 
   1445   *) Extend the uni2asc/asc2uni => OPENSSL_uni2asc/OPENSSL_asc2uni
   1446      renaming to all platforms (within the 0.9.8 branch, this was
   1447      done conditionally on Netware platforms to avoid a name clash).
   1448      [Guenter <lists (a] gknw.net>]
   1449 
   1450   *) Add ECDHE and PSK support to DTLS.
   1451      [Michael Tuexen <tuexen (a] fh-muenster.de>]
   1452 
   1453   *) Add CHECKED_STACK_OF macro to safestack.h, otherwise safestack can't
   1454      be used on C++.
   1455      [Steve Henson]
   1456 
   1457   *) Add "missing" function EVP_MD_flags() (without this the only way to
   1458      retrieve a digest flags is by accessing the structure directly. Update
   1459      EVP_MD_do_all*() and EVP_CIPHER_do_all*() to include the name a digest
   1460      or cipher is registered as in the "from" argument. Print out all
   1461      registered digests in the dgst usage message instead of manually 
   1462      attempting to work them out.
   1463      [Steve Henson]
   1464 
   1465   *) If no SSLv2 ciphers are used don't use an SSLv2 compatible client hello:
   1466      this allows the use of compression and extensions. Change default cipher
   1467      string to remove SSLv2 ciphersuites. This effectively avoids ancient SSLv2
   1468      by default unless an application cipher string requests it.
   1469      [Steve Henson]
   1470 
   1471   *) Alter match criteria in PKCS12_parse(). It used to try to use local
   1472      key ids to find matching certificates and keys but some PKCS#12 files
   1473      don't follow the (somewhat unwritten) rules and this strategy fails.
   1474      Now just gather all certificates together and the first private key
   1475      then look for the first certificate that matches the key.
   1476      [Steve Henson]
   1477 
   1478   *) Support use of registered digest and cipher names for dgst and cipher
   1479      commands instead of having to add each one as a special case. So now
   1480      you can do:
   1481 
   1482         openssl sha256 foo
   1483 
   1484      as well as:
   1485 
   1486         openssl dgst -sha256 foo
   1487 
   1488      and this works for ENGINE based algorithms too.
   1489 
   1490      [Steve Henson]
   1491 
   1492   *) Update Gost ENGINE to support parameter files.
   1493      [Victor B. Wagner <vitus (a] cryptocom.ru>]
   1494 
   1495   *) Support GeneralizedTime in ca utility. 
   1496      [Oliver Martin <oliver (a] volatilevoid.net>, Steve Henson]
   1497 
   1498   *) Enhance the hash format used for certificate directory links. The new
   1499      form uses the canonical encoding (meaning equivalent names will work
   1500      even if they aren't identical) and uses SHA1 instead of MD5. This form
   1501      is incompatible with the older format and as a result c_rehash should
   1502      be used to rebuild symbolic links.
   1503      [Steve Henson]
   1504 
   1505   *) Make PKCS#8 the default write format for private keys, replacing the
   1506      traditional format. This form is standardised, more secure and doesn't
   1507      include an implicit MD5 dependency.
   1508      [Steve Henson]
   1509 
   1510   *) Add a $gcc_devteam_warn option to Configure. The idea is that any code
   1511      committed to OpenSSL should pass this lot as a minimum.
   1512      [Steve Henson]
   1513 
   1514   *) Add session ticket override functionality for use by EAP-FAST.
   1515      [Jouni Malinen <j (a] w1.fi>]
   1516 
   1517   *) Modify HMAC functions to return a value. Since these can be implemented
   1518      in an ENGINE errors can occur.
   1519      [Steve Henson]
   1520 
   1521   *) Type-checked OBJ_bsearch_ex.
   1522      [Ben Laurie]
   1523 
   1524   *) Type-checked OBJ_bsearch. Also some constification necessitated
   1525      by type-checking.  Still to come: TXT_DB, bsearch(?),
   1526      OBJ_bsearch_ex, qsort, CRYPTO_EX_DATA, ASN1_VALUE, ASN1_STRING,
   1527      CONF_VALUE.
   1528      [Ben Laurie]
   1529 
   1530   *) New function OPENSSL_gmtime_adj() to add a specific number of days and
   1531      seconds to a tm structure directly, instead of going through OS
   1532      specific date routines. This avoids any issues with OS routines such
   1533      as the year 2038 bug. New *_adj() functions for ASN1 time structures
   1534      and X509_time_adj_ex() to cover the extended range. The existing
   1535      X509_time_adj() is still usable and will no longer have any date issues.
   1536      [Steve Henson]
   1537 
   1538   *) Delta CRL support. New use deltas option which will attempt to locate
   1539      and search any appropriate delta CRLs available.
   1540 
   1541      This work was sponsored by Google.
   1542      [Steve Henson]
   1543 
   1544   *) Support for CRLs partitioned by reason code. Reorganise CRL processing
   1545      code and add additional score elements. Validate alternate CRL paths
   1546      as part of the CRL checking and indicate a new error "CRL path validation
   1547      error" in this case. Applications wanting additional details can use
   1548      the verify callback and check the new "parent" field. If this is not
   1549      NULL CRL path validation is taking place. Existing applications wont
   1550      see this because it requires extended CRL support which is off by
   1551      default.
   1552 
   1553      This work was sponsored by Google.
   1554      [Steve Henson]
   1555 
   1556   *) Support for freshest CRL extension.
   1557 
   1558      This work was sponsored by Google.
   1559      [Steve Henson]
   1560 
   1561   *) Initial indirect CRL support. Currently only supported in the CRLs
   1562      passed directly and not via lookup. Process certificate issuer
   1563      CRL entry extension and lookup CRL entries by bother issuer name
   1564      and serial number. Check and process CRL issuer entry in IDP extension.
   1565 
   1566      This work was sponsored by Google.
   1567      [Steve Henson]
   1568 
   1569   *) Add support for distinct certificate and CRL paths. The CRL issuer
   1570      certificate is validated separately in this case. Only enabled if
   1571      an extended CRL support flag is set: this flag will enable additional
   1572      CRL functionality in future.
   1573 
   1574      This work was sponsored by Google.
   1575      [Steve Henson]
   1576 
   1577   *) Add support for policy mappings extension.
   1578 
   1579      This work was sponsored by Google.
   1580      [Steve Henson]
   1581 
   1582   *) Fixes to pathlength constraint, self issued certificate handling,
   1583      policy processing to align with RFC3280 and PKITS tests.
   1584 
   1585      This work was sponsored by Google.
   1586      [Steve Henson]
   1587 
   1588   *) Support for name constraints certificate extension. DN, email, DNS
   1589      and URI types are currently supported.
   1590 
   1591      This work was sponsored by Google.
   1592      [Steve Henson]
   1593 
   1594   *) To cater for systems that provide a pointer-based thread ID rather
   1595      than numeric, deprecate the current numeric thread ID mechanism and
   1596      replace it with a structure and associated callback type. This
   1597      mechanism allows a numeric "hash" to be extracted from a thread ID in
   1598      either case, and on platforms where pointers are larger than 'long',
   1599      mixing is done to help ensure the numeric 'hash' is usable even if it
   1600      can't be guaranteed unique. The default mechanism is to use "&errno"
   1601      as a pointer-based thread ID to distinguish between threads.
   1602 
   1603      Applications that want to provide their own thread IDs should now use
   1604      CRYPTO_THREADID_set_callback() to register a callback that will call
   1605      either CRYPTO_THREADID_set_numeric() or CRYPTO_THREADID_set_pointer().
   1606 
   1607      Note that ERR_remove_state() is now deprecated, because it is tied
   1608      to the assumption that thread IDs are numeric.  ERR_remove_state(0)
   1609      to free the current thread's error state should be replaced by
   1610      ERR_remove_thread_state(NULL).
   1611 
   1612      (This new approach replaces the functions CRYPTO_set_idptr_callback(),
   1613      CRYPTO_get_idptr_callback(), and CRYPTO_thread_idptr() that existed in
   1614      OpenSSL 0.9.9-dev between June 2006 and August 2008. Also, if an
   1615      application was previously providing a numeric thread callback that
   1616      was inappropriate for distinguishing threads, then uniqueness might
   1617      have been obtained with &errno that happened immediately in the
   1618      intermediate development versions of OpenSSL; this is no longer the
   1619      case, the numeric thread callback will now override the automatic use
   1620      of &errno.)
   1621      [Geoff Thorpe, with help from Bodo Moeller]
   1622 
   1623   *) Initial support for different CRL issuing certificates. This covers a
   1624      simple case where the self issued certificates in the chain exist and
   1625      the real CRL issuer is higher in the existing chain.
   1626 
   1627      This work was sponsored by Google.
   1628      [Steve Henson]
   1629 
   1630   *) Removed effectively defunct crypto/store from the build.
   1631      [Ben Laurie]
   1632 
   1633   *) Revamp of STACK to provide stronger type-checking. Still to come:
   1634      TXT_DB, bsearch(?), OBJ_bsearch, qsort, CRYPTO_EX_DATA, ASN1_VALUE,
   1635      ASN1_STRING, CONF_VALUE.
   1636      [Ben Laurie]
   1637 
   1638   *) Add a new SSL_MODE_RELEASE_BUFFERS mode flag to release unused buffer
   1639      RAM on SSL connections.  This option can save about 34k per idle SSL.
   1640      [Nick Mathewson]
   1641 
   1642   *) Revamp of LHASH to provide stronger type-checking. Still to come:
   1643      STACK, TXT_DB, bsearch, qsort.
   1644      [Ben Laurie]
   1645 
   1646   *) Initial support for Cryptographic Message Syntax (aka CMS) based
   1647      on RFC3850, RFC3851 and RFC3852. New cms directory and cms utility,
   1648      support for data, signedData, compressedData, digestedData and
   1649      encryptedData, envelopedData types included. Scripts to check against
   1650      RFC4134 examples draft and interop and consistency checks of many
   1651      content types and variants.
   1652      [Steve Henson]
   1653 
   1654   *) Add options to enc utility to support use of zlib compression BIO.
   1655      [Steve Henson]
   1656 
   1657   *) Extend mk1mf to support importing of options and assembly language
   1658      files from Configure script, currently only included in VC-WIN32.
   1659      The assembly language rules can now optionally generate the source
   1660      files from the associated perl scripts.
   1661      [Steve Henson]
   1662 
   1663   *) Implement remaining functionality needed to support GOST ciphersuites.
   1664      Interop testing has been performed using CryptoPro implementations.
   1665      [Victor B. Wagner <vitus (a] cryptocom.ru>]
   1666 
   1667   *) s390x assembler pack.
   1668      [Andy Polyakov]
   1669 
   1670   *) ARMv4 assembler pack. ARMv4 refers to v4 and later ISA, not CPU
   1671      "family."
   1672      [Andy Polyakov]
   1673 
   1674   *) Implement Opaque PRF Input TLS extension as specified in
   1675      draft-rescorla-tls-opaque-prf-input-00.txt.  Since this is not an
   1676      official specification yet and no extension type assignment by
   1677      IANA exists, this extension (for now) will have to be explicitly
   1678      enabled when building OpenSSL by providing the extension number
   1679      to use.  For example, specify an option
   1680 
   1681          -DTLSEXT_TYPE_opaque_prf_input=0x9527
   1682 
   1683      to the "config" or "Configure" script to enable the extension,
   1684      assuming extension number 0x9527 (which is a completely arbitrary
   1685      and unofficial assignment based on the MD5 hash of the Internet
   1686      Draft).  Note that by doing so, you potentially lose
   1687      interoperability with other TLS implementations since these might
   1688      be using the same extension number for other purposes.
   1689 
   1690      SSL_set_tlsext_opaque_prf_input(ssl, src, len) is used to set the
   1691      opaque PRF input value to use in the handshake.  This will create
   1692      an interal copy of the length-'len' string at 'src', and will
   1693      return non-zero for success.
   1694 
   1695      To get more control and flexibility, provide a callback function
   1696      by using
   1697 
   1698           SSL_CTX_set_tlsext_opaque_prf_input_callback(ctx, cb)
   1699           SSL_CTX_set_tlsext_opaque_prf_input_callback_arg(ctx, arg)
   1700 
   1701      where
   1702 
   1703           int (*cb)(SSL *, void *peerinput, size_t len, void *arg);
   1704           void *arg;
   1705 
   1706      Callback function 'cb' will be called in handshakes, and is
   1707      expected to use SSL_set_tlsext_opaque_prf_input() as appropriate.
   1708      Argument 'arg' is for application purposes (the value as given to
   1709      SSL_CTX_set_tlsext_opaque_prf_input_callback_arg() will directly
   1710      be provided to the callback function).  The callback function
   1711      has to return non-zero to report success: usually 1 to use opaque
   1712      PRF input just if possible, or 2 to enforce use of the opaque PRF
   1713      input.  In the latter case, the library will abort the handshake
   1714      if opaque PRF input is not successfully negotiated.
   1715 
   1716      Arguments 'peerinput' and 'len' given to the callback function
   1717      will always be NULL and 0 in the case of a client.  A server will
   1718      see the client's opaque PRF input through these variables if
   1719      available (NULL and 0 otherwise).  Note that if the server
   1720      provides an opaque PRF input, the length must be the same as the
   1721      length of the client's opaque PRF input.
   1722 
   1723      Note that the callback function will only be called when creating
   1724      a new session (session resumption can resume whatever was
   1725      previously negotiated), and will not be called in SSL 2.0
   1726      handshakes; thus, SSL_CTX_set_options(ctx, SSL_OP_NO_SSLv2) or
   1727      SSL_set_options(ssl, SSL_OP_NO_SSLv2) is especially recommended
   1728      for applications that need to enforce opaque PRF input.
   1729 
   1730      [Bodo Moeller]
   1731 
   1732   *) Update ssl code to support digests other than SHA1+MD5 for handshake
   1733      MAC. 
   1734 
   1735      [Victor B. Wagner <vitus (a] cryptocom.ru>]
   1736 
   1737   *) Add RFC4507 support to OpenSSL. This includes the corrections in
   1738      RFC4507bis. The encrypted ticket format is an encrypted encoded
   1739      SSL_SESSION structure, that way new session features are automatically
   1740      supported.
   1741 
   1742      If a client application caches session in an SSL_SESSION structure
   1743      support is transparent because tickets are now stored in the encoded
   1744      SSL_SESSION.
   1745      
   1746      The SSL_CTX structure automatically generates keys for ticket
   1747      protection in servers so again support should be possible
   1748      with no application modification.
   1749 
   1750      If a client or server wishes to disable RFC4507 support then the option
   1751      SSL_OP_NO_TICKET can be set.
   1752 
   1753      Add a TLS extension debugging callback to allow the contents of any client
   1754      or server extensions to be examined.
   1755 
   1756      This work was sponsored by Google.
   1757      [Steve Henson]
   1758 
   1759   *) Final changes to avoid use of pointer pointer casts in OpenSSL.
   1760      OpenSSL should now compile cleanly on gcc 4.2
   1761      [Peter Hartley <pdh (a] utter.chaos.org.uk>, Steve Henson]
   1762 
   1763   *) Update SSL library to use new EVP_PKEY MAC API. Include generic MAC
   1764      support including streaming MAC support: this is required for GOST
   1765      ciphersuite support.
   1766      [Victor B. Wagner <vitus (a] cryptocom.ru>, Steve Henson]
   1767 
   1768   *) Add option -stream to use PKCS#7 streaming in smime utility. New
   1769      function i2d_PKCS7_bio_stream() and PEM_write_PKCS7_bio_stream()
   1770      to output in BER and PEM format.
   1771      [Steve Henson]
   1772 
   1773   *) Experimental support for use of HMAC via EVP_PKEY interface. This
   1774      allows HMAC to be handled via the EVP_DigestSign*() interface. The
   1775      EVP_PKEY "key" in this case is the HMAC key, potentially allowing
   1776      ENGINE support for HMAC keys which are unextractable. New -mac and
   1777      -macopt options to dgst utility.
   1778      [Steve Henson]
   1779 
   1780   *) New option -sigopt to dgst utility. Update dgst to use
   1781      EVP_Digest{Sign,Verify}*. These two changes make it possible to use
   1782      alternative signing paramaters such as X9.31 or PSS in the dgst 
   1783      utility.
   1784      [Steve Henson]
   1785 
   1786   *) Change ssl_cipher_apply_rule(), the internal function that does
   1787      the work each time a ciphersuite string requests enabling
   1788      ("foo+bar"), moving ("+foo+bar"), disabling ("-foo+bar", or
   1789      removing ("!foo+bar") a class of ciphersuites: Now it maintains
   1790      the order of disabled ciphersuites such that those ciphersuites
   1791      that most recently went from enabled to disabled not only stay
   1792      in order with respect to each other, but also have higher priority
   1793      than other disabled ciphersuites the next time ciphersuites are
   1794      enabled again.
   1795 
   1796      This means that you can now say, e.g., "PSK:-PSK:HIGH" to enable
   1797      the same ciphersuites as with "HIGH" alone, but in a specific
   1798      order where the PSK ciphersuites come first (since they are the
   1799      most recently disabled ciphersuites when "HIGH" is parsed).
   1800 
   1801      Also, change ssl_create_cipher_list() (using this new
   1802      funcionality) such that between otherwise identical
   1803      cihpersuites, ephemeral ECDH is preferred over ephemeral DH in
   1804      the default order.
   1805      [Bodo Moeller]
   1806 
   1807   *) Change ssl_create_cipher_list() so that it automatically
   1808      arranges the ciphersuites in reasonable order before starting
   1809      to process the rule string.  Thus, the definition for "DEFAULT"
   1810      (SSL_DEFAULT_CIPHER_LIST) now is just "ALL:!aNULL:!eNULL", but
   1811      remains equivalent to "AES:ALL:!aNULL:!eNULL:+aECDH:+kRSA:+RC4:@STRENGTH".
   1812      This makes it much easier to arrive at a reasonable default order
   1813      in applications for which anonymous ciphers are OK (meaning
   1814      that you can't actually use DEFAULT).
   1815      [Bodo Moeller; suggested by Victor Duchovni]
   1816 
   1817   *) Split the SSL/TLS algorithm mask (as used for ciphersuite string
   1818      processing) into multiple integers instead of setting
   1819      "SSL_MKEY_MASK" bits, "SSL_AUTH_MASK" bits, "SSL_ENC_MASK",
   1820      "SSL_MAC_MASK", and "SSL_SSL_MASK" bits all in a single integer.
   1821      (These masks as well as the individual bit definitions are hidden
   1822      away into the non-exported interface ssl/ssl_locl.h, so this
   1823      change to the definition of the SSL_CIPHER structure shouldn't
   1824      affect applications.)  This give us more bits for each of these
   1825      categories, so there is no longer a need to coagulate AES128 and
   1826      AES256 into a single algorithm bit, and to coagulate Camellia128
   1827      and Camellia256 into a single algorithm bit, which has led to all
   1828      kinds of kludges.
   1829 
   1830      Thus, among other things, the kludge introduced in 0.9.7m and
   1831      0.9.8e for masking out AES256 independently of AES128 or masking
   1832      out Camellia256 independently of AES256 is not needed here in 0.9.9.
   1833 
   1834      With the change, we also introduce new ciphersuite aliases that
   1835      so far were missing: "AES128", "AES256", "CAMELLIA128", and
   1836      "CAMELLIA256".
   1837      [Bodo Moeller]
   1838 
   1839   *) Add support for dsa-with-SHA224 and dsa-with-SHA256.
   1840      Use the leftmost N bytes of the signature input if the input is
   1841      larger than the prime q (with N being the size in bytes of q).
   1842      [Nils Larsch]
   1843 
   1844   *) Very *very* experimental PKCS#7 streaming encoder support. Nothing uses
   1845      it yet and it is largely untested.
   1846      [Steve Henson]
   1847 
   1848   *) Add support for the ecdsa-with-SHA224/256/384/512 signature types.
   1849      [Nils Larsch]
   1850 
   1851   *) Initial incomplete changes to avoid need for function casts in OpenSSL
   1852      some compilers (gcc 4.2 and later) reject their use. Safestack is
   1853      reimplemented.  Update ASN1 to avoid use of legacy functions. 
   1854      [Steve Henson]
   1855 
   1856   *) Win32/64 targets are linked with Winsock2.
   1857      [Andy Polyakov]
   1858 
   1859   *) Add an X509_CRL_METHOD structure to allow CRL processing to be redirected
   1860      to external functions. This can be used to increase CRL handling 
   1861      efficiency especially when CRLs are very large by (for example) storing
   1862      the CRL revoked certificates in a database.
   1863      [Steve Henson]
   1864 
   1865   *) Overhaul of by_dir code. Add support for dynamic loading of CRLs so
   1866      new CRLs added to a directory can be used. New command line option
   1867      -verify_return_error to s_client and s_server. This causes real errors
   1868      to be returned by the verify callback instead of carrying on no matter
   1869      what. This reflects the way a "real world" verify callback would behave.
   1870      [Steve Henson]
   1871 
   1872   *) GOST engine, supporting several GOST algorithms and public key formats.
   1873      Kindly donated by Cryptocom.
   1874      [Cryptocom]
   1875 
   1876   *) Partial support for Issuing Distribution Point CRL extension. CRLs
   1877      partitioned by DP are handled but no indirect CRL or reason partitioning
   1878      (yet). Complete overhaul of CRL handling: now the most suitable CRL is
   1879      selected via a scoring technique which handles IDP and AKID in CRLs.
   1880      [Steve Henson]
   1881 
   1882   *) New X509_STORE_CTX callbacks lookup_crls() and lookup_certs() which
   1883      will ultimately be used for all verify operations: this will remove the
   1884      X509_STORE dependency on certificate verification and allow alternative
   1885      lookup methods.  X509_STORE based implementations of these two callbacks.
   1886      [Steve Henson]
   1887 
   1888   *) Allow multiple CRLs to exist in an X509_STORE with matching issuer names.
   1889      Modify get_crl() to find a valid (unexpired) CRL if possible.
   1890      [Steve Henson]
   1891 
   1892   *) New function X509_CRL_match() to check if two CRLs are identical. Normally
   1893      this would be called X509_CRL_cmp() but that name is already used by
   1894      a function that just compares CRL issuer names. Cache several CRL 
   1895      extensions in X509_CRL structure and cache CRLDP in X509.
   1896      [Steve Henson]
   1897 
   1898   *) Store a "canonical" representation of X509_NAME structure (ASN1 Name)
   1899      this maps equivalent X509_NAME structures into a consistent structure.
   1900      Name comparison can then be performed rapidly using memcmp().
   1901      [Steve Henson]
   1902 
   1903   *) Non-blocking OCSP request processing. Add -timeout option to ocsp 
   1904      utility.
   1905      [Steve Henson]
   1906 
   1907   *) Allow digests to supply their own micalg string for S/MIME type using
   1908      the ctrl EVP_MD_CTRL_MICALG.
   1909      [Steve Henson]
   1910 
   1911   *) During PKCS7 signing pass the PKCS7 SignerInfo structure to the
   1912      EVP_PKEY_METHOD before and after signing via the EVP_PKEY_CTRL_PKCS7_SIGN
   1913      ctrl. It can then customise the structure before and/or after signing
   1914      if necessary.
   1915      [Steve Henson]
   1916 
   1917   *) New function OBJ_add_sigid() to allow application defined signature OIDs
   1918      to be added to OpenSSLs internal tables. New function OBJ_sigid_free()
   1919      to free up any added signature OIDs.
   1920      [Steve Henson]
   1921 
   1922   *) New functions EVP_CIPHER_do_all(), EVP_CIPHER_do_all_sorted(),
   1923      EVP_MD_do_all() and EVP_MD_do_all_sorted() to enumerate internal
   1924      digest and cipher tables. New options added to openssl utility:
   1925      list-message-digest-algorithms and list-cipher-algorithms.
   1926      [Steve Henson]
   1927 
   1928   *) Change the array representation of binary polynomials: the list
   1929      of degrees of non-zero coefficients is now terminated with -1.
   1930      Previously it was terminated with 0, which was also part of the
   1931      value; thus, the array representation was not applicable to
   1932      polynomials where t^0 has coefficient zero.  This change makes
   1933      the array representation useful in a more general context.
   1934      [Douglas Stebila]
   1935 
   1936   *) Various modifications and fixes to SSL/TLS cipher string
   1937      handling.  For ECC, the code now distinguishes between fixed ECDH
   1938      with RSA certificates on the one hand and with ECDSA certificates
   1939      on the other hand, since these are separate ciphersuites.  The
   1940      unused code for Fortezza ciphersuites has been removed.
   1941 
   1942      For consistency with EDH, ephemeral ECDH is now called "EECDH"
   1943      (not "ECDHE").  For consistency with the code for DH
   1944      certificates, use of ECDH certificates is now considered ECDH
   1945      authentication, not RSA or ECDSA authentication (the latter is
   1946      merely the CA's signing algorithm and not actively used in the
   1947      protocol).
   1948 
   1949      The temporary ciphersuite alias "ECCdraft" is no longer
   1950      available, and ECC ciphersuites are no longer excluded from "ALL"
   1951      and "DEFAULT".  The following aliases now exist for RFC 4492
   1952      ciphersuites, most of these by analogy with the DH case:
   1953 
   1954          kECDHr   - ECDH cert, signed with RSA
   1955          kECDHe   - ECDH cert, signed with ECDSA
   1956          kECDH    - ECDH cert (signed with either RSA or ECDSA)
   1957          kEECDH   - ephemeral ECDH
   1958          ECDH     - ECDH cert or ephemeral ECDH
   1959 
   1960          aECDH    - ECDH cert
   1961          aECDSA   - ECDSA cert
   1962          ECDSA    - ECDSA cert
   1963 
   1964          AECDH    - anonymous ECDH
   1965          EECDH    - non-anonymous ephemeral ECDH (equivalent to "kEECDH:-AECDH")
   1966 
   1967      [Bodo Moeller]
   1968 
   1969   *) Add additional S/MIME capabilities for AES and GOST ciphers if supported.
   1970      Use correct micalg parameters depending on digest(s) in signed message.
   1971      [Steve Henson]
   1972 
   1973   *) Add engine support for EVP_PKEY_ASN1_METHOD. Add functions to process
   1974      an ENGINE asn1 method. Support ENGINE lookups in the ASN1 code.
   1975      [Steve Henson]
   1976 
   1977   *) Initial engine support for EVP_PKEY_METHOD. New functions to permit
   1978      an engine to register a method. Add ENGINE lookups for methods and
   1979      functional reference processing.
   1980      [Steve Henson]
   1981 
   1982   *) New functions EVP_Digest{Sign,Verify)*. These are enchance versions of
   1983      EVP_{Sign,Verify}* which allow an application to customise the signature
   1984      process.
   1985      [Steve Henson]
   1986 
   1987   *) New -resign option to smime utility. This adds one or more signers
   1988      to an existing PKCS#7 signedData structure. Also -md option to use an
   1989      alternative message digest algorithm for signing.
   1990      [Steve Henson]
   1991 
   1992   *) Tidy up PKCS#7 routines and add new functions to make it easier to
   1993      create PKCS7 structures containing multiple signers. Update smime
   1994      application to support multiple signers.
   1995      [Steve Henson]
   1996 
   1997   *) New -macalg option to pkcs12 utility to allow setting of an alternative
   1998      digest MAC.
   1999      [Steve Henson]
   2000 
   2001   *) Initial support for PKCS#5 v2.0 PRFs other than default SHA1 HMAC.
   2002      Reorganize PBE internals to lookup from a static table using NIDs,
   2003      add support for HMAC PBE OID translation. Add a EVP_CIPHER ctrl:
   2004      EVP_CTRL_PBE_PRF_NID this allows a cipher to specify an alternative
   2005      PRF which will be automatically used with PBES2.
   2006      [Steve Henson]
   2007 
   2008   *) Replace the algorithm specific calls to generate keys in "req" with the
   2009      new API.
   2010      [Steve Henson]
   2011 
   2012   *) Update PKCS#7 enveloped data routines to use new API. This is now
   2013      supported by any public key method supporting the encrypt operation. A
   2014      ctrl is added to allow the public key algorithm to examine or modify
   2015      the PKCS#7 RecipientInfo structure if it needs to: for RSA this is
   2016      a no op.
   2017      [Steve Henson]
   2018 
   2019   *) Add a ctrl to asn1 method to allow a public key algorithm to express
   2020      a default digest type to use. In most cases this will be SHA1 but some
   2021      algorithms (such as GOST) need to specify an alternative digest. The
   2022      return value indicates how strong the prefernce is 1 means optional and
   2023      2 is mandatory (that is it is the only supported type). Modify
   2024      ASN1_item_sign() to accept a NULL digest argument to indicate it should
   2025      use the default md. Update openssl utilities to use the default digest
   2026      type for signing if it is not explicitly indicated.
   2027      [Steve Henson]
   2028 
   2029   *) Use OID cross reference table in ASN1_sign() and ASN1_verify(). New 
   2030      EVP_MD flag EVP_MD_FLAG_PKEY_METHOD_SIGNATURE. This uses the relevant
   2031      signing method from the key type. This effectively removes the link
   2032      between digests and public key types.
   2033      [Steve Henson]
   2034 
   2035   *) Add an OID cross reference table and utility functions. Its purpose is to
   2036      translate between signature OIDs such as SHA1WithrsaEncryption and SHA1,
   2037      rsaEncryption. This will allow some of the algorithm specific hackery
   2038      needed to use the correct OID to be removed. 
   2039      [Steve Henson]
   2040 
   2041   *) Remove algorithm specific dependencies when setting PKCS7_SIGNER_INFO
   2042      structures for PKCS7_sign(). They are now set up by the relevant public
   2043      key ASN1 method.
   2044      [Steve Henson]
   2045 
   2046   *) Add provisional EC pkey method with support for ECDSA and ECDH.
   2047      [Steve Henson]
   2048 
   2049   *) Add support for key derivation (agreement) in the API, DH method and
   2050      pkeyutl.
   2051      [Steve Henson]
   2052 
   2053   *) Add DSA pkey method and DH pkey methods, extend DH ASN1 method to support
   2054      public and private key formats. As a side effect these add additional 
   2055      command line functionality not previously available: DSA signatures can be
   2056      generated and verified using pkeyutl and DH key support and generation in
   2057      pkey, genpkey.
   2058      [Steve Henson]
   2059 
   2060   *) BeOS support.
   2061      [Oliver Tappe <zooey (a] hirschkaefer.de>]
   2062 
   2063   *) New make target "install_html_docs" installs HTML renditions of the
   2064      manual pages.
   2065      [Oliver Tappe <zooey (a] hirschkaefer.de>]
   2066 
   2067   *) New utility "genpkey" this is analagous to "genrsa" etc except it can
   2068      generate keys for any algorithm. Extend and update EVP_PKEY_METHOD to
   2069      support key and parameter generation and add initial key generation
   2070      functionality for RSA.
   2071      [Steve Henson]
   2072 
   2073   *) Add functions for main EVP_PKEY_method operations. The undocumented
   2074      functions EVP_PKEY_{encrypt,decrypt} have been renamed to
   2075      EVP_PKEY_{encrypt,decrypt}_old. 
   2076      [Steve Henson]
   2077 
   2078   *) Initial definitions for EVP_PKEY_METHOD. This will be a high level public
   2079      key API, doesn't do much yet.
   2080      [Steve Henson]
   2081 
   2082   *) New function EVP_PKEY_asn1_get0_info() to retrieve information about
   2083      public key algorithms. New option to openssl utility:
   2084      "list-public-key-algorithms" to print out info.
   2085      [Steve Henson]
   2086 
   2087   *) Implement the Supported Elliptic Curves Extension for
   2088      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
   2089      [Douglas Stebila]
   2090 
   2091   *) Don't free up OIDs in OBJ_cleanup() if they are in use by EVP_MD or
   2092      EVP_CIPHER structures to avoid later problems in EVP_cleanup().
   2093      [Steve Henson]
   2094 
   2095   *) New utilities pkey and pkeyparam. These are similar to algorithm specific
   2096      utilities such as rsa, dsa, dsaparam etc except they process any key
   2097      type.
   2098      [Steve Henson]
   2099 
   2100   *) Transfer public key printing routines to EVP_PKEY_ASN1_METHOD. New 
   2101      functions EVP_PKEY_print_public(), EVP_PKEY_print_private(),
   2102      EVP_PKEY_print_param() to print public key data from an EVP_PKEY
   2103      structure.
   2104      [Steve Henson]
   2105 
   2106   *) Initial support for pluggable public key ASN1.
   2107      De-spaghettify the public key ASN1 handling. Move public and private
   2108      key ASN1 handling to a new EVP_PKEY_ASN1_METHOD structure. Relocate
   2109      algorithm specific handling to a single module within the relevant
   2110      algorithm directory. Add functions to allow (near) opaque processing
   2111      of public and private key structures.
   2112      [Steve Henson]
   2113 
   2114   *) Implement the Supported Point Formats Extension for
   2115      ECC ciphersuites from draft-ietf-tls-ecc-12.txt.
   2116      [Douglas Stebila]
   2117 
   2118   *) Add initial support for RFC 4279 PSK TLS ciphersuites. Add members
   2119      for the psk identity [hint] and the psk callback functions to the
   2120      SSL_SESSION, SSL and SSL_CTX structure.
   2121      
   2122      New ciphersuites:
   2123          PSK-RC4-SHA, PSK-3DES-EDE-CBC-SHA, PSK-AES128-CBC-SHA,
   2124          PSK-AES256-CBC-SHA
   2125  
   2126      New functions:
   2127          SSL_CTX_use_psk_identity_hint
   2128          SSL_get_psk_identity_hint
   2129          SSL_get_psk_identity
   2130          SSL_use_psk_identity_hint
   2131 
   2132      [Mika Kousa and Pasi Eronen of Nokia Corporation]
   2133 
   2134   *) Add RFC 3161 compliant time stamp request creation, response generation
   2135      and response verification functionality.
   2136      [Zoltn Glzik <zglozik (a] opentsa.org>, The OpenTSA Project]
   2137 
   2138   *) Add initial support for TLS extensions, specifically for the server_name
   2139      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
   2140      have new members for a host name.  The SSL data structure has an
   2141      additional member SSL_CTX *initial_ctx so that new sessions can be
   2142      stored in that context to allow for session resumption, even after the
   2143      SSL has been switched to a new SSL_CTX in reaction to a client's
   2144      server_name extension.
   2145 
   2146      New functions (subject to change):
   2147 
   2148          SSL_get_servername()
   2149          SSL_get_servername_type()
   2150          SSL_set_SSL_CTX()
   2151 
   2152      New CTRL codes and macros (subject to change):
   2153 
   2154          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
   2155                                  - SSL_CTX_set_tlsext_servername_callback()
   2156          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
   2157                                       - SSL_CTX_set_tlsext_servername_arg()
   2158          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
   2159 
   2160      openssl s_client has a new '-servername ...' option.
   2161 
   2162      openssl s_server has new options '-servername_host ...', '-cert2 ...',
   2163      '-key2 ...', '-servername_fatal' (subject to change).  This allows
   2164      testing the HostName extension for a specific single host name ('-cert'
   2165      and '-key' remain fallbacks for handshakes without HostName
   2166      negotiation).  If the unrecogninzed_name alert has to be sent, this by
   2167      default is a warning; it becomes fatal with the '-servername_fatal'
   2168      option.
   2169 
   2170      [Peter Sylvester,  Remy Allais, Christophe Renou]
   2171 
   2172   *) Whirlpool hash implementation is added.
   2173      [Andy Polyakov]
   2174 
   2175   *) BIGNUM code on 64-bit SPARCv9 targets is switched from bn(64,64) to
   2176      bn(64,32). Because of instruction set limitations it doesn't have
   2177      any negative impact on performance. This was done mostly in order
   2178      to make it possible to share assembler modules, such as bn_mul_mont
   2179      implementations, between 32- and 64-bit builds without hassle.
   2180      [Andy Polyakov]
   2181 
   2182   *) Move code previously exiled into file crypto/ec/ec2_smpt.c
   2183      to ec2_smpl.c, and no longer require the OPENSSL_EC_BIN_PT_COMP
   2184      macro.
   2185      [Bodo Moeller]
   2186 
   2187   *) New candidate for BIGNUM assembler implementation, bn_mul_mont,
   2188      dedicated Montgomery multiplication procedure, is introduced.
   2189      BN_MONT_CTX is modified to allow bn_mul_mont to reach for higher
   2190      "64-bit" performance on certain 32-bit targets.
   2191      [Andy Polyakov]
   2192 
   2193   *) New option SSL_OP_NO_COMP to disable use of compression selectively
   2194      in SSL structures. New SSL ctrl to set maximum send fragment size. 
   2195      Save memory by seeting the I/O buffer sizes dynamically instead of
   2196      using the maximum available value.
   2197      [Steve Henson]
   2198 
   2199   *) New option -V for 'openssl ciphers'. This prints the ciphersuite code
   2200      in addition to the text details.
   2201      [Bodo Moeller]
   2202 
   2203   *) Very, very preliminary EXPERIMENTAL support for printing of general
   2204      ASN1 structures. This currently produces rather ugly output and doesn't
   2205      handle several customised structures at all.
   2206      [Steve Henson]
   2207 
   2208   *) Integrated support for PVK file format and some related formats such
   2209      as MS PUBLICKEYBLOB and PRIVATEKEYBLOB. Command line switches to support
   2210      these in the 'rsa' and 'dsa' utilities.
   2211      [Steve Henson]
   2212 
   2213   *) Support for PKCS#1 RSAPublicKey format on rsa utility command line.
   2214      [Steve Henson]
   2215 
   2216   *) Remove the ancient ASN1_METHOD code. This was only ever used in one
   2217      place for the (very old) "NETSCAPE" format certificates which are now
   2218      handled using new ASN1 code equivalents.
   2219      [Steve Henson]
   2220 
   2221   *) Let the TLSv1_method() etc. functions return a 'const' SSL_METHOD
   2222      pointer and make the SSL_METHOD parameter in SSL_CTX_new,
   2223      SSL_CTX_set_ssl_version and SSL_set_ssl_method 'const'.
   2224      [Nils Larsch]
   2225 
   2226   *) Modify CRL distribution points extension code to print out previously
   2227      unsupported fields. Enhance extension setting code to allow setting of
   2228      all fields.
   2229      [Steve Henson]
   2230 
   2231   *) Add print and set support for Issuing Distribution Point CRL extension.
   2232      [Steve Henson]
   2233 
   2234   *) Change 'Configure' script to enable Camellia by default.
   2235      [NTT]
   2236 
   2237  Changes between 0.9.8m and 0.9.8n [24 Mar 2010]
   2238 
   2239   *) When rejecting SSL/TLS records due to an incorrect version number, never
   2240      update s->server with a new major version number.  As of
   2241      - OpenSSL 0.9.8m if 'short' is a 16-bit type,
   2242      - OpenSSL 0.9.8f if 'short' is longer than 16 bits,
   2243      the previous behavior could result in a read attempt at NULL when
   2244      receiving specific incorrect SSL/TLS records once record payload
   2245      protection is active.  (CVE-2010-0740)
   2246      [Bodo Moeller, Adam Langley <agl (a] chromium.org>]
   2247 
   2248   *) Fix for CVE-2010-0433 where some kerberos enabled versions of OpenSSL 
   2249      could be crashed if the relevant tables were not present (e.g. chrooted).
   2250      [Tomas Hoger <thoger (a] redhat.com>]
   2251 
   2252  Changes between 0.9.8l and 0.9.8m [25 Feb 2010]
   2253 
   2254   *) Always check bn_wexpend() return values for failure.  (CVE-2009-3245)
   2255      [Martin Olsson, Neel Mehta]
   2256 
   2257   *) Fix X509_STORE locking: Every 'objs' access requires a lock (to
   2258      accommodate for stack sorting, always a write lock!).
   2259      [Bodo Moeller]
   2260 
   2261   *) On some versions of WIN32 Heap32Next is very slow. This can cause
   2262      excessive delays in the RAND_poll(): over a minute. As a workaround
   2263      include a time check in the inner Heap32Next loop too.
   2264      [Steve Henson]
   2265 
   2266   *) The code that handled flushing of data in SSL/TLS originally used the
   2267      BIO_CTRL_INFO ctrl to see if any data was pending first. This caused
   2268      the problem outlined in PR#1949. The fix suggested there however can
   2269      trigger problems with buggy BIO_CTRL_WPENDING (e.g. some versions
   2270      of Apache). So instead simplify the code to flush unconditionally.
   2271      This should be fine since flushing with no data to flush is a no op.
   2272      [Steve Henson]
   2273 
   2274   *) Handle TLS versions 2.0 and later properly and correctly use the
   2275      highest version of TLS/SSL supported. Although TLS >= 2.0 is some way
   2276      off ancient servers have a habit of sticking around for a while...
   2277      [Steve Henson]
   2278 
   2279   *) Modify compression code so it frees up structures without using the
   2280      ex_data callbacks. This works around a problem where some applications
   2281      call CRYPTO_cleanup_all_ex_data() before application exit (e.g. when
   2282      restarting) then use compression (e.g. SSL with compression) later.
   2283      This results in significant per-connection memory leaks and
   2284      has caused some security issues including CVE-2008-1678 and
   2285      CVE-2009-4355.
   2286      [Steve Henson]
   2287 
   2288   *) Constify crypto/cast (i.e., <openssl/cast.h>): a CAST_KEY doesn't
   2289      change when encrypting or decrypting.
   2290      [Bodo Moeller]
   2291 
   2292   *) Add option SSL_OP_LEGACY_SERVER_CONNECT which will allow clients to
   2293      connect and renegotiate with servers which do not support RI.
   2294      Until RI is more widely deployed this option is enabled by default.
   2295      [Steve Henson]
   2296 
   2297   *) Add "missing" ssl ctrls to clear options and mode.
   2298      [Steve Henson]
   2299 
   2300   *) If client attempts to renegotiate and doesn't support RI respond with
   2301      a no_renegotiation alert as required by RFC5746.  Some renegotiating
   2302      TLS clients will continue a connection gracefully when they receive
   2303      the alert. Unfortunately OpenSSL mishandled this alert and would hang
   2304      waiting for a server hello which it will never receive. Now we treat a
   2305      received no_renegotiation alert as a fatal error. This is because
   2306      applications requesting a renegotiation might well expect it to succeed
   2307      and would have no code in place to handle the server denying it so the
   2308      only safe thing to do is to terminate the connection.
