Make a note of the "up" functions.

[openssl.git] / CHANGES

 OpenSSL CHANGES
 _______________

 Changes between 0.9.6 and 0.9.7  [xx XXX 2001]

     OpenSSL 0.9.6a/0.9.6b (bugfix releases, 5 Apr 2001 and 9 July 2001)
     and OpenSSL 0.9.7 were developped in parallel, based on OpenSSL 0.9.6.  

     Change log entries are tagged as follows:
         -) applies to 0.9.6a/0.9.6b/0.9.6c only
         *) applies to 0.9.6a/0.9.6b/0.9.6c and 0.9.7
         +) applies to 0.9.7 only

  +) Give DH, DSA, and RSA types their own "**_up()" function to increment
     reference counts. This performs normal REF_PRINT/REF_CHECK macros on
     the operation, and provides a more encapsulated way for external code
     (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
     to use these functions rather than manually incrementing the counts.
     [Geoff Thorpe]

  *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
     This function was broken, as the check for a new client hello message
     to handle SGC did not allow these large messages.
     (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
     [Lutz Jaenicke]

  *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
     [Lutz Jaenicke]

  +) Add EVP test program.
     [Ben Laurie]

  +) Add symmetric cipher support to ENGINE. Expect the API to change!
     [Ben Laurie]

  +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
     X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
     X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
     These allow a CRL to be built without having to access X509_CRL fields
     directly. Modify 'ca' application to use new functions.
     [Steve Henson]

  *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
     for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
     [Lutz Jaenicke]

  *) Rework the configuration and shared library support for Tru64 Unix.
     The configuration part makes use of modern compiler features and
     still retains old compiler behavior for those that run older versions
     of the OS.  The shared library support part includes a variant that
     uses the RPATH feature, and is available through the speciel
     configuration target "alpha-cc-rpath", which will never be selected
     automatically.
     [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]

  *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
     with the same message size as in ssl3_get_certificate_request().
     Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
     messages might inadvertently be reject as too long.
     [Petr Lampa <lampa@fee.vutbr.cz>]

  +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
     bug workarounds. Rollback attack detection is a security feature.
     The problem will only arise on OpenSSL servers when TLSv1 is not
     available (sslv3_server_method() or SSL_OP_NO_TLSv1).
     Software authors not wanting to support TLSv1 will have special reasons
     for their choice and can explicitly enable this option.
     [Bodo Moeller, Lutz Jaenicke]

  +) Rationalise EVP so it can be extended: don't include a union of
     cipher/digest structures, add init/cleanup functions. This also reduces
     the number of header dependencies.
     [Ben Laurie]

  +) Make DES key schedule conform to the usual scheme, as well as
     correcting its structure. This means that calls to DES functions
     now have to pass a pointer to a des_key_schedule instead of a
     plain des_key_schedule (which was actually always a pointer
     anyway).
     [Ben Laurie]

  +) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
     [Andy Polyakov]

  *) Modified SSL library such that the verify_callback that has been set
     specificly for an SSL object with SSL_set_verify() is actually being
     used. Before the change, a verify_callback set with this function was
     ignored and the verify_callback() set in the SSL_CTX at the time of
     the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
     to allow the necessary settings.
     [Lutz Jaenicke]

  +) Initial reduction of linker bloat: the use of some functions, such as
     PEM causes large amounts of unused functions to be linked in due to
     poor organisation. For example pem_all.c contains every PEM function
     which has a knock on effect of linking in large amounts of (unused)
     ASN1 code. Grouping together similar functions and splitting unrelated
     functions prevents this.
     [Steve Henson]

  *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
     explicitely to NULL, as at least on Solaris 8 this seems not always to be
     done automatically (in contradiction to the requirements of the C
     standard). This made problems when used from OpenSSH.
     [Lutz Jaenicke]

  *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
     dh->length and always used

          BN_rand_range(priv_key, dh->p).

     BN_rand_range() is not necessary for Diffie-Hellman, and this
     specific range makes Diffie-Hellman unnecessarily inefficient if
     dh->length (recommended exponent length) is much smaller than the
     length of dh->p.  We could use BN_rand_range() if the order of
     the subgroup was stored in the DH structure, but we only have
     dh->length.

     So switch back to

          BN_rand(priv_key, l, ...)

     where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
     otherwise.
     [Bodo Moeller]

  *) In

          RSA_eay_public_encrypt
          RSA_eay_private_decrypt
          RSA_eay_private_encrypt (signing)
          RSA_eay_public_decrypt (signature verification)

     (default implementations for RSA_public_encrypt,
     RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
     always reject numbers >= n.
     [Bodo Moeller]

  *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
     to synchronize access to 'locking_thread'.  This is necessary on
     systems where access to 'locking_thread' (an 'unsigned long'
     variable) is not atomic.
     [Bodo Moeller]

  *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
     *before* setting the 'crypto_lock_rand' flag.  The previous code had
     a race condition if 0 is a valid thread ID.
     [Travis Vitek <vitek@roguewave.com>]

  +) Cleanup of EVP macros.
     [Ben Laurie]

  +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
     correct _ecb suffix.
     [Ben Laurie]

  +) Add initial OCSP responder support to ocsp application. The
     revocation information is handled using the text based index
     use by the ca application. The responder can either handle
     requests generated internally, supplied in files (for example
     via a CGI script) or using an internal minimal server.
     [Steve Henson]

  +) Add configuration choices to get zlib compression for TLS.
     [Richard Levitte]

  +) Changes to Kerberos SSL for RFC 2712 compliance:
     1.  Implemented real KerberosWrapper, instead of just using
         KRB5 AP_REQ message.  [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
     2.  Implemented optional authenticator field of KerberosWrapper.

     Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
     and authenticator structs; see crypto/krb5/.

     Generalized Kerberos calls to support multiple Kerberos libraries.
     [Vern Staats <staatsvr@asc.hpc.mil>,
      Jeffrey Altman <jaltman@columbia.edu>
      via Richard Levitte]

  +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
     already does with RSA. testdsa.h now has 'priv_key/pub_key'
     values for each of the key sizes rather than having just
     parameters (and 'speed' generating keys each time).
     [Geoff Thorpe]

  -) OpenSSL 0.9.6b released [9 July 2001]

  *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
     to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
     Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
     PRNG state recovery was possible based on the output of
     one PRNG request appropriately sized to gain knowledge on
     'md' followed by enough consecutive 1-byte PRNG requests
     to traverse all of 'state'.

     1. When updating 'md_local' (the current thread's copy of 'md')
        during PRNG output generation, hash all of the previous
        'md_local' value, not just the half used for PRNG output.

     2. Make the number of bytes from 'state' included into the hash
        independent from the number of PRNG bytes requested.

     The first measure alone would be sufficient to avoid
     Markku-Juhani's attack.  (Actually it had never occurred
     to me that the half of 'md_local' used for chaining was the
     half from which PRNG output bytes were taken -- I had always
     assumed that the secret half would be used.)  The second
     measure makes sure that additional data from 'state' is never
     mixed into 'md_local' in small portions; this heuristically
     further strengthens the PRNG.
     [Bodo Moeller]
  
  +) Speed up EVP routines.
     Before:
encrypt
type              8 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
des-cbc           4408.85k     5560.51k     5778.46k     5862.20k     5825.16k
des-cbc           4389.55k     5571.17k     5792.23k     5846.91k     5832.11k
des-cbc           4394.32k     5575.92k     5807.44k     5848.37k     5841.30k
decrypt
des-cbc           3482.66k     5069.49k     5496.39k     5614.16k     5639.28k
des-cbc           3480.74k     5068.76k     5510.34k     5609.87k     5635.52k
des-cbc           3483.72k     5067.62k     5504.60k     5708.01k     5724.80k
     After:
encrypt
des-cbc           4660.16k     5650.19k     5807.19k     5827.13k     5783.32k
decrypt
des-cbc           3624.96k     5258.21k     5530.91k     5624.30k     5628.26k
     [Ben Laurie]

  *) Fix crypto/bn/asm/mips3.s.
     [Andy Polyakov]

  *) When only the key is given to "enc", the IV is undefined. Print out
     an error message in this case.
     [Lutz Jaenicke]

  +) Added the OS2-EMX target.
     ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]

  +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
     to support NCONF routines in extension code. New function CONF_set_nconf()
     to allow functions which take an NCONF to also handle the old LHASH
     structure: this means that the old CONF compatible routines can be
     retained (in particular wrt extensions) without having to duplicate the
     code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
     [Steve Henson]

  *) Handle special case when X509_NAME is empty in X509 printing routines.
     [Steve Henson]

  *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
     positive and less than q.
     [Bodo Moeller]

  +) Enhance the general user interface with mechanisms for inner control
     and with pssibilities to have yes/no kind of prompts.
     [Richard Levitte]

  +) Change all calls to low level digest routines in the library and
     applications to use EVP. Add missing calls to HMAC_cleanup() and
     don't assume HMAC_CTX can be copied using memcpy().
     [Verdon Walker <VWalker@novell.com>, Steve Henson]

  +) Add the possibility to control engines through control names but with
     arbitrary arguments instead of just a string.
     Change the key loaders to take a UI_METHOD instead of a callback
     function pointer.  NOTE: this breaks binary compatibility with earlier
     versions of OpenSSL [engine].
     Addapt the nCipher code for these new conditions and add a card insertion
     callback.
     [Richard Levitte]

  +) Enhance the general user interface with mechanisms to better support
     dialog box interfaces, application-defined prompts, the possibility
     to use defaults (for example default passwords from somewhere else)
     and interrupts/cancelations.
     [Richard Levitte]

  *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
     used: it isn't thread safe and the add_lock_callback should handle
     that itself.
     [Paul Rose <Paul.Rose@bridge.com>]

  *) Verify that incoming data obeys the block size in
     ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
     [Bodo Moeller]

  +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
     attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
     [Steve Henson]

  *) Fix OAEP check.
     [Ulf Möller, Bodo Möller]

  *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
     RSA encryption was accidentily removed in s3_srvr.c in OpenSSL 0.9.5
     when fixing the server behaviour for backwards-compatible 'client
     hello' messages.  (Note that the attack is impractical against
     SSL 3.0 and TLS 1.0 anyway because length and version checking
     means that the probability of guessing a valid ciphertext is
     around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
     paper.)