   2309      [Steve Henson]
   2310 
   2311   *) Add ctrl macro SSL_get_secure_renegotiation_support() which returns 1 if
   2312      peer supports secure renegotiation and 0 otherwise. Print out peer
   2313      renegotiation support in s_client/s_server.
   2314      [Steve Henson]
   2315 
   2316   *) Replace the highly broken and deprecated SPKAC certification method with
   2317      the updated NID creation version. This should correctly handle UTF8.
   2318      [Steve Henson]
   2319 
   2320   *) Implement RFC5746. Re-enable renegotiation but require the extension
   2321      as needed. Unfortunately, SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION
   2322      turns out to be a bad idea. It has been replaced by
   2323      SSL_OP_ALLOW_UNSAFE_LEGACY_RENEGOTIATION which can be set with
   2324      SSL_CTX_set_options(). This is really not recommended unless you
   2325      know what you are doing.
   2326      [Eric Rescorla <ekr (a] networkresonance.com>, Ben Laurie, Steve Henson]
   2327 
   2328   *) Fixes to stateless session resumption handling. Use initial_ctx when
   2329      issuing and attempting to decrypt tickets in case it has changed during
   2330      servername handling. Use a non-zero length session ID when attempting
   2331      stateless session resumption: this makes it possible to determine if
   2332      a resumption has occurred immediately after receiving server hello
   2333      (several places in OpenSSL subtly assume this) instead of later in
   2334      the handshake.
   2335      [Steve Henson]
   2336 
   2337   *) The functions ENGINE_ctrl(), OPENSSL_isservice(),
   2338      CMS_get1_RecipientRequest() and RAND_bytes() can return <=0 on error
   2339      fixes for a few places where the return code is not checked
   2340      correctly.
   2341      [Julia Lawall <julia (a] diku.dk>]
   2342 
   2343   *) Add --strict-warnings option to Configure script to include devteam
   2344      warnings in other configurations.
   2345      [Steve Henson]
   2346 
   2347   *) Add support for --libdir option and LIBDIR variable in makefiles. This
   2348      makes it possible to install openssl libraries in locations which
   2349      have names other than "lib", for example "/usr/lib64" which some
   2350      systems need.
   2351      [Steve Henson, based on patch from Jeremy Utley]
   2352 
   2353   *) Don't allow the use of leading 0x80 in OIDs. This is a violation of
   2354      X690 8.9.12 and can produce some misleading textual output of OIDs.
   2355      [Steve Henson, reported by Dan Kaminsky]
   2356 
   2357   *) Delete MD2 from algorithm tables. This follows the recommendation in
   2358      several standards that it is not used in new applications due to
   2359      several cryptographic weaknesses. For binary compatibility reasons
   2360      the MD2 API is still compiled in by default.
   2361      [Steve Henson]
   2362 
   2363   *) Add compression id to {d2i,i2d}_SSL_SESSION so it is correctly saved
   2364      and restored.
   2365      [Steve Henson]
   2366 
   2367   *) Rename uni2asc and asc2uni functions to OPENSSL_uni2asc and
   2368      OPENSSL_asc2uni conditionally on Netware platforms to avoid a name
   2369      clash.
   2370      [Guenter <lists (a] gknw.net>]
   2371 
   2372   *) Fix the server certificate chain building code to use X509_verify_cert(),
   2373      it used to have an ad-hoc builder which was unable to cope with anything
   2374      other than a simple chain.
   2375      [David Woodhouse <dwmw2 (a] infradead.org>, Steve Henson]
   2376 
   2377   *) Don't check self signed certificate signatures in X509_verify_cert()
   2378      by default (a flag can override this): it just wastes time without
   2379      adding any security. As a useful side effect self signed root CAs
   2380      with non-FIPS digests are now usable in FIPS mode.
   2381      [Steve Henson]
   2382 
   2383   *) In dtls1_process_out_of_seq_message() the check if the current message
   2384      is already buffered was missing. For every new message was memory
   2385      allocated, allowing an attacker to perform an denial of service attack
   2386      with sending out of seq handshake messages until there is no memory
   2387      left. Additionally every future messege was buffered, even if the
   2388      sequence number made no sense and would be part of another handshake.
   2389      So only messages with sequence numbers less than 10 in advance will be
   2390      buffered.  (CVE-2009-1378)
   2391      [Robin Seggelmann, discovered by Daniel Mentz] 	
   2392 
   2393   *) Records are buffered if they arrive with a future epoch to be
   2394      processed after finishing the corresponding handshake. There is
   2395      currently no limitation to this buffer allowing an attacker to perform
   2396      a DOS attack with sending records with future epochs until there is no
   2397      memory left. This patch adds the pqueue_size() function to detemine
   2398      the size of a buffer and limits the record buffer to 100 entries.
   2399      (CVE-2009-1377)
   2400      [Robin Seggelmann, discovered by Daniel Mentz] 	
   2401 
   2402   *) Keep a copy of frag->msg_header.frag_len so it can be used after the
   2403      parent structure is freed.  (CVE-2009-1379)
   2404      [Daniel Mentz] 	
   2405 
   2406   *) Handle non-blocking I/O properly in SSL_shutdown() call.
   2407      [Darryl Miles <darryl-mailinglists (a] netbauds.net>]
   2408 
   2409   *) Add 2.5.4.* OIDs
   2410      [Ilya O. <vrghost (a] gmail.com>]
   2411 
   2412  Changes between 0.9.8k and 0.9.8l  [5 Nov 2009]
   2413 
   2414   *) Disable renegotiation completely - this fixes a severe security
   2415      problem (CVE-2009-3555) at the cost of breaking all
   2416      renegotiation. Renegotiation can be re-enabled by setting
   2417      SSL3_FLAGS_ALLOW_UNSAFE_LEGACY_RENEGOTIATION in s3->flags at
   2418      run-time. This is really not recommended unless you know what
   2419      you're doing.
   2420      [Ben Laurie]
   2421 
   2422  Changes between 0.9.8j and 0.9.8k  [25 Mar 2009]
   2423 
   2424   *) Don't set val to NULL when freeing up structures, it is freed up by
   2425      underlying code. If sizeof(void *) > sizeof(long) this can result in
   2426      zeroing past the valid field. (CVE-2009-0789)
   2427      [Paolo Ganci <Paolo.Ganci (a] AdNovum.CH>]
   2428 
   2429   *) Fix bug where return value of CMS_SignerInfo_verify_content() was not
   2430      checked correctly. This would allow some invalid signed attributes to
   2431      appear to verify correctly. (CVE-2009-0591)
   2432      [Ivan Nestlerode <inestlerode (a] us.ibm.com>]
   2433 
   2434   *) Reject UniversalString and BMPString types with invalid lengths. This
   2435      prevents a crash in ASN1_STRING_print_ex() which assumes the strings have
   2436      a legal length. (CVE-2009-0590)
   2437      [Steve Henson]
   2438 
   2439   *) Set S/MIME signing as the default purpose rather than setting it 
   2440      unconditionally. This allows applications to override it at the store
   2441      level.
   2442      [Steve Henson]
   2443 
   2444   *) Permit restricted recursion of ASN1 strings. This is needed in practice
   2445      to handle some structures.
   2446      [Steve Henson]
   2447 
   2448   *) Improve efficiency of mem_gets: don't search whole buffer each time
   2449      for a '\n'
   2450      [Jeremy Shapiro <jnshapir (a] us.ibm.com>]
   2451 
   2452   *) New -hex option for openssl rand.
   2453      [Matthieu Herrb]
   2454 
   2455   *) Print out UTF8String and NumericString when parsing ASN1.
   2456      [Steve Henson]
   2457 
   2458   *) Support NumericString type for name components.
   2459      [Steve Henson]
   2460 
   2461   *) Allow CC in the environment to override the automatically chosen
   2462      compiler. Note that nothing is done to ensure flags work with the
   2463      chosen compiler.
   2464      [Ben Laurie]
   2465 
   2466  Changes between 0.9.8i and 0.9.8j  [07 Jan 2009]
   2467 
   2468   *) Properly check EVP_VerifyFinal() and similar return values
   2469      (CVE-2008-5077).
   2470      [Ben Laurie, Bodo Moeller, Google Security Team]
   2471 
   2472   *) Enable TLS extensions by default.
   2473      [Ben Laurie]
   2474 
   2475   *) Allow the CHIL engine to be loaded, whether the application is
   2476      multithreaded or not. (This does not release the developer from the
   2477      obligation to set up the dynamic locking callbacks.)
   2478      [Sander Temme <sander (a] temme.net>]
   2479 
   2480   *) Use correct exit code if there is an error in dgst command.
   2481      [Steve Henson; problem pointed out by Roland Dirlewanger]
   2482 
   2483   *) Tweak Configure so that you need to say "experimental-jpake" to enable
   2484      JPAKE, and need to use -DOPENSSL_EXPERIMENTAL_JPAKE in applications.
   2485      [Bodo Moeller]
   2486 
   2487   *) Add experimental JPAKE support, including demo authentication in
   2488      s_client and s_server.
   2489      [Ben Laurie]
   2490 
   2491   *) Set the comparison function in v3_addr_canonize().
   2492      [Rob Austein <sra (a] hactrn.net>]
   2493 
   2494   *) Add support for XMPP STARTTLS in s_client.
   2495      [Philip Paeps <philip (a] freebsd.org>]
   2496 
   2497   *) Change the server-side SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG behavior
   2498      to ensure that even with this option, only ciphersuites in the
   2499      server's preference list will be accepted.  (Note that the option
   2500      applies only when resuming a session, so the earlier behavior was
   2501      just about the algorithm choice for symmetric cryptography.)
   2502      [Bodo Moeller]
   2503 
   2504  Changes between 0.9.8h and 0.9.8i  [15 Sep 2008]
   2505 
   2506   *) Fix NULL pointer dereference if a DTLS server received
   2507      ChangeCipherSpec as first record (CVE-2009-1386).
   2508      [PR #1679]
   2509 
   2510   *) Fix a state transitition in s3_srvr.c and d1_srvr.c
   2511      (was using SSL3_ST_CW_CLNT_HELLO_B, should be ..._ST_SW_SRVR_...).
   2512      [Nagendra Modadugu]
   2513 
   2514   *) The fix in 0.9.8c that supposedly got rid of unsafe
   2515      double-checked locking was incomplete for RSA blinding,
   2516      addressing just one layer of what turns out to have been
   2517      doubly unsafe triple-checked locking.
   2518 
   2519      So now fix this for real by retiring the MONT_HELPER macro
   2520      in crypto/rsa/rsa_eay.c.
   2521 
   2522      [Bodo Moeller; problem pointed out by Marius Schilder]
   2523 
   2524   *) Various precautionary measures:
   2525 
   2526      - Avoid size_t integer overflow in HASH_UPDATE (md32_common.h).
   2527 
   2528      - Avoid a buffer overflow in d2i_SSL_SESSION() (ssl_asn1.c).
   2529        (NB: This would require knowledge of the secret session ticket key
   2530        to exploit, in which case you'd be SOL either way.)
   2531 
   2532      - Change bn_nist.c so that it will properly handle input BIGNUMs
   2533        outside the expected range.
   2534 
   2535      - Enforce the 'num' check in BN_div() (bn_div.c) for non-BN_DEBUG
   2536        builds.
   2537 
   2538      [Neel Mehta, Bodo Moeller]
   2539 
   2540   *) Allow engines to be "soft loaded" - i.e. optionally don't die if
   2541      the load fails. Useful for distros.
   2542      [Ben Laurie and the FreeBSD team]
   2543 
   2544   *) Add support for Local Machine Keyset attribute in PKCS#12 files.
   2545      [Steve Henson]
   2546 
   2547   *) Fix BN_GF2m_mod_arr() top-bit cleanup code.
   2548      [Huang Ying]
   2549 
   2550   *) Expand ENGINE to support engine supplied SSL client certificate functions.
   2551 
   2552      This work was sponsored by Logica.
   2553      [Steve Henson]
   2554 
   2555   *) Add CryptoAPI ENGINE to support use of RSA and DSA keys held in Windows
   2556      keystores. Support for SSL/TLS client authentication too.
   2557      Not compiled unless enable-capieng specified to Configure.
   2558 
   2559      This work was sponsored by Logica.
   2560      [Steve Henson]
   2561 
   2562   *) Fix bug in X509_ATTRIBUTE creation: dont set attribute using
   2563      ASN1_TYPE_set1 if MBSTRING flag set. This bug would crash certain
   2564      attribute creation routines such as certifcate requests and PKCS#12
   2565      files.
   2566      [Steve Henson]
   2567 
   2568  Changes between 0.9.8g and 0.9.8h  [28 May 2008]
   2569 
   2570   *) Fix flaw if 'Server Key exchange message' is omitted from a TLS
   2571      handshake which could lead to a cilent crash as found using the
   2572      Codenomicon TLS test suite (CVE-2008-1672) 
   2573      [Steve Henson, Mark Cox]
   2574 
   2575   *) Fix double free in TLS server name extensions which could lead to
   2576      a remote crash found by Codenomicon TLS test suite (CVE-2008-0891) 
   2577      [Joe Orton]
   2578 
   2579   *) Clear error queue in SSL_CTX_use_certificate_chain_file()
   2580 
   2581      Clear the error queue to ensure that error entries left from
   2582      older function calls do not interfere with the correct operation.
   2583      [Lutz Jaenicke, Erik de Castro Lopo]
   2584 
   2585   *) Remove root CA certificates of commercial CAs:
   2586 
   2587      The OpenSSL project does not recommend any specific CA and does not
   2588      have any policy with respect to including or excluding any CA.
   2589      Therefore it does not make any sense to ship an arbitrary selection
   2590      of root CA certificates with the OpenSSL software.
   2591      [Lutz Jaenicke]
   2592 
   2593   *) RSA OAEP patches to fix two separate invalid memory reads.
   2594      The first one involves inputs when 'lzero' is greater than
   2595      'SHA_DIGEST_LENGTH' (it would read about SHA_DIGEST_LENGTH bytes
   2596      before the beginning of from). The second one involves inputs where
   2597      the 'db' section contains nothing but zeroes (there is a one-byte
   2598      invalid read after the end of 'db').
   2599      [Ivan Nestlerode <inestlerode (a] us.ibm.com>]
   2600 
   2601   *) Partial backport from 0.9.9-dev:
   2602 
   2603      Introduce bn_mul_mont (dedicated Montgomery multiplication
   2604      procedure) as a candidate for BIGNUM assembler implementation.
   2605      While 0.9.9-dev uses assembler for various architectures, only
   2606      x86_64 is available by default here in the 0.9.8 branch, and
   2607      32-bit x86 is available through a compile-time setting.
   2608 
   2609      To try the 32-bit x86 assembler implementation, use Configure
   2610      option "enable-montasm" (which exists only for this backport).
   2611 
   2612      As "enable-montasm" for 32-bit x86 disclaims code stability
   2613      anyway, in this constellation we activate additional code
   2614      backported from 0.9.9-dev for further performance improvements,
   2615      namely BN_from_montgomery_word.  (To enable this otherwise,
   2616      e.g. x86_64, try "-DMONT_FROM_WORD___NON_DEFAULT_0_9_8_BUILD".)
   2617 
   2618      [Andy Polyakov (backport partially by Bodo Moeller)]
   2619 
   2620   *) Add TLS session ticket callback. This allows an application to set
   2621      TLS ticket cipher and HMAC keys rather than relying on hardcoded fixed
   2622      values. This is useful for key rollover for example where several key
   2623      sets may exist with different names.
   2624      [Steve Henson]
   2625 
   2626   *) Reverse ENGINE-internal logic for caching default ENGINE handles.
   2627      This was broken until now in 0.9.8 releases, such that the only way
   2628      a registered ENGINE could be used (assuming it initialises
   2629      successfully on the host) was to explicitly set it as the default
   2630      for the relevant algorithms. This is in contradiction with 0.9.7
   2631      behaviour and the documentation. With this fix, when an ENGINE is
   2632      registered into a given algorithm's table of implementations, the
   2633      'uptodate' flag is reset so that auto-discovery will be used next
   2634      time a new context for that algorithm attempts to select an
   2635      implementation.
   2636      [Ian Lister (tweaked by Geoff Thorpe)]
   2637 
   2638   *) Backport of CMS code to OpenSSL 0.9.8. This differs from the 0.9.9
   2639      implemention in the following ways:
   2640 
   2641      Lack of EVP_PKEY_ASN1_METHOD means algorithm parameters have to be
   2642      hard coded.
   2643 
   2644      Lack of BER streaming support means one pass streaming processing is
   2645      only supported if data is detached: setting the streaming flag is
   2646      ignored for embedded content.
   2647 
   2648      CMS support is disabled by default and must be explicitly enabled
   2649      with the enable-cms configuration option.
   2650      [Steve Henson]
   2651 
   2652   *) Update the GMP engine glue to do direct copies between BIGNUM and
   2653      mpz_t when openssl and GMP use the same limb size. Otherwise the
   2654      existing "conversion via a text string export" trick is still used.
   2655      [Paul Sheer <paulsheer (a] gmail.com>]
   2656 
   2657   *) Zlib compression BIO. This is a filter BIO which compressed and
   2658      uncompresses any data passed through it.
   2659      [Steve Henson]
   2660 
   2661   *) Add AES_wrap_key() and AES_unwrap_key() functions to implement
   2662      RFC3394 compatible AES key wrapping.
   2663      [Steve Henson]
   2664 
   2665   *) Add utility functions to handle ASN1 structures. ASN1_STRING_set0():
   2666      sets string data without copying. X509_ALGOR_set0() and
   2667      X509_ALGOR_get0(): set and retrieve X509_ALGOR (AlgorithmIdentifier)
   2668      data. Attribute function X509at_get0_data_by_OBJ(): retrieves data
   2669      from an X509_ATTRIBUTE structure optionally checking it occurs only
   2670      once. ASN1_TYPE_set1(): set and ASN1_TYPE structure copying supplied
   2671      data.
   2672      [Steve Henson]
   2673 
   2674   *) Fix BN flag handling in RSA_eay_mod_exp() and BN_MONT_CTX_set()
   2675      to get the expected BN_FLG_CONSTTIME behavior.
   2676      [Bodo Moeller (Google)]
   2677   
   2678   *) Netware support:
   2679 
   2680      - fixed wrong usage of ioctlsocket() when build for LIBC BSD sockets
   2681      - fixed do_tests.pl to run the test suite with CLIB builds too (CLIB_OPT)
   2682      - added some more tests to do_tests.pl
   2683      - fixed RunningProcess usage so that it works with newer LIBC NDKs too
   2684      - removed usage of BN_LLONG for CLIB builds to avoid runtime dependency
   2685      - added new Configure targets netware-clib-bsdsock, netware-clib-gcc,
   2686        netware-clib-bsdsock-gcc, netware-libc-bsdsock-gcc
   2687      - various changes to netware.pl to enable gcc-cross builds on Win32
   2688        platform
   2689      - changed crypto/bio/b_sock.c to work with macro functions (CLIB BSD)
   2690      - various changes to fix missing prototype warnings
   2691      - fixed x86nasm.pl to create correct asm files for NASM COFF output
   2692      - added AES, WHIRLPOOL and CPUID assembler code to build files
   2693      - added missing AES assembler make rules to mk1mf.pl
   2694      - fixed order of includes in apps/ocsp.c so that e_os.h settings apply
   2695      [Guenter Knauf <eflash (a] gmx.net>]
   2696 
   2697   *) Implement certificate status request TLS extension defined in RFC3546.
   2698      A client can set the appropriate parameters and receive the encoded
   2699      OCSP response via a callback. A server can query the supplied parameters
   2700      and set the encoded OCSP response in the callback. Add simplified examples
   2701      to s_client and s_server.
   2702      [Steve Henson]
   2703 
   2704  Changes between 0.9.8f and 0.9.8g  [19 Oct 2007]
   2705 
   2706   *) Fix various bugs:
   2707      + Binary incompatibility of ssl_ctx_st structure
   2708      + DTLS interoperation with non-compliant servers
   2709      + Don't call get_session_cb() without proposed session
   2710      + Fix ia64 assembler code
   2711      [Andy Polyakov, Steve Henson]
   2712 
   2713  Changes between 0.9.8e and 0.9.8f  [11 Oct 2007]
   2714 
   2715   *) DTLS Handshake overhaul. There were longstanding issues with
   2716      OpenSSL DTLS implementation, which were making it impossible for
   2717      RFC 4347 compliant client to communicate with OpenSSL server.
   2718      Unfortunately just fixing these incompatibilities would "cut off"
   2719      pre-0.9.8f clients. To allow for hassle free upgrade post-0.9.8e
   2720      server keeps tolerating non RFC compliant syntax. The opposite is
   2721      not true, 0.9.8f client can not communicate with earlier server.
   2722      This update even addresses CVE-2007-4995.
   2723      [Andy Polyakov]
   2724 
   2725   *) Changes to avoid need for function casts in OpenSSL: some compilers
   2726      (gcc 4.2 and later) reject their use.
   2727      [Kurt Roeckx <kurt (a] roeckx.be>, Peter Hartley <pdh (a] utter.chaos.org.uk>,
   2728       Steve Henson]
   2729   
   2730   *) Add RFC4507 support to OpenSSL. This includes the corrections in
   2731      RFC4507bis. The encrypted ticket format is an encrypted encoded
   2732      SSL_SESSION structure, that way new session features are automatically
   2733      supported.
   2734 
   2735      If a client application caches session in an SSL_SESSION structure
   2736      support is transparent because tickets are now stored in the encoded
   2737      SSL_SESSION.
   2738      
   2739      The SSL_CTX structure automatically generates keys for ticket
   2740      protection in servers so again support should be possible
   2741      with no application modification.
   2742 
   2743      If a client or server wishes to disable RFC4507 support then the option
   2744      SSL_OP_NO_TICKET can be set.
   2745 
   2746      Add a TLS extension debugging callback to allow the contents of any client
   2747      or server extensions to be examined.
   2748 
   2749      This work was sponsored by Google.
   2750      [Steve Henson]
   2751 
   2752   *) Add initial support for TLS extensions, specifically for the server_name
   2753      extension so far.  The SSL_SESSION, SSL_CTX, and SSL data structures now
   2754      have new members for a host name.  The SSL data structure has an
   2755      additional member SSL_CTX *initial_ctx so that new sessions can be
   2756      stored in that context to allow for session resumption, even after the
   2757      SSL has been switched to a new SSL_CTX in reaction to a client's
   2758      server_name extension.
   2759 
   2760      New functions (subject to change):
   2761 
   2762          SSL_get_servername()
   2763          SSL_get_servername_type()
   2764          SSL_set_SSL_CTX()
   2765 
   2766      New CTRL codes and macros (subject to change):
   2767 
   2768          SSL_CTRL_SET_TLSEXT_SERVERNAME_CB
   2769                                  - SSL_CTX_set_tlsext_servername_callback()
   2770          SSL_CTRL_SET_TLSEXT_SERVERNAME_ARG
   2771                                       - SSL_CTX_set_tlsext_servername_arg()
   2772          SSL_CTRL_SET_TLSEXT_HOSTNAME           - SSL_set_tlsext_host_name()
   2773 
   2774      openssl s_client has a new '-servername ...' option.
   2775 
   2776      openssl s_server has new options '-servername_host ...', '-cert2 ...',
   2777      '-key2 ...', '-servername_fatal' (subject to change).  This allows
   2778      testing the HostName extension for a specific single host name ('-cert'
   2779      and '-key' remain fallbacks for handshakes without HostName
   2780      negotiation).  If the unrecogninzed_name alert has to be sent, this by
   2781      default is a warning; it becomes fatal with the '-servername_fatal'
   2782      option.
   2783 
   2784      [Peter Sylvester,  Remy Allais, Christophe Renou, Steve Henson]
   2785 
   2786   *) Add AES and SSE2 assembly language support to VC++ build.
   2787      [Steve Henson]
   2788 
   2789   *) Mitigate attack on final subtraction in Montgomery reduction.
   2790      [Andy Polyakov]
   2791 
   2792   *) Fix crypto/ec/ec_mult.c to work properly with scalars of value 0
   2793      (which previously caused an internal error).
   2794      [Bodo Moeller]
   2795 
   2796   *) Squeeze another 10% out of IGE mode when in != out.
   2797      [Ben Laurie]
   2798 
   2799   *) AES IGE mode speedup.
   2800      [Dean Gaudet (Google)]
   2801 
   2802   *) Add the Korean symmetric 128-bit cipher SEED (see
   2803      http://www.kisa.or.kr/kisa/seed/jsp/seed_eng.jsp) and
   2804      add SEED ciphersuites from RFC 4162:
   2805 
   2806         TLS_RSA_WITH_SEED_CBC_SHA      =  "SEED-SHA"
   2807         TLS_DHE_DSS_WITH_SEED_CBC_SHA  =  "DHE-DSS-SEED-SHA"
   2808         TLS_DHE_RSA_WITH_SEED_CBC_SHA  =  "DHE-RSA-SEED-SHA"
   2809         TLS_DH_anon_WITH_SEED_CBC_SHA  =  "ADH-SEED-SHA"
   2810 
   2811      To minimize changes between patchlevels in the OpenSSL 0.9.8
   2812      series, SEED remains excluded from compilation unless OpenSSL
   2813      is configured with 'enable-seed'.
   2814      [KISA, Bodo Moeller]
   2815 
   2816   *) Mitigate branch prediction attacks, which can be practical if a
   2817      single processor is shared, allowing a spy process to extract
   2818      information.  For detailed background information, see
   2819      http://eprint.iacr.org/2007/039 (O. Aciicmez, S. Gueron,
   2820      J.-P. Seifert, "New Branch Prediction Vulnerabilities in OpenSSL
   2821      and Necessary Software Countermeasures").  The core of the change
   2822      are new versions BN_div_no_branch() and
   2823      BN_mod_inverse_no_branch() of BN_div() and BN_mod_inverse(),
   2824      respectively, which are slower, but avoid the security-relevant
   2825      conditional branches.  These are automatically called by BN_div()
   2826      and BN_mod_inverse() if the flag BN_FLG_CONSTTIME is set for one
   2827      of the input BIGNUMs.  Also, BN_is_bit_set() has been changed to
   2828      remove a conditional branch.
   2829 
   2830      BN_FLG_CONSTTIME is the new name for the previous
   2831      BN_FLG_EXP_CONSTTIME flag, since it now affects more than just
   2832      modular exponentiation.  (Since OpenSSL 0.9.7h, setting this flag
   2833      in the exponent causes BN_mod_exp_mont() to use the alternative
   2834      implementation in BN_mod_exp_mont_consttime().)  The old name
   2835      remains as a deprecated alias.
   2836 
   2837      Similary, RSA_FLAG_NO_EXP_CONSTTIME is replaced by a more general
   2838      RSA_FLAG_NO_CONSTTIME flag since the RSA implementation now uses
   2839      constant-time implementations for more than just exponentiation.
   2840      Here too the old name is kept as a deprecated alias.
   2841 
   2842      BN_BLINDING_new() will now use BN_dup() for the modulus so that
   2843      the BN_BLINDING structure gets an independent copy of the
   2844      modulus.  This means that the previous "BIGNUM *m" argument to
   2845      BN_BLINDING_new() and to BN_BLINDING_create_param() now
   2846      essentially becomes "const BIGNUM *m", although we can't actually
   2847      change this in the header file before 0.9.9.  It allows
   2848      RSA_setup_blinding() to use BN_with_flags() on the modulus to
   2849      enable BN_FLG_CONSTTIME.
   2850 
   2851      [Matthew D Wood (Intel Corp)]
   2852 
   2853   *) In the SSL/TLS server implementation, be strict about session ID
   2854      context matching (which matters if an application uses a single
   2855      external cache for different purposes).  Previously,
   2856      out-of-context reuse was forbidden only if SSL_VERIFY_PEER was
   2857      set.  This did ensure strict client verification, but meant that,
   2858      with applications using a single external cache for quite
   2859      different requirements, clients could circumvent ciphersuite
   2860      restrictions for a given session ID context by starting a session
   2861      in a different context.
   2862      [Bodo Moeller]
   2863 
   2864   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
   2865      a ciphersuite string such as "DEFAULT:RSA" cannot enable
   2866      authentication-only ciphersuites.
   2867      [Bodo Moeller]
   2868 
   2869   *) Update the SSL_get_shared_ciphers() fix CVE-2006-3738 which was
   2870      not complete and could lead to a possible single byte overflow
   2871      (CVE-2007-5135) [Ben Laurie]
   2872 
   2873  Changes between 0.9.8d and 0.9.8e  [23 Feb 2007]
   2874 
   2875   *) Since AES128 and AES256 (and similarly Camellia128 and
   2876      Camellia256) share a single mask bit in the logic of
   2877      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
   2878      kludge to work properly if AES128 is available and AES256 isn't
   2879      (or if Camellia128 is available and Camellia256 isn't).
   2880      [Victor Duchovni]
   2881 
   2882   *) Fix the BIT STRING encoding generated by crypto/ec/ec_asn1.c
   2883      (within i2d_ECPrivateKey, i2d_ECPKParameters, i2d_ECParameters):
   2884      When a point or a seed is encoded in a BIT STRING, we need to
   2885      prevent the removal of trailing zero bits to get the proper DER
   2886      encoding.  (By default, crypto/asn1/a_bitstr.c assumes the case
   2887      of a NamedBitList, for which trailing 0 bits need to be removed.)
   2888      [Bodo Moeller]
   2889 
   2890   *) Have SSL/TLS server implementation tolerate "mismatched" record
   2891      protocol version while receiving ClientHello even if the
   2892      ClientHello is fragmented.  (The server can't insist on the
   2893      particular protocol version it has chosen before the ServerHello
   2894      message has informed the client about his choice.)
   2895      [Bodo Moeller]
   2896 
   2897   *) Add RFC 3779 support.
   2898      [Rob Austein for ARIN, Ben Laurie]
   2899 
   2900   *) Load error codes if they are not already present instead of using a
   2901      static variable. This allows them to be cleanly unloaded and reloaded.
   2902      Improve header file function name parsing.
   2903      [Steve Henson]
   2904 
   2905   *) extend SMTP and IMAP protocol emulation in s_client to use EHLO
   2906      or CAPABILITY handshake as required by RFCs.
   2907      [Goetz Babin-Ebell]
   2908 
   2909  Changes between 0.9.8c and 0.9.8d  [28 Sep 2006]
   2910 
   2911   *) Introduce limits to prevent malicious keys being able to
   2912      cause a denial of service.  (CVE-2006-2940)
   2913      [Steve Henson, Bodo Moeller]
   2914 
   2915   *) Fix ASN.1 parsing of certain invalid structures that can result
   2916      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
   2917 
   2918   *) Fix buffer overflow in SSL_get_shared_ciphers() function. 
   2919      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
   2920 
   2921   *) Fix SSL client code which could crash if connecting to a
   2922      malicious SSLv2 server.  (CVE-2006-4343)
   2923      [Tavis Ormandy and Will Drewry, Google Security Team]
   2924 
   2925   *) Since 0.9.8b, ciphersuite strings naming explicit ciphersuites
   2926      match only those.  Before that, "AES256-SHA" would be interpreted
   2927      as a pattern and match "AES128-SHA" too (since AES128-SHA got
   2928      the same strength classification in 0.9.7h) as we currently only
   2929      have a single AES bit in the ciphersuite description bitmap.
   2930      That change, however, also applied to ciphersuite strings such as
   2931      "RC4-MD5" that intentionally matched multiple ciphersuites --
   2932      namely, SSL 2.0 ciphersuites in addition to the more common ones
   2933      from SSL 3.0/TLS 1.0.
   2934 
   2935      So we change the selection algorithm again: Naming an explicit
   2936      ciphersuite selects this one ciphersuite, and any other similar
   2937      ciphersuite (same bitmap) from *other* protocol versions.
   2938      Thus, "RC4-MD5" again will properly select both the SSL 2.0
   2939      ciphersuite and the SSL 3.0/TLS 1.0 ciphersuite.
   2940 
   2941      Since SSL 2.0 does not have any ciphersuites for which the
   2942      128/256 bit distinction would be relevant, this works for now.
   2943      The proper fix will be to use different bits for AES128 and
   2944      AES256, which would have avoided the problems from the beginning;
   2945      however, bits are scarce, so we can only do this in a new release
   2946      (not just a patchlevel) when we can change the SSL_CIPHER
   2947      definition to split the single 'unsigned long mask' bitmap into
   2948      multiple values to extend the available space.
   2949 
   2950      [Bodo Moeller]
   2951 
   2952  Changes between 0.9.8b and 0.9.8c  [05 Sep 2006]
   2953 
   2954   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
   2955      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
   2956 
   2957   *) Add AES IGE and biIGE modes.
   2958      [Ben Laurie]
   2959 
   2960   *) Change the Unix randomness entropy gathering to use poll() when
   2961      possible instead of select(), since the latter has some
   2962      undesirable limitations.
   2963      [Darryl Miles via Richard Levitte and Bodo Moeller]
   2964 
   2965   *) Disable "ECCdraft" ciphersuites more thoroughly.  Now special
   2966      treatment in ssl/ssl_ciph.s makes sure that these ciphersuites
   2967      cannot be implicitly activated as part of, e.g., the "AES" alias.
   2968      However, please upgrade to OpenSSL 0.9.9[-dev] for
   2969      non-experimental use of the ECC ciphersuites to get TLS extension
   2970      support, which is required for curve and point format negotiation
   2971      to avoid potential handshake problems.
   2972      [Bodo Moeller]
   2973 
   2974   *) Disable rogue ciphersuites:
   2975 
   2976       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
   2977       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
   2978       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
   2979 
   2980      The latter two were purportedly from
   2981      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
   2982      appear there.
   2983 
   2984      Also deactivate the remaining ciphersuites from
   2985      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
   2986      unofficial, and the ID has long expired.
   2987      [Bodo Moeller]
   2988 
   2989   *) Fix RSA blinding Heisenbug (problems sometimes occured on
   2990      dual-core machines) and other potential thread-safety issues.
   2991      [Bodo Moeller]
   2992 
   2993   *) Add the symmetric cipher Camellia (128-bit, 192-bit, 256-bit key
   2994      versions), which is now available for royalty-free use
   2995      (see http://info.isl.ntt.co.jp/crypt/eng/info/chiteki.html).
   2996      Also, add Camellia TLS ciphersuites from RFC 4132.
   2997 
   2998      To minimize changes between patchlevels in the OpenSSL 0.9.8
   2999      series, Camellia remains excluded from compilation unless OpenSSL
   3000      is configured with 'enable-camellia'.
   3001      [NTT]
   3002 
   3003   *) Disable the padding bug check when compression is in use. The padding
   3004      bug check assumes the first packet is of even length, this is not
   3005      necessarily true if compresssion is enabled and can result in false
   3006      positives causing handshake failure. The actual bug test is ancient
   3007      code so it is hoped that implementations will either have fixed it by
   3008      now or any which still have the bug do not support compression.
   3009      [Steve Henson]
   3010 
   3011  Changes between 0.9.8a and 0.9.8b  [04 May 2006]
   3012 
   3013   *) When applying a cipher rule check to see if string match is an explicit
   3014      cipher suite and only match that one cipher suite if it is.
   3015      [Steve Henson]
   3016 
   3017   *) Link in manifests for VC++ if needed.
   3018      [Austin Ziegler <halostatue (a] gmail.com>]
   3019 
   3020   *) Update support for ECC-based TLS ciphersuites according to
   3021      draft-ietf-tls-ecc-12.txt with proposed changes (but without
   3022      TLS extensions, which are supported starting with the 0.9.9
   3023      branch, not in the OpenSSL 0.9.8 branch).
   3024      [Douglas Stebila]
   3025 
   3026   *) New functions EVP_CIPHER_CTX_new() and EVP_CIPHER_CTX_free() to support
   3027      opaque EVP_CIPHER_CTX handling.
   3028      [Steve Henson]
   3029 
   3030   *) Fixes and enhancements to zlib compression code. We now only use
   3031      "zlib1.dll" and use the default __cdecl calling convention on Win32
   3032      to conform with the standards mentioned here:
   3033            http://www.zlib.net/DLL_FAQ.txt
   3034      Static zlib linking now works on Windows and the new --with-zlib-include
   3035      --with-zlib-lib options to Configure can be used to supply the location
   3036      of the headers and library. Gracefully handle case where zlib library
   3037      can't be loaded.
   3038      [Steve Henson]
   3039 
   3040   *) Several fixes and enhancements to the OID generation code. The old code
   3041      sometimes allowed invalid OIDs (1.X for X >= 40 for example), couldn't
   3042      handle numbers larger than ULONG_MAX, truncated printing and had a
   3043      non standard OBJ_obj2txt() behaviour.
   3044      [Steve Henson]
   3045 
   3046   *) Add support for building of engines under engine/ as shared libraries
   3047      under VC++ build system.
   3048      [Steve Henson]
   3049 
   3050   *) Corrected the numerous bugs in the Win32 path splitter in DSO.
   3051      Hopefully, we will not see any false combination of paths any more.
   3052      [Richard Levitte]
   3053 
   3054  Changes between 0.9.8 and 0.9.8a  [11 Oct 2005]
   3055 
   3056   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
   3057      (part of SSL_OP_ALL).  This option used to disable the
   3058      countermeasure against man-in-the-middle protocol-version
   3059      rollback in the SSL 2.0 server implementation, which is a bad
   3060      idea.  (CVE-2005-2969)
   3061 
   3062      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
   3063      for Information Security, National Institute of Advanced Industrial
   3064      Science and Technology [AIST], Japan)]
   3065 
   3066   *) Add two function to clear and return the verify parameter flags.
   3067      [Steve Henson]
   3068 
   3069   *) Keep cipherlists sorted in the source instead of sorting them at
   3070      runtime, thus removing the need for a lock.
   3071      [Nils Larsch]
   3072 
   3073   *) Avoid some small subgroup attacks in Diffie-Hellman.
   3074      [Nick Mathewson and Ben Laurie]
   3075 
   3076   *) Add functions for well-known primes.
   3077      [Nick Mathewson]
   3078 
   3079   *) Extended Windows CE support.
   3080      [Satoshi Nakamura and Andy Polyakov]
   3081 
   3082   *) Initialize SSL_METHOD structures at compile time instead of during
   3083      runtime, thus removing the need for a lock.
   3084      [Steve Henson]
   3085 
   3086   *) Make PKCS7_decrypt() work even if no certificate is supplied by
   3087      attempting to decrypt each encrypted key in turn. Add support to
   3088      smime utility.
   3089      [Steve Henson]
   3090 
   3091  Changes between 0.9.7h and 0.9.8  [05 Jul 2005]
   3092 
   3093   [NB: OpenSSL 0.9.7i and later 0.9.7 patch levels were released after
   3094   OpenSSL 0.9.8.]
   3095 
   3096   *) Add libcrypto.pc and libssl.pc for those who feel they need them.
   3097      [Richard Levitte]
   3098 
   3099   *) Change CA.sh and CA.pl so they don't bundle the CSR and the private
   3100      key into the same file any more.
   3101      [Richard Levitte]
   3102 
   3103   *) Add initial support for Win64, both IA64 and AMD64/x64 flavors.
   3104      [Andy Polyakov]
   3105 
   3106   *) Add -utf8 command line and config file option to 'ca'.
   3107      [Stefan <stf (a] udoma.org]
   3108 
   3109   *) Removed the macro des_crypt(), as it seems to conflict with some
   3110      libraries.  Use DES_crypt().
   3111      [Richard Levitte]
   3112 
   3113   *) Correct naming of the 'chil' and '4758cca' ENGINEs. This
   3114      involves renaming the source and generated shared-libs for
   3115      both. The engines will accept the corrected or legacy ids
   3116      ('ncipher' and '4758_cca' respectively) when binding. NB,
   3117      this only applies when building 'shared'.
   3118      [Corinna Vinschen <vinschen (a] redhat.com> and Geoff Thorpe]
   3119 
   3120   *) Add attribute functions to EVP_PKEY structure. Modify
   3121      PKCS12_create() to recognize a CSP name attribute and
   3122      use it. Make -CSP option work again in pkcs12 utility.
   3123      [Steve Henson]
   3124 
   3125   *) Add new functionality to the bn blinding code:
   3126      - automatic re-creation of the BN_BLINDING parameters after
   3127        a fixed number of uses (currently 32)
   3128      - add new function for parameter creation
   3129      - introduce flags to control the update behaviour of the
   3130        BN_BLINDING parameters
   3131      - hide BN_BLINDING structure
   3132      Add a second BN_BLINDING slot to the RSA structure to improve
   3133      performance when a single RSA object is shared among several
   3134      threads.
   3135      [Nils Larsch]
   3136 
   3137   *) Add support for DTLS.
   3138      [Nagendra Modadugu <nagendra (a] cs.stanford.edu> and Ben Laurie]
   3139 
   3140   *) Add support for DER encoded private keys (SSL_FILETYPE_ASN1)
   3141      to SSL_CTX_use_PrivateKey_file() and SSL_use_PrivateKey_file()
   3142      [Walter Goulet]
   3143 
   3144   *) Remove buggy and incompletet DH cert support from
   3145      ssl/ssl_rsa.c and ssl/s3_both.c
   3146      [Nils Larsch]
   3147 
   3148   *) Use SHA-1 instead of MD5 as the default digest algorithm for
   3149      the apps/openssl applications.
   3150      [Nils Larsch]
   3151 
   3152   *) Compile clean with "-Wall -Wmissing-prototypes
   3153      -Wstrict-prototypes -Wmissing-declarations -Werror". Currently
   3154      DEBUG_SAFESTACK must also be set.
   3155      [Ben Laurie]
   3156 
   3157   *) Change ./Configure so that certain algorithms can be disabled by default.
   3158      The new counterpiece to "no-xxx" is "enable-xxx".
   3159 
   3160      The patented RC5 and MDC2 algorithms will now be disabled unless
   3161      "enable-rc5" and "enable-mdc2", respectively, are specified.
   3162 
   3163      (IDEA remains enabled despite being patented.  This is because IDEA
   3164      is frequently required for interoperability, and there is no license
   3165      fee for non-commercial use.  As before, "no-idea" can be used to
   3166      avoid this algorithm.)
   3167 
   3168      [Bodo Moeller]
   3169 
   3170   *) Add processing of proxy certificates (see RFC 3820).  This work was
   3171      sponsored by KTH (The Royal Institute of Technology in Stockholm) and
   3172      EGEE (Enabling Grids for E-science in Europe).
   3173      [Richard Levitte]
   3174 
   3175   *) RC4 performance overhaul on modern architectures/implementations, such
   3176      as Intel P4, IA-64 and AMD64.
   3177      [Andy Polyakov]
   3178 
   3179   *) New utility extract-section.pl. This can be used specify an alternative
   3180      section number in a pod file instead of having to treat each file as
   3181      a separate case in Makefile. This can be done by adding two lines to the
   3182      pod file:
   3183 
   3184      =for comment openssl_section:XXX
   3185 
   3186      The blank line is mandatory.
   3187 
   3188      [Steve Henson]
   3189 
   3190   *) New arguments -certform, -keyform and -pass for s_client and s_server
   3191      to allow alternative format key and certificate files and passphrase
   3192      sources.
   3193      [Steve Henson]
   3194 
   3195   *) New structure X509_VERIFY_PARAM which combines current verify parameters,
   3196      update associated structures and add various utility functions.
   3197 
   3198      Add new policy related verify parameters, include policy checking in 
   3199      standard verify code. Enhance 'smime' application with extra parameters
   3200      to support policy checking and print out.
   3201      [Steve Henson]
   3202 
   3203   *) Add a new engine to support VIA PadLock ACE extensions in the VIA C3
   3204      Nehemiah processors. These extensions support AES encryption in hardware
   3205      as well as RNG (though RNG support is currently disabled).
   3206      [Michal Ludvig <michal (a] logix.cz>, with help from Andy Polyakov]
   3207 
   3208   *) Deprecate BN_[get|set]_params() functions (they were ignored internally).
   3209      [Geoff Thorpe]
   3210 
   3211   *) New FIPS 180-2 algorithms, SHA-224/-256/-384/-512 are implemented.
   3212      [Andy Polyakov and a number of other people]
   3213 
   3214   *) Improved PowerPC platform support. Most notably BIGNUM assembler
   3215      implementation contributed by IBM.
   3216      [Suresh Chari, Peter Waltenberg, Andy Polyakov]
   3217 
   3218   *) The new 'RSA_generate_key_ex' function now takes a BIGNUM for the public
   3219      exponent rather than 'unsigned long'. There is a corresponding change to
   3220      the new 'rsa_keygen' element of the RSA_METHOD structure.
   3221      [Jelte Jansen, Geoff Thorpe]
   3222 
   3223   *) Functionality for creating the initial serial number file is now
   3224      moved from CA.pl to the 'ca' utility with a new option -create_serial.
   3225 
   3226      (Before OpenSSL 0.9.7e, CA.pl used to initialize the serial
   3227      number file to 1, which is bound to cause problems.  To avoid
   3228      the problems while respecting compatibility between different 0.9.7
   3229      patchlevels, 0.9.7e  employed 'openssl x509 -next_serial' in
   3230      CA.pl for serial number initialization.  With the new release 0.9.8,
   3231      we can fix the problem directly in the 'ca' utility.)
   3232      [Steve Henson]
   3233 
   3234   *) Reduced header interdepencies by declaring more opaque objects in
   3235      ossl_typ.h. As a consequence, including some headers (eg. engine.h) will
   3236      give fewer recursive includes, which could break lazy source code - so
   3237      this change is covered by the OPENSSL_NO_DEPRECATED symbol. As always,
   3238      developers should define this symbol when building and using openssl to
   3239      ensure they track the recommended behaviour, interfaces, [etc], but
   3240      backwards-compatible behaviour prevails when this isn't defined.
   3241      [Geoff Thorpe]
   3242 
   3243   *) New function X509_POLICY_NODE_print() which prints out policy nodes.
   3244      [Steve Henson]
   3245 
   3246   *) Add new EVP function EVP_CIPHER_CTX_rand_key and associated functionality.
   3247      This will generate a random key of the appropriate length based on the 
   3248      cipher context. The EVP_CIPHER can provide its own random key generation
   3249      routine to support keys of a specific form. This is used in the des and 
   3250      3des routines to generate a key of the correct parity. Update S/MIME
   3251      code to use new functions and hence generate correct parity DES keys.
   3252      Add EVP_CHECK_DES_KEY #define to return an error if the key is not 
   3253      valid (weak or incorrect parity).
   3254      [Steve Henson]
   3255 
   3256   *) Add a local set of CRLs that can be used by X509_verify_cert() as well
   3257      as looking them up. This is useful when the verified structure may contain
   3258      CRLs, for example PKCS#7 signedData. Modify PKCS7_verify() to use any CRLs
   3259      present unless the new PKCS7_NO_CRL flag is asserted.
   3260      [Steve Henson]
   3261 
   3262   *) Extend ASN1 oid configuration module. It now additionally accepts the
   3263      syntax:
   3264 
   3265      shortName = some long name, 1.2.3.4
   3266      [Steve Henson]
   3267 
   3268   *) Reimplemented the BN_CTX implementation. There is now no more static
   3269      limitation on the number of variables it can handle nor the depth of the
   3270      "stack" handling for BN_CTX_start()/BN_CTX_end() pairs. The stack
   3271      information can now expand as required, and rather than having a single
   3272      static array of bignums, BN_CTX now uses a linked-list of such arrays
   3273      allowing it to expand on demand whilst maintaining the usefulness of
   3274      BN_CTX's "bundling".
   3275      [Geoff Thorpe]
   3276 
   3277   *) Add a missing BN_CTX parameter to the 'rsa_mod_exp' callback in RSA_METHOD
   3278      to allow all RSA operations to function using a single BN_CTX.
   3279      [Geoff Thorpe]
   3280 
   3281   *) Preliminary support for certificate policy evaluation and checking. This
   3282      is initially intended to pass the tests outlined in "Conformance Testing
   3283      of Relying Party Client Certificate Path Processing Logic" v1.07.
   3284      [Steve Henson]
   3285 
   3286   *) bn_dup_expand() has been deprecated, it was introduced in 0.9.7 and
   3287      remained unused and not that useful. A variety of other little bignum
   3288      tweaks and fixes have also been made continuing on from the audit (see
   3289      below).
   3290      [Geoff Thorpe]
   3291 
   3292   *) Constify all or almost all d2i, c2i, s2i and r2i functions, along with
   3293      associated ASN1, EVP and SSL functions and old ASN1 macros.
   3294      [Richard Levitte]
   3295 
   3296   *) BN_zero() only needs to set 'top' and 'neg' to zero for correct results,
   3297      and this should never fail. So the return value from the use of
   3298      BN_set_word() (which can fail due to needless expansion) is now deprecated;
   3299      if OPENSSL_NO_DEPRECATED is defined, BN_zero() is a void macro.
   3300      [Geoff Thorpe]
   3301 
   3302   *) BN_CTX_get() should return zero-valued bignums, providing the same
   3303      initialised value as BN_new().
   3304      [Geoff Thorpe, suggested by Ulf Mller]
   3305 
   3306   *) Support for inhibitAnyPolicy certificate extension.
   3307      [Steve Henson]
   3308 
   3309   *) An audit of the BIGNUM code is underway, for which debugging code is
   3310      enabled when BN_DEBUG is defined. This makes stricter enforcements on what
   3311      is considered valid when processing BIGNUMs, and causes execution to
   3312      assert() when a problem is discovered. If BN_DEBUG_RAND is defined,
   3313      further steps are taken to deliberately pollute unused data in BIGNUM
   3314      structures to try and expose faulty code further on. For now, openssl will
   3315      (in its default mode of operation) continue to tolerate the inconsistent
   3316      forms that it has tolerated in the past, but authors and packagers should
   3317      consider trying openssl and their own applications when compiled with
   3318      these debugging symbols defined. It will help highlight potential bugs in
   3319      their own code, and will improve the test coverage for OpenSSL itself. At
   3320      some point, these tighter rules will become openssl's default to improve
   3321      maintainability, though the assert()s and other overheads will remain only
   3322      in debugging configurations. See bn.h for more details.
   3323      [Geoff Thorpe, Nils Larsch, Ulf Mller]
   3324 
   3325   *) BN_CTX_init() has been deprecated, as BN_CTX is an opaque structure
   3326      that can only be obtained through BN_CTX_new() (which implicitly
   3327      initialises it). The presence of this function only made it possible
   3328      to overwrite an existing structure (and cause memory leaks).
   3329      [Geoff Thorpe]
   3330 
   3331   *) Because of the callback-based approach for implementing LHASH as a
   3332      template type, lh_insert() adds opaque objects to hash-tables and
   3333      lh_doall() or lh_doall_arg() are typically used with a destructor callback
   3334      to clean up those corresponding objects before destroying the hash table
   3335      (and losing the object pointers). So some over-zealous constifications in
   3336      LHASH have been relaxed so that lh_insert() does not take (nor store) the
   3337      objects as "const" and the lh_doall[_arg] callback wrappers are not
   3338      prototyped to have "const" restrictions on the object pointers they are
   3339      given (and so aren't required to cast them away any more).
   3340      [Geoff Thorpe]
   3341 
   3342   *) The tmdiff.h API was so ugly and minimal that our own timing utility
   3343      (speed) prefers to use its own implementation. The two implementations
   3344      haven't been consolidated as yet (volunteers?) but the tmdiff API has had
   3345      its object type properly exposed (MS_TM) instead of casting to/from "char
   3346      *". This may still change yet if someone realises MS_TM and "ms_time_***"
   3347      aren't necessarily the greatest nomenclatures - but this is what was used
   3348      internally to the implementation so I've used that for now.
   3349      [Geoff Thorpe]
   3350 
   3351   *) Ensure that deprecated functions do not get compiled when
   3352      OPENSSL_NO_DEPRECATED is defined. Some "openssl" subcommands and a few of
   3353      the self-tests were still using deprecated key-generation functions so
   3354      these have been updated also.
   3355      [Geoff Thorpe]
   3356 
   3357   *) Reorganise PKCS#7 code to separate the digest location functionality
   3358      into PKCS7_find_digest(), digest addtion into PKCS7_bio_add_digest().
   3359      New function PKCS7_set_digest() to set the digest type for PKCS#7
   3360      digestedData type. Add additional code to correctly generate the
   3361      digestedData type and add support for this type in PKCS7 initialization
   3362      functions.
   3363      [Steve Henson]
   3364 
   3365   *) New function PKCS7_set0_type_other() this initializes a PKCS7 
   3366      structure of type "other".
   3367      [Steve Henson]
   3368 
   3369   *) Fix prime generation loop in crypto/bn/bn_prime.pl by making
   3370      sure the loop does correctly stop and breaking ("division by zero")
   3371      modulus operations are not performed. The (pre-generated) prime
   3372      table crypto/bn/bn_prime.h was already correct, but it could not be
   3373      re-generated on some platforms because of the "division by zero"
   3374      situation in the script.
   3375      [Ralf S. Engelschall]
   3376 
   3377   *) Update support for ECC-based TLS ciphersuites according to
   3378      draft-ietf-tls-ecc-03.txt: the KDF1 key derivation function with
   3379      SHA-1 now is only used for "small" curves (where the
   3380      representation of a field element takes up to 24 bytes); for
   3381      larger curves, the field element resulting from ECDH is directly
   3382      used as premaster secret.
   3383      [Douglas Stebila (Sun Microsystems Laboratories)]
   3384 
   3385   *) Add code for kP+lQ timings to crypto/ec/ectest.c, and add SEC2
   3386      curve secp160r1 to the tests.
   3387      [Douglas Stebila (Sun Microsystems Laboratories)]
   3388 
   3389   *) Add the possibility to load symbols globally with DSO.
   3390      [Gtz Babin-Ebell <babin-ebell (a] trustcenter.de> via Richard Levitte]
   3391 
   3392   *) Add the functions ERR_set_mark() and ERR_pop_to_mark() for better
   3393      control of the error stack.
   3394      [Richard Levitte]
   3395 
   3396   *) Add support for STORE in ENGINE.
   3397      [Richard Levitte]
   3398 
   3399   *) Add the STORE type.  The intention is to provide a common interface
   3400      to certificate and key stores, be they simple file-based stores, or
   3401      HSM-type store, or LDAP stores, or...
   3402      NOTE: The code is currently UNTESTED and isn't really used anywhere.
   3403      [Richard Levitte]
   3404 
   3405   *) Add a generic structure called OPENSSL_ITEM.  This can be used to
   3406      pass a list of arguments to any function as well as provide a way
   3407      for a function to pass data back to the caller.
   3408      [Richard Levitte]
   3409 
   3410   *) Add the functions BUF_strndup() and BUF_memdup().  BUF_strndup()
   3411      works like BUF_strdup() but can be used to duplicate a portion of
   3412      a string.  The copy gets NUL-terminated.  BUF_memdup() duplicates
   3413      a memory area.
   3414      [Richard Levitte]
   3415 
   3416   *) Add the function sk_find_ex() which works like sk_find(), but will
   3417      return an index to an element even if an exact match couldn't be
   3418      found.  The index is guaranteed to point at the element where the
   3419      searched-for key would be inserted to preserve sorting order.
   3420      [Richard Levitte]
   3421 
   3422   *) Add the function OBJ_bsearch_ex() which works like OBJ_bsearch() but
   3423      takes an extra flags argument for optional functionality.  Currently,
   3424      the following flags are defined:
   3425 
   3426 	OBJ_BSEARCH_VALUE_ON_NOMATCH
   3427 	This one gets OBJ_bsearch_ex() to return a pointer to the first
   3428 	element where the comparing function returns a negative or zero
   3429 	number.
   3430 
   3431 	OBJ_BSEARCH_FIRST_VALUE_ON_MATCH
   3432 	This one gets OBJ_bsearch_ex() to return a pointer to the first
   3433 	element where the comparing function returns zero.  This is useful
   3434 	if there are more than one element where the comparing function
   3435 	returns zero.
   3436      [Richard Levitte]
   3437 
   3438   *) Make it possible to create self-signed certificates with 'openssl ca'
   3439      in such a way that the self-signed certificate becomes part of the
   3440      CA database and uses the same mechanisms for serial number generation
   3441      as all other certificate signing.  The new flag '-selfsign' enables
   3442      this functionality.  Adapt CA.sh and CA.pl.in.
   3443      [Richard Levitte]
   3444 
   3445   *) Add functionality to check the public key of a certificate request
   3446      against a given private.  This is useful to check that a certificate
   3447      request can be signed by that key (self-signing).
   3448      [Richard Levitte]
   3449 
   3450   *) Make it possible to have multiple active certificates with the same
   3451      subject in the CA index file.  This is done only if the keyword
   3452      'unique_subject' is set to 'no' in the main CA section (default
   3453      if 'CA_default') of the configuration file.  The value is saved
   3454      with the database itself in a separate index attribute file,
   3455      named like the index file with '.attr' appended to the name.
   3456      [Richard Levitte]
   3457 
   3458   *) Generate muti valued AVAs using '+' notation in config files for
   3459      req and dirName.
   3460      [Steve Henson]
   3461 
   3462   *) Support for nameConstraints certificate extension.
   3463      [Steve Henson]
   3464 
   3465   *) Support for policyConstraints certificate extension.
   3466      [Steve Henson]
   3467 
   3468   *) Support for policyMappings certificate extension.
   3469      [Steve Henson]
   3470 
   3471   *) Make sure the default DSA_METHOD implementation only uses its
   3472      dsa_mod_exp() and/or bn_mod_exp() handlers if they are non-NULL,
   3473      and change its own handlers to be NULL so as to remove unnecessary
   3474      indirection. This lets alternative implementations fallback to the
   3475      default implementation more easily.
   3476      [Geoff Thorpe]
   3477 
   3478   *) Support for directoryName in GeneralName related extensions
   3479      in config files.
   3480      [Steve Henson]
   3481 
   3482   *) Make it possible to link applications using Makefile.shared.
   3483      Make that possible even when linking against static libraries!
   3484      [Richard Levitte]
   3485 
   3486   *) Support for single pass processing for S/MIME signing. This now
   3487      means that S/MIME signing can be done from a pipe, in addition
   3488      cleartext signing (multipart/signed type) is effectively streaming
   3489      and the signed data does not need to be all held in memory.
   3490 
   3491      This is done with a new flag PKCS7_STREAM. When this flag is set
   3492      PKCS7_sign() only initializes the PKCS7 structure and the actual signing
   3493      is done after the data is output (and digests calculated) in
   3494      SMIME_write_PKCS7().
   3495      [Steve Henson]
   3496 
   3497   *) Add full support for -rpath/-R, both in shared libraries and
   3498      applications, at least on the platforms where it's known how
   3499      to do it.
   3500      [Richard Levitte]
   3501 
   3502   *) In crypto/ec/ec_mult.c, implement fast point multiplication with
   3503      precomputation, based on wNAF splitting: EC_GROUP_precompute_mult()
   3504      will now compute a table of multiples of the generator that
   3505      makes subsequent invocations of EC_POINTs_mul() or EC_POINT_mul()
   3506      faster (notably in the case of a single point multiplication,
   3507      scalar * generator).
   3508      [Nils Larsch, Bodo Moeller]
   3509 
   3510   *) IPv6 support for certificate extensions. The various extensions
   3511      which use the IP:a.b.c.d can now take IPv6 addresses using the
   3512      formats of RFC1884 2.2 . IPv6 addresses are now also displayed
   3513      correctly.
   3514      [Steve Henson]
   3515 
   3516   *) Added an ENGINE that implements RSA by performing private key
   3517      exponentiations with the GMP library. The conversions to and from
   3518      GMP's mpz_t format aren't optimised nor are any montgomery forms
   3519      cached, and on x86 it appears OpenSSL's own performance has caught up.
   3520      However there are likely to be other architectures where GMP could
   3521      provide a boost. This ENGINE is not built in by default, but it can be
   3522      specified at Configure time and should be accompanied by the necessary
   3523      linker additions, eg;
   3524          ./config -DOPENSSL_USE_GMP -lgmp
   3525      [Geoff Thorpe]
   3526 
   3527   *) "openssl engine" will not display ENGINE/DSO load failure errors when
   3528      testing availability of engines with "-t" - the old behaviour is
   3529      produced by increasing the feature's verbosity with "-tt".
   3530      [Geoff Thorpe]
   3531 
   3532   *) ECDSA routines: under certain error conditions uninitialized BN objects
   3533      could be freed. Solution: make sure initialization is performed early
   3534      enough. (Reported and fix supplied by Nils Larsch <nla (a] trustcenter.de>
   3535      via PR#459)
   3536      [Lutz Jaenicke]
   3537 
   3538   *) Key-generation can now be implemented in RSA_METHOD, DSA_METHOD
   3539      and DH_METHOD (eg. by ENGINE implementations) to override the normal
   3540      software implementations. For DSA and DH, parameter generation can
   3541      also be overriden by providing the appropriate method callbacks.
   3542      [Geoff Thorpe]
   3543 
   3544   *) Change the "progress" mechanism used in key-generation and
   3545      primality testing to functions that take a new BN_GENCB pointer in
   3546      place of callback/argument pairs. The new API functions have "_ex"
   3547      postfixes and the older functions are reimplemented as wrappers for
   3548      the new ones. The OPENSSL_NO_DEPRECATED symbol can be used to hide
   3549      declarations of the old functions to help (graceful) attempts to
   3550      migrate to the new functions. Also, the new key-generation API
   3551      functions operate on a caller-supplied key-structure and return
   3552      success/failure rather than returning a key or NULL - this is to
   3553      help make "keygen" another member function of RSA_METHOD etc.
   3554 
   3555      Example for using the new callback interface:
   3556 
   3557           int (*my_callback)(int a, int b, BN_GENCB *cb) = ...;
   3558           void *my_arg = ...;
   3559           BN_GENCB my_cb;
   3560 
   3561           BN_GENCB_set(&my_cb, my_callback, my_arg);
   3562 
   3563           return BN_is_prime_ex(some_bignum, BN_prime_checks, NULL, &cb);
   3564           /* For the meaning of a, b in calls to my_callback(), see the
   3565            * documentation of the function that calls the callback.
   3566            * cb will point to my_cb; my_arg can be retrieved as cb->arg.
   3567            * my_callback should return 1 if it wants BN_is_prime_ex()
   3568            * to continue, or 0 to stop.
   3569            */
   3570 
   3571      [Geoff Thorpe]
   3572 
   3573   *) Change the ZLIB compression method to be stateful, and make it
   3574      available to TLS with the number defined in 
   3575      draft-ietf-tls-compression-04.txt.
   3576      [Richard Levitte]
   3577 
   3578   *) Add the ASN.1 structures and functions for CertificatePair, which
   3579      is defined as follows (according to X.509_4thEditionDraftV6.pdf):
   3580 
   3581      CertificatePair ::= SEQUENCE {
   3582         forward		[0]	Certificate OPTIONAL,
   3583         reverse		[1]	Certificate OPTIONAL,
   3584         -- at least one of the pair shall be present -- }
   3585 
   3586      Also implement the PEM functions to read and write certificate
   3587      pairs, and defined the PEM tag as "CERTIFICATE PAIR".
   3588 
   3589      This needed to be defined, mostly for the sake of the LDAP
   3590      attribute crossCertificatePair, but may prove useful elsewhere as
   3591      well.
   3592      [Richard Levitte]
   3593 
   3594   *) Make it possible to inhibit symlinking of shared libraries in
   3595      Makefile.shared, for Cygwin's sake.
   3596      [Richard Levitte]
   3597 
   3598   *) Extend the BIGNUM API by creating a function 
   3599           void BN_set_negative(BIGNUM *a, int neg);
   3600      and a macro that behave like
   3601           int  BN_is_negative(const BIGNUM *a);
   3602 
   3603      to avoid the need to access 'a->neg' directly in applications.
   3604      [Nils Larsch]
   3605 
   3606   *) Implement fast modular reduction for pseudo-Mersenne primes
   3607      used in NIST curves (crypto/bn/bn_nist.c, crypto/ec/ecp_nist.c).
   3608      EC_GROUP_new_curve_GFp() will now automatically use this
   3609      if applicable.
   3610      [Nils Larsch <nla (a] trustcenter.de>]
   3611 
   3612   *) Add new lock type (CRYPTO_LOCK_BN).
   3613      [Bodo Moeller]
   3614 
   3615   *) Change the ENGINE framework to automatically load engines
   3616      dynamically from specific directories unless they could be
   3617      found to already be built in or loaded.  Move all the
   3618      current engines except for the cryptodev one to a new
   3619      directory engines/.
   3620      The engines in engines/ are built as shared libraries if
   3621      the "shared" options was given to ./Configure or ./config.
   3622      Otherwise, they are inserted in libcrypto.a.
   3623      /usr/local/ssl/engines is the default directory for dynamic
   3624      engines, but that can be overriden at configure time through
   3625      the usual use of --prefix and/or --openssldir, and at run
   3626      time with the environment variable OPENSSL_ENGINES.
   3627      [Geoff Thorpe and Richard Levitte]
   3628 
   3629   *) Add Makefile.shared, a helper makefile to build shared
   3630      libraries.  Addapt Makefile.org.
   3631      [Richard Levitte]
   3632 
   3633   *) Add version info to Win32 DLLs.
   3634      [Peter 'Luna' Runestig" <peter (a] runestig.com>]
   3635 
   3636   *) Add new 'medium level' PKCS#12 API. Certificates and keys
   3637      can be added using this API to created arbitrary PKCS#12
   3638      files while avoiding the low level API.
   3639 
   3640      New options to PKCS12_create(), key or cert can be NULL and
   3641      will then be omitted from the output file. The encryption
   3642      algorithm NIDs can be set to -1 for no encryption, the mac
   3643      iteration count can be set to 0 to omit the mac.
   3644 
   3645      Enhance pkcs12 utility by making the -nokeys and -nocerts
   3646      options work when creating a PKCS#12 file. New option -nomac
   3647      to omit the mac, NONE can be set for an encryption algorithm.
   3648      New code is modified to use the enhanced PKCS12_create()
   3649      instead of the low level API.
   3650      [Steve Henson]
   3651 
   3652   *) Extend ASN1 encoder to support indefinite length constructed
   3653      encoding. This can output sequences tags and octet strings in
   3654      this form. Modify pk7_asn1.c to support indefinite length
   3655      encoding. This is experimental and needs additional code to
   3656      be useful, such as an ASN1 bio and some enhanced streaming
   3657      PKCS#7 code.
   3658 
   3659      Extend template encode functionality so that tagging is passed
   3660      down to the template encoder.
   3661      [Steve Henson]
   3662 
   3663   *) Let 'openssl req' fail if an argument to '-newkey' is not
   3664      recognized instead of using RSA as a default.
   3665      [Bodo Moeller]
   3666 
   3667   *) Add support for ECC-based ciphersuites from draft-ietf-tls-ecc-01.txt.
   3668      As these are not official, they are not included in "ALL";
   3669      the "ECCdraft" ciphersuite group alias can be used to select them.
   3670      [Vipul Gupta and Sumit Gupta (Sun Microsystems Laboratories)]
   3671 
   3672   *) Add ECDH engine support.
   3673      [Nils Gura and Douglas Stebila (Sun Microsystems Laboratories)]
   3674 
   3675   *) Add ECDH in new directory crypto/ecdh/.
   3676      [Douglas Stebila (Sun Microsystems Laboratories)]
   3677 
   3678   *) Let BN_rand_range() abort with an error after 100 iterations
   3679      without success (which indicates a broken PRNG).
   3680      [Bodo Moeller]
   3681 
   3682   *) Change BN_mod_sqrt() so that it verifies that the input value
   3683      is really the square of the return value.  (Previously,
   3684      BN_mod_sqrt would show GIGO behaviour.)
   3685      [Bodo Moeller]
   3686 
   3687   *) Add named elliptic curves over binary fields from X9.62, SECG,
   3688      and WAP/WTLS; add OIDs that were still missing.
   3689 
   3690      [Sheueling Chang Shantz and Douglas Stebila
   3691      (Sun Microsystems Laboratories)]
   3692 
   3693   *) Extend the EC library for elliptic curves over binary fields
   3694      (new files ec2_smpl.c, ec2_smpt.c, ec2_mult.c in crypto/ec/).
   3695      New EC_METHOD:
   3696 
   3697           EC_GF2m_simple_method
   3698 
   3699      New API functions:
   3700 
   3701           EC_GROUP_new_curve_GF2m
   3702           EC_GROUP_set_curve_GF2m
   3703           EC_GROUP_get_curve_GF2m
   3704           EC_POINT_set_affine_coordinates_GF2m
   3705           EC_POINT_get_affine_coordinates_GF2m
   3706           EC_POINT_set_compressed_coordinates_GF2m
   3707 
   3708      Point compression for binary fields is disabled by default for
   3709      patent reasons (compile with OPENSSL_EC_BIN_PT_COMP defined to
   3710      enable it).
   3711 
   3712      As binary polynomials are represented as BIGNUMs, various members
   3713      of the EC_GROUP and EC_POINT data structures can be shared
   3714      between the implementations for prime fields and binary fields;
   3715      the above ..._GF2m functions (except for EX_GROUP_new_curve_GF2m)
   3716      are essentially identical to their ..._GFp counterparts.
   3717      (For simplicity, the '..._GFp' prefix has been dropped from
   3718      various internal method names.)
   3719 
   3720      An internal 'field_div' method (similar to 'field_mul' and
   3721      'field_sqr') has been added; this is used only for binary fields.
   3722 
   3723      [Sheueling Chang Shantz and Douglas Stebila
   3724      (Sun Microsystems Laboratories)]
   3725 
   3726   *) Optionally dispatch EC_POINT_mul(), EC_POINT_precompute_mult()
   3727      through methods ('mul', 'precompute_mult').
   3728 
   3729      The generic implementations (now internally called 'ec_wNAF_mul'
   3730      and 'ec_wNAF_precomputed_mult') remain the default if these
   3731      methods are undefined.
   3732 
   3733      [Sheueling Chang Shantz and Douglas Stebila
   3734      (Sun Microsystems Laboratories)]
   3735 
   3736   *) New function EC_GROUP_get_degree, which is defined through
   3737      EC_METHOD.  For curves over prime fields, this returns the bit
   3738      length of the modulus.
   3739 
   3740      [Sheueling Chang Shantz and Douglas Stebila
   3741      (Sun Microsystems Laboratories)]
   3742 
   3743   *) New functions EC_GROUP_dup, EC_POINT_dup.
   3744      (These simply call ..._new  and ..._copy).
   3745 
   3746      [Sheueling Chang Shantz and Douglas Stebila
   3747      (Sun Microsystems Laboratories)]
   3748 
   3749   *) Add binary polynomial arithmetic software in crypto/bn/bn_gf2m.c.
   3750      Polynomials are represented as BIGNUMs (where the sign bit is not
   3751      used) in the following functions [macros]:  
   3752 
   3753           BN_GF2m_add
   3754           BN_GF2m_sub             [= BN_GF2m_add]
   3755           BN_GF2m_mod             [wrapper for BN_GF2m_mod_arr]
   3756           BN_GF2m_mod_mul         [wrapper for BN_GF2m_mod_mul_arr]
   3757           BN_GF2m_mod_sqr         [wrapper for BN_GF2m_mod_sqr_arr]
   3758           BN_GF2m_mod_inv
   3759           BN_GF2m_mod_exp         [wrapper for BN_GF2m_mod_exp_arr]
   3760           BN_GF2m_mod_sqrt        [wrapper for BN_GF2m_mod_sqrt_arr]
   3761           BN_GF2m_mod_solve_quad  [wrapper for BN_GF2m_mod_solve_quad_arr]
   3762           BN_GF2m_cmp             [= BN_ucmp]
   3763 
   3764      (Note that only the 'mod' functions are actually for fields GF(2^m).
   3765      BN_GF2m_add() is misnomer, but this is for the sake of consistency.)
   3766 
   3767      For some functions, an the irreducible polynomial defining a
   3768      field can be given as an 'unsigned int[]' with strictly
   3769      decreasing elements giving the indices of those bits that are set;
   3770      i.e., p[] represents the polynomial
   3771           f(t) = t^p[0] + t^p[1] + ... + t^p[k]
   3772      where
   3773           p[0] > p[1] > ... > p[k] = 0.
   3774      This applies to the following functions:
   3775 
   3776           BN_GF2m_mod_arr
   3777           BN_GF2m_mod_mul_arr
   3778           BN_GF2m_mod_sqr_arr
   3779           BN_GF2m_mod_inv_arr        [wrapper for BN_GF2m_mod_inv]
   3780           BN_GF2m_mod_div_arr        [wrapper for BN_GF2m_mod_div]
   3781           BN_GF2m_mod_exp_arr
   3782           BN_GF2m_mod_sqrt_arr
   3783           BN_GF2m_mod_solve_quad_arr
   3784           BN_GF2m_poly2arr
   3785           BN_GF2m_arr2poly
   3786 
   3787      Conversion can be performed by the following functions:
   3788 
   3789           BN_GF2m_poly2arr
   3790           BN_GF2m_arr2poly
   3791 
   3792      bntest.c has additional tests for binary polynomial arithmetic.
   3793 
   3794      Two implementations for BN_GF2m_mod_div() are available.
   3795      The default algorithm simply uses BN_GF2m_mod_inv() and
   3796      BN_GF2m_mod_mul().  The alternative algorithm is compiled in only
   3797      if OPENSSL_SUN_GF2M_DIV is defined (patent pending; read the
   3798      copyright notice in crypto/bn/bn_gf2m.c before enabling it).
   3799 
   3800      [Sheueling Chang Shantz and Douglas Stebila
   3801      (Sun Microsystems Laboratories)]
   3802 
   3803   *) Add new error code 'ERR_R_DISABLED' that can be used when some
   3804      functionality is disabled at compile-time.
   3805      [Douglas Stebila <douglas.stebila (a] sun.com>]
   3806 
   3807   *) Change default behaviour of 'openssl asn1parse' so that more
   3808      information is visible when viewing, e.g., a certificate:
   3809 
   3810      Modify asn1_parse2 (crypto/asn1/asn1_par.c) so that in non-'dump'
   3811      mode the content of non-printable OCTET STRINGs is output in a
   3812      style similar to INTEGERs, but with '[HEX DUMP]' prepended to
   3813      avoid the appearance of a printable string.
   3814      [Nils Larsch <nla (a] trustcenter.de>]
   3815 
   3816   *) Add 'asn1_flag' and 'asn1_form' member to EC_GROUP with access
   3817      functions
   3818           EC_GROUP_set_asn1_flag()
   3819           EC_GROUP_get_asn1_flag()
   3820           EC_GROUP_set_point_conversion_form()
   3821           EC_GROUP_get_point_conversion_form()
   3822      These control ASN1 encoding details:
   3823      - Curves (i.e., groups) are encoded explicitly unless asn1_flag
   3824        has been set to OPENSSL_EC_NAMED_CURVE.
   3825      - Points are encoded in uncompressed form by default; options for
   3826        asn1_for are as for point2oct, namely
   3827           POINT_CONVERSION_COMPRESSED
   3828           POINT_CONVERSION_UNCOMPRESSED
   3829           POINT_CONVERSION_HYBRID
   3830 
   3831      Also add 'seed' and 'seed_len' members to EC_GROUP with access
   3832      functions
   3833           EC_GROUP_set_seed()
   3834           EC_GROUP_get0_seed()
   3835           EC_GROUP_get_seed_len()
   3836      This is used only for ASN1 purposes (so far).
   3837      [Nils Larsch <nla (a] trustcenter.de>]
   3838 
   3839   *) Add 'field_type' member to EC_METHOD, which holds the NID
   3840      of the appropriate field type OID.  The new function
   3841      EC_METHOD_get_field_type() returns this value.
   3842      [Nils Larsch <nla (a] trustcenter.de>]
   3843 
   3844   *) Add functions 
   3845           EC_POINT_point2bn()
   3846           EC_POINT_bn2point()
   3847           EC_POINT_point2hex()
   3848           EC_POINT_hex2point()
   3849      providing useful interfaces to EC_POINT_point2oct() and
   3850      EC_POINT_oct2point().
   3851      [Nils Larsch <nla (a] trustcenter.de>]
   3852 
   3853   *) Change internals of the EC library so that the functions
   3854           EC_GROUP_set_generator()
   3855           EC_GROUP_get_generator()
   3856           EC_GROUP_get_order()
   3857           EC_GROUP_get_cofactor()
   3858      are implemented directly in crypto/ec/ec_lib.c and not dispatched
   3859      to methods, which would lead to unnecessary code duplication when
   3860      adding different types of curves.
   3861      [Nils Larsch <nla (a] trustcenter.de> with input by Bodo Moeller]
   3862 
   3863   *) Implement compute_wNAF (crypto/ec/ec_mult.c) without BIGNUM
   3864      arithmetic, and such that modified wNAFs are generated
   3865      (which avoid length expansion in many cases).
   3866      [Bodo Moeller]
   3867 
   3868   *) Add a function EC_GROUP_check_discriminant() (defined via
   3869      EC_METHOD) that verifies that the curve discriminant is non-zero.
   3870 
   3871      Add a function EC_GROUP_check() that makes some sanity tests
   3872      on a EC_GROUP, its generator and order.  This includes
   3873      EC_GROUP_check_discriminant().
   3874      [Nils Larsch <nla (a] trustcenter.de>]
   3875 
   3876   *) Add ECDSA in new directory crypto/ecdsa/.
   3877 
   3878      Add applications 'openssl ecparam' and 'openssl ecdsa'
   3879      (these are based on 'openssl dsaparam' and 'openssl dsa').
   3880 
   3881      ECDSA support is also included in various other files across the
   3882      library.  Most notably,
   3883      - 'openssl req' now has a '-newkey ecdsa:file' option;
   3884      - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
   3885      - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
   3886        d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
   3887        them suitable for ECDSA where domain parameters must be
   3888        extracted before the specific public key;
   3889      - ECDSA engine support has been added.
   3890      [Nils Larsch <nla (a] trustcenter.de>]
   3891 
   3892   *) Include some named elliptic curves, and add OIDs from X9.62,
   3893      SECG, and WAP/WTLS.  Each curve can be obtained from the new
   3894      function
   3895           EC_GROUP_new_by_curve_name(),
   3896      and the list of available named curves can be obtained with
   3897           EC_get_builtin_curves().
   3898      Also add a 'curve_name' member to EC_GROUP objects, which can be
   3899      accessed via
   3900          EC_GROUP_set_curve_name()
   3901          EC_GROUP_get_curve_name()
   3902      [Nils Larsch <larsch (a] trustcenter.de, Bodo Moeller]
   3903  
   3904   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
   3905      was actually never needed) and in BN_mul().  The removal in BN_mul()
   3906      required a small change in bn_mul_part_recursive() and the addition
   3907      of the functions bn_cmp_part_words(), bn_sub_part_words() and
   3908      bn_add_part_words(), which do the same thing as bn_cmp_words(),
   3909      bn_sub_words() and bn_add_words() except they take arrays with
   3910      differing sizes.
   3911      [Richard Levitte]
   3912 
   3913  Changes between 0.9.7l and 0.9.7m  [23 Feb 2007]
   3914 
   3915   *) Cleanse PEM buffers before freeing them since they may contain 
   3916      sensitive data.
   3917      [Benjamin Bennett <ben (a] psc.edu>]
   3918 
   3919   *) Include "!eNULL" in SSL_DEFAULT_CIPHER_LIST to make sure that
   3920      a ciphersuite string such as "DEFAULT:RSA" cannot enable
   3921      authentication-only ciphersuites.
   3922      [Bodo Moeller]
   3923 
   3924   *) Since AES128 and AES256 share a single mask bit in the logic of
   3925      ssl/ssl_ciph.c, the code for masking out disabled ciphers needs a
   3926      kludge to work properly if AES128 is available and AES256 isn't.
   3927      [Victor Duchovni]
   3928 
   3929   *) Expand security boundary to match 1.1.1 module.
   3930      [Steve Henson]
   3931 
   3932   *) Remove redundant features: hash file source, editing of test vectors
   3933      modify fipsld to use external fips_premain.c signature.
   3934      [Steve Henson]
   3935 
   3936   *) New perl script mkfipsscr.pl to create shell scripts or batch files to
   3937      run algorithm test programs.
   3938      [Steve Henson]
   3939 
   3940   *) Make algorithm test programs more tolerant of whitespace.
   3941      [Steve Henson]
   3942 
   3943   *) Have SSL/TLS server implementation tolerate "mismatched" record
   3944      protocol version while receiving ClientHello even if the
   3945      ClientHello is fragmented.  (The server can't insist on the
   3946      particular protocol version it has chosen before the ServerHello
   3947      message has informed the client about his choice.)
   3948      [Bodo Moeller]
   3949 
   3950   *) Load error codes if they are not already present instead of using a
   3951      static variable. This allows them to be cleanly unloaded and reloaded.
   3952      [Steve Henson]
   3953 
   3954  Changes between 0.9.7k and 0.9.7l  [28 Sep 2006]
   3955 
   3956   *) Introduce limits to prevent malicious keys being able to
   3957      cause a denial of service.  (CVE-2006-2940)
   3958      [Steve Henson, Bodo Moeller]
   3959 
   3960   *) Fix ASN.1 parsing of certain invalid structures that can result
   3961      in a denial of service.  (CVE-2006-2937)  [Steve Henson]
   3962 
   3963   *) Fix buffer overflow in SSL_get_shared_ciphers() function. 
   3964      (CVE-2006-3738) [Tavis Ormandy and Will Drewry, Google Security Team]
   3965 
   3966   *) Fix SSL client code which could crash if connecting to a
   3967      malicious SSLv2 server.  (CVE-2006-4343)
   3968      [Tavis Ormandy and Will Drewry, Google Security Team]
   3969 
   3970   *) Change ciphersuite string processing so that an explicit
   3971      ciphersuite selects this one ciphersuite (so that "AES256-SHA"
   3972      will no longer include "AES128-SHA"), and any other similar
   3973      ciphersuite (same bitmap) from *other* protocol versions (so that
   3974      "RC4-MD5" will still include both the SSL 2.0 ciphersuite and the
   3975      SSL 3.0/TLS 1.0 ciphersuite).  This is a backport combining
   3976      changes from 0.9.8b and 0.9.8d.
   3977      [Bodo Moeller]
   3978 
   3979  Changes between 0.9.7j and 0.9.7k  [05 Sep 2006]
   3980 
   3981   *) Avoid PKCS #1 v1.5 signature attack discovered by Daniel Bleichenbacher
   3982      (CVE-2006-4339)  [Ben Laurie and Google Security Team]
   3983 
   3984   *) Change the Unix randomness entropy gathering to use poll() when
   3985      possible instead of select(), since the latter has some
   3986      undesirable limitations.
   3987      [Darryl Miles via Richard Levitte and Bodo Moeller]
   3988 
   3989   *) Disable rogue ciphersuites:
   3990 
   3991       - SSLv2 0x08 0x00 0x80 ("RC4-64-MD5")
   3992       - SSLv3/TLSv1 0x00 0x61 ("EXP1024-RC2-CBC-MD5")
   3993       - SSLv3/TLSv1 0x00 0x60 ("EXP1024-RC4-MD5")
   3994 
   3995      The latter two were purportedly from
   3996      draft-ietf-tls-56-bit-ciphersuites-0[01].txt, but do not really
   3997      appear there.
   3998 
   3999      Also deactive the remaining ciphersuites from
   4000      draft-ietf-tls-56-bit-ciphersuites-01.txt.  These are just as
   4001      unofficial, and the ID has long expired.
   4002      [Bodo Moeller]
   4003 
   4004   *) Fix RSA blinding Heisenbug (problems sometimes occured on
   4005      dual-core machines) and other potential thread-safety issues.
   4006      [Bodo Moeller]
   4007 
   4008  Changes between 0.9.7i and 0.9.7j  [04 May 2006]
   4009 
   4010   *) Adapt fipsld and the build system to link against the validated FIPS
   4011      module in FIPS mode.
   4012      [Steve Henson]
   4013 
   4014   *) Fixes for VC++ 2005 build under Windows.
   4015      [Steve Henson]
   4016 
   4017   *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make 
   4018      from a Windows bash shell such as MSYS. It is autodetected from the
   4019      "config" script when run from a VC++ environment. Modify standard VC++
   4020      build to use fipscanister.o from the GNU make build. 
   4021      [Steve Henson]
   4022 
   4023  Changes between 0.9.7h and 0.9.7i  [14 Oct 2005]
   4024 
   4025   *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
   4026      The value now differs depending on if you build for FIPS or not.
   4027      BEWARE!  A program linked with a shared FIPSed libcrypto can't be
   4028      safely run with a non-FIPSed libcrypto, as it may crash because of
   4029      the difference induced by this change.
   4030      [Andy Polyakov]
   4031 
   4032  Changes between 0.9.7g and 0.9.7h  [11 Oct 2005]
   4033 
   4034   *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
   4035      (part of SSL_OP_ALL).  This option used to disable the
   4036      countermeasure against man-in-the-middle protocol-version
   4037      rollback in the SSL 2.0 server implementation, which is a bad
   4038      idea.  (CVE-2005-2969)
   4039 
   4040      [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
   4041      for Information Security, National Institute of Advanced Industrial
   4042      Science and Technology [AIST], Japan)]
   4043 
   4044   *) Minimal support for X9.31 signatures and PSS padding modes. This is
   4045      mainly for FIPS compliance and not fully integrated at this stage.
   4046      [Steve Henson]
   4047 
   4048   *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
   4049      the exponentiation using a fixed-length exponent.  (Otherwise,
   4050      the information leaked through timing could expose the secret key
   4051      after many signatures; cf. Bleichenbacher's attack on DSA with
   4052      biased k.)
   4053      [Bodo Moeller]
   4054 
   4055   *) Make a new fixed-window mod_exp implementation the default for
   4056      RSA, DSA, and DH private-key operations so that the sequence of
   4057      squares and multiplies and the memory access pattern are
   4058      independent of the particular secret key.  This will mitigate
   4059      cache-timing and potential related attacks.
   4060 
   4061      BN_mod_exp_mont_consttime() is the new exponentiation implementation,
   4062      and this is automatically used by BN_mod_exp_mont() if the new flag
   4063      BN_FLG_EXP_CONSTTIME is set for the exponent.  RSA, DSA, and DH
   4064      will use this BN flag for private exponents unless the flag
   4065      RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
   4066      DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
   4067 
   4068      [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
   4069 
   4070   *) Change the client implementation for SSLv23_method() and
   4071      SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
   4072      Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
   4073      (Previously, the SSL 2.0 backwards compatible Client Hello
   4074      message format would be used even with SSL_OP_NO_SSLv2.)
   4075      [Bodo Moeller]
   4076 
   4077   *) Add support for smime-type MIME parameter in S/MIME messages which some
   4078      clients need.
   4079      [Steve Henson]
   4080 
   4081   *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
   4082      a threadsafe manner. Modify rsa code to use new function and add calls
   4083      to dsa and dh code (which had race conditions before).
   4084      [Steve Henson]
   4085 
   4086   *) Include the fixed error library code in the C error file definitions
   4087      instead of fixing them up at runtime. This keeps the error code
   4088      structures constant.
   4089      [Steve Henson]
   4090 
   4091  Changes between 0.9.7f and 0.9.7g  [11 Apr 2005]
   4092 
   4093   [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
   4094   OpenSSL 0.9.8.]
   4095 
   4096   *) Fixes for newer kerberos headers. NB: the casts are needed because
   4097      the 'length' field is signed on one version and unsigned on another
   4098      with no (?) obvious way to tell the difference, without these VC++
   4099      complains. Also the "definition" of FAR (blank) is no longer included
   4100      nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
   4101      some needed definitions.
   4102      [Steve Henson]
   4103 
   4104   *) Undo Cygwin change.
   4105      [Ulf Mller]
   4106 
   4107   *) Added support for proxy certificates according to RFC 3820.
   4108      Because they may be a security thread to unaware applications,
   4109      they must be explicitely allowed in run-time.  See
   4110      docs/HOWTO/proxy_certificates.txt for further information.
   4111      [Richard Levitte]
   4112 
   4113  Changes between 0.9.7e and 0.9.7f  [22 Mar 2005]
   4114 
   4115   *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
   4116      server and client random values. Previously
   4117      (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
   4118      less random data when sizeof(time_t) > 4 (some 64 bit platforms).
   4119 
   4120      This change has negligible security impact because:
   4121 
   4122      1. Server and client random values still have 24 bytes of pseudo random
   4123         data.
   4124 
   4125      2. Server and client random values are sent in the clear in the initial
   4126         handshake.
   4127 
   4128      3. The master secret is derived using the premaster secret (48 bytes in
   4129         size for static RSA ciphersuites) as well as client server and random
   4130         values.
   4131 
   4132      The OpenSSL team would like to thank the UK NISCC for bringing this issue
   4133      to our attention. 
   4134 
   4135      [Stephen Henson, reported by UK NISCC]
   4136 
   4137   *) Use Windows randomness collection on Cygwin.
   4138      [Ulf Mller]
   4139 
   4140   *) Fix hang in EGD/PRNGD query when communication socket is closed
   4141      prematurely by EGD/PRNGD.
   4142      [Darren Tucker <dtucker (a] zip.com.au> via Lutz Jnicke, resolves #1014]
   4143 
   4144   *) Prompt for pass phrases when appropriate for PKCS12 input format.
   4145      [Steve Henson]
   4146 
   4147   *) Back-port of selected performance improvements from development
   4148      branch, as well as improved support for PowerPC platforms.
   4149      [Andy Polyakov]
   4150 
   4151   *) Add lots of checks for memory allocation failure, error codes to indicate
   4152      failure and freeing up memory if a failure occurs.
   4153      [Nauticus Networks SSL Team <openssl (a] nauticusnet.com>, Steve Henson]
   4154 
   4155   *) Add new -passin argument to dgst.
   4156      [Steve Henson]
   4157 
   4158   *) Perform some character comparisons of different types in X509_NAME_cmp:
   4159      this is needed for some certificates that reencode DNs into UTF8Strings
   4160      (in violation of RFC3280) and can't or wont issue name rollover
   4161      certificates.
   4162      [Steve Henson]
   4163 
   4164   *) Make an explicit check during certificate validation to see that
   4165      the CA setting in each certificate on the chain is correct.  As a
   4166      side effect always do the following basic checks on extensions,
   4167      not just when there's an associated purpose to the check:
   4168 
   4169       - if there is an unhandled critical extension (unless the user
   4170         has chosen to ignore this fault)
   4171       - if the path length has been exceeded (if one is set at all)
   4172       - that certain extensions fit the associated purpose (if one has
   4173         been given)
   4174      [Richard Levitte]
   4175 
   4176  Changes between 0.9.7d and 0.9.7e  [25 Oct 2004]
   4177 
   4178   *) Avoid a race condition when CRLs are checked in a multi threaded 
   4179      environment. This would happen due to the reordering of the revoked
   4180      entries during signature checking and serial number lookup. Now the
   4181      encoding is cached and the serial number sort performed under a lock.
   4182      Add new STACK function sk_is_sorted().
   4183      [Steve Henson]
   4184 
   4185   *) Add Delta CRL to the extension code.
   4186      [Steve Henson]
   4187 
   4188   *) Various fixes to s3_pkt.c so alerts are sent properly.
   4189      [David Holmes <d.holmes (a] f5.com>]
   4190 
   4191   *) Reduce the chances of duplicate issuer name and serial numbers (in
   4192      violation of RFC3280) using the OpenSSL certificate creation utilities.
   4193      This is done by creating a random 64 bit value for the initial serial
   4194      number when a serial number file is created or when a self signed
   4195      certificate is created using 'openssl req -x509'. The initial serial
   4196      number file is created using 'openssl x509 -next_serial' in CA.pl
   4197      rather than being initialized to 1.
   4198      [Steve Henson]
   4199 
   4200  Changes between 0.9.7c and 0.9.7d  [17 Mar 2004]
   4201 
   4202   *) Fix null-pointer assignment in do_change_cipher_spec() revealed           
   4203      by using the Codenomicon TLS Test Tool (CVE-2004-0079)                    
   4204      [Joe Orton, Steve Henson]   
   4205 
   4206   *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
   4207      (CVE-2004-0112)
   4208      [Joe Orton, Steve Henson]   
   4209 
   4210   *) Make it possible to have multiple active certificates with the same
   4211      subject in the CA index file.  This is done only if the keyword
   4212      'unique_subject' is set to 'no' in the main CA section (default
   4213      if 'CA_default') of the configuration file.  The value is saved
   4214      with the database itself in a separate index attribute file,
   4215      named like the index file with '.attr' appended to the name.
   4216      [Richard Levitte]
   4217 
   4218   *) X509 verify fixes. Disable broken certificate workarounds when 
   4219      X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
   4220      keyUsage extension present. Don't accept CRLs with unhandled critical
   4221      extensions: since verify currently doesn't process CRL extensions this
   4222      rejects a CRL with *any* critical extensions. Add new verify error codes
   4223      for these cases.
   4224      [Steve Henson]
   4225 
   4226   *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
   4227      A clarification of RFC2560 will require the use of OCTET STRINGs and 
   4228      some implementations cannot handle the current raw format. Since OpenSSL
   4229      copies and compares OCSP nonces as opaque blobs without any attempt at
   4230      parsing them this should not create any compatibility issues.
   4231      [Steve Henson]
   4232 
   4233   *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
   4234      calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
   4235      this HMAC (and other) operations are several times slower than OpenSSL
   4236      < 0.9.7.
   4237      [Steve Henson]
   4238 
   4239   *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
   4240      [Peter Sylvester <Peter.Sylvester (a] EdelWeb.fr>]
   4241 
   4242   *) Use the correct content when signing type "other".
   4243      [Steve Henson]
   4244 
   4245  Changes between 0.9.7b and 0.9.7c  [30 Sep 2003]
   4246 
   4247   *) Fix various bugs revealed by running the NISCC test suite:
   4248 
   4249      Stop out of bounds reads in the ASN1 code when presented with
   4250      invalid tags (CVE-2003-0543 and CVE-2003-0544).
   4251      
   4252      Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
   4253 
   4254      If verify callback ignores invalid public key errors don't try to check
   4255      certificate signature with the NULL public key.
   4256 
   4257      [Steve Henson]
   4258 
   4259   *) New -ignore_err option in ocsp application to stop the server
   4260      exiting on the first error in a request.
   4261      [Steve Henson]
   4262 
   4263   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
   4264      if the server requested one: as stated in TLS 1.0 and SSL 3.0
   4265      specifications.
   4266      [Steve Henson]
   4267 
   4268   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
   4269      extra data after the compression methods not only for TLS 1.0
   4270      but also for SSL 3.0 (as required by the specification).
   4271      [Bodo Moeller; problem pointed out by Matthias Loepfe]
   4272 
   4273   *) Change X509_certificate_type() to mark the key as exported/exportable
   4274      when it's 512 *bits* long, not 512 bytes.
   4275      [Richard Levitte]
   4276 
   4277   *) Change AES_cbc_encrypt() so it outputs exact multiple of
   4278      blocks during encryption.
   4279      [Richard Levitte]
   4280 
   4281   *) Various fixes to base64 BIO and non blocking I/O. On write 
   4282      flushes were not handled properly if the BIO retried. On read
   4283      data was not being buffered properly and had various logic bugs.
   4284      This also affects blocking I/O when the data being decoded is a
   4285      certain size.
   4286      [Steve Henson]
   4287 
   4288   *) Various S/MIME bugfixes and compatibility changes:
   4289      output correct application/pkcs7 MIME type if
   4290      PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
   4291      Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
   4292      of files as .eml work). Correctly handle very long lines in MIME
   4293      parser.
   4294      [Steve Henson]
   4295 
   4296  Changes between 0.9.7a and 0.9.7b  [10 Apr 2003]
   4297 
   4298   *) Countermeasure against the Klima-Pokorny-Rosa extension of
   4299      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
   4300      a protocol version number mismatch like a decryption error
   4301      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
   4302      [Bodo Moeller]
   4303 
   4304   *) Turn on RSA blinding by default in the default implementation
   4305      to avoid a timing attack. Applications that don't want it can call
   4306      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
   4307      They would be ill-advised to do so in most cases.
   4308      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
   4309 
   4310   *) Change RSA blinding code so that it works when the PRNG is not
   4311      seeded (in this case, the secret RSA exponent is abused as
   4312      an unpredictable seed -- if it is not unpredictable, there
   4313      is no point in blinding anyway).  Make RSA blinding thread-safe
   4314      by remembering the creator's thread ID in rsa->blinding and
   4315      having all other threads use local one-time blinding factors
   4316      (this requires more computation than sharing rsa->blinding, but
   4317      avoids excessive locking; and if an RSA object is not shared
   4318      between threads, blinding will still be very fast).
   4319      [Bodo Moeller]
   4320 
   4321   *) Fixed a typo bug that would cause ENGINE_set_default() to set an
   4322      ENGINE as defaults for all supported algorithms irrespective of
   4323      the 'flags' parameter. 'flags' is now honoured, so applications
   4324      should make sure they are passing it correctly.
   4325      [Geoff Thorpe]
   4326 
   4327   *) Target "mingw" now allows native Windows code to be generated in
   4328      the Cygwin environment as well as with the MinGW compiler.
   4329      [Ulf Moeller] 
   4330 
   4331  Changes between 0.9.7 and 0.9.7a  [19 Feb 2003]
   4332 
   4333   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
   4334      via timing by performing a MAC computation even if incorrrect
   4335      block cipher padding has been found.  This is a countermeasure
   4336      against active attacks where the attacker has to distinguish
   4337      between bad padding and a MAC verification error. (CVE-2003-0078)
   4338 
   4339      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
   4340      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
   4341      Martin Vuagnoux (EPFL, Ilion)]
   4342 
   4343   *) Make the no-err option work as intended.  The intention with no-err
   4344      is not to have the whole error stack handling routines removed from
   4345      libcrypto, it's only intended to remove all the function name and
   4346      reason texts, thereby removing some of the footprint that may not
   4347      be interesting if those errors aren't displayed anyway.
   4348 
   4349      NOTE: it's still possible for any application or module to have it's
   4350      own set of error texts inserted.  The routines are there, just not
   4351      used by default when no-err is given.
   4352      [Richard Levitte]
   4353 
   4354   *) Add support for FreeBSD on IA64.
   4355      [dirk.meyer (a] dinoex.sub.org via Richard Levitte, resolves #454]
   4356 
   4357   *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
   4358      Kerberos function mit_des_cbc_cksum().  Before this change,
   4359      the value returned by DES_cbc_cksum() was like the one from
   4360      mit_des_cbc_cksum(), except the bytes were swapped.
   4361      [Kevin Greaney <Kevin.Greaney (a] hp.com> and Richard Levitte]
   4362 
   4363   *) Allow an application to disable the automatic SSL chain building.
   4364      Before this a rather primitive chain build was always performed in
   4365      ssl3_output_cert_chain(): an application had no way to send the 
   4366      correct chain if the automatic operation produced an incorrect result.
   4367 
   4368      Now the chain builder is disabled if either:
   4369 
   4370      1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
   4371 
   4372      2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
   4373 
   4374      The reasoning behind this is that an application would not want the
   4375      auto chain building to take place if extra chain certificates are
   4376      present and it might also want a means of sending no additional
   4377      certificates (for example the chain has two certificates and the
   4378      root is omitted).
   4379      [Steve Henson]
   4380 
   4381   *) Add the possibility to build without the ENGINE framework.
   4382      [Steven Reddie <smr (a] essemer.com.au> via Richard Levitte]
   4383 
   4384   *) Under Win32 gmtime() can return NULL: check return value in
   4385      OPENSSL_gmtime(). Add error code for case where gmtime() fails.
   4386      [Steve Henson]
   4387 
   4388   *) DSA routines: under certain error conditions uninitialized BN objects
   4389      could be freed. Solution: make sure initialization is performed early
   4390      enough. (Reported and fix supplied by Ivan D Nestlerode <nestler (a] MIT.EDU>,
   4391      Nils Larsch <nla (a] trustcenter.de> via PR#459)
   4392      [Lutz Jaenicke]
   4393 
   4394   *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
   4395      checked on reconnect on the client side, therefore session resumption
   4396      could still fail with a "ssl session id is different" error. This
   4397      behaviour is masked when SSL_OP_ALL is used due to
   4398      SSL_OP_MICROSOFT_SESS_ID_BUG being set.
   4399      Behaviour observed by Crispin Flowerday <crispin (a] flowerday.cx> as
   4400      followup to PR #377.
   4401      [Lutz Jaenicke]
   4402 
   4403   *) IA-32 assembler support enhancements: unified ELF targets, support
   4404      for SCO/Caldera platforms, fix for Cygwin shared build.
   4405      [Andy Polyakov]
   4406 
   4407   *) Add support for FreeBSD on sparc64.  As a consequence, support for
   4408      FreeBSD on non-x86 processors is separate from x86 processors on
   4409      the config script, much like the NetBSD support.
   4410      [Richard Levitte & Kris Kennaway <kris (a] obsecurity.org>]
   4411 
   4412  Changes between 0.9.6h and 0.9.7  [31 Dec 2002]
   4413 
   4414   [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
   4415   OpenSSL 0.9.7.]
   4416 
   4417   *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
   4418      code (06) was taken as the first octet of the session ID and the last
   4419      octet was ignored consequently. As a result SSLv2 client side session
   4420      caching could not have worked due to the session ID mismatch between
   4421      client and server.
   4422      Behaviour observed by Crispin Flowerday <crispin (a] flowerday.cx> as
   4423      PR #377.
   4424      [Lutz Jaenicke]
   4425 
   4426   *) Change the declaration of needed Kerberos libraries to use EX_LIBS
   4427      instead of the special (and badly supported) LIBKRB5.  LIBKRB5 is
   4428      removed entirely.
   4429      [Richard Levitte]
   4430 
   4431   *) The hw_ncipher.c engine requires dynamic locks.  Unfortunately, it
   4432      seems that in spite of existing for more than a year, many application
   4433      author have done nothing to provide the necessary callbacks, which
   4434      means that this particular engine will not work properly anywhere.
   4435      This is a very unfortunate situation which forces us, in the name
   4436      of usability, to give the hw_ncipher.c a static lock, which is part
   4437      of libcrypto.
   4438      NOTE: This is for the 0.9.7 series ONLY.  This hack will never
   4439      appear in 0.9.8 or later.  We EXPECT application authors to have
   4440      dealt properly with this when 0.9.8 is released (unless we actually
   4441      make such changes in the libcrypto locking code that changes will
   4442      have to be made anyway).
   4443      [Richard Levitte]
   4444 
   4445   *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
   4446      octets have been read, EOF or an error occurs. Without this change
   4447      some truncated ASN1 structures will not produce an error.
   4448      [Steve Henson]
   4449 
   4450   *) Disable Heimdal support, since it hasn't been fully implemented.
   4451      Still give the possibility to force the use of Heimdal, but with
   4452      warnings and a request that patches get sent to openssl-dev.
   4453      [Richard Levitte]
   4454 
   4455   *) Add the VC-CE target, introduce the WINCE sysname, and add
   4456      INSTALL.WCE and appropriate conditionals to make it build.
   4457      [Steven Reddie <smr (a] essemer.com.au> via Richard Levitte]
   4458 
   4459   *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
   4460      cygssl-x.y.z.dll, where x, y and z are the major, minor and
   4461      edit numbers of the version.
   4462      [Corinna Vinschen <vinschen (a] redhat.com> and Richard Levitte]
   4463 
   4464   *) Introduce safe string copy and catenation functions
   4465      (BUF_strlcpy() and BUF_strlcat()).
   4466      [Ben Laurie (CHATS) and Richard Levitte]
   4467 
   4468   *) Avoid using fixed-size buffers for one-line DNs.
   4469      [Ben Laurie (CHATS)]
   4470 
   4471   *) Add BUF_MEM_grow_clean() to avoid information leakage when
   4472      resizing buffers containing secrets, and use where appropriate.
   4473      [Ben Laurie (CHATS)]
   4474 
   4475   *) Avoid using fixed size buffers for configuration file location.
   4476      [Ben Laurie (CHATS)]
   4477 
   4478   *) Avoid filename truncation for various CA files.
   4479      [Ben Laurie (CHATS)]
   4480 
   4481   *) Use sizeof in preference to magic numbers.
   4482      [Ben Laurie (CHATS)]
   4483 
   4484   *) Avoid filename truncation in cert requests.
   4485      [Ben Laurie (CHATS)]
   4486 
   4487   *) Add assertions to check for (supposedly impossible) buffer
   4488      overflows.
   4489      [Ben Laurie (CHATS)]
   4490 
   4491   *) Don't cache truncated DNS entries in the local cache (this could
   4492      potentially lead to a spoofing attack).
   4493      [Ben Laurie (CHATS)]
   4494 
   4495   *) Fix various buffers to be large enough for hex/decimal
   4496      representations in a platform independent manner.
   4497      [Ben Laurie (CHATS)]
   4498 
   4499   *) Add CRYPTO_realloc_clean() to avoid information leakage when
   4500      resizing buffers containing secrets, and use where appropriate.
   4501      [Ben Laurie (CHATS)]
   4502 
   4503   *) Add BIO_indent() to avoid much slightly worrying code to do
   4504      indents.
   4505      [Ben Laurie (CHATS)]
   4506 
   4507   *) Convert sprintf()/BIO_puts() to BIO_printf().
   4508      [Ben Laurie (CHATS)]
   4509 
   4510   *) buffer_gets() could terminate with the buffer only half
   4511      full. Fixed.
   4512      [Ben Laurie (CHATS)]
   4513 
   4514   *) Add assertions to prevent user-supplied crypto functions from
   4515      overflowing internal buffers by having large block sizes, etc.
   4516      [Ben Laurie (CHATS)]
   4517 
   4518   *) New OPENSSL_assert() macro (similar to assert(), but enabled
   4519      unconditionally).
   4520      [Ben Laurie (CHATS)]
   4521 
   4522   *) Eliminate unused copy of key in RC4.
   4523      [Ben Laurie (CHATS)]
   4524 
   4525   *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
   4526      [Ben Laurie (CHATS)]
   4527 
   4528   *) Fix off-by-one error in EGD path.
   4529      [Ben Laurie (CHATS)]
   4530 
   4531   *) If RANDFILE path is too long, ignore instead of truncating.
   4532      [Ben Laurie (CHATS)]
   4533 
   4534   *) Eliminate unused and incorrectly sized X.509 structure
   4535      CBCParameter.
   4536      [Ben Laurie (CHATS)]
   4537 
   4538   *) Eliminate unused and dangerous function knumber().
   4539      [Ben Laurie (CHATS)]
   4540 
   4541   *) Eliminate unused and dangerous structure, KSSL_ERR.
   4542      [Ben Laurie (CHATS)]
   4543 
   4544   *) Protect against overlong session ID context length in an encoded
   4545      session object. Since these are local, this does not appear to be
   4546      exploitable.
   4547      [Ben Laurie (CHATS)]
   4548 
   4549   *) Change from security patch (see 0.9.6e below) that did not affect
   4550      the 0.9.6 release series:
   4551 
   4552      Remote buffer overflow in SSL3 protocol - an attacker could
   4553      supply an oversized master key in Kerberos-enabled versions.
   4554      (CVE-2002-0657)
   4555      [Ben Laurie (CHATS)]
   4556 
   4557   *) Change the SSL kerb5 codes to match RFC 2712.
   4558      [Richard Levitte]
   4559 
   4560   *) Make -nameopt work fully for req and add -reqopt switch.
   4561      [Michael Bell <michael.bell (a] rz.hu-berlin.de>, Steve Henson]
   4562 
   4563   *) The "block size" for block ciphers in CFB and OFB mode should be 1.
   4564      [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve (a] opera.com>]
   4565 
   4566   *) Make sure tests can be performed even if the corresponding algorithms
   4567      have been removed entirely.  This was also the last step to make
   4568      OpenSSL compilable with DJGPP under all reasonable conditions.
   4569      [Richard Levitte, Doug Kaufman <dkaufman (a] rahul.net>]
   4570 
   4571   *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
   4572      to allow version independent disabling of normally unselected ciphers,
   4573      which may be activated as a side-effect of selecting a single cipher.
   4574 
   4575      (E.g., cipher list string "RSA" enables ciphersuites that are left
   4576      out of "ALL" because they do not provide symmetric encryption.
   4577      "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
   4578      [Lutz Jaenicke, Bodo Moeller]
   4579 
   4580   *) Add appropriate support for separate platform-dependent build
   4581      directories.  The recommended way to make a platform-dependent
   4582      build directory is the following (tested on Linux), maybe with
   4583      some local tweaks:
   4584 
   4585 	# Place yourself outside of the OpenSSL source tree.  In
   4586 	# this example, the environment variable OPENSSL_SOURCE
   4587 	# is assumed to contain the absolute OpenSSL source directory.
   4588 	mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
   4589 	cd objtree/"`uname -s`-`uname -r`-`uname -m`"
   4590 	(cd $OPENSSL_SOURCE; find . -type f) | while read F; do
   4591 		mkdir -p `dirname $F`
   4592 		ln -s $OPENSSL_SOURCE/$F $F
   4593 	done
   4594 
   4595      To be absolutely sure not to disturb the source tree, a "make clean"
   4596      is a good thing.  If it isn't successfull, don't worry about it,
   4597      it probably means the source directory is very clean.
   4598      [Richard Levitte]
   4599 
   4600   *) Make sure any ENGINE control commands make local copies of string
   4601      pointers passed to them whenever necessary. Otherwise it is possible
   4602      the caller may have overwritten (or deallocated) the original string
   4603      data when a later ENGINE operation tries to use the stored values.
   4604      [Gtz Babin-Ebell <babinebell (a] trustcenter.de>]
   4605 
   4606   *) Improve diagnostics in file reading and command-line digests.
   4607      [Ben Laurie aided and abetted by Solar Designer <solar (a] openwall.com>]
   4608 
   4609   *) Add AES modes CFB and OFB to the object database.  Correct an
   4610      error in AES-CFB decryption.
   4611      [Richard Levitte]
   4612 
   4613   *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this 
   4614      allows existing EVP_CIPHER_CTX structures to be reused after
   4615      calling EVP_*Final(). This behaviour is used by encryption
   4616      BIOs and some applications. This has the side effect that
   4617      applications must explicitly clean up cipher contexts with
   4618      EVP_CIPHER_CTX_cleanup() or they will leak memory.
   4619      [Steve Henson]
   4620 
   4621   *) Check the values of dna and dnb in bn_mul_recursive before calling
   4622      bn_mul_comba (a non zero value means the a or b arrays do not contain
   4623      n2 elements) and fallback to bn_mul_normal if either is not zero.
   4624      [Steve Henson]
   4625 
   4626   *) Fix escaping of non-ASCII characters when using the -subj option
   4627      of the "openssl req" command line tool. (Robert Joop <joop (a] fokus.gmd.de>)
   4628      [Lutz Jaenicke]
   4629 
   4630   *) Make object definitions compliant to LDAP (RFC2256): SN is the short
   4631      form for "surname", serialNumber has no short form.
   4632      Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
   4633      therefore remove "mail" short name for "internet 7".
   4634      The OID for unique identifiers in X509 certificates is
   4635      x500UniqueIdentifier, not uniqueIdentifier.
   4636      Some more OID additions. (Michael Bell <michael.bell (a] rz.hu-berlin.de>)
   4637      [Lutz Jaenicke]
   4638 
   4639   *) Add an "init" command to the ENGINE config module and auto initialize
   4640      ENGINEs. Without any "init" command the ENGINE will be initialized 
   4641      after all ctrl commands have been executed on it. If init=1 the 
   4642      ENGINE is initailized at that point (ctrls before that point are run
   4643      on the uninitialized ENGINE and after on the initialized one). If
   4644      init=0 then the ENGINE will not be iniatialized at all.
   4645      [Steve Henson]
   4646 
   4647   *) Fix the 'app_verify_callback' interface so that the user-defined
   4648      argument is actually passed to the callback: In the
   4649      SSL_CTX_set_cert_verify_callback() prototype, the callback
   4650      declaration has been changed from
   4651           int (*cb)()
   4652      into
   4653           int (*cb)(X509_STORE_CTX *,void *);
   4654      in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
   4655           i=s->ctx->app_verify_callback(&ctx)
   4656      has been changed into
   4657           i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
   4658 
   4659      To update applications using SSL_CTX_set_cert_verify_callback(),
   4660      a dummy argument can be added to their callback functions.
   4661      [D. K. Smetters <smetters (a] parc.xerox.com>]
   4662 
   4663   *) Added the '4758cca' ENGINE to support IBM 4758 cards.
   4664      [Maurice Gittens <maurice (a] gittens.nl>, touchups by Geoff Thorpe]
   4665 
   4666   *) Add and OPENSSL_LOAD_CONF define which will cause
   4667      OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
   4668      This allows older applications to transparently support certain
   4669      OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
   4670      Two new functions OPENSSL_add_all_algorithms_noconf() which will never
   4671      load the config file and OPENSSL_add_all_algorithms_conf() which will
   4672      always load it have also been added.
   4673      [Steve Henson]
   4674 
   4675   *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
   4676      Adjust NIDs and EVP layer.
   4677      [Stephen Sprunk <stephen (a] sprunk.org> and Richard Levitte]
   4678 
   4679   *) Config modules support in openssl utility.
   4680 
   4681      Most commands now load modules from the config file,
   4682      though in a few (such as version) this isn't done 
   4683      because it couldn't be used for anything.
   4684 
   4685      In the case of ca and req the config file used is
   4686      the same as the utility itself: that is the -config
   4687      command line option can be used to specify an
   4688      alternative file.
   4689      [Steve Henson]
   4690 
   4691   *) Move default behaviour from OPENSSL_config(). If appname is NULL
   4692      use "openssl_conf" if filename is NULL use default openssl config file.
   4693      [Steve Henson]
   4694 
   4695   *) Add an argument to OPENSSL_config() to allow the use of an alternative
   4696      config section name. Add a new flag to tolerate a missing config file
   4697      and move code to CONF_modules_load_file().
   4698      [Steve Henson]
   4699 
   4700   *) Support for crypto accelerator cards from Accelerated Encryption
   4701      Processing, www.aep.ie.  (Use engine 'aep')
   4702      The support was copied from 0.9.6c [engine] and adapted/corrected
   4703      to work with the new engine framework.
   4704      [AEP Inc. and Richard Levitte]
   4705 
   4706   *) Support for SureWare crypto accelerator cards from Baltimore
   4707      Technologies.  (Use engine 'sureware')
   4708      The support was copied from 0.9.6c [engine] and adapted
   4709      to work with the new engine framework.
   4710      [Richard Levitte]
   4711 
   4712   *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
   4713      make the newer ENGINE framework commands for the CHIL engine work.
   4714      [Toomas Kiisk <vix (a] cyber.ee> and Richard Levitte]
   4715 
   4716   *) Make it possible to produce shared libraries on ReliantUNIX.
   4717      [Robert Dahlem <Robert.Dahlem (a] ffm2.siemens.de> via Richard Levitte]
   4718 
   4719   *) Add the configuration target debug-linux-ppro.
   4720      Make 'openssl rsa' use the general key loading routines
   4721      implemented in apps.c, and make those routines able to
   4722      handle the key format FORMAT_NETSCAPE and the variant
   4723      FORMAT_IISSGC.
   4724      [Toomas Kiisk <vix (a] cyber.ee> via Richard Levitte]
   4725 
   4726  *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
   4727      [Toomas Kiisk <vix (a] cyber.ee> via Richard Levitte]
   4728 
   4729   *) Add -keyform to rsautl, and document -engine.
   4730      [Richard Levitte, inspired by Toomas Kiisk <vix (a] cyber.ee>]
   4731 
   4732   *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
   4733      BIO_R_NO_SUCH_FILE error code rather than the generic
   4734      ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
   4735      [Ben Laurie]
   4736 
   4737   *) Add new functions
   4738           ERR_peek_last_error
   4739           ERR_peek_last_error_line
   4740           ERR_peek_last_error_line_data.
   4741      These are similar to
   4742           ERR_peek_error
   4743           ERR_peek_error_line
   4744           ERR_peek_error_line_data,
   4745      but report on the latest error recorded rather than the first one
   4746      still in the error queue.
   4747      [Ben Laurie, Bodo Moeller]
   4748         
   4749   *) default_algorithms option in ENGINE config module. This allows things
   4750      like:
   4751      default_algorithms = ALL
   4752      default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
   4753      [Steve Henson]
   4754 
   4755   *) Prelminary ENGINE config module.
   4756      [Steve Henson]
   4757 
   4758   *) New experimental application configuration code.
   4759      [Steve Henson]
   4760 
   4761   *) Change the AES code to follow the same name structure as all other
   4762      symmetric ciphers, and behave the same way.  Move everything to
   4763      the directory crypto/aes, thereby obsoleting crypto/rijndael.
   4764      [Stephen Sprunk <stephen (a] sprunk.org> and Richard Levitte]
   4765 
   4766   *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
   4767      [Ben Laurie and Theo de Raadt]
   4768 
   4769   *) Add option to output public keys in req command.
   4770      [Massimiliano Pala madwolf (a] openca.org]
   4771 
   4772   *) Use wNAFs in EC_POINTs_mul() for improved efficiency
   4773      (up to about 10% better than before for P-192 and P-224).
   4774      [Bodo Moeller]
   4775 
   4776   *) New functions/macros
   4777 
   4778           SSL_CTX_set_msg_callback(ctx, cb)
   4779           SSL_CTX_set_msg_callback_arg(ctx, arg)
   4780           SSL_set_msg_callback(ssl, cb)
   4781           SSL_set_msg_callback_arg(ssl, arg)
   4782 
   4783      to request calling a callback function
   4784 
   4785           void cb(int write_p, int version, int content_type,
   4786                   const void *buf, size_t len, SSL *ssl, void *arg)
   4787 
   4788      whenever a protocol message has been completely received
   4789      (write_p == 0) or sent (write_p == 1).  Here 'version' is the
   4790      protocol version  according to which the SSL library interprets
   4791      the current protocol message (SSL2_VERSION, SSL3_VERSION, or
   4792      TLS1_VERSION).  'content_type' is 0 in the case of SSL 2.0, or
   4793      the content type as defined in the SSL 3.0/TLS 1.0 protocol
   4794      specification (change_cipher_spec(20), alert(21), handshake(22)).
   4795      'buf' and 'len' point to the actual message, 'ssl' to the
   4796      SSL object, and 'arg' is the application-defined value set by
   4797      SSL[_CTX]_set_msg_callback_arg().
   4798 
   4799      'openssl s_client' and 'openssl s_server' have new '-msg' options
   4800      to enable a callback that displays all protocol messages.
   4801      [Bodo Moeller]
   4802 
   4803   *) Change the shared library support so shared libraries are built as
   4804      soon as the corresponding static library is finished, and thereby get
   4805      openssl and the test programs linked against the shared library.
   4806      This still only happens when the keyword "shard" has been given to
   4807      the configuration scripts.
   4808 
   4809      NOTE: shared library support is still an experimental thing, and
   4810      backward binary compatibility is still not guaranteed.
   4811      ["Maciej W. Rozycki" <macro (a] ds2.pg.gda.pl> and Richard Levitte]
   4812 
   4813   *) Add support for Subject Information Access extension.
   4814      [Peter Sylvester <Peter.Sylvester (a] EdelWeb.fr>]
   4815 
   4816   *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
   4817      additional bytes when new memory had to be allocated, not just
   4818      when reusing an existing buffer.
   4819      [Bodo Moeller]
   4820 
   4821   *) New command line and configuration option 'utf8' for the req command.
   4822      This allows field values to be specified as UTF8 strings.
   4823      [Steve Henson]
   4824 
   4825   *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
   4826      runs for the former and machine-readable output for the latter.
   4827      [Ben Laurie]
   4828 
   4829   *) Add '-noemailDN' option to 'openssl ca'.  This prevents inclusion
   4830      of the e-mail address in the DN (i.e., it will go into a certificate
   4831      extension only).  The new configuration file option 'email_in_dn = no'
   4832      has the same effect.
   4833      [Massimiliano Pala madwolf (a] openca.org]
   4834 
   4835   *) Change all functions with names starting with des_ to be starting
   4836      with DES_ instead.  Add wrappers that are compatible with libdes,
   4837      but are named _ossl_old_des_*.  Finally, add macros that map the
   4838      des_* symbols to the corresponding _ossl_old_des_* if libdes
   4839      compatibility is desired.  If OpenSSL 0.9.6c compatibility is
   4840      desired, the des_* symbols will be mapped to DES_*, with one
   4841      exception.
   4842 
   4843      Since we provide two compatibility mappings, the user needs to
   4844      define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
   4845      compatibility is desired.  The default (i.e., when that macro
   4846      isn't defined) is OpenSSL 0.9.6c compatibility.
   4847 
   4848      There are also macros that enable and disable the support of old
   4849      des functions altogether.  Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
   4850      and OPENSSL_DISABLE_OLD_DES_SUPPORT.  If none or both of those
   4851      are defined, the default will apply: to support the old des routines.
   4852 
   4853      In either case, one must include openssl/des.h to get the correct
   4854      definitions.  Do not try to just include openssl/des_old.h, that
   4855      won't work.
   4856 
   4857      NOTE: This is a major break of an old API into a new one.  Software
   4858      authors are encouraged to switch to the DES_ style functions.  Some
   4859      time in the future, des_old.h and the libdes compatibility functions
   4860      will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
   4861      default), and then completely removed.
   4862      [Richard Levitte]
   4863 
   4864   *) Test for certificates which contain unsupported critical extensions.
   4865      If such a certificate is found during a verify operation it is 
   4866      rejected by default: this behaviour can be overridden by either
   4867      handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
   4868      by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
   4869      X509_supported_extension() has also been added which returns 1 if a
   4870      particular extension is supported.
   4871      [Steve Henson]
   4872 
   4873   *) Modify the behaviour of EVP cipher functions in similar way to digests
   4874      to retain compatibility with existing code.
   4875      [Steve Henson]
   4876 
   4877   *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
   4878      compatibility with existing code. In particular the 'ctx' parameter does
   4879      not have to be to be initialized before the call to EVP_DigestInit() and
   4880      it is tidied up after a call to EVP_DigestFinal(). New function
   4881      EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
   4882      EVP_MD_CTX_copy() changed to not require the destination to be
   4883      initialized valid and new function EVP_MD_CTX_copy_ex() added which
   4884      requires the destination to be valid.
   4885 
   4886      Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
   4887      EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
   4888      [Steve Henson]
   4889 
   4890   *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
   4891      so that complete 'Handshake' protocol structures are kept in memory
   4892      instead of overwriting 'msg_type' and 'length' with 'body' data.
   4893      [Bodo Moeller]
   4894 
   4895   *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
   4896      [Massimo Santin via Richard Levitte]
   4897 
   4898   *) Major restructuring to the underlying ENGINE code. This includes
   4899      reduction of linker bloat, separation of pure "ENGINE" manipulation
   4900      (initialisation, etc) from functionality dealing with implementations
   4901      of specific crypto iterfaces. This change also introduces integrated
   4902      support for symmetric ciphers and digest implementations - so ENGINEs
   4903      can now accelerate these by providing EVP_CIPHER and EVP_MD
   4904      implementations of their own. This is detailed in crypto/engine/README
   4905      as it couldn't be adequately described here. However, there are a few
   4906      API changes worth noting - some RSA, DSA, DH, and RAND functions that
   4907      were changed in the original introduction of ENGINE code have now
   4908      reverted back - the hooking from this code to ENGINE is now a good
   4909      deal more passive and at run-time, operations deal directly with
   4910      RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
   4911      dereferencing through an ENGINE pointer any more. Also, the ENGINE
   4912      functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
   4913      they were not being used by the framework as there is no concept of a
   4914      BIGNUM_METHOD and they could not be generalised to the new
   4915      'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
   4916      ENGINE_cpy() has been removed as it cannot be consistently defined in
   4917      the new code.
   4918      [Geoff Thorpe]
   4919 
   4920   *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
   4921      [Steve Henson]
   4922 
   4923   *) Change mkdef.pl to sort symbols that get the same entry number,
   4924      and make sure the automatically generated functions ERR_load_*
   4925      become part of libeay.num as well.
   4926      [Richard Levitte]
   4927 
   4928   *) New function SSL_renegotiate_pending().  This returns true once
   4929      renegotiation has been requested (either SSL_renegotiate() call
   4930      or HelloRequest/ClientHello receveived from the peer) and becomes
   4931      false once a handshake has been completed.
   4932      (For servers, SSL_renegotiate() followed by SSL_do_handshake()
   4933      sends a HelloRequest, but does not ensure that a handshake takes
   4934      place.  SSL_renegotiate_pending() is useful for checking if the
   4935      client has followed the request.)
   4936      [Bodo Moeller]
   4937 
   4938   *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
   4939      By default, clients may request session resumption even during
   4940      renegotiation (if session ID contexts permit); with this option,
   4941      session resumption is possible only in the first handshake.
   4942 
   4943      SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL.  This makes
   4944      more bits available for options that should not be part of
   4945      SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
   4946      [Bodo Moeller]
   4947 
   4948   *) Add some demos for certificate and certificate request creation.
   4949      [Steve Henson]
   4950 
   4951   *) Make maximum certificate chain size accepted from the peer application
   4952      settable (SSL*_get/set_max_cert_list()), as proposed by
   4953      "Douglas E. Engert" <deengert (a] anl.gov>.
   4954      [Lutz Jaenicke]
   4955 
   4956   *) Add support for shared libraries for Unixware-7
   4957      (Boyd Lynn Gerber <gerberb (a] zenez.com>).
   4958      [Lutz Jaenicke]
   4959 
   4960   *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
   4961      be done prior to destruction. Use this to unload error strings from
   4962      ENGINEs that load their own error strings. NB: This adds two new API
   4963      functions to "get" and "set" this destroy handler in an ENGINE.
   4964      [Geoff Thorpe]
   4965 
   4966   *) Alter all existing ENGINE implementations (except "openssl" and
   4967      "openbsd") to dynamically instantiate their own error strings. This
   4968      makes them more flexible to be built both as statically-linked ENGINEs
   4969      and self-contained shared-libraries loadable via the "dynamic" ENGINE.
   4970      Also, add stub code to each that makes building them as self-contained
   4971      shared-libraries easier (see README.ENGINE).
   4972      [Geoff Thorpe]
   4973 
   4974   *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
   4975      implementations into applications that are completely implemented in
   4976      self-contained shared-libraries. The "dynamic" ENGINE exposes control
   4977      commands that can be used to configure what shared-library to load and
   4978      to control aspects of the way it is handled. Also, made an update to
   4979      the README.ENGINE file that brings its information up-to-date and
   4980      provides some information and instructions on the "dynamic" ENGINE
   4981      (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
   4982      [Geoff Thorpe]
   4983 
   4984   *) Make it possible to unload ranges of ERR strings with a new
   4985      "ERR_unload_strings" function.
   4986      [Geoff Thorpe]
   4987 
   4988   *) Add a copy() function to EVP_MD.
   4989      [Ben Laurie]
   4990 
   4991   *) Make EVP_MD routines take a context pointer instead of just the
   4992      md_data void pointer.
   4993      [Ben Laurie]
   4994 
   4995   *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
   4996      that the digest can only process a single chunk of data
   4997      (typically because it is provided by a piece of
   4998      hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
   4999      is only going to provide a single chunk of data, and hence the
   5000      framework needn't accumulate the data for oneshot drivers.
   5001      [Ben Laurie]
   5002 
   5003   *) As with "ERR", make it possible to replace the underlying "ex_data"
   5004      functions. This change also alters the storage and management of global
   5005      ex_data state - it's now all inside ex_data.c and all "class" code (eg.
   5006      RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
   5007      index counters. The API functions that use this state have been changed
   5008      to take a "class_index" rather than pointers to the class's local STACK
   5009      and counter, and there is now an API function to dynamically create new
   5010      classes. This centralisation allows us to (a) plug a lot of the
   5011      thread-safety problems that existed, and (b) makes it possible to clean
   5012      up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
   5013      such data would previously have always leaked in application code and
   5014      workarounds were in place to make the memory debugging turn a blind eye
   5015      to it. Application code that doesn't use this new function will still
   5016      leak as before, but their memory debugging output will announce it now
   5017      rather than letting it slide.
   5018 
   5019      Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
   5020      induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
   5021      has a return value to indicate success or failure.
   5022      [Geoff Thorpe]
   5023 
   5024   *) Make it possible to replace the underlying "ERR" functions such that the
   5025      global state (2 LHASH tables and 2 locks) is only used by the "default"
   5026      implementation. This change also adds two functions to "get" and "set"
   5027      the implementation prior to it being automatically set the first time
   5028      any other ERR function takes place. Ie. an application can call "get",
   5029      pass the return value to a module it has just loaded, and that module
   5030      can call its own "set" function using that value. This means the
   5031      module's "ERR" operations will use (and modify) the error state in the
   5032      application and not in its own statically linked copy of OpenSSL code.
   5033      [Geoff Thorpe]
   5034 
   5035   *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
   5036      reference counts. This performs normal REF_PRINT/REF_CHECK macros on
   5037      the operation, and provides a more encapsulated way for external code
   5038      (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
   5039      to use these functions rather than manually incrementing the counts.
   5040 
   5041      Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
   5042      [Geoff Thorpe]
   5043 
   5044   *) Add EVP test program.
   5045      [Ben Laurie]
   5046 
   5047   *) Add symmetric cipher support to ENGINE. Expect the API to change!
   5048      [Ben Laurie]
   5049 
   5050   *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
   5051      X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
   5052      X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
   5053      These allow a CRL to be built without having to access X509_CRL fields
   5054      directly. Modify 'ca' application to use new functions.
   5055      [Steve Henson]
   5056 
   5057   *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
   5058      bug workarounds. Rollback attack detection is a security feature.
   5059      The problem will only arise on OpenSSL servers when TLSv1 is not
   5060      available (sslv3_server_method() or SSL_OP_NO_TLSv1).
   5061      Software authors not wanting to support TLSv1 will have special reasons
   5062      for their choice and can explicitly enable this option.
   5063      [Bodo Moeller, Lutz Jaenicke]
   5064 
   5065   *) Rationalise EVP so it can be extended: don't include a union of
   5066      cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
   5067      (similar to those existing for EVP_CIPHER_CTX).
   5068      Usage example:
   5069 
   5070          EVP_MD_CTX md;
   5071 
   5072          EVP_MD_CTX_init(&md);             /* new function call */
   5073          EVP_DigestInit(&md, EVP_sha1());
   5074          EVP_DigestUpdate(&md, in, len);
   5075          EVP_DigestFinal(&md, out, NULL);
   5076          EVP_MD_CTX_cleanup(&md);          /* new function call */
   5077 
   5078      [Ben Laurie]
   5079 
   5080   *) Make DES key schedule conform to the usual scheme, as well as
   5081      correcting its structure. This means that calls to DES functions
   5082      now have to pass a pointer to a des_key_schedule instead of a
   5083      plain des_key_schedule (which was actually always a pointer
   5084      anyway): E.g.,
   5085 
   5086          des_key_schedule ks;
   5087 
   5088 	 des_set_key_checked(..., &ks);
   5089 	 des_ncbc_encrypt(..., &ks, ...);
   5090 
   5091      (Note that a later change renames 'des_...' into 'DES_...'.)
   5092      [Ben Laurie]
   5093 
   5094   *) Initial reduction of linker bloat: the use of some functions, such as
   5095      PEM causes large amounts of unused functions to be linked in due to
   5096      poor organisation. For example pem_all.c contains every PEM function
   5097      which has a knock on effect of linking in large amounts of (unused)
   5098      ASN1 code. Grouping together similar functions and splitting unrelated
   5099      functions prevents this.
   5100      [Steve Henson]
   5101 
   5102   *) Cleanup of EVP macros.
   5103      [Ben Laurie]
   5104 
   5105   *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
   5106      correct _ecb suffix.
   5107      [Ben Laurie]
   5108 
   5109   *) Add initial OCSP responder support to ocsp application. The
   5110      revocation information is handled using the text based index
   5111      use by the ca application. The responder can either handle
   5112      requests generated internally, supplied in files (for example
   5113      via a CGI script) or using an internal minimal server.
   5114      [Steve Henson]
   5115 
   5116   *) Add configuration choices to get zlib compression for TLS.
   5117      [Richard Levitte]
   5118 
   5119   *) Changes to Kerberos SSL for RFC 2712 compliance:
   5120      1.  Implemented real KerberosWrapper, instead of just using
   5121          KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw (a] sxw.org.uk>]
   5122      2.  Implemented optional authenticator field of KerberosWrapper.
   5123 
   5124      Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
   5125      and authenticator structs; see crypto/krb5/.
   5126 
   5127      Generalized Kerberos calls to support multiple Kerberos libraries.
   5128      [Vern Staats <staatsvr (a] asc.hpc.mil>,
   5129       Jeffrey Altman <jaltman (a] columbia.edu>
   5130       via Richard Levitte]
   5131 
   5132   *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
   5133      already does with RSA. testdsa.h now has 'priv_key/pub_key'
   5134      values for each of the key sizes rather than having just
   5135      parameters (and 'speed' generating keys each time).
   5136      [Geoff Thorpe]
   5137 
   5138   *) Speed up EVP routines.
   5139      Before:
   5140 encrypt
   5141 type              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
   5142 des-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
   5143 des-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
   5144 des-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
   5145 decrypt
   5146 des-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
   5147 des-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
   5148 des-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
   5149      After:
   5150 encrypt
   5151 des-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
   5152 decrypt
   5153 des-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
   5154      [Ben Laurie]
   5155 
   5156   *) Added the OS2-EMX target.
   5157      ["Brian Havard" <brianh (a] kheldar.apana.org.au> and Richard Levitte]
   5158 
   5159   *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
   5160      to support NCONF routines in extension code. New function CONF_set_nconf()
   5161      to allow functions which take an NCONF to also handle the old LHASH
   5162      structure: this means that the old CONF compatible routines can be
   5163      retained (in particular wrt extensions) without having to duplicate the
   5164      code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
   5165      [Steve Henson]
   5166 
   5167   *) Enhance the general user interface with mechanisms for inner control
   5168      and with possibilities to have yes/no kind of prompts.
   5169      [Richard Levitte]
   5170 
   5171   *) Change all calls to low level digest routines in the library and
   5172      applications to use EVP. Add missing calls to HMAC_cleanup() and
   5173      don't assume HMAC_CTX can be copied using memcpy().
   5174      [Verdon Walker <VWalker (a] novell.com>, Steve Henson]
   5175 
   5176   *) Add the possibility to control engines through control names but with
   5177      arbitrary arguments instead of just a string.
   5178      Change the key loaders to take a UI_METHOD instead of a callback
   5179      function pointer.  NOTE: this breaks binary compatibility with earlier
   5180      versions of OpenSSL [engine].
   5181      Adapt the nCipher code for these new conditions and add a card insertion
   5182      callback.
   5183      [Richard Levitte]
   5184 
   5185   *) Enhance the general user interface with mechanisms to better support
   5186      dialog box interfaces, application-defined prompts, the possibility
   5187      to use defaults (for example default passwords from somewhere else)
   5188      and interrupts/cancellations.
   5189      [Richard Levitte]
   5190 
   5191   *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
   5192      attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
   5193      [Steve Henson]
   5194 
   5195   *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
   5196      tidy up some unnecessarily weird code in 'sk_new()').
   5197      [Geoff, reported by Diego Tartara <dtartara (a] novamens.com>]
   5198 
   5199   *) Change the key loading routines for ENGINEs to use the same kind
   5200      callback (pem_password_cb) as all other routines that need this
   5201      kind of callback.
   5202      [Richard Levitte]
   5203 
   5204   *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
   5205      256 bit (=32 byte) keys. Of course seeding with more entropy bytes
   5206      than this minimum value is recommended.
   5207      [Lutz Jaenicke]
   5208 
   5209   *) New random seeder for OpenVMS, using the system process statistics
   5210      that are easily reachable.
   5211      [Richard Levitte]
   5212 
   5213   *) Windows apparently can't transparently handle global
   5214      variables defined in DLLs. Initialisations such as:
   5215 
   5216         const ASN1_ITEM *it = &ASN1_INTEGER_it;
   5217 
   5218      wont compile. This is used by the any applications that need to
   5219      declare their own ASN1 modules. This was fixed by adding the option
   5220      EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
   5221      needed for static libraries under Win32.
   5222      [Steve Henson]
   5223 
   5224   *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
   5225      setting of purpose and trust fields. New X509_STORE trust and
   5226      purpose functions and tidy up setting in other SSL functions.
   5227      [Steve Henson]
   5228 
   5229   *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
   5230      structure. These are inherited by X509_STORE_CTX when it is 
   5231      initialised. This allows various defaults to be set in the
   5232      X509_STORE structure (such as flags for CRL checking and custom
   5233      purpose or trust settings) for functions which only use X509_STORE_CTX
   5234      internally such as S/MIME.
   5235 
   5236      Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
   5237      trust settings if they are not set in X509_STORE. This allows X509_STORE
   5238      purposes and trust (in S/MIME for example) to override any set by default.
   5239 
   5240      Add command line options for CRL checking to smime, s_client and s_server
   5241      applications.
   5242      [Steve Henson]
   5243 
   5244   *) Initial CRL based revocation checking. If the CRL checking flag(s)
   5245      are set then the CRL is looked up in the X509_STORE structure and
   5246      its validity and signature checked, then if the certificate is found
   5247      in the CRL the verify fails with a revoked error.
   5248 
   5249      Various new CRL related callbacks added to X509_STORE_CTX structure.
   5250 
   5251      Command line options added to 'verify' application to support this.
   5252 
   5253      This needs some additional work, such as being able to handle multiple
   5254      CRLs with different times, extension based lookup (rather than just
   5255      by subject name) and ultimately more complete V2 CRL extension
   5256      handling.
   5257      [Steve Henson]
   5258 
   5259   *) Add a general user interface API (crypto/ui/).  This is designed
   5260      to replace things like des_read_password and friends (backward
   5261      compatibility functions using this new API are provided).
   5262      The purpose is to remove prompting functions from the DES code
   5263      section as well as provide for prompting through dialog boxes in
   5264      a window system and the like.
   5265      [Richard Levitte]
   5266 
   5267   *) Add "ex_data" support to ENGINE so implementations can add state at a
   5268      per-structure level rather than having to store it globally.
   5269      [Geoff]
   5270 
   5271   *) Make it possible for ENGINE structures to be copied when retrieved by
   5272      ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
   5273      This causes the "original" ENGINE structure to act like a template,
   5274      analogous to the RSA vs. RSA_METHOD type of separation. Because of this
   5275      operational state can be localised to each ENGINE structure, despite the
   5276      fact they all share the same "methods". New ENGINE structures returned in
   5277      this case have no functional references and the return value is the single
   5278      structural reference. This matches the single structural reference returned
   5279      by ENGINE_by_id() normally, when it is incremented on the pre-existing
   5280      ENGINE structure.
   5281      [Geoff]
   5282 
   5283   *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
   5284      needs to match any other type at all we need to manually clear the
   5285      tag cache.
   5286      [Steve Henson]
   5287 
   5288   *) Changes to the "openssl engine" utility to include;
   5289      - verbosity levels ('-v', '-vv', and '-vvv') that provide information
   5290        about an ENGINE's available control commands.
   5291      - executing control commands from command line arguments using the
   5292        '-pre' and '-post' switches. '-post' is only used if '-t' is
   5293        specified and the ENGINE is successfully initialised. The syntax for
   5294        the individual commands are colon-separated, for example;
   5295 	 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
   5296      [Geoff]
   5297 
   5298   *) New dynamic control command support for ENGINEs. ENGINEs can now
   5299      declare their own commands (numbers), names (strings), descriptions,
   5300      and input types for run-time discovery by calling applications. A
   5301      subset of these commands are implicitly classed as "executable"
   5302      depending on their input type, and only these can be invoked through
   5303      the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
   5304      can be based on user input, config files, etc). The distinction is
   5305      that "executable" commands cannot return anything other than a boolean
   5306      result and can only support numeric or string input, whereas some
   5307      discoverable commands may only be for direct use through
   5308      ENGINE_ctrl(), eg. supporting the exchange of binary data, function
   5309      pointers, or other custom uses. The "executable" commands are to
   5310      support parameterisations of ENGINE behaviour that can be
   5311      unambiguously defined by ENGINEs and used consistently across any
   5312      OpenSSL-based application. Commands have been added to all the
   5313      existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
   5314      control over shared-library paths without source code alterations.
   5315      [Geoff]
   5316 
   5317   *) Changed all ENGINE implementations to dynamically allocate their
   5318      ENGINEs rather than declaring them statically. Apart from this being
   5319      necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
   5320      this also allows the implementations to compile without using the
   5321      internal engine_int.h header.
   5322      [Geoff]
   5323 
   5324   *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
   5325      'const' value. Any code that should be able to modify a RAND_METHOD
   5326      should already have non-const pointers to it (ie. they should only
   5327      modify their own ones).
   5328      [Geoff]
   5329 
   5330   *) Made a variety of little tweaks to the ENGINE code.
   5331      - "atalla" and "ubsec" string definitions were moved from header files
   5332        to C code. "nuron" string definitions were placed in variables
   5333        rather than hard-coded - allowing parameterisation of these values
   5334        later on via ctrl() commands.
   5335      - Removed unused "#if 0"'d code.
   5336      - Fixed engine list iteration code so it uses ENGINE_free() to release
   5337        structural references.
   5338      - Constified the RAND_METHOD element of ENGINE structures.
   5339      - Constified various get/set functions as appropriate and added
   5340        missing functions (including a catch-all ENGINE_cpy that duplicates
   5341        all ENGINE values onto a new ENGINE except reference counts/state).
   5342      - Removed NULL parameter checks in get/set functions. Setting a method
   5343        or function to NULL is a way of cancelling out a previously set
   5344        value.  Passing a NULL ENGINE parameter is just plain stupid anyway
   5345        and doesn't justify the extra error symbols and code.
   5346      - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
   5347        flags from engine_int.h to engine.h.
   5348      - Changed prototypes for ENGINE handler functions (init(), finish(),
   5349        ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
   5350      [Geoff]
   5351 
   5352   *) Implement binary inversion algorithm for BN_mod_inverse in addition
   5353      to the algorithm using long division.  The binary algorithm can be
   5354      used only if the modulus is odd.  On 32-bit systems, it is faster
   5355      only for relatively small moduli (roughly 20-30% for 128-bit moduli,
   5356      roughly 5-15% for 256-bit moduli), so we use it only for moduli
   5357      up to 450 bits.  In 64-bit environments, the binary algorithm
   5358      appears to be advantageous for much longer moduli; here we use it
   5359      for moduli up to 2048 bits.
   5360      [Bodo Moeller]
   5361 
   5362   *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
   5363      could not support the combine flag in choice fields.
   5364      [Steve Henson]
   5365 
   5366   *) Add a 'copy_extensions' option to the 'ca' utility. This copies
   5367      extensions from a certificate request to the certificate.
   5368      [Steve Henson]
   5369 
   5370   *) Allow multiple 'certopt' and 'nameopt' options to be separated
   5371      by commas. Add 'namopt' and 'certopt' options to the 'ca' config
   5372      file: this allows the display of the certificate about to be
   5373      signed to be customised, to allow certain fields to be included
   5374      or excluded and extension details. The old system didn't display
   5375      multicharacter strings properly, omitted fields not in the policy
   5376      and couldn't display additional details such as extensions.
   5377      [Steve Henson]
   5378 
   5379   *) Function EC_POINTs_mul for multiple scalar multiplication
   5380      of an arbitrary number of elliptic curve points
   5381           \sum scalars[i]*points[i],
   5382      optionally including the generator defined for the EC_GROUP:
   5383           scalar*generator +  \sum scalars[i]*points[i].
   5384 
   5385      EC_POINT_mul is a simple wrapper function for the typical case
   5386      that the point list has just one item (besides the optional
   5387      generator).
   5388      [Bodo Moeller]
   5389 
   5390   *) First EC_METHODs for curves over GF(p):
   5391 
   5392      EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
   5393      operations and provides various method functions that can also
   5394      operate with faster implementations of modular arithmetic.     
   5395 
   5396      EC_GFp_mont_method() reuses most functions that are part of
   5397      EC_GFp_simple_method, but uses Montgomery arithmetic.
   5398 
   5399      [Bodo Moeller; point addition and point doubling
   5400      implementation directly derived from source code provided by
   5401      Lenka Fibikova <fibikova (a] exp-math.uni-essen.de>]
   5402 
   5403   *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
   5404      crypto/ec/ec_lib.c):
   5405 
   5406      Curves are EC_GROUP objects (with an optional group generator)
   5407      based on EC_METHODs that are built into the library.
   5408 
   5409      Points are EC_POINT objects based on EC_GROUP objects.
   5410 
   5411      Most of the framework would be able to handle curves over arbitrary
   5412      finite fields, but as there are no obvious types for fields other
   5413      than GF(p), some functions are limited to that for now.
   5414      [Bodo Moeller]
   5415 
   5416   *) Add the -HTTP option to s_server.  It is similar to -WWW, but requires
   5417      that the file contains a complete HTTP response.
   5418      [Richard Levitte]
   5419 
   5420   *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
   5421      change the def and num file printf format specifier from "%-40sXXX"
   5422      to "%-39s XXX". The latter will always guarantee a space after the
   5423      field while the former will cause them to run together if the field
   5424      is 40 of more characters long.
   5425      [Steve Henson]
   5426 
   5427   *) Constify the cipher and digest 'method' functions and structures
   5428      and modify related functions to take constant EVP_MD and EVP_CIPHER
   5429      pointers.
   5430      [Steve Henson]
   5431 
   5432   *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
   5433      in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
   5434      [Bodo Moeller]
   5435 
   5436   *) Modify EVP_Digest*() routines so they now return values. Although the
   5437      internal software routines can never fail additional hardware versions
   5438      might.
   5439      [Steve Henson]
   5440 
   5441   *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
   5442 
   5443      Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
   5444      (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
   5445 
   5446      ASN1 error codes
   5447           ERR_R_NESTED_ASN1_ERROR
   5448           ...
   5449           ERR_R_MISSING_ASN1_EOS
   5450      were 4 .. 9, conflicting with
   5451           ERR_LIB_RSA (= ERR_R_RSA_LIB)
   5452           ...
   5453           ERR_LIB_PEM (= ERR_R_PEM_LIB).
   5454      They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
   5455 
   5456      Add new error code 'ERR_R_INTERNAL_ERROR'.
   5457      [Bodo Moeller]
   5458 
   5459   *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
   5460      suffices.
   5461      [Bodo Moeller]
   5462 
   5463   *) New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
   5464      sets the subject name for a new request or supersedes the
   5465      subject name in a given request. Formats that can be parsed are
   5466           'CN=Some Name, OU=myOU, C=IT'
   5467      and
   5468           'CN=Some Name/OU=myOU/C=IT'.
   5469 
   5470      Add options '-batch' and '-verbose' to 'openssl req'.
   5471      [Massimiliano Pala <madwolf (a] hackmasters.net>]
   5472 
   5473   *) Introduce the possibility to access global variables through
   5474      functions on platform were that's the best way to handle exporting
   5475      global variables in shared libraries.  To enable this functionality,
   5476      one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
   5477      "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
   5478      is normally done by Configure or something similar).
   5479 
   5480      To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
   5481      in the source file (foo.c) like this:
   5482 
   5483 	OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
   5484 	OPENSSL_IMPLEMENT_GLOBAL(double,bar);
   5485 
   5486      To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
   5487      and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
   5488 
   5489 	OPENSSL_DECLARE_GLOBAL(int,foo);
   5490 	#define foo OPENSSL_GLOBAL_REF(foo)
   5491 	OPENSSL_DECLARE_GLOBAL(double,bar);
   5492 	#define bar OPENSSL_GLOBAL_REF(bar)
   5493 
   5494      The #defines are very important, and therefore so is including the
   5495      header file everywhere where the defined globals are used.
   5496 
   5497      The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
   5498      of ASN.1 items, but that structure is a bit different.
   5499 
   5500      The largest change is in util/mkdef.pl which has been enhanced with
   5501      better and easier to understand logic to choose which symbols should
   5502      go into the Windows .def files as well as a number of fixes and code
   5503      cleanup (among others, algorithm keywords are now sorted
   5504      lexicographically to avoid constant rewrites).
   5505      [Richard Levitte]
   5506 
   5507   *) In BN_div() keep a copy of the sign of 'num' before writing the
   5508      result to 'rm' because if rm==num the value will be overwritten
   5509      and produce the wrong result if 'num' is negative: this caused
   5510      problems with BN_mod() and BN_nnmod().
   5511      [Steve Henson]
   5512 
   5513   *) Function OCSP_request_verify(). This checks the signature on an
   5514      OCSP request and verifies the signer certificate. The signer
   5515      certificate is just checked for a generic purpose and OCSP request
   5516      trust settings.
   5517      [Steve Henson]
   5518 
   5519   *) Add OCSP_check_validity() function to check the validity of OCSP
   5520      responses. OCSP responses are prepared in real time and may only
   5521      be a few seconds old. Simply checking that the current time lies
   5522      between thisUpdate and nextUpdate max reject otherwise valid responses
   5523      caused by either OCSP responder or client clock inaccuracy. Instead
   5524      we allow thisUpdate and nextUpdate to fall within a certain period of
   5525      the current time. The age of the response can also optionally be
   5526      checked. Two new options -validity_period and -status_age added to
   5527      ocsp utility.
   5528      [Steve Henson]
   5529 
   5530   *) If signature or public key algorithm is unrecognized print out its
   5531      OID rather that just UNKNOWN.
   5532      [Steve Henson]
   5533 
   5534   *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
   5535      OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
   5536      ID to be generated from the issuer certificate alone which can then be
   5537      passed to OCSP_id_issuer_cmp().
   5538      [Steve Henson]
   5539 
   5540   *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
   5541      ASN1 modules to export functions returning ASN1_ITEM pointers
   5542      instead of the ASN1_ITEM structures themselves. This adds several
   5543      new macros which allow the underlying ASN1 function/structure to
   5544      be accessed transparently. As a result code should not use ASN1_ITEM
   5545      references directly (such as &X509_it) but instead use the relevant
   5546      macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
   5547      use of the new ASN1 code on platforms where exporting structures
   5548      is problematical (for example in shared libraries) but exporting
   5549      functions returning pointers to structures is not.
   5550      [Steve Henson]
   5551 
   5552   *) Add support for overriding the generation of SSL/TLS session IDs.
   5553      These callbacks can be registered either in an SSL_CTX or per SSL.
   5554      The purpose of this is to allow applications to control, if they wish,
   5555      the arbitrary values chosen for use as session IDs, particularly as it
   5556      can be useful for session caching in multiple-server environments. A
   5557      command-line switch for testing this (and any client code that wishes
   5558      to use such a feature) has been added to "s_server".
   5559      [Geoff Thorpe, Lutz Jaenicke]
   5560 
   5561   *) Modify mkdef.pl to recognise and parse preprocessor conditionals
   5562      of the form '#if defined(...) || defined(...) || ...' and
   5563      '#if !defined(...) && !defined(...) && ...'.  This also avoids
   5564      the growing number of special cases it was previously handling.
   5565      [Richard Levitte]
   5566 
   5567   *) Make all configuration macros available for application by making
   5568      sure they are available in opensslconf.h, by giving them names starting
   5569      with "OPENSSL_" to avoid conflicts with other packages and by making
   5570      sure e_os2.h will cover all platform-specific cases together with
   5571      opensslconf.h.
   5572      Additionally, it is now possible to define configuration/platform-
   5573      specific names (called "system identities").  In the C code, these
   5574      are prefixed with "OPENSSL_SYSNAME_".  e_os2.h will create another
   5575      macro with the name beginning with "OPENSSL_SYS_", which is determined
   5576      from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
   5577      what is available.
   5578      [Richard Levitte]
   5579 
   5580   *) New option -set_serial to 'req' and 'x509' this allows the serial
   5581      number to use to be specified on the command line. Previously self
   5582      signed certificates were hard coded with serial number 0 and the 
   5583      CA options of 'x509' had to use a serial number in a file which was
   5584      auto incremented.
   5585      [Steve Henson]
   5586 
   5587   *) New options to 'ca' utility to support V2 CRL entry extensions.
   5588      Currently CRL reason, invalidity date and hold instruction are
   5589      supported. Add new CRL extensions to V3 code and some new objects.
   5590      [Steve Henson]
   5591 
   5592   *) New function EVP_CIPHER_CTX_set_padding() this is used to
   5593      disable standard block padding (aka PKCS#5 padding) in the EVP
   5594      API, which was previously mandatory. This means that the data is
   5595      not padded in any way and so the total length much be a multiple
   5596      of the block size, otherwise an error occurs.
   5597      [Steve Henson]
   5598 
   5599   *) Initial (incomplete) OCSP SSL support.
   5600      [Steve Henson]
   5601 
   5602   *) New function OCSP_parse_url(). This splits up a URL into its host,
   5603      port and path components: primarily to parse OCSP URLs. New -url
   5604      option to ocsp utility.
   5605      [Steve Henson]
   5606 
   5607   *) New nonce behavior. The return value of OCSP_check_nonce() now 
   5608      reflects the various checks performed. Applications can decide
   5609      whether to tolerate certain situations such as an absent nonce
   5610      in a response when one was present in a request: the ocsp application
   5611      just prints out a warning. New function OCSP_add1_basic_nonce()
   5612      this is to allow responders to include a nonce in a response even if
   5613      the request is nonce-less.
   5614      [Steve Henson]
   5615 
   5616   *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
   5617      skipped when using openssl x509 multiple times on a single input file,
   5618      e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
   5619      [Bodo Moeller]
   5620 
   5621   *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
   5622      set string type: to handle setting ASN1_TIME structures. Fix ca
   5623      utility to correctly initialize revocation date of CRLs.
   5624      [Steve Henson]
   5625 
   5626   *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
   5627      the clients preferred ciphersuites and rather use its own preferences.
   5628      Should help to work around M$ SGC (Server Gated Cryptography) bug in
   5629      Internet Explorer by ensuring unchanged hash method during stepup.
   5630      (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
   5631      [Lutz Jaenicke]
   5632 
   5633   *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
   5634      to aes and add a new 'exist' option to print out symbols that don't
   5635      appear to exist.
   5636      [Steve Henson]
   5637 
   5638   *) Additional options to ocsp utility to allow flags to be set and
   5639      additional certificates supplied.
   5640      [Steve Henson]
   5641 
   5642   *) Add the option -VAfile to 'openssl ocsp', so the user can give the
   5643      OCSP client a number of certificate to only verify the response
   5644      signature against.
   5645      [Richard Levitte]
   5646 
   5647   *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
   5648      handle the new API. Currently only ECB, CBC modes supported. Add new
   5649      AES OIDs.
   5650 
   5651      Add TLS AES ciphersuites as described in RFC3268, "Advanced
   5652      Encryption Standard (AES) Ciphersuites for Transport Layer
   5653      Security (TLS)".  (In beta versions of OpenSSL 0.9.7, these were
   5654      not enabled by default and were not part of the "ALL" ciphersuite
   5655      alias because they were not yet official; they could be
   5656      explicitly requested by specifying the "AESdraft" ciphersuite
   5657      group alias.  In the final release of OpenSSL 0.9.7, the group
   5658      alias is called "AES" and is part of "ALL".)
   5659      [Ben Laurie, Steve  Henson, Bodo Moeller]
   5660 
   5661   *) New function OCSP_copy_nonce() to copy nonce value (if present) from
   5662      request to response.
   5663      [Steve Henson]
   5664 
   5665   *) Functions for OCSP responders. OCSP_request_onereq_count(),
   5666      OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
   5667      extract information from a certificate request. OCSP_response_create()
   5668      creates a response and optionally adds a basic response structure.
   5669      OCSP_basic_add1_status() adds a complete single response to a basic
   5670      response and returns the OCSP_SINGLERESP structure just added (to allow
   5671      extensions to be included for example). OCSP_basic_add1_cert() adds a
   5672      certificate to a basic response and OCSP_basic_sign() signs a basic
   5673      response with various flags. New helper functions ASN1_TIME_check()
   5674      (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
   5675      (converts ASN1_TIME to GeneralizedTime).
   5676      [Steve Henson]
   5677 
   5678   *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
   5679      in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
   5680      structure from a certificate. X509_pubkey_digest() digests the public_key
   5681      contents: this is used in various key identifiers. 
   5682      [Steve Henson]
   5683 
   5684   *) Make sk_sort() tolerate a NULL argument.
   5685      [Steve Henson reported by Massimiliano Pala <madwolf (a] comune.modena.it>]
   5686 
   5687   *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
   5688      passed by the function are trusted implicitly. If any of them signed the
   5689      response then it is assumed to be valid and is not verified.
   5690      [Steve Henson]
   5691 
   5692   *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
   5693      to data. This was previously part of the PKCS7 ASN1 code. This
   5694      was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
   5695      [Steve Henson, reported by Kenneth R. Robinette
   5696 				<support (a] securenetterm.com>]
   5697 
   5698   *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
   5699      routines: without these tracing memory leaks is very painful.
   5700      Fix leaks in PKCS12 and PKCS7 routines.
   5701      [Steve Henson]
   5702 
   5703   *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
   5704      Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
   5705      effectively meant GeneralizedTime would never be used. Now it
   5706      is initialised to -1 but X509_time_adj() now has to check the value
   5707      and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
   5708      V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
   5709      [Steve Henson, reported by Kenneth R. Robinette
   5710 				<support (a] securenetterm.com>]
   5711 
   5712   *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
   5713      result in a zero length in the ASN1_INTEGER structure which was
   5714      not consistent with the structure when d2i_ASN1_INTEGER() was used
   5715      and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
   5716      to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
   5717      where it did not print out a minus for negative ASN1_INTEGER.
   5718      [Steve Henson]
   5719 
   5720   *) Add summary printout to ocsp utility. The various functions which
   5721      convert status values to strings have been renamed to:
   5722      OCSP_response_status_str(), OCSP_cert_status_str() and
   5723      OCSP_crl_reason_str() and are no longer static. New options
   5724      to verify nonce values and to disable verification. OCSP response
   5725      printout format cleaned up.
   5726      [Steve Henson]
   5727 
   5728   *) Add additional OCSP certificate checks. These are those specified
   5729      in RFC2560. This consists of two separate checks: the CA of the
   5730      certificate being checked must either be the OCSP signer certificate
   5731      or the issuer of the OCSP signer certificate. In the latter case the
   5732      OCSP signer certificate must contain the OCSP signing extended key
   5733      usage. This check is performed by attempting to match the OCSP
   5734      signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
   5735      in the OCSP_CERTID structures of the response.
   5736      [Steve Henson]
   5737 
   5738   *) Initial OCSP certificate verification added to OCSP_basic_verify()
   5739      and related routines. This uses the standard OpenSSL certificate
   5740      verify routines to perform initial checks (just CA validity) and
   5741      to obtain the certificate chain. Then additional checks will be
   5742      performed on the chain. Currently the root CA is checked to see
   5743      if it is explicitly trusted for OCSP signing. This is used to set
   5744      a root CA as a global signing root: that is any certificate that
   5745      chains to that CA is an acceptable OCSP signing certificate.
   5746      [Steve Henson]
   5747 
   5748   *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
   5749      extensions from a separate configuration file.
   5750      As when reading extensions from the main configuration file,
   5751      the '-extensions ...' option may be used for specifying the
   5752      section to use.
   5753      [Massimiliano Pala <madwolf (a] comune.modena.it>]
   5754 
   5755   *) New OCSP utility. Allows OCSP requests to be generated or
   5756      read. The request can be sent to a responder and the output
   5757      parsed, outputed or printed in text form. Not complete yet:
   5758      still needs to check the OCSP response validity.
   5759      [Steve Henson]
   5760 
   5761   *) New subcommands for 'openssl ca':
   5762      'openssl ca -status <serial>' prints the status of the cert with
   5763      the given serial number (according to the index file).
   5764      'openssl ca -updatedb' updates the expiry status of certificates
   5765      in the index file.
   5766      [Massimiliano Pala <madwolf (a] comune.modena.it>]
   5767 
   5768   *) New '-newreq-nodes' command option to CA.pl.  This is like
   5769      '-newreq', but calls 'openssl req' with the '-nodes' option
   5770      so that the resulting key is not encrypted.
   5771      [Damien Miller <djm (a] mindrot.org>]
   5772 
   5773   *) New configuration for the GNU Hurd.
   5774      [Jonathan Bartlett <johnnyb (a] wolfram.com> via Richard Levitte]
   5775 
   5776   *) Initial code to implement OCSP basic response verify. This
   5777      is currently incomplete. Currently just finds the signer's
   5778      certificate and verifies the signature on the response.
   5779      [Steve Henson]
   5780 
   5781   *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
   5782      value of OPENSSLDIR.  This is available via the new '-d' option
   5783      to 'openssl version', and is also included in 'openssl version -a'.
   5784      [Bodo Moeller]
   5785 
   5786   *) Allowing defining memory allocation callbacks that will be given
   5787      file name and line number information in additional arguments
   5788      (a const char* and an int).  The basic functionality remains, as
   5789      well as the original possibility to just replace malloc(),
   5790      realloc() and free() by functions that do not know about these
   5791      additional arguments.  To register and find out the current
   5792      settings for extended allocation functions, the following
   5793      functions are provided:
   5794 
   5795 	CRYPTO_set_mem_ex_functions
   5796 	CRYPTO_set_locked_mem_ex_functions
   5797 	CRYPTO_get_mem_ex_functions
   5798 	CRYPTO_get_locked_mem_ex_functions
   5799 
   5800      These work the same way as CRYPTO_set_mem_functions and friends.
   5801      CRYPTO_get_[locked_]mem_functions now writes 0 where such an
   5802      extended allocation function is enabled.
   5803      Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
   5804      a conventional allocation function is enabled.
   5805      [Richard Levitte, Bodo Moeller]
   5806 
   5807   *) Finish off removing the remaining LHASH function pointer casts.
   5808      There should no longer be any prototype-casting required when using
   5809      the LHASH abstraction, and any casts that remain are "bugs". See
   5810      the callback types and macros at the head of lhash.h for details
   5811      (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
   5812      [Geoff Thorpe]
   5813 
   5814   *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
   5815      If /dev/[u]random devices are not available or do not return enough
   5816      entropy, EGD style sockets (served by EGD or PRNGD) will automatically
   5817      be queried.
   5818      The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
   5819      /etc/entropy will be queried once each in this sequence, quering stops
   5820      when enough entropy was collected without querying more sockets.
   5821      [Lutz Jaenicke]
   5822 
   5823   *) Change the Unix RAND_poll() variant to be able to poll several
   5824      random devices, as specified by DEVRANDOM, until a sufficient amount
   5825      of data has been collected.   We spend at most 10 ms on each file
   5826      (select timeout) and read in non-blocking mode.  DEVRANDOM now
   5827      defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
   5828      (previously it was just the string "/dev/urandom"), so on typical
   5829      platforms the 10 ms delay will never occur.
   5830      Also separate out the Unix variant to its own file, rand_unix.c.
   5831      For VMS, there's a currently-empty rand_vms.c.
   5832      [Richard Levitte]
   5833 
   5834   *) Move OCSP client related routines to ocsp_cl.c. These
   5835      provide utility functions which an application needing
   5836      to issue a request to an OCSP responder and analyse the
   5837      response will typically need: as opposed to those which an
   5838      OCSP responder itself would need which will be added later.
   5839 
   5840      OCSP_request_sign() signs an OCSP request with an API similar
   5841      to PKCS7_sign(). OCSP_response_status() returns status of OCSP
   5842      response. OCSP_response_get1_basic() extracts basic response
   5843      from response. OCSP_resp_find_status(): finds and extracts status
   5844      information from an OCSP_CERTID structure (which will be created
   5845      when the request structure is built). These are built from lower
   5846      level functions which work on OCSP_SINGLERESP structures but
   5847      wont normally be used unless the application wishes to examine
   5848      extensions in the OCSP response for example.
   5849 
   5850      Replace nonce routines with a pair of functions.
   5851      OCSP_request_add1_nonce() adds a nonce value and optionally
   5852      generates a random value. OCSP_check_nonce() checks the
   5853      validity of the nonce in an OCSP response.
   5854      [Steve Henson]
   5855 
   5856   *) Change function OCSP_request_add() to OCSP_request_add0_id().
   5857      This doesn't copy the supplied OCSP_CERTID and avoids the
   5858      need to free up the newly created id. Change return type
   5859      to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
   5860      This can then be used to add extensions to the request.
   5861      Deleted OCSP_request_new(), since most of its functionality
   5862      is now in OCSP_REQUEST_new() (and the case insensitive name
   5863      clash) apart from the ability to set the request name which
   5864      will be added elsewhere.
   5865      [Steve Henson]
   5866 
   5867   *) Update OCSP API. Remove obsolete extensions argument from
   5868      various functions. Extensions are now handled using the new
   5869      OCSP extension code. New simple OCSP HTTP function which 
   5870      can be used to send requests and parse the response.
   5871      [Steve Henson]
   5872 
   5873   *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
   5874      ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
   5875      uses the special reorder version of SET OF to sort the attributes
   5876      and reorder them to match the encoded order. This resolves a long
   5877      standing problem: a verify on a PKCS7 structure just after signing
   5878      it used to fail because the attribute order did not match the
   5879      encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
   5880      it uses the received order. This is necessary to tolerate some broken
   5881      software that does not order SET OF. This is handled by encoding
   5882      as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
   5883      to produce the required SET OF.
   5884      [Steve Henson]
   5885 
   5886   *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
   5887      OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
   5888      files to get correct declarations of the ASN.1 item variables.
   5889      [Richard Levitte]
   5890 
   5891   *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
   5892      PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
   5893      asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
   5894      NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
   5895      New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
   5896      ASN1_ITEM and no wrapper functions.
   5897      [Steve Henson]
   5898 
   5899   *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
   5900      replace the old function pointer based I/O routines. Change most of
   5901      the *_d2i_bio() and *_d2i_fp() functions to use these.
   5902      [Steve Henson]
   5903 
   5904   *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
   5905      lines, recognice more "algorithms" that can be deselected, and make
   5906      it complain about algorithm deselection that isn't recognised.
   5907      [Richard Levitte]
   5908 
   5909   *) New ASN1 functions to handle dup, sign, verify, digest, pack and
   5910      unpack operations in terms of ASN1_ITEM. Modify existing wrappers
   5911      to use new functions. Add NO_ASN1_OLD which can be set to remove
   5912      some old style ASN1 functions: this can be used to determine if old
   5913      code will still work when these eventually go away.
   5914      [Steve Henson]
   5915 
   5916   *) New extension functions for OCSP structures, these follow the
   5917      same conventions as certificates and CRLs.
   5918      [Steve Henson]
   5919 
   5920   *) New function X509V3_add1_i2d(). This automatically encodes and
   5921      adds an extension. Its behaviour can be customised with various
   5922      flags to append, replace or delete. Various wrappers added for
   5923      certifcates and CRLs.
   5924      [Steve Henson]
   5925 
   5926   *) Fix to avoid calling the underlying ASN1 print routine when
   5927      an extension cannot be parsed. Correct a typo in the
   5928      OCSP_SERVICELOC extension. Tidy up print OCSP format.
   5929      [Steve Henson]
   5930 
   5931   *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
   5932      entries for variables.
   5933      [Steve Henson]
   5934 
   5935   *) Add functionality to apps/openssl.c for detecting locking
   5936      problems: As the program is single-threaded, all we have
   5937      to do is register a locking callback using an array for
   5938      storing which locks are currently held by the program.
   5939      [Bodo Moeller]
   5940 
   5941   *) Use a lock around the call to CRYPTO_get_ex_new_index() in
   5942      SSL_get_ex_data_X509_STORE_idx(), which is used in
   5943      ssl_verify_cert_chain() and thus can be called at any time
   5944      during TLS/SSL handshakes so that thread-safety is essential.
   5945      Unfortunately, the ex_data design is not at all suited
   5946      for multi-threaded use, so it probably should be abolished.
   5947      [Bodo Moeller]
   5948 
   5949   *) Added Broadcom "ubsec" ENGINE to OpenSSL.
   5950      [Broadcom, tweaked and integrated by Geoff Thorpe]
   5951 
   5952   *) Move common extension printing code to new function
   5953      X509V3_print_extensions(). Reorganise OCSP print routines and
   5954      implement some needed OCSP ASN1 functions. Add OCSP extensions.
   5955      [Steve Henson]
   5956 
   5957   *) New function X509_signature_print() to remove duplication in some
   5958      print routines.
   5959      [Steve Henson]
   5960 
   5961   *) Add a special meaning when SET OF and SEQUENCE OF flags are both
   5962      set (this was treated exactly the same as SET OF previously). This
   5963      is used to reorder the STACK representing the structure to match the
   5964      encoding. This will be used to get round a problem where a PKCS7
   5965      structure which was signed could not be verified because the STACK
   5966      order did not reflect the encoded order.
   5967      [Steve Henson]
   5968 
   5969   *) Reimplement the OCSP ASN1 module using the new code.
   5970      [Steve Henson]
   5971 
   5972   *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
   5973      for its ASN1 operations. The old style function pointers still exist
   5974      for now but they will eventually go away.
   5975      [Steve Henson]
   5976 
   5977   *) Merge in replacement ASN1 code from the ASN1 branch. This almost
   5978      completely replaces the old ASN1 functionality with a table driven
   5979      encoder and decoder which interprets an ASN1_ITEM structure describing
   5980      the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
   5981      largely maintained. Almost all of the old asn1_mac.h macro based ASN1
   5982      has also been converted to the new form.
   5983      [Steve Henson]
   5984 
   5985   *) Change BN_mod_exp_recp so that negative moduli are tolerated
   5986      (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
   5987      so that BN_mod_exp_mont and BN_mod_exp_mont_word work
   5988      for negative moduli.
   5989      [Bodo Moeller]
   5990 
   5991   *) Fix BN_uadd and BN_usub: Always return non-negative results instead
   5992      of not touching the result's sign bit.
   5993      [Bodo Moeller]
   5994 
   5995   *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
   5996      set.
   5997      [Bodo Moeller]
   5998 
   5999   *) Changed the LHASH code to use prototypes for callbacks, and created
   6000      macros to declare and implement thin (optionally static) functions
   6001      that provide type-safety and avoid function pointer casting for the
   6002      type-specific callbacks.
   6003      [Geoff Thorpe]
   6004 
   6005   *) Added Kerberos Cipher Suites to be used with TLS, as written in
   6006      RFC 2712.
   6007      [Veers Staats <staatsvr (a] asc.hpc.mil>,
   6008       Jeffrey Altman <jaltman (a] columbia.edu>, via Richard Levitte]
   6009 
   6010   *) Reformat the FAQ so the different questions and answers can be divided
   6011      in sections depending on the subject.
   6012      [Richard Levitte]
   6013 
   6014   *) Have the zlib compression code load ZLIB.DLL dynamically under
   6015      Windows.
   6016      [Richard Levitte]
   6017 
   6018   *) New function BN_mod_sqrt for computing square roots modulo a prime
   6019      (using the probabilistic Tonelli-Shanks algorithm unless
   6020      p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
   6021      be handled deterministically).
   6022      [Lenka Fibikova <fibikova (a] exp-math.uni-essen.de>, Bodo Moeller]
   6023 
   6024   *) Make BN_mod_inverse faster by explicitly handling small quotients
   6025      in the Euclid loop. (Speed gain about 20% for small moduli [256 or
   6026      512 bits], about 30% for larger ones [1024 or 2048 bits].)
   6027      [Bodo Moeller]
   6028 
   6029   *) New function BN_kronecker.
   6030      [Bodo Moeller]
   6031 
   6032   *) Fix BN_gcd so that it works on negative inputs; the result is
   6033      positive unless both parameters are zero.
   6034      Previously something reasonably close to an infinite loop was
   6035      possible because numbers could be growing instead of shrinking
   6036      in the implementation of Euclid's algorithm.
   6037      [Bodo Moeller]
   6038 
   6039   *) Fix BN_is_word() and BN_is_one() macros to take into account the
   6040      sign of the number in question.
   6041 
   6042      Fix BN_is_word(a,w) to work correctly for w == 0.
   6043 
   6044      The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
   6045      because its test if the absolute value of 'a' equals 'w'.
   6046      Note that BN_abs_is_word does *not* handle w == 0 reliably;
   6047      it exists mostly for use in the implementations of BN_is_zero(),
   6048      BN_is_one(), and BN_is_word().
   6049      [Bodo Moeller]
   6050 
   6051   *) New function BN_swap.
   6052      [Bodo Moeller]
   6053 
   6054   *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
   6055      the exponentiation functions are more likely to produce reasonable
   6056      results on negative inputs.
   6057      [Bodo Moeller]
   6058 
   6059   *) Change BN_mod_mul so that the result is always non-negative.
   6060      Previously, it could be negative if one of the factors was negative;
   6061      I don't think anyone really wanted that behaviour.
   6062      [Bodo Moeller]
   6063 
   6064   *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
   6065      (except for exponentiation, which stays in crypto/bn/bn_exp.c,
   6066      and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
   6067      and add new functions:
   6068 
   6069           BN_nnmod
   6070           BN_mod_sqr
   6071           BN_mod_add
   6072           BN_mod_add_quick
   6073           BN_mod_sub
   6074           BN_mod_sub_quick
   6075           BN_mod_lshift1
   6076           BN_mod_lshift1_quick
   6077           BN_mod_lshift
   6078           BN_mod_lshift_quick
   6079 
   6080      These functions always generate non-negative results.
   6081 
   6082      BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder  r
   6083      such that  |m| < r < 0,  BN_nnmod will output  rem + |m|  instead).
   6084 
   6085      BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
   6086      BN_mod_XXX(r, a, [b,] m, ctx), but requires that  a  [and  b]
   6087      be reduced modulo  m.
   6088      [Lenka Fibikova <fibikova (a] exp-math.uni-essen.de>, Bodo Moeller]
   6089 
   6090 #if 0
   6091      The following entry accidentily appeared in the CHANGES file
   6092      distributed with OpenSSL 0.9.7.  The modifications described in
   6093      it do *not* apply to OpenSSL 0.9.7.
   6094 
   6095   *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
   6096      was actually never needed) and in BN_mul().  The removal in BN_mul()
   6097      required a small change in bn_mul_part_recursive() and the addition
   6098      of the functions bn_cmp_part_words(), bn_sub_part_words() and
   6099      bn_add_part_words(), which do the same thing as bn_cmp_words(),
   6100      bn_sub_words() and bn_add_words() except they take arrays with
   6101      differing sizes.
   6102      [Richard Levitte]
   6103 #endif
   6104 
   6105   *) In 'openssl passwd', verify passwords read from the terminal
   6106      unless the '-salt' option is used (which usually means that
   6107      verification would just waste user's time since the resulting
   6108      hash is going to be compared with some given password hash)
   6109      or the new '-noverify' option is used.
   6110 
   6111      This is an incompatible change, but it does not affect
   6112      non-interactive use of 'openssl passwd' (passwords on the command
   6113      line, '-stdin' option, '-in ...' option) and thus should not
   6114      cause any problems.
   6115      [Bodo Moeller]
   6116 
   6117   *) Remove all references to RSAref, since there's no more need for it.
   6118      [Richard Levitte]
   6119 
   6120   *) Make DSO load along a path given through an environment variable
   6121      (SHLIB_PATH) with shl_load().
   6122      [Richard Levitte]
   6123 
   6124   *) Constify the ENGINE code as a result of BIGNUM constification.
   6125      Also constify the RSA code and most things related to it.  In a
   6126      few places, most notable in the depth of the ASN.1 code, ugly
   6127      casts back to non-const were required (to be solved at a later
   6128      time)
   6129      [Richard Levitte]
   6130 
   6131   *) Make it so the openssl application has all engines loaded by default.
   6132      [Richard Levitte]
   6133 
   6134   *) Constify the BIGNUM routines a little more.
   6135      [Richard Levitte]
   6136 
   6137   *) Add the following functions:
   6138 
   6139 	ENGINE_load_cswift()
   6140 	ENGINE_load_chil()
   6141 	ENGINE_load_atalla()
   6142 	ENGINE_load_nuron()
   6143 	ENGINE_load_builtin_engines()
   6144 
   6145      That way, an application can itself choose if external engines that
   6146      are built-in in OpenSSL shall ever be used or not.  The benefit is
   6147      that applications won't have to be linked with libdl or other dso
   6148      libraries unless it's really needed.
   6149 
   6150      Changed 'openssl engine' to load all engines on demand.
   6151      Changed the engine header files to avoid the duplication of some
   6152      declarations (they differed!).
   6153      [Richard Levitte]
   6154 
   6155   *) 'openssl engine' can now list capabilities.
   6156      [Richard Levitte]
   6157 
   6158   *) Better error reporting in 'openssl engine'.
   6159      [Richard Levitte]
   6160 
   6161   *) Never call load_dh_param(NULL) in s_server.
   6162      [Bodo Moeller]
   6163 
   6164   *) Add engine application.  It can currently list engines by name and
   6165      identity, and test if they are actually available.
   6166      [Richard Levitte]
   6167 
   6168   *) Improve RPM specification file by forcing symbolic linking and making
   6169      sure the installed documentation is also owned by root.root.
   6170      [Damien Miller <djm (a] mindrot.org>]
   6171 
   6172   *) Give the OpenSSL applications more possibilities to make use of
   6173      keys (public as well as private) handled by engines.
   6174      [Richard Levitte]
   6175 
   6176   *) Add OCSP code that comes from CertCo.
   6177      [Richard Levitte]
   6178 
   6179   *) Add VMS support for the Rijndael code.
   6180      [Richard Levitte]
   6181 
   6182   *) Added untested support for Nuron crypto accelerator.
   6183      [Ben Laurie]
   6184 
   6185   *) Add support for external cryptographic devices.  This code was
   6186      previously distributed separately as the "engine" branch.
   6187      [Geoff Thorpe, Richard Levitte]
   6188 
   6189   *) Rework the filename-translation in the DSO code. It is now possible to
   6190      have far greater control over how a "name" is turned into a filename
   6191      depending on the operating environment and any oddities about the
   6192      different shared library filenames on each system.
   6193      [Geoff Thorpe]
   6194 
   6195   *) Support threads on FreeBSD-elf in Configure.
   6196      [Richard Levitte]
   6197 
   6198   *) Fix for SHA1 assembly problem with MASM: it produces
   6199      warnings about corrupt line number information when assembling
   6200      with debugging information. This is caused by the overlapping
   6201      of two sections.
   6202      [Bernd Matthes <mainbug (a] celocom.de>, Steve Henson]
   6203 
   6204   *) NCONF changes.
   6205      NCONF_get_number() has no error checking at all.  As a replacement,
   6206      NCONF_get_number_e() is defined (_e for "error checking") and is
   6207      promoted strongly.  The old NCONF_get_number is kept around for
   6208      binary backward compatibility.
   6209      Make it possible for methods to load from something other than a BIO,
   6210      by providing a function pointer that is given a name instead of a BIO.
   6211      For example, this could be used to load configuration data from an
   6212      LDAP server.
   6213      [Richard Levitte]
   6214 
   6215   *) Fix for non blocking accept BIOs. Added new I/O special reason
   6216      BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
   6217      with non blocking I/O was not possible because no retry code was
   6218      implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
   6219      this case.
   6220      [Steve Henson]
   6221 
   6222   *) Added the beginnings of Rijndael support.
   6223      [Ben Laurie]
   6224 
   6225   *) Fix for bug in DirectoryString mask setting. Add support for
   6226      X509_NAME_print_ex() in 'req' and X509_print_ex() function
   6227      to allow certificate printing to more controllable, additional
   6228      'certopt' option to 'x509' to allow new printing options to be
   6229      set.
   6230      [Steve Henson]
   6231 
   6232   *) Clean old EAY MD5 hack from e_os.h.
   6233      [Richard Levitte]
   6234 
   6235  Changes between 0.9.6l and 0.9.6m  [17 Mar 2004]
   6236 
   6237   *) Fix null-pointer assignment in do_change_cipher_spec() revealed
   6238      by using the Codenomicon TLS Test Tool (CVE-2004-0079)
   6239      [Joe Orton, Steve Henson]
   6240 
   6241  Changes between 0.9.6k and 0.9.6l  [04 Nov 2003]
   6242 
   6243   *) Fix additional bug revealed by the NISCC test suite:
   6244 
   6245      Stop bug triggering large recursion when presented with
   6246      certain ASN.1 tags (CVE-2003-0851)
   6247      [Steve Henson]
   6248 
   6249  Changes between 0.9.6j and 0.9.6k  [30 Sep 2003]
   6250 
   6251   *) Fix various bugs revealed by running the NISCC test suite:
   6252 
   6253      Stop out of bounds reads in the ASN1 code when presented with
   6254      invalid tags (CVE-2003-0543 and CVE-2003-0544).
   6255      
   6256      If verify callback ignores invalid public key errors don't try to check
   6257      certificate signature with the NULL public key.
   6258 
   6259      [Steve Henson]
   6260 
   6261   *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
   6262      if the server requested one: as stated in TLS 1.0 and SSL 3.0
   6263      specifications.
   6264      [Steve Henson]
   6265 
   6266   *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
   6267      extra data after the compression methods not only for TLS 1.0
   6268      but also for SSL 3.0 (as required by the specification).
   6269      [Bodo Moeller; problem pointed out by Matthias Loepfe]
   6270 
   6271   *) Change X509_certificate_type() to mark the key as exported/exportable
   6272      when it's 512 *bits* long, not 512 bytes.
   6273      [Richard Levitte]
   6274 
   6275  Changes between 0.9.6i and 0.9.6j  [10 Apr 2003]
   6276 
   6277   *) Countermeasure against the Klima-Pokorny-Rosa extension of
   6278      Bleichbacher's attack on PKCS #1 v1.5 padding: treat
   6279      a protocol version number mismatch like a decryption error
   6280      in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
   6281      [Bodo Moeller]
   6282 
   6283   *) Turn on RSA blinding by default in the default implementation
   6284      to avoid a timing attack. Applications that don't want it can call
   6285      RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
   6286      They would be ill-advised to do so in most cases.
   6287      [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
   6288 
   6289   *) Change RSA blinding code so that it works when the PRNG is not
   6290      seeded (in this case, the secret RSA exponent is abused as
   6291      an unpredictable seed -- if it is not unpredictable, there
   6292      is no point in blinding anyway).  Make RSA blinding thread-safe
   6293      by remembering the creator's thread ID in rsa->blinding and
   6294      having all other threads use local one-time blinding factors
   6295      (this requires more computation than sharing rsa->blinding, but
   6296      avoids excessive locking; and if an RSA object is not shared
   6297      between threads, blinding will still be very fast).
   6298      [Bodo Moeller]
   6299 
   6300  Changes between 0.9.6h and 0.9.6i  [19 Feb 2003]
   6301 
   6302   *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
   6303      via timing by performing a MAC computation even if incorrrect
   6304      block cipher padding has been found.  This is a countermeasure
   6305      against active attacks where the attacker has to distinguish
   6306      between bad padding and a MAC verification error. (CVE-2003-0078)
   6307 
   6308      [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
   6309      Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
   6310      Martin Vuagnoux (EPFL, Ilion)]
   6311 
   6312  Changes between 0.9.6g and 0.9.6h  [5 Dec 2002]
   6313 
   6314   *) New function OPENSSL_cleanse(), which is used to cleanse a section of
   6315      memory from it's contents.  This is done with a counter that will
   6316      place alternating values in each byte.  This can be used to solve
   6317      two issues: 1) the removal of calls to memset() by highly optimizing
   6318      compilers, and 2) cleansing with other values than 0, since those can
   6319      be read through on certain media, for example a swap space on disk.
   6320      [Geoff Thorpe]
   6321 
   6322   *) Bugfix: client side session caching did not work with external caching,
   6323      because the session->cipher setting was not restored when reloading
   6324      from the external cache. This problem was masked, when
   6325      SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
   6326      (Found by Steve Haslam <steve (a] araqnid.ddts.net>.)
   6327      [Lutz Jaenicke]
   6328 
   6329   *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
   6330      length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
   6331      [Zeev Lieber <zeev-l (a] yahoo.com>]
   6332 
   6333   *) Undo an undocumented change introduced in 0.9.6e which caused
   6334      repeated calls to OpenSSL_add_all_ciphers() and 
   6335      OpenSSL_add_all_digests() to be ignored, even after calling
   6336      EVP_cleanup().
   6337      [Richard Levitte]
   6338 
   6339   *) Change the default configuration reader to deal with last line not
   6340      being properly terminated.
   6341      [Richard Levitte]
   6342 
   6343   *) Change X509_NAME_cmp() so it applies the special rules on handling
   6344      DN values that are of type PrintableString, as well as RDNs of type
   6345      emailAddress where the value has the type ia5String.
   6346      [stefank (a] valicert.com via Richard Levitte]
   6347 
   6348   *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
   6349      the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
   6350      doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
   6351      the bitwise-OR of the two for use by the majority of applications
   6352      wanting this behaviour, and update the docs. The documented
   6353      behaviour and actual behaviour were inconsistent and had been
   6354      changing anyway, so this is more a bug-fix than a behavioural
   6355      change.
   6356      [Geoff Thorpe, diagnosed by Nadav Har'El]
   6357 
   6358   *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
   6359      (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
   6360      [Bodo Moeller]
   6361 
   6362   *) Fix initialization code race conditions in
   6363         SSLv23_method(),  SSLv23_client_method(),   SSLv23_server_method(),
   6364         SSLv2_method(),   SSLv2_client_method(),    SSLv2_server_method(),
   6365         SSLv3_method(),   SSLv3_client_method(),    SSLv3_server_method(),
   6366         TLSv1_method(),   TLSv1_client_method(),    TLSv1_server_method(),
   6367         ssl2_get_cipher_by_char(),
   6368         ssl3_get_cipher_by_char().
   6369      [Patrick McCormick <patrick (a] tellme.com>, Bodo Moeller]
   6370 
   6371   *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
   6372      the cached sessions are flushed, as the remove_cb() might use ex_data
   6373      contents. Bug found by Sam Varshavchik <mrsam (a] courier-mta.com>
   6374      (see [openssl.org #212]).
   6375      [Geoff Thorpe, Lutz Jaenicke]
   6376 
   6377   *) Fix typo in OBJ_txt2obj which incorrectly passed the content
   6378      length, instead of the encoding length to d2i_ASN1_OBJECT.
   6379      [Steve Henson]
   6380 
   6381  Changes between 0.9.6f and 0.9.6g  [9 Aug 2002]
   6382 
   6383   *) [In 0.9.6g-engine release:]
   6384      Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
   6385      [Lynn Gazis <lgazis (a] rainbow.com>]
   6386 
   6387  Changes between 0.9.6e and 0.9.6f  [8 Aug 2002]
   6388 
   6389   *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
   6390      and get fix the header length calculation.
   6391      [Florian Weimer <Weimer (a] CERT.Uni-Stuttgart.DE>,
   6392 	Alon Kantor <alonk (a] checkpoint.com> (and others),
   6393 	Steve Henson]
   6394 
   6395   *) Use proper error handling instead of 'assertions' in buffer
   6396      overflow checks added in 0.9.6e.  This prevents DoS (the
   6397      assertions could call abort()).
   6398      [Arne Ansper <arne (a] ats.cyber.ee>, Bodo Moeller]
   6399 
   6400  Changes between 0.9.6d and 0.9.6e  [30 Jul 2002]
   6401 
   6402   *) Add various sanity checks to asn1_get_length() to reject
   6403      the ASN1 length bytes if they exceed sizeof(long), will appear
   6404      negative or the content length exceeds the length of the
   6405      supplied buffer.
   6406      [Steve Henson, Adi Stav <stav (a] mercury.co.il>, James Yonan <jim (a] ntlp.com>]
   6407 
   6408   *) Fix cipher selection routines: ciphers without encryption had no flags
   6409      for the cipher strength set and where therefore not handled correctly
   6410      by the selection routines (PR #130).
   6411      [Lutz Jaenicke]
   6412 
   6413   *) Fix EVP_dsa_sha macro.
   6414      [Nils Larsch]
   6415 
   6416   *) New option
   6417           SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
   6418      for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
   6419      that was added in OpenSSL 0.9.6d.
   6420 
   6421      As the countermeasure turned out to be incompatible with some
   6422      broken SSL implementations, the new option is part of SSL_OP_ALL.
   6423      SSL_OP_ALL is usually employed when compatibility with weird SSL
   6424      implementations is desired (e.g. '-bugs' option to 's_client' and
   6425      's_server'), so the new option is automatically set in many
   6426      applications.
   6427      [Bodo Moeller]
   6428 
   6429   *) Changes in security patch:
   6430 
   6431      Changes marked "(CHATS)" were sponsored by the Defense Advanced
   6432      Research Projects Agency (DARPA) and Air Force Research Laboratory,
   6433      Air Force Materiel Command, USAF, under agreement number
   6434      F30602-01-2-0537.
   6435 
   6436   *) Add various sanity checks to asn1_get_length() to reject
   6437      the ASN1 length bytes if they exceed sizeof(long), will appear
   6438      negative or the content length exceeds the length of the
   6439      supplied buffer. (CVE-2002-0659)
   6440      [Steve Henson, Adi Stav <stav (a] mercury.co.il>, James Yonan <jim (a] ntlp.com>]
   6441 
   6442   *) Assertions for various potential buffer overflows, not known to
   6443      happen in practice.
   6444      [Ben Laurie (CHATS)]
   6445 
   6446   *) Various temporary buffers to hold ASCII versions of integers were
   6447      too small for 64 bit platforms. (CVE-2002-0655)
   6448      [Matthew Byng-Maddick <mbm (a] aldigital.co.uk> and Ben Laurie (CHATS)>
   6449 
   6450   *) Remote buffer overflow in SSL3 protocol - an attacker could
   6451      supply an oversized session ID to a client. (CVE-2002-0656)
   6452      [Ben Laurie (CHATS)]
   6453 
   6454   *) Remote buffer overflow in SSL2 protocol - an attacker could
   6455      supply an oversized client master key. (CVE-2002-0656)
   6456      [Ben Laurie (CHATS)]
   6457 
   6458  Changes between 0.9.6c and 0.9.6d  [9 May 2002]
   6459 
   6460   *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
   6461      encoded as NULL) with id-dsa-with-sha1.
   6462      [Nils Larsch <nla (a] trustcenter.de>; problem pointed out by Bodo Moeller]
   6463 
   6464   *) Check various X509_...() return values in apps/req.c.
   6465      [Nils Larsch <nla (a] trustcenter.de>]
   6466 
   6467   *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
   6468      an end-of-file condition would erronously be flagged, when the CRLF
   6469      was just at the end of a processed block. The bug was discovered when
   6470      processing data through a buffering memory BIO handing the data to a
   6471      BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
   6472      <ptsekov (a] syntrex.com> and Nedelcho Stanev.
   6473      [Lutz Jaenicke]
   6474 
   6475   *) Implement a countermeasure against a vulnerability recently found
   6476      in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
   6477      before application data chunks to avoid the use of known IVs
   6478      with data potentially chosen by the attacker.
   6479      [Bodo Moeller]
   6480 
   6481   *) Fix length checks in ssl3_get_client_hello().
   6482      [Bodo Moeller]
   6483 
   6484   *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
   6485      to prevent ssl3_read_internal() from incorrectly assuming that
   6486      ssl3_read_bytes() found application data while handshake
   6487      processing was enabled when in fact s->s3->in_read_app_data was
   6488      merely automatically cleared during the initial handshake.
   6489      [Bodo Moeller; problem pointed out by Arne Ansper <arne (a] ats.cyber.ee>]
   6490 
   6491   *) Fix object definitions for Private and Enterprise: they were not
   6492      recognized in their shortname (=lowercase) representation. Extend
   6493      obj_dat.pl to issue an error when using undefined keywords instead
   6494      of silently ignoring the problem (Svenning Sorensen
   6495      <sss (a] sss.dnsalias.net>).
   6496      [Lutz Jaenicke]
   6497 
   6498   *) Fix DH_generate_parameters() so that it works for 'non-standard'
   6499      generators, i.e. generators other than 2 and 5.  (Previously, the
   6500      code did not properly initialise the 'add' and 'rem' values to
   6501      BN_generate_prime().)
   6502 
   6503      In the new general case, we do not insist that 'generator' is
   6504      actually a primitive root: This requirement is rather pointless;
   6505      a generator of the order-q subgroup is just as good, if not
   6506      better.
   6507      [Bodo Moeller]
   6508  
   6509   *) Map new X509 verification errors to alerts. Discovered and submitted by
   6510      Tom Wu <tom (a] arcot.com>.
   6511      [Lutz Jaenicke]
   6512 
   6513   *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
   6514      returning non-zero before the data has been completely received
   6515      when using non-blocking I/O.
   6516      [Bodo Moeller; problem pointed out by John Hughes]
   6517 
   6518   *) Some of the ciphers missed the strength entry (SSL_LOW etc).
   6519      [Ben Laurie, Lutz Jaenicke]
   6520 
   6521   *) Fix bug in SSL_clear(): bad sessions were not removed (found by
   6522      Yoram Zahavi <YoramZ (a] gilian.com>).
   6523      [Lutz Jaenicke]
   6524 
   6525   *) Add information about CygWin 1.3 and on, and preserve proper
   6526      configuration for the versions before that.
   6527      [Corinna Vinschen <vinschen (a] redhat.com> and Richard Levitte]
   6528 
   6529   *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
   6530      check whether we deal with a copy of a session and do not delete from
   6531      the cache in this case. Problem reported by "Izhar Shoshani Levi"
   6532      <izhar (a] checkpoint.com>.
   6533      [Lutz Jaenicke]
   6534 
   6535   *) Do not store session data into the internal session cache, if it
   6536      is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
   6537      flag is set). Proposed by Aslam <aslam (a] funk.com>.
   6538      [Lutz Jaenicke]
   6539 
   6540   *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
   6541      value is 0.
   6542      [Richard Levitte]
   6543 
   6544   *) [In 0.9.6d-engine release:]
   6545      Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
   6546      [Toomas Kiisk <vix (a] cyber.ee> via Richard Levitte]
   6547 
   6548   *) Add the configuration target linux-s390x.
   6549      [Neale Ferguson <Neale.Ferguson (a] SoftwareAG-USA.com> via Richard Levitte]
   6550 
   6551   *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
   6552      ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
   6553      variable as an indication that a ClientHello message has been
   6554      received.  As the flag value will be lost between multiple
   6555      invocations of ssl3_accept when using non-blocking I/O, the
   6556      function may not be aware that a handshake has actually taken
   6557      place, thus preventing a new session from being added to the
   6558      session cache.
   6559 
   6560      To avoid this problem, we now set s->new_session to 2 instead of
   6561      using a local variable.
   6562      [Lutz Jaenicke, Bodo Moeller]
   6563 
   6564   *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
   6565      if the SSL_R_LENGTH_MISMATCH error is detected.
   6566      [Geoff Thorpe, Bodo Moeller]
   6567 
   6568   *) New 'shared_ldflag' column in Configure platform table.
   6569      [Richard Levitte]
   6570 
   6571   *) Fix EVP_CIPHER_mode macro.
   6572      ["Dan S. Camper" <dan (a] bti.net>]
   6573 
   6574   *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
   6575      type, we must throw them away by setting rr->length to 0.
   6576      [D P Chang <dpc (a] qualys.com>]
   6577 
   6578  Changes between 0.9.6b and 0.9.6c  [21 dec 2001]
   6579 
   6580   *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
   6581      <Dominikus.Scherkl (a] biodata.com>.  (The previous implementation
   6582      worked incorrectly for those cases where  range = 10..._2  and
   6583      3*range  is two bits longer than  range.)
   6584      [Bodo Moeller]
   6585 
   6586   *) Only add signing time to PKCS7 structures if it is not already
   6587      present.
   6588      [Steve Henson]
   6589 
   6590   *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
   6591      OBJ_ld_ce should be OBJ_id_ce.
   6592      Also some ip-pda OIDs in crypto/objects/objects.txt were
   6593      incorrect (cf. RFC 3039).
   6594      [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
   6595 
   6596   *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
   6597      returns early because it has nothing to do.
   6598      [Andy Schneider <andy.schneider (a] bjss.co.uk>]
   6599 
   6600   *) [In 0.9.6c-engine release:]
   6601      Fix mutex callback return values in crypto/engine/hw_ncipher.c.
   6602      [Andy Schneider <andy.schneider (a] bjss.co.uk>]
   6603 
   6604   *) [In 0.9.6c-engine release:]
   6605      Add support for Cryptographic Appliance's keyserver technology.
   6606      (Use engine 'keyclient')
   6607      [Cryptographic Appliances and Geoff Thorpe]
   6608 
   6609   *) Add a configuration entry for OS/390 Unix.  The C compiler 'c89'
   6610      is called via tools/c89.sh because arguments have to be
   6611      rearranged (all '-L' options must appear before the first object
   6612      modules).
   6613      [Richard Shapiro <rshapiro (a] abinitio.com>]
   6614 
   6615   *) [In 0.9.6c-engine release:]
   6616      Add support for Broadcom crypto accelerator cards, backported
   6617      from 0.9.7.
   6618      [Broadcom, Nalin Dahyabhai <nalin (a] redhat.com>, Mark Cox]
   6619 
   6620   *) [In 0.9.6c-engine release:]
   6621      Add support for SureWare crypto accelerator cards from 
   6622      Baltimore Technologies.  (Use engine 'sureware')
   6623      [Baltimore Technologies and Mark Cox]
   6624 
   6625   *) [In 0.9.6c-engine release:]
   6626      Add support for crypto accelerator cards from Accelerated
   6627      Encryption Processing, www.aep.ie.  (Use engine 'aep')
   6628      [AEP Inc. and Mark Cox]
   6629 
   6630   *) Add a configuration entry for gcc on UnixWare.
   6631      [Gary Benson <gbenson (a] redhat.com>]
   6632 
   6633   *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
   6634      messages are stored in a single piece (fixed-length part and
   6635      variable-length part combined) and fix various bugs found on the way.
   6636      [Bodo Moeller]
   6637 
   6638   *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
   6639      instead.  BIO_gethostbyname() does not know what timeouts are
   6640      appropriate, so entries would stay in cache even when they have
   6641      become invalid.
   6642      [Bodo Moeller; problem pointed out by Rich Salz <rsalz (a] zolera.com>
   6643 
   6644   *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
   6645      faced with a pathologically small ClientHello fragment that does
   6646      not contain client_version: Instead of aborting with an error,
   6647      simply choose the highest available protocol version (i.e.,
   6648      TLS 1.0 unless it is disabled).  In practice, ClientHello
   6649      messages are never sent like this, but this change gives us
   6650      strictly correct behaviour at least for TLS.
   6651      [Bodo Moeller]
   6652 
   6653   *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
   6654      never resets s->method to s->ctx->method when called from within
   6655      one of the SSL handshake functions.
   6656      [Bodo Moeller; problem pointed out by Niko Baric]
   6657 
   6658   *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
   6659      (sent using the client's version number) if client_version is
   6660      smaller than the protocol version in use.  Also change
   6661      ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
   6662      the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
   6663      the client will at least see that alert.
   6664      [Bodo Moeller]
   6665 
   6666   *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
   6667      correctly.
   6668      [Bodo Moeller]
   6669 
   6670   *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
   6671      client receives HelloRequest while in a handshake.
   6672      [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider (a] bjss.co.uk>]
   6673 
   6674   *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
   6675      should end in 'break', not 'goto end' which circuments various
   6676      cleanups done in state SSL_ST_OK.   But session related stuff
   6677      must be disabled for SSL_ST_OK in the case that we just sent a
   6678      HelloRequest.
   6679 
   6680      Also avoid some overhead by not calling ssl_init_wbio_buffer()
   6681      before just sending a HelloRequest.
   6682      [Bodo Moeller, Eric Rescorla <ekr (a] rtfm.com>]
   6683 
   6684   *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
   6685      reveal whether illegal block cipher padding was found or a MAC
   6686      verification error occured.  (Neither SSLerr() codes nor alerts
   6687      are directly visible to potential attackers, but the information
   6688      may leak via logfiles.)
   6689 
   6690      Similar changes are not required for the SSL 2.0 implementation
   6691      because the number of padding bytes is sent in clear for SSL 2.0,
   6692      and the extra bytes are just ignored.  However ssl/s2_pkt.c
   6693      failed to verify that the purported number of padding bytes is in
   6694      the legal range.
   6695      [Bodo Moeller]
   6696 
   6697   *) Add OpenUNIX-8 support including shared libraries
   6698      (Boyd Lynn Gerber <gerberb (a] zenez.com>).
   6699      [Lutz Jaenicke]
   6700 
   6701   *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
   6702      'wristwatch attack' using huge encoding parameters (cf.
   6703      James H. Manger's CRYPTO 2001 paper).  Note that the
   6704      RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
   6705      encoding parameters and hence was not vulnerable.
   6706      [Bodo Moeller]
   6707 
   6708   *) BN_sqr() bug fix.
   6709      [Ulf Mller, reported by Jim Ellis <jim.ellis (a] cavium.com>]
   6710 
   6711   *) Rabin-Miller test analyses assume uniformly distributed witnesses,
   6712      so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
   6713      followed by modular reduction.
   6714      [Bodo Moeller; pointed out by Adam Young <AYoung1 (a] NCSUS.JNJ.COM>]
   6715 
   6716   *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
   6717      equivalent based on BN_pseudo_rand() instead of BN_rand().
   6718      [Bodo Moeller]
   6719 
   6720   *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
   6721      This function was broken, as the check for a new client hello message
   6722      to handle SGC did not allow these large messages.
   6723      (Tracked down by "Douglas E. Engert" <deengert (a] anl.gov>.)
   6724      [Lutz Jaenicke]
   6725 
   6726   *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
   6727      [Lutz Jaenicke]
   6728 
   6729   *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
   6730      for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton (a] netopia.com>).
   6731      [Lutz Jaenicke]
   6732 
   6733   *) Rework the configuration and shared library support for Tru64 Unix.
   6734      The configuration part makes use of modern compiler features and
   6735      still retains old compiler behavior for those that run older versions
   6736      of the OS.  The shared library support part includes a variant that
   6737      uses the RPATH feature, and is available through the special
   6738      configuration target "alpha-cc-rpath", which will never be selected
   6739      automatically.
   6740      [Tim Mooney <mooney (a] dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
   6741 
   6742   *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
   6743      with the same message size as in ssl3_get_certificate_request().
   6744      Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
   6745      messages might inadvertently be reject as too long.
   6746      [Petr Lampa <lampa (a] fee.vutbr.cz>]
   6747 
   6748   *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
   6749      [Andy Polyakov]
   6750 
   6751   *) Modified SSL library such that the verify_callback that has been set
   6752      specificly for an SSL object with SSL_set_verify() is actually being
   6753      used. Before the change, a verify_callback set with this function was
   6754      ignored and the verify_callback() set in the SSL_CTX at the time of
   6755      the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
   6756      to allow the necessary settings.
   6757      [Lutz Jaenicke]
   6758 
   6759   *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
   6760      explicitly to NULL, as at least on Solaris 8 this seems not always to be
   6761      done automatically (in contradiction to the requirements of the C
   6762      standard). This made problems when used from OpenSSH.
   6763      [Lutz Jaenicke]
   6764 
   6765   *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
   6766      dh->length and always used
   6767 
   6768           BN_rand_range(priv_key, dh->p).
   6769 
   6770      BN_rand_range() is not necessary for Diffie-Hellman, and this
   6771      specific range makes Diffie-Hellman unnecessarily inefficient if
   6772      dh->length (recommended exponent length) is much smaller than the
   6773      length of dh->p.  We could use BN_rand_range() if the order of
   6774      the subgroup was stored in the DH structure, but we only have
   6775      dh->length.
   6776 
   6777      So switch back to
   6778 
   6779           BN_rand(priv_key, l, ...)
   6780 
   6781      where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
   6782      otherwise.
   6783      [Bodo Moeller]
   6784 
   6785   *) In
   6786 
   6787           RSA_eay_public_encrypt
   6788           RSA_eay_private_decrypt
   6789           RSA_eay_private_encrypt (signing)
   6790           RSA_eay_public_decrypt (signature verification)
   6791 
   6792      (default implementations for RSA_public_encrypt,
   6793      RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
   6794      always reject numbers >= n.
   6795      [Bodo Moeller]
   6796 
   6797   *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
   6798      to synchronize access to 'locking_thread'.  This is necessary on
   6799      systems where access to 'locking_thread' (an 'unsigned long'
   6800      variable) is not atomic.
   6801      [Bodo Moeller]
   6802 
   6803   *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
   6804      *before* setting the 'crypto_lock_rand' flag.  The previous code had
   6805      a race condition if 0 is a valid thread ID.
   6806      [Travis Vitek <vitek (a] roguewave.com>]
   6807 
   6808   *) Add support for shared libraries under Irix.
   6809      [Albert Chin-A-Young <china (a] thewrittenword.com>]
   6810 
   6811   *) Add configuration option to build on Linux on both big-endian and
   6812      little-endian MIPS.
   6813      [Ralf Baechle <ralf (a] uni-koblenz.de>]
   6814 
   6815   *) Add the possibility to create shared libraries on HP-UX.
   6816      [Richard Levitte]
   6817 
   6818  Changes between 0.9.6a and 0.9.6b  [9 Jul 2001]
   6819 
   6820   *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
   6821      to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
   6822      Markku-Juhani O. Saarinen <markku-juhani.saarinen (a] nokia.com>:
   6823      PRNG state recovery was possible based on the output of
   6824      one PRNG request appropriately sized to gain knowledge on
   6825      'md' followed by enough consecutive 1-byte PRNG requests
   6826      to traverse all of 'state'.
   6827 
   6828      1. When updating 'md_local' (the current thread's copy of 'md')
   6829         during PRNG output generation, hash all of the previous
   6830         'md_local' value, not just the half used for PRNG output.
   6831 
   6832      2. Make the number of bytes from 'state' included into the hash
   6833         independent from the number of PRNG bytes requested.
   6834 
   6835      The first measure alone would be sufficient to avoid
   6836      Markku-Juhani's attack.  (Actually it had never occurred
   6837      to me that the half of 'md_local' used for chaining was the
   6838      half from which PRNG output bytes were taken -- I had always
   6839      assumed that the secret half would be used.)  The second
   6840      measure makes sure that additional data from 'state' is never
   6841      mixed into 'md_local' in small portions; this heuristically
   6842      further strengthens the PRNG.
   6843      [Bodo Moeller]
   6844 
   6845   *) Fix crypto/bn/asm/mips3.s.
   6846      [Andy Polyakov]
   6847 
   6848   *) When only the key is given to "enc", the IV is undefined. Print out
   6849      an error message in this case.
   6850      [Lutz Jaenicke]
   6851 
   6852   *) Handle special case when X509_NAME is empty in X509 printing routines.
   6853      [Steve Henson]
   6854 
   6855   *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
   6856      positive and less than q.
   6857      [Bodo Moeller]
   6858 
   6859   *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
   6860      used: it isn't thread safe and the add_lock_callback should handle
   6861      that itself.
   6862      [Paul Rose <Paul.Rose (a] bridge.com>]
   6863 
   6864   *) Verify that incoming data obeys the block size in
   6865      ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
   6866      [Bodo Moeller]
   6867 
   6868   *) Fix OAEP check.
   6869      [Ulf Mller, Bodo Mller]
   6870 
   6871   *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
   6872      RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
   6873      when fixing the server behaviour for backwards-compatible 'client
   6874      hello' messages.  (Note that the attack is impractical against
   6875      SSL 3.0 and TLS 1.0 anyway because length and version checking
   6876      means that the probability of guessing a valid ciphertext is
   6877      around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
   6878      paper.)
   6879 
   6880      Before 0.9.5, the countermeasure (hide the error by generating a
   6881      random 'decryption result') did not work properly because
   6882      ERR_clear_error() was missing, meaning that SSL_get_error() would
   6883      detect the supposedly ignored error.
   6884 
   6885      Both problems are now fixed.
   6886      [Bodo Moeller]
   6887 
   6888   *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
   6889      (previously it was 1024).
   6890      [Bodo Moeller]
   6891 
   6892   *) Fix for compatibility mode trust settings: ignore trust settings
   6893      unless some valid trust or reject settings are present.
   6894      [Steve Henson]
   6895 
   6896   *) Fix for blowfish EVP: its a variable length cipher.
   6897      [Steve Henson]
   6898 
   6899   *) Fix various bugs related to DSA S/MIME verification. Handle missing
   6900      parameters in DSA public key structures and return an error in the
   6901      DSA routines if parameters are absent.
   6902      [Steve Henson]
   6903 
   6904   *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
   6905      in the current directory if neither $RANDFILE nor $HOME was set.
   6906      RAND_file_name() in 0.9.6a returned NULL in this case.  This has
   6907      caused some confusion to Windows users who haven't defined $HOME.
   6908      Thus RAND_file_name() is changed again: e_os.h can define a
   6909      DEFAULT_HOME, which will be used if $HOME is not set.
   6910      For Windows, we use "C:"; on other platforms, we still require
   6911      environment variables.
   6912 
   6913   *) Move 'if (!initialized) RAND_poll()' into regions protected by
   6914      CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
   6915      having multiple threads call RAND_poll() concurrently.
   6916      [Bodo Moeller]
   6917 
   6918   *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
   6919      combination of a flag and a thread ID variable.
   6920      Otherwise while one thread is in ssleay_rand_bytes (which sets the
   6921      flag), *other* threads can enter ssleay_add_bytes without obeying
   6922      the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
   6923      that they do not hold after the first thread unsets add_do_not_lock).
   6924      [Bodo Moeller]
   6925 
   6926   *) Change bctest again: '-x' expressions are not available in all
   6927      versions of 'test'.
   6928      [Bodo Moeller]
   6929 
   6930  Changes between 0.9.6 and 0.9.6a  [5 Apr 2001]
   6931 
   6932   *) Fix a couple of memory leaks in PKCS7_dataDecode()
   6933      [Steve Henson, reported by Heyun Zheng <hzheng (a] atdsprint.com>]
   6934 
   6935   *) Change Configure and Makefiles to provide EXE_EXT, which will contain
   6936      the default extension for executables, if any.  Also, make the perl
   6937      scripts that use symlink() to test if it really exists and use "cp"
   6938      if it doesn't.  All this made OpenSSL compilable and installable in
   6939      CygWin.
   6940      [Richard Levitte]
   6941 
   6942   *) Fix for asn1_GetSequence() for indefinite length constructed data.
   6943      If SEQUENCE is length is indefinite just set c->slen to the total
   6944      amount of data available.
   6945      [Steve Henson, reported by shige (a] FreeBSD.org]
   6946      [This change does not apply to 0.9.7.]
   6947 
   6948   *) Change bctest to avoid here-documents inside command substitution
   6949      (workaround for FreeBSD /bin/sh bug).
   6950      For compatibility with Ultrix, avoid shell functions (introduced
   6951      in the bctest version that searches along $PATH).
   6952      [Bodo Moeller]
   6953 
   6954   *) Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
   6955      with des_encrypt() defined on some operating systems, like Solaris
   6956      and UnixWare.
   6957      [Richard Levitte]
   6958 
   6959   *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
   6960      On the Importance of Eliminating Errors in Cryptographic
   6961      Computations, J. Cryptology 14 (2001) 2, 101-119,
   6962      http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
   6963      [Ulf Moeller]
   6964   
   6965   *) MIPS assembler BIGNUM division bug fix. 
   6966      [Andy Polyakov]
   6967 
   6968   *) Disabled incorrect Alpha assembler code.
   6969      [Richard Levitte]
   6970 
   6971   *) Fix PKCS#7 decode routines so they correctly update the length
   6972      after reading an EOC for the EXPLICIT tag.
   6973      [Steve Henson]
   6974      [This change does not apply to 0.9.7.]
   6975 
   6976   *) Fix bug in PKCS#12 key generation routines. This was triggered
   6977      if a 3DES key was generated with a 0 initial byte. Include
   6978      PKCS12_BROKEN_KEYGEN compilation option to retain the old
   6979      (but broken) behaviour.
   6980      [Steve Henson]
   6981 
   6982   *) Enhance bctest to search for a working bc along $PATH and print
   6983      it when found.
   6984      [Tim Rice <tim (a] multitalents.net> via Richard Levitte]
   6985 
   6986   *) Fix memory leaks in err.c: free err_data string if necessary;
   6987      don't write to the wrong index in ERR_set_error_data.
   6988      [Bodo Moeller]
   6989 
   6990   *) Implement ssl23_peek (analogous to ssl23_read), which previously
   6991      did not exist.
   6992      [Bodo Moeller]
   6993 
   6994   *) Replace rdtsc with _emit statements for VC++ version 5.
   6995      [Jeremy Cooper <jeremy (a] baymoo.org>]
   6996 
   6997   *) Make it possible to reuse SSLv2 sessions.
   6998      [Richard Levitte]
   6999 
   7000   *) In copy_email() check for >= 0 as a return value for
   7001      X509_NAME_get_index_by_NID() since 0 is a valid index.
   7002      [Steve Henson reported by Massimiliano Pala <madwolf (a] opensca.org>]
   7003 
   7004   *) Avoid coredump with unsupported or invalid public keys by checking if
   7005      X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
   7006      PKCS7_verify() fails with non detached data.
   7007      [Steve Henson]
   7008 
   7009   *) Don't use getenv in library functions when run as setuid/setgid.
   7010      New function OPENSSL_issetugid().
   7011      [Ulf Moeller]
   7012 
   7013   *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
   7014      due to incorrect handling of multi-threading:
   7015 
   7016      1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
   7017 
   7018      2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
   7019 
   7020      3. Count how many times MemCheck_off() has been called so that
   7021         nested use can be treated correctly.  This also avoids 
   7022         inband-signalling in the previous code (which relied on the
   7023         assumption that thread ID 0 is impossible).
   7024      [Bodo Moeller]
   7025 
   7026   *) Add "-rand" option also to s_client and s_server.
   7027      [Lutz Jaenicke]
   7028 
   7029   *) Fix CPU detection on Irix 6.x.
   7030      [Kurt Hockenbury <khockenb (a] stevens-tech.edu> and
   7031       "Bruce W. Forsberg" <bruce.forsberg (a] baesystems.com>]
   7032 
   7033   *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
   7034      was empty.
   7035      [Steve Henson]
   7036      [This change does not apply to 0.9.7.]
   7037 
   7038   *) Use the cached encoding of an X509_NAME structure rather than
   7039      copying it. This is apparently the reason for the libsafe "errors"
   7040      but the code is actually correct.
   7041      [Steve Henson]
   7042 
   7043   *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
   7044      Bleichenbacher's DSA attack.
   7045      Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
   7046      to be set and top=0 forces the highest bit to be set; top=-1 is new
   7047      and leaves the highest bit random.
   7048      [Ulf Moeller, Bodo Moeller]
   7049 
   7050   *) In the NCONF_...-based implementations for CONF_... queries
   7051      (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
   7052      a temporary CONF structure with the data component set to NULL
   7053      (which gives segmentation faults in lh_retrieve).
   7054      Instead, use NULL for the CONF pointer in CONF_get_string and
   7055      CONF_get_number (which may use environment variables) and directly
   7056      return NULL from CONF_get_section.
   7057      [Bodo Moeller]
   7058 
   7059   *) Fix potential buffer overrun for EBCDIC.
   7060      [Ulf Moeller]
   7061 
   7062   *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
   7063      keyUsage if basicConstraints absent for a CA.
   7064      [Steve Henson]
   7065 
   7066   *) Make SMIME_write_PKCS7() write mail header values with a format that
   7067      is more generally accepted (no spaces before the semicolon), since
   7068      some programs can't parse those values properly otherwise.  Also make
   7069      sure BIO's that break lines after each write do not create invalid
   7070      headers.
   7071      [Richard Levitte]
   7072 
   7073   *) Make the CRL encoding routines work with empty SEQUENCE OF. The
   7074      macros previously used would not encode an empty SEQUENCE OF
   7075      and break the signature.
   7076      [Steve Henson]
   7077      [This change does not apply to 0.9.7.]
   7078 
   7079   *) Zero the premaster secret after deriving the master secret in
   7080      DH ciphersuites.
   7081      [Steve Henson]
   7082 
   7083   *) Add some EVP_add_digest_alias registrations (as found in
   7084      OpenSSL_add_all_digests()) to SSL_library_init()
   7085      aka OpenSSL_add_ssl_algorithms().  This provides improved
   7086      compatibility with peers using X.509 certificates
   7087      with unconventional AlgorithmIdentifier OIDs.
   7088      [Bodo Moeller]
   7089 
   7090   *) Fix for Irix with NO_ASM.
   7091      ["Bruce W. Forsberg" <bruce.forsberg (a] baesystems.com>]
   7092 
   7093   *) ./config script fixes.
   7094      [Ulf Moeller, Richard Levitte]
   7095 
   7096   *) Fix 'openssl passwd -1'.
   7097      [Bodo Moeller]
   7098 
   7099   *) Change PKCS12_key_gen_asc() so it can cope with non null
   7100      terminated strings whose length is passed in the passlen
   7101      parameter, for example from PEM callbacks. This was done
   7102      by adding an extra length parameter to asc2uni().
   7103      [Steve Henson, reported by <oddissey (a] samsung.co.kr>]
   7104 
   7105   *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
   7106      call failed, free the DSA structure.
   7107      [Bodo Moeller]
   7108 
   7109   *) Fix to uni2asc() to cope with zero length Unicode strings.
   7110      These are present in some PKCS#12 files.
   7111      [Steve Henson]
   7112 
   7113   *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
   7114      Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
   7115      when writing a 32767 byte record.
   7116      [Bodo Moeller; problem reported by Eric Day <eday (a] concentric.net>]
   7117 
   7118   *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
   7119      obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
   7120 
   7121      (RSA objects have a reference count access to which is protected
   7122      by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
   7123      so they are meant to be shared between threads.)
   7124      [Bodo Moeller, Geoff Thorpe; original patch submitted by
   7125      "Reddie, Steven" <Steven.Reddie (a] ca.com>]
   7126 
   7127   *) Fix a deadlock in CRYPTO_mem_leaks().
   7128      [Bodo Moeller]
   7129 
   7130   *) Use better test patterns in bntest.
   7131      [Ulf Mller]
   7132 
   7133   *) rand_win.c fix for Borland C.
   7134      [Ulf Mller]
   7135  
   7136   *) BN_rshift bugfix for n == 0.
   7137      [Bodo Moeller]
   7138 
   7139   *) Add a 'bctest' script that checks for some known 'bc' bugs
   7140      so that 'make test' does not abort just because 'bc' is broken.
   7141      [Bodo Moeller]
   7142 
   7143   *) Store verify_result within SSL_SESSION also for client side to
   7144      avoid potential security hole. (Re-used sessions on the client side
   7145      always resulted in verify_result==X509_V_OK, not using the original
   7146      result of the server certificate verification.)
   7147      [Lutz Jaenicke]
   7148 
   7149   *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
   7150      SSL3_RT_APPLICATION_DATA, return 0.
   7151      Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
   7152      [Bodo Moeller]
   7153 
   7154   *) Fix SSL_peek:
   7155      Both ssl2_peek and ssl3_peek, which were totally broken in earlier
   7156      releases, have been re-implemented by renaming the previous
   7157      implementations of ssl2_read and ssl3_read to ssl2_read_internal
   7158      and ssl3_read_internal, respectively, and adding 'peek' parameters
   7159      to them.  The new ssl[23]_{read,peek} functions are calls to
   7160      ssl[23]_read_internal with the 'peek' flag set appropriately.
   7161      A 'peek' parameter has also been added to ssl3_read_bytes, which
   7162      does the actual work for ssl3_read_internal.
   7163      [Bodo Moeller]
   7164 
   7165   *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
   7166      the method-specific "init()" handler. Also clean up ex_data after
   7167      calling the method-specific "finish()" handler. Previously, this was
   7168      happening the other way round.
   7169      [Geoff Thorpe]
   7170 
   7171   *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
   7172      The previous value, 12, was not always sufficient for BN_mod_exp().
   7173      [Bodo Moeller]
   7174 
   7175   *) Make sure that shared libraries get the internal name engine with
   7176      the full version number and not just 0.  This should mark the
   7177      shared libraries as not backward compatible.  Of course, this should
   7178      be changed again when we can guarantee backward binary compatibility.
   7179      [Richard Levitte]
   7180 
   7181   *) Fix typo in get_cert_by_subject() in by_dir.c
   7182      [Jean-Marc Desperrier <jean-marc.desperrier (a] certplus.com>]
   7183 
   7184   *) Rework the system to generate shared libraries:
   7185 
   7186      - Make note of the expected extension for the shared libraries and
   7187        if there is a need for symbolic links from for example libcrypto.so.0
   7188        to libcrypto.so.0.9.7.  There is extended info in Configure for
   7189        that.
   7190 
   7191      - Make as few rebuilds of the shared libraries as possible.
   7192 
   7193      - Still avoid linking the OpenSSL programs with the shared libraries.
   7194 
   7195      - When installing, install the shared libraries separately from the
   7196        static ones.
   7197      [Richard Levitte]
   7198 
   7199   *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
   7200 
   7201      Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
   7202      and not in SSL_clear because the latter is also used by the
   7203      accept/connect functions; previously, the settings made by
   7204      SSL_set_read_ahead would be lost during the handshake.
   7205      [Bodo Moeller; problems reported by Anders Gertz <gertz (a] epact.se>]     
   7206 
   7207   *) Correct util/mkdef.pl to be selective about disabled algorithms.
   7208      Previously, it would create entries for disableed algorithms no
   7209      matter what.
   7210      [Richard Levitte]
   7211 
   7212   *) Added several new manual pages for SSL_* function.
   7213      [Lutz Jaenicke]
   7214 
   7215  Changes between 0.9.5a and 0.9.6  [24 Sep 2000]
   7216 
   7217   *) In ssl23_get_client_hello, generate an error message when faced
   7218      with an initial SSL 3.0/TLS record that is too small to contain the
   7219      first two bytes of the ClientHello message, i.e. client_version.
   7220      (Note that this is a pathologic case that probably has never happened
   7221      in real life.)  The previous approach was to use the version number
   7222      from the record header as a substitute; but our protocol choice
   7223      should not depend on that one because it is not authenticated
   7224      by the Finished messages.
   7225      [Bodo Moeller]
   7226 
   7227   *) More robust randomness gathering functions for Windows.
   7228      [Jeffrey Altman <jaltman (a] columbia.edu>]
   7229 
   7230   *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
   7231      not set then we don't setup the error code for issuer check errors
   7232      to avoid possibly overwriting other errors which the callback does
   7233      handle. If an application does set the flag then we assume it knows
   7234      what it is doing and can handle the new informational codes
   7235      appropriately.
   7236      [Steve Henson]
   7237 
   7238   *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
   7239      a general "ANY" type, as such it should be able to decode anything
   7240      including tagged types. However it didn't check the class so it would
   7241      wrongly interpret tagged types in the same way as their universal
   7242      counterpart and unknown types were just rejected. Changed so that the
   7243      tagged and unknown types are handled in the same way as a SEQUENCE:
   7244      that is the encoding is stored intact. There is also a new type
   7245      "V_ASN1_OTHER" which is used when the class is not universal, in this
   7246      case we have no idea what the actual type is so we just lump them all
   7247      together.
   7248      [Steve Henson]
   7249 
   7250   *) On VMS, stdout may very well lead to a file that is written to
   7251      in a record-oriented fashion.  That means that every write() will
   7252      write a separate record, which will be read separately by the
   7253      programs trying to read from it.  This can be very confusing.
   7254 
   7255      The solution is to put a BIO filter in the way that will buffer
   7256      text until a linefeed is reached, and then write everything a
   7257      line at a time, so every record written will be an actual line,
   7258      not chunks of lines and not (usually doesn't happen, but I've
   7259      seen it once) several lines in one record.  BIO_f_linebuffer() is
   7260      the answer.
   7261 
   7262      Currently, it's a VMS-only method, because that's where it has
   7263      been tested well enough.
   7264      [Richard Levitte]
   7265 
   7266   *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
   7267      it can return incorrect results.
   7268      (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
   7269      but it was in 0.9.6-beta[12].)
   7270      [Bodo Moeller]
   7271 
   7272   *) Disable the check for content being present when verifying detached
   7273      signatures in pk7_smime.c. Some versions of Netscape (wrongly)
   7274      include zero length content when signing messages.
   7275      [Steve Henson]
   7276 
   7277   *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
   7278      BIO_ctrl (for BIO pairs).
   7279      [Bodo Mller]
   7280 
   7281   *) Add DSO method for VMS.
   7282      [Richard Levitte]
   7283 
   7284   *) Bug fix: Montgomery multiplication could produce results with the
   7285      wrong sign.
   7286      [Ulf Mller]
   7287 
   7288   *) Add RPM specification openssl.spec and modify it to build three
   7289      packages.  The default package contains applications, application
   7290      documentation and run-time libraries.  The devel package contains
   7291      include files, static libraries and function documentation.  The
   7292      doc package contains the contents of the doc directory.  The original
   7293      openssl.spec was provided by Damien Miller <djm (a] mindrot.org>.
   7294      [Richard Levitte]
   7295      
   7296   *) Add a large number of documentation files for many SSL routines.
   7297      [Lutz Jaenicke <Lutz.Jaenicke (a] aet.TU-Cottbus.DE>]
   7298 
   7299   *) Add a configuration entry for Sony News 4.
   7300      [NAKAJI Hiroyuki <nakaji (a] tutrp.tut.ac.jp>]
   7301 
   7302   *) Don't set the two most significant bits to one when generating a
   7303      random number < q in the DSA library.
   7304      [Ulf Mller]
   7305 
   7306   *) New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
   7307      behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
   7308      the underlying transport is blocking) if a handshake took place.
   7309      (The default behaviour is needed by applications such as s_client
   7310      and s_server that use select() to determine when to use SSL_read;
   7311      but for applications that know in advance when to expect data, it
   7312      just makes things more complicated.)
   7313      [Bodo Moeller]
   7314 
   7315   *) Add RAND_egd_bytes(), which gives control over the number of bytes read
   7316      from EGD.
   7317      [Ben Laurie]
   7318 
   7319   *) Add a few more EBCDIC conditionals that make `req' and `x509'
   7320      work better on such systems.
   7321      [Martin Kraemer <Martin.Kraemer (a] MchP.Siemens.De>]
   7322 
   7323   *) Add two demo programs for PKCS12_parse() and PKCS12_create().
   7324      Update PKCS12_parse() so it copies the friendlyName and the
   7325      keyid to the certificates aux info.
   7326      [Steve Henson]
   7327 
   7328   *) Fix bug in PKCS7_verify() which caused an infinite loop
   7329      if there was more than one signature.
   7330      [Sven Uszpelkat <su (a] celocom.de>]
   7331 
   7332   *) Major change in util/mkdef.pl to include extra information
   7333      about each symbol, as well as presentig variables as well
   7334      as functions.  This change means that there's n more need
   7335      to rebuild the .num files when some algorithms are excluded.
   7336      [Richard Levitte]
   7337 
   7338   *) Allow the verify time to be set by an application,
   7339      rather than always using the current time.
   7340      [Steve Henson]
   7341   
   7342   *) Phase 2 verify code reorganisation. The certificate
   7343      verify code now looks up an issuer certificate by a
   7344      number of criteria: subject name, authority key id
   7345      and key usage. It also verifies self signed certificates
   7346      by the same criteria. The main comparison function is
   7347      X509_check_issued() which performs these checks.
   7348  
   7349      Lot of changes were necessary in order to support this
   7350      without completely rewriting the lookup code.
   7351  
   7352      Authority and subject key identifier are now cached.
   7353  
   7354      The LHASH 'certs' is X509_STORE has now been replaced
   7355      by a STACK_OF(X509_OBJECT). This is mainly because an
   7356      LHASH can't store or retrieve multiple objects with
   7357      the same hash value.
   7358 
   7359      As a result various functions (which were all internal
   7360      use only) have changed to handle the new X509_STORE
   7361      structure. This will break anything that messed round
   7362      with X509_STORE internally.
   7363  
   7364      The functions X509_STORE_add_cert() now checks for an
   7365      exact match, rather than just subject name.
   7366  
   7367      The X509_STORE API doesn't directly support the retrieval
   7368      of multiple certificates matching a given criteria, however
   7369      this can be worked round by performing a lookup first
   7370      (which will fill the cache with candidate certificates)
   7371      and then examining the cache for matches. This is probably
   7372      the best we can do without throwing out X509_LOOKUP
   7373      entirely (maybe later...).
   7374  
   7375      The X509_VERIFY_CTX structure has been enhanced considerably.
   7376  
   7377      All certificate lookup operations now go via a get_issuer()
   7378      callback. Although this currently uses an X509_STORE it
   7379      can be replaced by custom lookups. This is a simple way
   7380      to bypass the X509_STORE hackery necessary to make this
   7381      work and makes it possible to use more efficient techniques
   7382      in future. A very simple version which uses a simple
   7383      STACK for its trusted certificate store is also provided
   7384      using X509_STORE_CTX_trusted_stack().
   7385  
   7386      The verify_cb() and verify() callbacks now have equivalents
   7387      in the X509_STORE_CTX structure.
   7388  
   7389      X509_STORE_CTX also has a 'flags' field which can be used
   7390      to customise the verify behaviour.
   7391      [Steve Henson]
   7392  
   7393   *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 
   7394      excludes S/MIME capabilities.
   7395      [Steve Henson]
   7396 
   7397   *) When a certificate request is read in keep a copy of the
   7398      original encoding of the signed data and use it when outputing
   7399      again. Signatures then use the original encoding rather than
   7400      a decoded, encoded version which may cause problems if the
   7401      request is improperly encoded.
   7402      [Steve Henson]
   7403 
   7404   *) For consistency with other BIO_puts implementations, call
   7405      buffer_write(b, ...) directly in buffer_puts instead of calling
   7406      BIO_write(b, ...).
   7407 
   7408      In BIO_puts, increment b->num_write as in BIO_write.
   7409      [Peter.Sylvester (a] EdelWeb.fr]
   7410 
   7411   *) Fix BN_mul_word for the case where the word is 0. (We have to use
   7412      BN_zero, we may not return a BIGNUM with an array consisting of
   7413      words set to zero.)
   7414      [Bodo Moeller]
   7415 
   7416   *) Avoid calling abort() from within the library when problems are
   7417      detected, except if preprocessor symbols have been defined
   7418      (such as REF_CHECK, BN_DEBUG etc.).
   7419      [Bodo Moeller]
   7420 
   7421   *) New openssl application 'rsautl'. This utility can be
   7422      used for low level RSA operations. DER public key
   7423      BIO/fp routines also added.
   7424      [Steve Henson]
   7425 
   7426   *) New Configure entry and patches for compiling on QNX 4.
   7427      [Andreas Schneider <andreas (a] ds3.etech.fh-hamburg.de>]
   7428 
   7429   *) A demo state-machine implementation was sponsored by
   7430      Nuron (http://www.nuron.com/) and is now available in
   7431      demos/state_machine.
   7432      [Ben Laurie]
   7433 
   7434   *) New options added to the 'dgst' utility for signature
   7435      generation and verification.
   7436      [Steve Henson]
   7437 
   7438   *) Unrecognized PKCS#7 content types are now handled via a
   7439      catch all ASN1_TYPE structure. This allows unsupported
   7440      types to be stored as a "blob" and an application can
   7441      encode and decode it manually.
   7442      [Steve Henson]
   7443 
   7444   *) Fix various signed/unsigned issues to make a_strex.c
   7445      compile under VC++.
   7446      [Oscar Jacobsson <oscar.jacobsson (a] celocom.com>]
   7447 
   7448   *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
   7449      length if passed a buffer. ASN1_INTEGER_to_BN failed
   7450      if passed a NULL BN and its argument was negative.
   7451      [Steve Henson, pointed out by Sven Heiberg <sven (a] tartu.cyber.ee>]
   7452 
   7453   *) Modification to PKCS#7 encoding routines to output definite
   7454      length encoding. Since currently the whole structures are in
   7455      memory there's not real point in using indefinite length 
   7456      constructed encoding. However if OpenSSL is compiled with
   7457      the flag PKCS7_INDEFINITE_ENCODING the old form is used.
   7458      [Steve Henson]
   7459 
   7460   *) Added BIO_vprintf() and BIO_vsnprintf().
   7461      [Richard Levitte]
   7462 
   7463   *) Added more prefixes to parse for in the the strings written
   7464      through a logging bio, to cover all the levels that are available
   7465      through syslog.  The prefixes are now:
   7466 
   7467 	PANIC, EMERG, EMR	=>	LOG_EMERG
   7468 	ALERT, ALR		=>	LOG_ALERT
   7469 	CRIT, CRI		=>	LOG_CRIT
   7470 	ERROR, ERR		=>	LOG_ERR
   7471 	WARNING, WARN, WAR	=>	LOG_WARNING
   7472 	NOTICE, NOTE, NOT	=>	LOG_NOTICE
   7473 	INFO, INF		=>	LOG_INFO
   7474 	DEBUG, DBG		=>	LOG_DEBUG
   7475 
   7476      and as before, if none of those prefixes are present at the
   7477      beginning of the string, LOG_ERR is chosen.
   7478 
   7479      On Win32, the LOG_* levels are mapped according to this:
   7480 
   7481 	LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR	=> EVENTLOG_ERROR_TYPE
   7482 	LOG_WARNING				=> EVENTLOG_WARNING_TYPE
   7483 	LOG_NOTICE, LOG_INFO, LOG_DEBUG		=> EVENTLOG_INFORMATION_TYPE
   7484 
   7485      [Richard Levitte]
   7486 
   7487   *) Made it possible to reconfigure with just the configuration
   7488      argument "reconf" or "reconfigure".  The command line arguments
   7489      are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
   7490      and are retrieved from there when reconfiguring.
   7491      [Richard Levitte]
   7492 
   7493   *) MD4 implemented.
   7494      [Assar Westerlund <assar (a] sics.se>, Richard Levitte]
   7495 
   7496   *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
   7497      [Richard Levitte]
   7498 
   7499   *) The obj_dat.pl script was messing up the sorting of object
   7500      names. The reason was that it compared the quoted version
   7501      of strings as a result "OCSP" > "OCSP Signing" because
   7502      " > SPACE. Changed script to store unquoted versions of
   7503      names and add quotes on output. It was also omitting some
   7504      names from the lookup table if they were given a default
   7505      value (that is if SN is missing it is given the same
   7506      value as LN and vice versa), these are now added on the
   7507      grounds that if an object has a name we should be able to
   7508      look it up. Finally added warning output when duplicate
   7509      short or long names are found.
   7510      [Steve Henson]
   7511 
   7512   *) Changes needed for Tandem NSK.
   7513      [Scott Uroff <scott (a] xypro.com>]
   7514 
   7515   *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
   7516      RSA_padding_check_SSLv23(), special padding was never detected
   7517      and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
   7518      version rollback attacks was not effective.
   7519 
   7520      In s23_clnt.c, don't use special rollback-attack detection padding
   7521      (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
   7522      client; similarly, in s23_srvr.c, don't do the rollback check if
   7523      SSL 2.0 is the only protocol enabled in the server.
   7524      [Bodo Moeller]
   7525 
   7526   *) Make it possible to get hexdumps of unprintable data with 'openssl
   7527      asn1parse'.  By implication, the functions ASN1_parse_dump() and
   7528      BIO_dump_indent() are added.
   7529      [Richard Levitte]
   7530 
   7531   *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
   7532      these print out strings and name structures based on various
   7533      flags including RFC2253 support and proper handling of
   7534      multibyte characters. Added options to the 'x509' utility 
   7535      to allow the various flags to be set.
   7536      [Steve Henson]
   7537 
   7538   *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
   7539      Also change the functions X509_cmp_current_time() and
   7540      X509_gmtime_adj() work with an ASN1_TIME structure,
   7541      this will enable certificates using GeneralizedTime in validity
   7542      dates to be checked.
   7543      [Steve Henson]
   7544 
   7545   *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
   7546      negative public key encodings) on by default,
   7547      NO_NEG_PUBKEY_BUG can be set to disable it.
   7548      [Steve Henson]
   7549 
   7550   *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
   7551      content octets. An i2c_ASN1_OBJECT is unnecessary because
   7552      the encoding can be trivially obtained from the structure.
   7553      [Steve Henson]
   7554 
   7555   *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
   7556      not read locks (CRYPTO_r_[un]lock).
   7557      [Bodo Moeller]
   7558 
   7559   *) A first attempt at creating official support for shared
   7560      libraries through configuration.  I've kept it so the
   7561      default is static libraries only, and the OpenSSL programs
   7562      are always statically linked for now, but there are
   7563      preparations for dynamic linking in place.
   7564      This has been tested on Linux and Tru64.
   7565      [Richard Levitte]
   7566 
   7567   *) Randomness polling function for Win9x, as described in:
   7568      Peter Gutmann, Software Generation of Practically Strong
   7569      Random Numbers.
   7570      [Ulf Mller]
   7571 
   7572   *) Fix so PRNG is seeded in req if using an already existing
   7573      DSA key.
   7574      [Steve Henson]
   7575 
   7576   *) New options to smime application. -inform and -outform
   7577      allow alternative formats for the S/MIME message including
   7578      PEM and DER. The -content option allows the content to be
   7579      specified separately. This should allow things like Netscape
   7580      form signing output easier to verify.
   7581      [Steve Henson]
   7582 
   7583   *) Fix the ASN1 encoding of tags using the 'long form'.
   7584      [Steve Henson]
   7585 
   7586   *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
   7587      STRING types. These convert content octets to and from the
   7588      underlying type. The actual tag and length octets are
   7589      already assumed to have been read in and checked. These
   7590      are needed because all other string types have virtually
   7591      identical handling apart from the tag. By having versions
   7592      of the ASN1 functions that just operate on content octets
   7593      IMPLICIT tagging can be handled properly. It also allows
   7594      the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
   7595      and ASN1_INTEGER are identical apart from the tag.
   7596      [Steve Henson]
   7597 
   7598   *) Change the handling of OID objects as follows:
   7599 
   7600      - New object identifiers are inserted in objects.txt, following
   7601        the syntax given in objects.README.
   7602      - objects.pl is used to process obj_mac.num and create a new
   7603        obj_mac.h.
   7604      - obj_dat.pl is used to create a new obj_dat.h, using the data in
   7605        obj_mac.h.
   7606 
   7607      This is currently kind of a hack, and the perl code in objects.pl
   7608      isn't very elegant, but it works as I intended.  The simplest way
   7609      to check that it worked correctly is to look in obj_dat.h and
   7610      check the array nid_objs and make sure the objects haven't moved
   7611      around (this is important!).  Additions are OK, as well as
   7612      consistent name changes. 
   7613      [Richard Levitte]
   7614 
   7615   *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
   7616      [Bodo Moeller]
   7617 
   7618   *) Addition of the command line parameter '-rand file' to 'openssl req'.
   7619      The given file adds to whatever has already been seeded into the
   7620      random pool through the RANDFILE configuration file option or
   7621      environment variable, or the default random state file.
   7622      [Richard Levitte]
   7623 
   7624   *) mkstack.pl now sorts each macro group into lexical order.
   7625      Previously the output order depended on the order the files
   7626      appeared in the directory, resulting in needless rewriting
   7627      of safestack.h .
   7628      [Steve Henson]
   7629 
   7630   *) Patches to make OpenSSL compile under Win32 again. Mostly
   7631      work arounds for the VC++ problem that it treats func() as
   7632      func(void). Also stripped out the parts of mkdef.pl that
   7633      added extra typesafe functions: these no longer exist.
   7634      [Steve Henson]
   7635 
   7636   *) Reorganisation of the stack code. The macros are now all 
   7637      collected in safestack.h . Each macro is defined in terms of
   7638      a "stack macro" of the form SKM_<name>(type, a, b). The 
   7639      DEBUG_SAFESTACK is now handled in terms of function casts,
   7640      this has the advantage of retaining type safety without the
   7641      use of additional functions. If DEBUG_SAFESTACK is not defined
   7642      then the non typesafe macros are used instead. Also modified the
   7643      mkstack.pl script to handle the new form. Needs testing to see
   7644      if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
   7645      the default if no major problems. Similar behaviour for ASN1_SET_OF
   7646      and PKCS12_STACK_OF.
   7647      [Steve Henson]
   7648 
   7649   *) When some versions of IIS use the 'NET' form of private key the
   7650      key derivation algorithm is different. Normally MD5(password) is
   7651      used as a 128 bit RC4 key. In the modified case
   7652      MD5(MD5(password) + "SGCKEYSALT")  is used insted. Added some
   7653      new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
   7654      as the old Netscape_RSA functions except they have an additional
   7655      'sgckey' parameter which uses the modified algorithm. Also added
   7656      an -sgckey command line option to the rsa utility. Thanks to 
   7657      Adrian Peck <bertie (a] ncipher.com> for posting details of the modified
   7658      algorithm to openssl-dev.
   7659      [Steve Henson]
   7660 
   7661   *) The evp_local.h macros were using 'c.##kname' which resulted in
   7662      invalid expansion on some systems (SCO 5.0.5 for example).
   7663      Corrected to 'c.kname'.
   7664      [Phillip Porch <root (a] theporch.com>]
   7665 
   7666   *) New X509_get1_email() and X509_REQ_get1_email() functions that return
   7667      a STACK of email addresses from a certificate or request, these look
   7668      in the subject name and the subject alternative name extensions and 
   7669      omit any duplicate addresses.
   7670      [Steve Henson]
   7671 
   7672   *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
   7673      This makes DSA verification about 2 % faster.
   7674      [Bodo Moeller]
   7675 
   7676   *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
   7677      (meaning that now 2^5 values will be precomputed, which is only 4 KB
   7678      plus overhead for 1024 bit moduli).
   7679      This makes exponentiations about 0.5 % faster for 1024 bit
   7680      exponents (as measured by "openssl speed rsa2048").
   7681      [Bodo Moeller]
   7682 
   7683   *) Rename memory handling macros to avoid conflicts with other
   7684      software:
   7685           Malloc         =>  OPENSSL_malloc
   7686           Malloc_locked  =>  OPENSSL_malloc_locked
   7687           Realloc        =>  OPENSSL_realloc
   7688           Free           =>  OPENSSL_free
   7689      [Richard Levitte]
   7690 
   7691   *) New function BN_mod_exp_mont_word for small bases (roughly 15%
   7692      faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
   7693      [Bodo Moeller]
   7694 
   7695   *) CygWin32 support.
   7696      [John Jarvie <jjarvie (a] newsguy.com>]
   7697 
   7698   *) The type-safe stack code has been rejigged. It is now only compiled
   7699      in when OpenSSL is configured with the DEBUG_SAFESTACK option and
   7700      by default all type-specific stack functions are "#define"d back to
   7701      standard stack functions. This results in more streamlined output
   7702      but retains the type-safety checking possibilities of the original
   7703      approach.
   7704      [Geoff Thorpe]
   7705 
   7706   *) The STACK code has been cleaned up, and certain type declarations
   7707      that didn't make a lot of sense have been brought in line. This has
   7708      also involved a cleanup of sorts in safestack.h to more correctly
   7709      map type-safe stack functions onto their plain stack counterparts.
   7710      This work has also resulted in a variety of "const"ifications of
   7711      lots of the code, especially "_cmp" operations which should normally
   7712      be prototyped with "const" parameters anyway.
   7713      [Geoff Thorpe]
   7714 
   7715   *) When generating bytes for the first time in md_rand.c, 'stir the pool'
   7716      by seeding with STATE_SIZE dummy bytes (with zero entropy count).
   7717      (The PRNG state consists of two parts, the large pool 'state' and 'md',
   7718      where all of 'md' is used each time the PRNG is used, but 'state'
   7719      is used only indexed by a cyclic counter. As entropy may not be
   7720      well distributed from the beginning, 'md' is important as a
   7721      chaining variable. However, the output function chains only half
   7722      of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
   7723      all of 'md', and seeding with STATE_SIZE dummy bytes will result
   7724      in all of 'state' being rewritten, with the new values depending
   7725      on virtually all of 'md'.  This overcomes the 80 bit limitation.)
   7726      [Bodo Moeller]
   7727 
   7728   *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
   7729      the handshake is continued after ssl_verify_cert_chain();
   7730      otherwise, if SSL_VERIFY_NONE is set, remaining error codes
   7731      can lead to 'unexplainable' connection aborts later.
   7732      [Bodo Moeller; problem tracked down by Lutz Jaenicke]
   7733 
   7734   *) Major EVP API cipher revision.
   7735      Add hooks for extra EVP features. This allows various cipher
   7736      parameters to be set in the EVP interface. Support added for variable
   7737      key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
   7738      setting of RC2 and RC5 parameters.
   7739 
   7740      Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
   7741      ciphers.
   7742 
   7743      Remove lots of duplicated code from the EVP library. For example *every*
   7744      cipher init() function handles the 'iv' in the same way according to the
   7745      cipher mode. They also all do nothing if the 'key' parameter is NULL and
   7746      for CFB and OFB modes they zero ctx->num.
   7747 
   7748      New functionality allows removal of S/MIME code RC2 hack.
   7749 
   7750      Most of the routines have the same form and so can be declared in terms
   7751      of macros.
   7752 
   7753      By shifting this to the top level EVP_CipherInit() it can be removed from
   7754      all individual ciphers. If the cipher wants to handle IVs or keys
   7755      differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
   7756      flags.
   7757 
   7758      Change lots of functions like EVP_EncryptUpdate() to now return a
   7759      value: although software versions of the algorithms cannot fail
   7760      any installed hardware versions can.
   7761      [Steve Henson]
   7762 
   7763   *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
   7764      this option is set, tolerate broken clients that send the negotiated
   7765      protocol version number instead of the requested protocol version
   7766      number.
   7767      [Bodo Moeller]
   7768 
   7769   *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
   7770      i.e. non-zero for export ciphersuites, zero otherwise.
   7771      Previous versions had this flag inverted, inconsistent with
   7772      rsa_tmp_cb (..._TMP_RSA_CB).
   7773      [Bodo Moeller; problem reported by Amit Chopra]
   7774 
   7775   *) Add missing DSA library text string. Work around for some IIS
   7776      key files with invalid SEQUENCE encoding.
   7777      [Steve Henson]
   7778 
   7779   *) Add a document (doc/standards.txt) that list all kinds of standards
   7780      and so on that are implemented in OpenSSL.
   7781      [Richard Levitte]
   7782 
   7783   *) Enhance c_rehash script. Old version would mishandle certificates
   7784      with the same subject name hash and wouldn't handle CRLs at all.
   7785      Added -fingerprint option to crl utility, to support new c_rehash
   7786      features.
   7787      [Steve Henson]
   7788 
   7789   *) Eliminate non-ANSI declarations in crypto.h and stack.h.
   7790      [Ulf Mller]
   7791 
   7792   *) Fix for SSL server purpose checking. Server checking was
   7793      rejecting certificates which had extended key usage present
   7794      but no ssl client purpose.
   7795      [Steve Henson, reported by Rene Grosser <grosser (a] hisolutions.com>]
   7796 
   7797   *) Make PKCS#12 code work with no password. The PKCS#12 spec
   7798      is a little unclear about how a blank password is handled.
   7799      Since the password in encoded as a BMPString with terminating
   7800      double NULL a zero length password would end up as just the
   7801      double NULL. However no password at all is different and is
   7802      handled differently in the PKCS#12 key generation code. NS
   7803      treats a blank password as zero length. MSIE treats it as no
   7804      password on export: but it will try both on import. We now do
   7805      the same: PKCS12_parse() tries zero length and no password if
   7806      the password is set to "" or NULL (NULL is now a valid password:
   7807      it wasn't before) as does the pkcs12 application.
   7808      [Steve Henson]
   7809 
   7810   *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
   7811      perror when PEM_read_bio_X509_REQ fails, the error message must
   7812      be obtained from the error queue.
   7813      [Bodo Moeller]
   7814 
   7815   *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
   7816      it in ERR_remove_state if appropriate, and change ERR_get_state
   7817      accordingly to avoid race conditions (this is necessary because
   7818      thread_hash is no longer constant once set).
   7819      [Bodo Moeller]
   7820 
   7821   *) Bugfix for linux-elf makefile.one.
   7822      [Ulf Mller]
   7823 
   7824   *) RSA_get_default_method() will now cause a default
   7825      RSA_METHOD to be chosen if one doesn't exist already.
   7826      Previously this was only set during a call to RSA_new()
   7827      or RSA_new_method(NULL) meaning it was possible for
   7828      RSA_get_default_method() to return NULL.
   7829      [Geoff Thorpe]
   7830 
   7831   *) Added native name translation to the existing DSO code
   7832      that will convert (if the flag to do so is set) filenames
   7833      that are sufficiently small and have no path information
   7834      into a canonical native form. Eg. "blah" converted to
   7835      "libblah.so" or "blah.dll" etc.
   7836      [Geoff Thorpe]
   7837 
   7838   *) New function ERR_error_string_n(e, buf, len) which is like
   7839      ERR_error_string(e, buf), but writes at most 'len' bytes
   7840      including the 0 terminator.  For ERR_error_string_n, 'buf'
   7841      may not be NULL.
   7842      [Damien Miller <djm (a] mindrot.org>, Bodo Moeller]
   7843 
   7844   *) CONF library reworked to become more general.  A new CONF
   7845      configuration file reader "class" is implemented as well as a
   7846      new functions (NCONF_*, for "New CONF") to handle it.  The now
   7847      old CONF_* functions are still there, but are reimplemented to
   7848      work in terms of the new functions.  Also, a set of functions
   7849      to handle the internal storage of the configuration data is
   7850      provided to make it easier to write new configuration file
   7851      reader "classes" (I can definitely see something reading a
   7852      configuration file in XML format, for example), called _CONF_*,
   7853      or "the configuration storage API"...
   7854 
   7855      The new configuration file reading functions are:
   7856 
   7857         NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
   7858         NCONF_get_section, NCONF_get_string, NCONF_get_numbre
   7859 
   7860         NCONF_default, NCONF_WIN32
   7861 
   7862         NCONF_dump_fp, NCONF_dump_bio
   7863 
   7864      NCONF_default and NCONF_WIN32 are method (or "class") choosers,
   7865      NCONF_new creates a new CONF object.  This works in the same way
   7866      as other interfaces in OpenSSL, like the BIO interface.
   7867      NCONF_dump_* dump the internal storage of the configuration file,
   7868      which is useful for debugging.  All other functions take the same
   7869      arguments as the old CONF_* functions wth the exception of the
   7870      first that must be a `CONF *' instead of a `LHASH *'.
   7871 
   7872      To make it easer to use the new classes with the old CONF_* functions,
   7873      the function CONF_set_default_method is provided.
   7874      [Richard Levitte]
   7875 
   7876   *) Add '-tls1' option to 'openssl ciphers', which was already
   7877      mentioned in the documentation but had not been implemented.
   7878      (This option is not yet really useful because even the additional
   7879      experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
   7880      [Bodo Moeller]
   7881 
   7882   *) Initial DSO code added into libcrypto for letting OpenSSL (and
   7883      OpenSSL-based applications) load shared libraries and bind to
   7884      them in a portable way.
   7885      [Geoff Thorpe, with contributions from Richard Levitte]
   7886 
   7887  Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]
   7888 
   7889   *) Make sure _lrotl and _lrotr are only used with MSVC.
   7890 
   7891   *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
   7892      (the default implementation of RAND_status).
   7893 
   7894   *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
   7895      to '-clrext' (= clear extensions), as intended and documented.
   7896      [Bodo Moeller; inconsistency pointed out by Michael Attili
   7897      <attili (a] amaxo.com>]
   7898 
   7899   *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
   7900      was larger than the MD block size.      
   7901      [Steve Henson, pointed out by Yost William <YostW (a] tce.com>]
   7902 
   7903   *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
   7904      fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
   7905      using the passed key: if the passed key was a private key the result
   7906      of X509_print(), for example, would be to print out all the private key
   7907      components.
   7908      [Steve Henson]
   7909 
   7910   *) des_quad_cksum() byte order bug fix.
   7911      [Ulf Mller, using the problem description in krb4-0.9.7, where
   7912       the solution is attributed to Derrick J Brashear <shadow (a] DEMENTIA.ORG>]
   7913 
   7914   *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
   7915      discouraged.
   7916      [Steve Henson, pointed out by Brian Korver <briank (a] cs.stanford.edu>]
   7917 
   7918   *) For easily testing in shell scripts whether some command
   7919      'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
   7920      returns with exit code 0 iff no command of the given name is available.
   7921      'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
   7922      the output goes to stdout and nothing is printed to stderr.
   7923      Additional arguments are always ignored.
   7924 
   7925      Since for each cipher there is a command of the same name,
   7926      the 'no-cipher' compilation switches can be tested this way.
   7927 
   7928      ('openssl no-XXX' is not able to detect pseudo-commands such
   7929      as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
   7930      [Bodo Moeller]
   7931 
   7932   *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
   7933      [Bodo Moeller]
   7934 
   7935   *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
   7936      is set; it will be thrown away anyway because each handshake creates
   7937      its own key.
   7938      ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
   7939      to parameters -- in previous versions (since OpenSSL 0.9.3) the
   7940      'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
   7941      you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
   7942      [Bodo Moeller]
   7943 
   7944   *) New s_client option -ign_eof: EOF at stdin is ignored, and
   7945      'Q' and 'R' lose their special meanings (quit/renegotiate).
   7946      This is part of what -quiet does; unlike -quiet, -ign_eof
   7947      does not suppress any output.
   7948      [Richard Levitte]
   7949 
   7950   *) Add compatibility options to the purpose and trust code. The
   7951      purpose X509_PURPOSE_ANY is "any purpose" which automatically
   7952      accepts a certificate or CA, this was the previous behaviour,
   7953      with all the associated security issues.
   7954 
   7955      X509_TRUST_COMPAT is the old trust behaviour: only and
   7956      automatically trust self signed roots in certificate store. A
   7957      new trust setting X509_TRUST_DEFAULT is used to specify that
   7958      a purpose has no associated trust setting and it should instead
   7959      use the value in the default purpose.
   7960      [Steve Henson]
   7961 
   7962   *) Fix the PKCS#8 DSA private key code so it decodes keys again
   7963      and fix a memory leak.
   7964      [Steve Henson]
   7965 
   7966   *) In util/mkerr.pl (which implements 'make errors'), preserve
   7967      reason strings from the previous version of the .c file, as
   7968      the default to have only downcase letters (and digits) in
   7969      automatically generated reasons codes is not always appropriate.
   7970      [Bodo Moeller]
   7971 
   7972   *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
   7973      using strerror.  Previously, ERR_reason_error_string() returned
   7974      library names as reason strings for SYSerr; but SYSerr is a special
   7975      case where small numbers are errno values, not library numbers.
   7976      [Bodo Moeller]
   7977 
   7978   *) Add '-dsaparam' option to 'openssl dhparam' application.  This
   7979      converts DSA parameters into DH parameters. (When creating parameters,
   7980      DSA_generate_parameters is used.)
   7981      [Bodo Moeller]
   7982 
   7983   *) Include 'length' (recommended exponent length) in C code generated
   7984      by 'openssl dhparam -C'.
   7985      [Bodo Moeller]
   7986 
   7987   *) The second argument to set_label in perlasm was already being used
   7988      so couldn't be used as a "file scope" flag. Moved to third argument
   7989      which was free.
   7990      [Steve Henson]
   7991 
   7992   *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
   7993      instead of RAND_bytes for encryption IVs and salts.
   7994      [Bodo Moeller]
   7995 
   7996   *) Include RAND_status() into RAND_METHOD instead of implementing
   7997      it only for md_rand.c  Otherwise replacing the PRNG by calling
   7998      RAND_set_rand_method would be impossible.
   7999      [Bodo Moeller]
   8000 
   8001   *) Don't let DSA_generate_key() enter an infinite loop if the random
   8002      number generation fails.
   8003      [Bodo Moeller]
   8004 
   8005   *) New 'rand' application for creating pseudo-random output.
   8006      [Bodo Moeller]
   8007 
   8008   *) Added configuration support for Linux/IA64
   8009      [Rolf Haberrecker <rolf (a] suse.de>]
   8010 
   8011   *) Assembler module support for Mingw32.
   8012      [Ulf Mller]
   8013 
   8014   *) Shared library support for HPUX (in shlib/).
   8015      [Lutz Jaenicke <Lutz.Jaenicke (a] aet.TU-Cottbus.DE> and Anonymous]
   8016 
   8017   *) Shared library support for Solaris gcc.
   8018      [Lutz Behnke <behnke (a] trustcenter.de>]
   8019 
   8020  Changes between 0.9.4 and 0.9.5  [28 Feb 2000]
   8021 
   8022   *) PKCS7_encrypt() was adding text MIME headers twice because they
   8023      were added manually and by SMIME_crlf_copy().
   8024      [Steve Henson]
   8025 
   8026   *) In bntest.c don't call BN_rand with zero bits argument.
   8027      [Steve Henson, pointed out by Andrew W. Gray <agray (a] iconsinc.com>]
   8028 
   8029   *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
   8030      case was implemented. This caused BN_div_recp() to fail occasionally.
   8031      [Ulf Mller]
   8032 
   8033   *) Add an optional second argument to the set_label() in the perl
   8034      assembly language builder. If this argument exists and is set
   8035      to 1 it signals that the assembler should use a symbol whose 
   8036      scope is the entire file, not just the current function. This
   8037      is needed with MASM which uses the format label:: for this scope.
   8038      [Steve Henson, pointed out by Peter Runestig <peter (a] runestig.com>]
   8039 
   8040   *) Change the ASN1 types so they are typedefs by default. Before
   8041      almost all types were #define'd to ASN1_STRING which was causing
   8042      STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
   8043      for example.
   8044      [Steve Henson]
   8045 
   8046   *) Change names of new functions to the new get1/get0 naming
   8047      convention: After 'get1', the caller owns a reference count
   8048      and has to call ..._free; 'get0' returns a pointer to some
   8049      data structure without incrementing reference counters.
   8050      (Some of the existing 'get' functions increment a reference
   8051      counter, some don't.)
   8052      Similarly, 'set1' and 'add1' functions increase reference
   8053      counters or duplicate objects.
   8054      [Steve Henson]
   8055 
   8056   *) Allow for the possibility of temp RSA key generation failure:
   8057      the code used to assume it always worked and crashed on failure.
   8058      [Steve Henson]
   8059 
   8060   *) Fix potential buffer overrun problem in BIO_printf().
   8061      [Ulf Mller, using public domain code by Patrick Powell; problem
   8062       pointed out by David Sacerdote <das33 (a] cornell.edu>]
   8063 
   8064   *) Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
   8065      RAND_egd() and RAND_status().  In the command line application,
   8066      the EGD socket can be specified like a seed file using RANDFILE
   8067      or -rand.
   8068      [Ulf Mller]
   8069 
   8070   *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
   8071      Some CAs (e.g. Verisign) distribute certificates in this form.
   8072      [Steve Henson]
   8073 
   8074   *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
   8075      list to exclude them. This means that no special compilation option
   8076      is needed to use anonymous DH: it just needs to be included in the
   8077      cipher list.
   8078      [Steve Henson]
   8079 
   8080   *) Change the EVP_MD_CTX_type macro so its meaning consistent with
   8081      EVP_MD_type. The old functionality is available in a new macro called
   8082      EVP_MD_md(). Change code that uses it and update docs.
   8083      [Steve Henson]
   8084 
   8085   *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
   8086      where the 'void *' argument is replaced by a function pointer argument.
   8087      Previously 'void *' was abused to point to functions, which works on
   8088      many platforms, but is not correct.  As these functions are usually
   8089      called by macros defined in OpenSSL header files, most source code
   8090      should work without changes.
   8091      [Richard Levitte]
   8092 
   8093   *) <openssl/opensslconf.h> (which is created by Configure) now contains
   8094      sections with information on -D... compiler switches used for
   8095      compiling the library so that applications can see them.  To enable
   8096      one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
   8097      must be defined.  E.g.,
   8098         #define OPENSSL_ALGORITHM_DEFINES
   8099         #include <openssl/opensslconf.h>
   8100      defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
   8101      [Richard Levitte, Ulf and Bodo Mller]
   8102 
   8103   *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
   8104      record layer.
   8105      [Bodo Moeller]
   8106 
   8107   *) Change the 'other' type in certificate aux info to a STACK_OF
   8108      X509_ALGOR. Although not an AlgorithmIdentifier as such it has
   8109      the required ASN1 format: arbitrary types determined by an OID.
   8110      [Steve Henson]
   8111 
   8112   *) Add some PEM_write_X509_REQ_NEW() functions and a command line
   8113      argument to 'req'. This is not because the function is newer or
   8114      better than others it just uses the work 'NEW' in the certificate
   8115      request header lines. Some software needs this.
   8116      [Steve Henson]
   8117 
   8118   *) Reorganise password command line arguments: now passwords can be
   8119      obtained from various sources. Delete the PEM_cb function and make
   8120      it the default behaviour: i.e. if the callback is NULL and the
   8121      usrdata argument is not NULL interpret it as a null terminated pass
   8122      phrase. If usrdata and the callback are NULL then the pass phrase
   8123      is prompted for as usual.
   8124      [Steve Henson]
   8125 
   8126   *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
   8127      the support is automatically enabled. The resulting binaries will
   8128      autodetect the card and use it if present.
   8129      [Ben Laurie and Compaq Inc.]
   8130 
   8131   *) Work around for Netscape hang bug. This sends certificate request
   8132      and server done in one record. Since this is perfectly legal in the
   8133      SSL/TLS protocol it isn't a "bug" option and is on by default. See
   8134      the bugs/SSLv3 entry for more info.
   8135      [Steve Henson]
   8136 
   8137   *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
   8138      [Andy Polyakov]
   8139 
   8140   *) Add -rand argument to smime and pkcs12 applications and read/write
   8141      of seed file.
   8142      [Steve Henson]
   8143 
   8144   *) New 'passwd' tool for crypt(3) and apr1 password hashes.
   8145      [Bodo Moeller]
   8146 
   8147   *) Add command line password options to the remaining applications.
   8148      [Steve Henson]
   8149 
   8150   *) Bug fix for BN_div_recp() for numerators with an even number of
   8151      bits.
   8152      [Ulf Mller]
   8153 
   8154   *) More tests in bntest.c, and changed test_bn output.
   8155      [Ulf Mller]
   8156 
   8157   *) ./config recognizes MacOS X now.
   8158      [Andy Polyakov]
   8159 
   8160   *) Bug fix for BN_div() when the first words of num and divsor are
   8161      equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
   8162      [Ulf Mller]
   8163 
   8164   *) Add support for various broken PKCS#8 formats, and command line
   8165      options to produce them.
   8166      [Steve Henson]
   8167 
   8168   *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
   8169      get temporary BIGNUMs from a BN_CTX.
   8170      [Ulf Mller]
   8171 
   8172   *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
   8173      for p == 0.
   8174      [Ulf Mller]
   8175 
   8176   *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
   8177      include a #define from the old name to the new. The original intent
   8178      was that statically linked binaries could for example just call
   8179      SSLeay_add_all_ciphers() to just add ciphers to the table and not
   8180      link with digests. This never worked becayse SSLeay_add_all_digests()
   8181      and SSLeay_add_all_ciphers() were in the same source file so calling
   8182      one would link with the other. They are now in separate source files.
   8183      [Steve Henson]
   8184 
   8185   *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
   8186      [Steve Henson]
   8187 
   8188   *) Use a less unusual form of the Miller-Rabin primality test (it used
   8189      a binary algorithm for exponentiation integrated into the Miller-Rabin
   8190      loop, our standard modexp algorithms are faster).
   8191      [Bodo Moeller]
   8192 
   8193   *) Support for the EBCDIC character set completed.
   8194      [Martin Kraemer <Martin.Kraemer (a] Mch.SNI.De>]
   8195 
   8196   *) Source code cleanups: use const where appropriate, eliminate casts,
   8197      use void * instead of char * in lhash.
   8198      [Ulf Mller] 
   8199 
   8200   *) Bugfix: ssl3_send_server_key_exchange was not restartable
   8201      (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
   8202      this the server could overwrite ephemeral keys that the client
   8203      has already seen).
   8204      [Bodo Moeller]
   8205 
   8206   *) Turn DSA_is_prime into a macro that calls BN_is_prime,
   8207      using 50 iterations of the Rabin-Miller test.
   8208 
   8209      DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
   8210      iterations of the Rabin-Miller test as required by the appendix
   8211      to FIPS PUB 186[-1]) instead of DSA_is_prime.
   8212      As BN_is_prime_fasttest includes trial division, DSA parameter
   8213      generation becomes much faster.
   8214 
   8215      This implies a change for the callback functions in DSA_is_prime
   8216      and DSA_generate_parameters: The callback function is called once
   8217      for each positive witness in the Rabin-Miller test, not just
   8218      occasionally in the inner loop; and the parameters to the
   8219      callback function now provide an iteration count for the outer
   8220      loop rather than for the current invocation of the inner loop.
   8221      DSA_generate_parameters additionally can call the callback
   8222      function with an 'iteration count' of -1, meaning that a
   8223      candidate has passed the trial division test (when q is generated 
   8224      from an application-provided seed, trial division is skipped).
   8225      [Bodo Moeller]
   8226 
   8227   *) New function BN_is_prime_fasttest that optionally does trial
   8228      division before starting the Rabin-Miller test and has
   8229      an additional BN_CTX * argument (whereas BN_is_prime always
   8230      has to allocate at least one BN_CTX).
   8231      'callback(1, -1, cb_arg)' is called when a number has passed the
   8232      trial division stage.
   8233      [Bodo Moeller]
   8234 
   8235   *) Fix for bug in CRL encoding. The validity dates weren't being handled
   8236      as ASN1_TIME.
   8237      [Steve Henson]
   8238 
   8239   *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
   8240      [Steve Henson]
   8241 
   8242   *) New function BN_pseudo_rand().
   8243      [Ulf Mller]
   8244 
   8245   *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
   8246      bignum version of BN_from_montgomery() with the working code from
   8247      SSLeay 0.9.0 (the word based version is faster anyway), and clean up
   8248      the comments.
   8249      [Ulf Mller]
   8250 
   8251   *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
   8252      made it impossible to use the same SSL_SESSION data structure in
   8253      SSL2 clients in multiple threads.
   8254      [Bodo Moeller]
   8255 
   8256   *) The return value of RAND_load_file() no longer counts bytes obtained
   8257      by stat().  RAND_load_file(..., -1) is new and uses the complete file
   8258      to seed the PRNG (previously an explicit byte count was required).
   8259      [Ulf Mller, Bodo Mller]
   8260 
   8261   *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
   8262      used (char *) instead of (void *) and had casts all over the place.
   8263      [Steve Henson]
   8264 
   8265   *) Make BN_generate_prime() return NULL on error if ret!=NULL.
   8266      [Ulf Mller]
   8267 
   8268   *) Retain source code compatibility for BN_prime_checks macro:
   8269      BN_is_prime(..., BN_prime_checks, ...) now uses
   8270      BN_prime_checks_for_size to determine the appropriate number of
   8271      Rabin-Miller iterations.
   8272      [Ulf Mller]
   8273 
   8274   *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
   8275      DH_CHECK_P_NOT_SAFE_PRIME.
   8276      (Check if this is true? OpenPGP calls them "strong".)
   8277      [Ulf Mller]
   8278 
   8279   *) Merge the functionality of "dh" and "gendh" programs into a new program
   8280      "dhparam". The old programs are retained for now but will handle DH keys
   8281      (instead of parameters) in future.
   8282      [Steve Henson]
   8283 
   8284   *) Make the ciphers, s_server and s_client programs check the return values
   8285      when a new cipher list is set.
   8286      [Steve Henson]
   8287 
   8288   *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
   8289      ciphers. Before when the 56bit ciphers were enabled the sorting was
   8290      wrong.
   8291 
   8292      The syntax for the cipher sorting has been extended to support sorting by
   8293      cipher-strength (using the strength_bits hard coded in the tables).
   8294      The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
   8295 
   8296      Fix a bug in the cipher-command parser: when supplying a cipher command
   8297      string with an "undefined" symbol (neither command nor alphanumeric
   8298      [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
   8299      an error is flagged.
   8300 
   8301      Due to the strength-sorting extension, the code of the
   8302      ssl_create_cipher_list() function was completely rearranged. I hope that
   8303      the readability was also increased :-)
   8304      [Lutz Jaenicke <Lutz.Jaenicke (a] aet.TU-Cottbus.DE>]
   8305 
   8306   *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
   8307      for the first serial number and places 2 in the serial number file. This
   8308      avoids problems when the root CA is created with serial number zero and
   8309      the first user certificate has the same issuer name and serial number
   8310      as the root CA.
   8311      [Steve Henson]
   8312 
   8313   *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
   8314      the new code. Add documentation for this stuff.
   8315      [Steve Henson]
   8316 
   8317   *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
   8318      X509_*() to X509at_*() on the grounds that they don't handle X509
   8319      structures and behave in an analagous way to the X509v3 functions:
   8320      they shouldn't be called directly but wrapper functions should be used
   8321      instead.
   8322 
   8323      So we also now have some wrapper functions that call the X509at functions
   8324      when passed certificate requests. (TO DO: similar things can be done with
   8325      PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
   8326      things. Some of these need some d2i or i2d and print functionality
   8327      because they handle more complex structures.)
   8328      [Steve Henson]
   8329 
   8330   *) Add missing #ifndefs that caused missing symbols when building libssl
   8331      as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
   8332      NO_RSA in ssl/s2*.c. 
   8333      [Kris Kennaway <kris (a] hub.freebsd.org>, modified by Ulf Mller]
   8334 
   8335   *) Precautions against using the PRNG uninitialized: RAND_bytes() now
   8336      has a return value which indicates the quality of the random data
   8337      (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
   8338      error queue. New function RAND_pseudo_bytes() generates output that is
   8339      guaranteed to be unique but not unpredictable. RAND_add is like
   8340      RAND_seed, but takes an extra argument for an entropy estimate
   8341      (RAND_seed always assumes full entropy).
   8342      [Ulf Mller]
   8343 
   8344   *) Do more iterations of Rabin-Miller probable prime test (specifically,
   8345      3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
   8346      instead of only 2 for all lengths; see BN_prime_checks_for_size definition
   8347      in crypto/bn/bn_prime.c for the complete table).  This guarantees a
   8348      false-positive rate of at most 2^-80 for random input.
   8349      [Bodo Moeller]
   8350 
   8351   *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
   8352      [Bodo Moeller]
   8353 
   8354   *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
   8355      in the 0.9.5 release), this returns the chain
   8356      from an X509_CTX structure with a dup of the stack and all
   8357      the X509 reference counts upped: so the stack will exist
   8358      after X509_CTX_cleanup() has been called. Modify pkcs12.c
   8359      to use this.
   8360 
   8361      Also make SSL_SESSION_print() print out the verify return
   8362      code.
   8363      [Steve Henson]
   8364 
   8365   *) Add manpage for the pkcs12 command. Also change the default
   8366      behaviour so MAC iteration counts are used unless the new
   8367      -nomaciter option is used. This improves file security and
   8368      only older versions of MSIE (4.0 for example) need it.
   8369      [Steve Henson]
   8370 
   8371   *) Honor the no-xxx Configure options when creating .DEF files.
   8372      [Ulf Mller]
   8373 
   8374   *) Add PKCS#10 attributes to field table: challengePassword, 
   8375      unstructuredName and unstructuredAddress. These are taken from
   8376      draft PKCS#9 v2.0 but are compatible with v1.2 provided no 
   8377      international characters are used.
   8378 
   8379      More changes to X509_ATTRIBUTE code: allow the setting of types
   8380      based on strings. Remove the 'loc' parameter when adding
   8381      attributes because these will be a SET OF encoding which is sorted
   8382      in ASN1 order.
   8383      [Steve Henson]
   8384 
   8385   *) Initial changes to the 'req' utility to allow request generation
   8386      automation. This will allow an application to just generate a template
   8387      file containing all the field values and have req construct the
   8388      request.
   8389 
   8390      Initial support for X509_ATTRIBUTE handling. Stacks of these are
   8391      used all over the place including certificate requests and PKCS#7
   8392      structures. They are currently handled manually where necessary with
   8393      some primitive wrappers for PKCS#7. The new functions behave in a
   8394      manner analogous to the X509 extension functions: they allow
   8395      attributes to be looked up by NID and added.
   8396 
   8397      Later something similar to the X509V3 code would be desirable to
   8398      automatically handle the encoding, decoding and printing of the
   8399      more complex types. The string types like challengePassword can
   8400      be handled by the string table functions.
   8401 
   8402      Also modified the multi byte string table handling. Now there is
   8403      a 'global mask' which masks out certain types. The table itself
   8404      can use the flag STABLE_NO_MASK to ignore the mask setting: this
   8405      is useful when for example there is only one permissible type
   8406      (as in countryName) and using the mask might result in no valid
   8407      types at all.
   8408      [Steve Henson]
   8409 
   8410   *) Clean up 'Finished' handling, and add functions SSL_get_finished and
   8411      SSL_get_peer_finished to allow applications to obtain the latest
   8412      Finished messages sent to the peer or expected from the peer,
   8413      respectively.  (SSL_get_peer_finished is usually the Finished message
   8414      actually received from the peer, otherwise the protocol will be aborted.)
   8415 
   8416      As the Finished message are message digests of the complete handshake
   8417      (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
   8418      be used for external authentication procedures when the authentication
   8419      provided by SSL/TLS is not desired or is not enough.
   8420      [Bodo Moeller]
   8421 
   8422   *) Enhanced support for Alpha Linux is added. Now ./config checks if
   8423      the host supports BWX extension and if Compaq C is present on the
   8424      $PATH. Just exploiting of the BWX extension results in 20-30%
   8425      performance kick for some algorithms, e.g. DES and RC4 to mention
   8426      a couple. Compaq C in turn generates ~20% faster code for MD5 and
   8427      SHA1.
   8428      [Andy Polyakov]
   8429 
   8430   *) Add support for MS "fast SGC". This is arguably a violation of the
   8431      SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
   8432      weak crypto and after checking the certificate is SGC a second one
   8433      with strong crypto. MS SGC stops the first handshake after receiving
   8434      the server certificate message and sends a second client hello. Since
   8435      a server will typically do all the time consuming operations before
   8436      expecting any further messages from the client (server key exchange
   8437      is the most expensive) there is little difference between the two.
   8438 
   8439      To get OpenSSL to support MS SGC we have to permit a second client
   8440      hello message after we have sent server done. In addition we have to
   8441      reset the MAC if we do get this second client hello.
   8442      [Steve Henson]
   8443 
   8444   *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
   8445      if a DER encoded private key is RSA or DSA traditional format. Changed
   8446      d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
   8447      format DER encoded private key. Newer code should use PKCS#8 format which
   8448      has the key type encoded in the ASN1 structure. Added DER private key
   8449      support to pkcs8 application.
   8450      [Steve Henson]
   8451 
   8452   *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
   8453      ciphersuites has been selected (as required by the SSL 3/TLS 1
   8454      specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
   8455      is set, we interpret this as a request to violate the specification
   8456      (the worst that can happen is a handshake failure, and 'correct'
   8457      behaviour would result in a handshake failure anyway).
   8458      [Bodo Moeller]
   8459 
   8460   *) In SSL_CTX_add_session, take into account that there might be multiple
   8461      SSL_SESSION structures with the same session ID (e.g. when two threads
   8462      concurrently obtain them from an external cache).
   8463      The internal cache can handle only one SSL_SESSION with a given ID,
   8464      so if there's a conflict, we now throw out the old one to achieve
   8465      consistency.
   8466      [Bodo Moeller]
   8467 
   8468   *) Add OIDs for idea and blowfish in CBC mode. This will allow both
   8469      to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
   8470      some routines that use cipher OIDs: some ciphers do not have OIDs
   8471      defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
   8472      example.
   8473      [Steve Henson]
   8474 
   8475   *) Simplify the trust setting structure and code. Now we just have
   8476      two sequences of OIDs for trusted and rejected settings. These will
   8477      typically have values the same as the extended key usage extension
   8478      and any application specific purposes.
   8479 
   8480      The trust checking code now has a default behaviour: it will just
   8481      check for an object with the same NID as the passed id. Functions can
   8482      be provided to override either the default behaviour or the behaviour
   8483      for a given id. SSL client, server and email already have functions
   8484      in place for compatibility: they check the NID and also return "trusted"
   8485      if the certificate is self signed.
   8486      [Steve Henson]
   8487 
   8488   *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
   8489      traditional format into an EVP_PKEY structure.
   8490      [Steve Henson]
   8491 
   8492   *) Add a password callback function PEM_cb() which either prompts for
   8493      a password if usr_data is NULL or otherwise assumes it is a null
   8494      terminated password. Allow passwords to be passed on command line
   8495      environment or config files in a few more utilities.
   8496      [Steve Henson]
   8497 
   8498   *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
   8499      keys. Add some short names for PKCS#8 PBE algorithms and allow them
   8500      to be specified on the command line for the pkcs8 and pkcs12 utilities.
   8501      Update documentation.
   8502      [Steve Henson]
   8503 
   8504   *) Support for ASN1 "NULL" type. This could be handled before by using
   8505      ASN1_TYPE but there wasn't any function that would try to read a NULL
   8506      and produce an error if it couldn't. For compatibility we also have
   8507      ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
   8508      don't allocate anything because they don't need to.
   8509      [Steve Henson]
   8510 
   8511   *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
   8512      for details.
   8513      [Andy Polyakov, Roy Woods <roy (a] centicsystems.ca>]
   8514 
   8515   *) Rebuild of the memory allocation routines used by OpenSSL code and
   8516      possibly others as well.  The purpose is to make an interface that
   8517      provide hooks so anyone can build a separate set of allocation and
   8518      deallocation routines to be used by OpenSSL, for example memory
   8519      pool implementations, or something else, which was previously hard
   8520      since Malloc(), Realloc() and Free() were defined as macros having
   8521      the values malloc, realloc and free, respectively (except for Win32
   8522      compilations).  The same is provided for memory debugging code.
   8523      OpenSSL already comes with functionality to find memory leaks, but
   8524      this gives people a chance to debug other memory problems.
   8525 
   8526      With these changes, a new set of functions and macros have appeared:
   8527 
   8528        CRYPTO_set_mem_debug_functions()	        [F]
   8529        CRYPTO_get_mem_debug_functions()         [F]
   8530        CRYPTO_dbg_set_options()	                [F]
   8531        CRYPTO_dbg_get_options()                 [F]
   8532        CRYPTO_malloc_debug_init()               [M]
   8533 
   8534      The memory debug functions are NULL by default, unless the library
   8535      is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
   8536      wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
   8537      gives the standard debugging functions that come with OpenSSL) or
   8538      CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
   8539      provided by the library user) must be used.  When the standard
   8540      debugging functions are used, CRYPTO_dbg_set_options can be used to
   8541      request additional information:
   8542      CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
   8543      the CRYPTO_MDEBUG_xxx macro when compiling the library.   
   8544 
   8545      Also, things like CRYPTO_set_mem_functions will always give the
   8546      expected result (the new set of functions is used for allocation
   8547      and deallocation) at all times, regardless of platform and compiler
   8548      options.
   8549 
   8550      To finish it up, some functions that were never use in any other
   8551      way than through macros have a new API and new semantic:
   8552 
   8553        CRYPTO_dbg_malloc()
   8554        CRYPTO_dbg_realloc()
   8555        CRYPTO_dbg_free()
   8556 
   8557      All macros of value have retained their old syntax.
   8558      [Richard Levitte and Bodo Moeller]
   8559 
   8560   *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
   8561      ordering of SMIMECapabilities wasn't in "strength order" and there
   8562      was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
   8563      algorithm.
   8564      [Steve Henson]
   8565 
   8566   *) Some ASN1 types with illegal zero length encoding (INTEGER,
   8567      ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
   8568      [Frans Heymans <fheymans (a] isaserver.be>, modified by Steve Henson]
   8569 
   8570   *) Merge in my S/MIME library for OpenSSL. This provides a simple
   8571      S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
   8572      functionality to handle multipart/signed properly) and a utility
   8573      called 'smime' to call all this stuff. This is based on code I
   8574      originally wrote for Celo who have kindly allowed it to be
   8575      included in OpenSSL.
   8576      [Steve Henson]
   8577 
   8578   *) Add variants des_set_key_checked and des_set_key_unchecked of
   8579      des_set_key (aka des_key_sched).  Global variable des_check_key
   8580      decides which of these is called by des_set_key; this way
   8581      des_check_key behaves as it always did, but applications and
   8582      the library itself, which was buggy for des_check_key == 1,
   8583      have a cleaner way to pick the version they need.
   8584      [Bodo Moeller]
   8585 
   8586   *) New function PKCS12_newpass() which changes the password of a
   8587      PKCS12 structure.
   8588      [Steve Henson]
   8589 
   8590   *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
   8591      dynamic mix. In both cases the ids can be used as an index into the
   8592      table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
   8593      functions so they accept a list of the field values and the
   8594      application doesn't need to directly manipulate the X509_TRUST
   8595      structure.
   8596      [Steve Henson]
   8597 
   8598   *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
   8599      need initialising.
   8600      [Steve Henson]
   8601 
   8602   *) Modify the way the V3 extension code looks up extensions. This now
   8603      works in a similar way to the object code: we have some "standard"
   8604      extensions in a static table which is searched with OBJ_bsearch()
   8605      and the application can add dynamic ones if needed. The file
   8606      crypto/x509v3/ext_dat.h now has the info: this file needs to be
   8607      updated whenever a new extension is added to the core code and kept
   8608      in ext_nid order. There is a simple program 'tabtest.c' which checks
   8609      this. New extensions are not added too often so this file can readily
   8610      be maintained manually.
   8611 
   8612      There are two big advantages in doing things this way. The extensions
   8613      can be looked up immediately and no longer need to be "added" using
   8614      X509V3_add_standard_extensions(): this function now does nothing.
   8615      [Side note: I get *lots* of email saying the extension code doesn't
   8616       work because people forget to call this function]
   8617      Also no dynamic allocation is done unless new extensions are added:
   8618      so if we don't add custom extensions there is no need to call
   8619      X509V3_EXT_cleanup().
   8620      [Steve Henson]
   8621 
   8622   *) Modify enc utility's salting as follows: make salting the default. Add a
   8623      magic header, so unsalted files fail gracefully instead of just decrypting
   8624      to garbage. This is because not salting is a big security hole, so people
   8625      should be discouraged from doing it.
   8626      [Ben Laurie]
   8627 
   8628   *) Fixes and enhancements to the 'x509' utility. It allowed a message
   8629      digest to be passed on the command line but it only used this
   8630      parameter when signing a certificate. Modified so all relevant
   8631      operations are affected by the digest parameter including the
   8632      -fingerprint and -x509toreq options. Also -x509toreq choked if a
   8633      DSA key was used because it didn't fix the digest.
   8634      [Steve Henson]
   8635 
   8636   *) Initial certificate chain verify code. Currently tests the untrusted
   8637      certificates for consistency with the verify purpose (which is set
   8638      when the X509_STORE_CTX structure is set up) and checks the pathlength.
   8639 
   8640      There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
   8641      this is because it will reject chains with invalid extensions whereas
   8642      every previous version of OpenSSL and SSLeay made no checks at all.
   8643 
   8644      Trust code: checks the root CA for the relevant trust settings. Trust
   8645      settings have an initial value consistent with the verify purpose: e.g.
   8646      if the verify purpose is for SSL client use it expects the CA to be
   8647      trusted for SSL client use. However the default value can be changed to
   8648      permit custom trust settings: one example of this would be to only trust
   8649      certificates from a specific "secure" set of CAs.
   8650 
   8651      Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
   8652      which should be used for version portability: especially since the
   8653      verify structure is likely to change more often now.
   8654 
   8655      SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
   8656      to set them. If not set then assume SSL clients will verify SSL servers
   8657      and vice versa.
   8658 
   8659      Two new options to the verify program: -untrusted allows a set of
   8660      untrusted certificates to be passed in and -purpose which sets the
   8661      intended purpose of the certificate. If a purpose is set then the
   8662      new chain verify code is used to check extension consistency.
   8663      [Steve Henson]
   8664 
   8665   *) Support for the authority information access extension.
   8666      [Steve Henson]
   8667 
   8668   *) Modify RSA and DSA PEM read routines to transparently handle
   8669      PKCS#8 format private keys. New *_PUBKEY_* functions that handle
   8670      public keys in a format compatible with certificate
   8671      SubjectPublicKeyInfo structures. Unfortunately there were already
   8672      functions called *_PublicKey_* which used various odd formats so
   8673      these are retained for compatibility: however the DSA variants were
   8674      never in a public release so they have been deleted. Changed dsa/rsa
   8675      utilities to handle the new format: note no releases ever handled public
   8676      keys so we should be OK.
   8677 
   8678      The primary motivation for this change is to avoid the same fiasco
   8679      that dogs private keys: there are several incompatible private key
   8680      formats some of which are standard and some OpenSSL specific and
   8681      require various evil hacks to allow partial transparent handling and
   8682      even then it doesn't work with DER formats. Given the option anything
   8683      other than PKCS#8 should be dumped: but the other formats have to
   8684      stay in the name of compatibility.
   8685 
   8686      With public keys and the benefit of hindsight one standard format 
   8687      is used which works with EVP_PKEY, RSA or DSA structures: though
   8688      it clearly returns an error if you try to read the wrong kind of key.
   8689 
   8690      Added a -pubkey option to the 'x509' utility to output the public key.
   8691      Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
   8692      (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
   8693      EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
   8694      that do the same as the EVP_PKEY_assign_*() except they up the
   8695      reference count of the added key (they don't "swallow" the
   8696      supplied key).
   8697      [Steve Henson]
   8698 
   8699   *) Fixes to crypto/x509/by_file.c the code to read in certificates and
   8700      CRLs would fail if the file contained no certificates or no CRLs:
   8701      added a new function to read in both types and return the number
   8702      read: this means that if none are read it will be an error. The
   8703      DER versions of the certificate and CRL reader would always fail
   8704      because it isn't possible to mix certificates and CRLs in DER format
   8705      without choking one or the other routine. Changed this to just read
   8706      a certificate: this is the best we can do. Also modified the code
   8707      in apps/verify.c to take notice of return codes: it was previously
   8708      attempting to read in certificates from NULL pointers and ignoring
   8709      any errors: this is one reason why the cert and CRL reader seemed
   8710      to work. It doesn't check return codes from the default certificate
   8711      routines: these may well fail if the certificates aren't installed.
   8712      [Steve Henson]
   8713 
   8714   *) Code to support otherName option in GeneralName.
   8715      [Steve Henson]
   8716 
   8717   *) First update to verify code. Change the verify utility
   8718      so it warns if it is passed a self signed certificate:
   8719      for consistency with the normal behaviour. X509_verify
   8720      has been modified to it will now verify a self signed
   8721      certificate if *exactly* the same certificate appears
   8722      in the store: it was previously impossible to trust a
   8723      single self signed certificate. This means that:
   8724      openssl verify ss.pem
   8725      now gives a warning about a self signed certificate but
   8726      openssl verify -CAfile ss.pem ss.pem
   8727      is OK.
   8728      [Steve Henson]
   8729 
   8730   *) For servers, store verify_result in SSL_SESSION data structure
   8731      (and add it to external session representation).
   8732      This is needed when client certificate verifications fails,
   8733      but an application-provided verification callback (set by
   8734      SSL_CTX_set_cert_verify_callback) allows accepting the session
   8735      anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
   8736      but returns 1): When the session is reused, we have to set
   8737      ssl->verify_result to the appropriate error code to avoid
   8738      security holes.
   8739      [Bodo Moeller, problem pointed out by Lutz Jaenicke]
   8740 
   8741   *) Fix a bug in the new PKCS#7 code: it didn't consider the
   8742      case in PKCS7_dataInit() where the signed PKCS7 structure
   8743      didn't contain any existing data because it was being created.
   8744      [Po-Cheng Chen <pocheng (a] nst.com.tw>, slightly modified by Steve Henson]
   8745 
   8746   *) Add a salt to the key derivation routines in enc.c. This
   8747      forms the first 8 bytes of the encrypted file. Also add a
   8748      -S option to allow a salt to be input on the command line.
   8749      [Steve Henson]
   8750 
   8751   *) New function X509_cmp(). Oddly enough there wasn't a function
   8752      to compare two certificates. We do this by working out the SHA1
   8753      hash and comparing that. X509_cmp() will be needed by the trust
   8754      code.
   8755      [Steve Henson]
   8756 
   8757   *) SSL_get1_session() is like SSL_get_session(), but increments
   8758      the reference count in the SSL_SESSION returned.
   8759      [Geoff Thorpe <geoff (a] eu.c2.net>]
   8760 
   8761   *) Fix for 'req': it was adding a null to request attributes.
   8762      Also change the X509_LOOKUP and X509_INFO code to handle
   8763      certificate auxiliary information.
   8764      [Steve Henson]
   8765 
   8766   *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
   8767      the 'enc' command.
   8768      [Steve Henson]
   8769 
   8770   *) Add the possibility to add extra information to the memory leak
   8771      detecting output, to form tracebacks, showing from where each
   8772      allocation was originated: CRYPTO_push_info("constant string") adds
   8773      the string plus current file name and line number to a per-thread
   8774      stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
   8775      is like calling CYRPTO_pop_info() until the stack is empty.
   8776      Also updated memory leak detection code to be multi-thread-safe.
   8777      [Richard Levitte]
   8778 
   8779   *) Add options -text and -noout to pkcs7 utility and delete the
   8780      encryption options which never did anything. Update docs.
   8781      [Steve Henson]
   8782 
   8783   *) Add options to some of the utilities to allow the pass phrase
   8784      to be included on either the command line (not recommended on
   8785      OSes like Unix) or read from the environment. Update the
   8786      manpages and fix a few bugs.
   8787      [Steve Henson]
   8788 
   8789   *) Add a few manpages for some of the openssl commands.
   8790      [Steve Henson]
   8791 
   8792   *) Fix the -revoke option in ca. It was freeing up memory twice,
   8793      leaking and not finding already revoked certificates.
   8794      [Steve Henson]
   8795 
   8796   *) Extensive changes to support certificate auxiliary information.
   8797      This involves the use of X509_CERT_AUX structure and X509_AUX
   8798      functions. An X509_AUX function such as PEM_read_X509_AUX()
   8799      can still read in a certificate file in the usual way but it
   8800      will also read in any additional "auxiliary information". By
   8801      doing things this way a fair degree of compatibility can be
   8802      retained: existing certificates can have this information added
   8803      using the new 'x509' options. 
   8804 
   8805      Current auxiliary information includes an "alias" and some trust
   8806      settings. The trust settings will ultimately be used in enhanced
   8807      certificate chain verification routines: currently a certificate
   8808      can only be trusted if it is self signed and then it is trusted
   8809      for all purposes.
   8810      [Steve Henson]
   8811 
   8812   *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
   8813      The problem was that one of the replacement routines had not been working
   8814      since SSLeay releases.  For now the offending routine has been replaced
   8815      with non-optimised assembler.  Even so, this now gives around 95%
   8816      performance improvement for 1024 bit RSA signs.
   8817      [Mark Cox]
   8818 
   8819   *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 
   8820      handling. Most clients have the effective key size in bits equal to
   8821      the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
   8822      A few however don't do this and instead use the size of the decrypted key
   8823      to determine the RC2 key length and the AlgorithmIdentifier to determine
   8824      the effective key length. In this case the effective key length can still
   8825      be 40 bits but the key length can be 168 bits for example. This is fixed
   8826      by manually forcing an RC2 key into the EVP_PKEY structure because the
   8827      EVP code can't currently handle unusual RC2 key sizes: it always assumes
   8828      the key length and effective key length are equal.
   8829      [Steve Henson]
   8830 
   8831   *) Add a bunch of functions that should simplify the creation of 
   8832      X509_NAME structures. Now you should be able to do:
   8833      X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
   8834      and have it automatically work out the correct field type and fill in
   8835      the structures. The more adventurous can try:
   8836      X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
   8837      and it will (hopefully) work out the correct multibyte encoding.
   8838      [Steve Henson]
   8839 
   8840   *) Change the 'req' utility to use the new field handling and multibyte
   8841      copy routines. Before the DN field creation was handled in an ad hoc
   8842      way in req, ca, and x509 which was rather broken and didn't support
   8843      BMPStrings or UTF8Strings. Since some software doesn't implement
   8844      BMPStrings or UTF8Strings yet, they can be enabled using the config file
   8845      using the dirstring_type option. See the new comment in the default
   8846      openssl.cnf for more info.
   8847      [Steve Henson]
   8848 
   8849   *) Make crypto/rand/md_rand.c more robust:
   8850      - Assure unique random numbers after fork().
   8851      - Make sure that concurrent threads access the global counter and
   8852        md serializably so that we never lose entropy in them
   8853        or use exactly the same state in multiple threads.
   8854        Access to the large state is not always serializable because
   8855        the additional locking could be a performance killer, and
   8856        md should be large enough anyway.
   8857      [Bodo Moeller]
   8858 
   8859   *) New file apps/app_rand.c with commonly needed functionality
   8860      for handling the random seed file.
   8861 
   8862      Use the random seed file in some applications that previously did not:
   8863           ca,
   8864           dsaparam -genkey (which also ignored its '-rand' option), 
   8865           s_client,
   8866           s_server,
   8867           x509 (when signing).
   8868      Except on systems with /dev/urandom, it is crucial to have a random
   8869      seed file at least for key creation, DSA signing, and for DH exchanges;
   8870      for RSA signatures we could do without one.
   8871 
   8872      gendh and gendsa (unlike genrsa) used to read only the first byte
   8873      of each file listed in the '-rand' option.  The function as previously
   8874      found in genrsa is now in app_rand.c and is used by all progra