     Before 0.9.5, the countermeasure (hide the error by generating a
     random 'decryption result') did not work properly because
     ERR_clear_error() was missing, meaning that SSL_get_error() would
     detect the supposedly ignored error.

     Both problems are now fixed.
     [Bodo Moeller]

  *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
     (previously it was 1024).
     [Bodo Moeller]

  +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
     tidy up some unecessarily weird code in 'sk_new()').
     [Geoff, reported by Diego Tartara <dtartara@novamens.com>]

  +) Change the key loading routines for ENGINEs to use the same kind
     callback (pem_password_cb) as all other routines that need this
     kind of callback.
     [Richard Levitte]

  *) Fix for compatibility mode trust settings: ignore trust settings
     unless some valid trust or reject settings are present.
     [Steve Henson]

  *) Fix for blowfish EVP: its a variable length cipher.
     [Steve Henson]

  +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
     256 bit (=32 byte) keys. Of course seeding with more entropy bytes
     than this minimum value is recommended.
     [Lutz Jaenicke]

  +) New random seeder for OpenVMS, using the system process statistics
     that are easily reachable.
     [Richard Levitte]

  +) Windows apparently can't transparently handle global
     variables defined in DLLs. Initialisations such as:

        const ASN1_ITEM *it = &ASN1_INTEGER_it;

     wont compile. This is used by the any applications that need to
     delcare their own ASN1 modules. This was fixed by adding the option
     EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
     needed for static libraries under Win32.
     [Steve Henson]

  +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
     setting of purpose and trust fields. New X509_STORE trust and
     purpose functions and tidy up setting in other SSL functions.
     [Steve Henson]

  +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
     structure. These are inherited by X509_STORE_CTX when it is 
     initialised. This allows various defaults to be set in the
     X509_STORE structure (such as flags for CRL checking and custom
     purpose or trust settings) for functions which only use X509_STORE_CTX
     internally such as S/MIME.

     Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
     trust settings if they are not set in X509_STORE. This allows X509_STORE
     purposes and trust (in S/MIME for example) to override any set by default.

     Add command line options for CRL checking to smime, s_client and s_server
     applications.
     [Steve Henson]

  +) Initial CRL based revocation checking. If the CRL checking flag(s)
     are set then the CRL is looked up in the X509_STORE structure and
     its validity and signature checked, then if the certificate is found
     in the CRL the verify fails with a revoked error.

     Various new CRL related callbacks added to X509_STORE_CTX structure.

     Command line options added to 'verify' application to support this.

     This needs some additional work, such as being able to handle multiple
     CRLs with different times, extension based lookup (rather than just
     by subject name) and ultimately more complete V2 CRL extension
     handling.
     [Steve Henson]

  +) Add a general user interface API (crypto/ui/).  This is designed
     to replace things like des_read_password and friends (backward
     compatibility functions using this new API are provided).
     The purpose is to remove prompting functions from the DES code
     section as well as provide for prompting through dialog boxes in
     a window system and the like.
     [Richard Levitte]

  *) Fix various bugs related to DSA S/MIME verification. Handle missing
     parameters in DSA public key structures and return an error in the
     DSA routines if parameters are absent.
     [Steve Henson]

  *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
     in the current directory if neither $RANDFILE nor $HOME was set.
     RAND_file_name() in 0.9.6a returned NULL in this case.  This has
     caused some confusion to Windows users who haven't defined $HOME.
     Thus RAND_file_name() is changed again: e_os.h can define a
     DEFAULT_HOME, which will be used if $HOME is not set.
     For Windows, we use "C:"; on other platforms, we still require
     environment variables.

  +) Add "ex_data" support to ENGINE so implementations can add state at a
     per-structure level rather than having to store it globally.
     [Geoff]

  +) Make it possible for ENGINE structures to be copied when retrieved by
     ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
     This causes the "original" ENGINE structure to act like a template,
     analogous to the RSA vs. RSA_METHOD type of separation. Because of this
     operational state can be localised to each ENGINE structure, despite the
     fact they all share the same "methods". New ENGINE structures returned in
     this case have no functional references and the return value is the single
     structural reference. This matches the single structural reference returned
     by ENGINE_by_id() normally, when it is incremented on the pre-existing
     ENGINE structure.
     [Geoff]

  +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
     needs to match any other type at all we need to manually clear the
     tag cache.
     [Steve Henson]

  +) Changes to the "openssl engine" utility to include;
     - verbosity levels ('-v', '-vv', and '-vvv') that provide information
       about an ENGINE's available control commands.
     - executing control commands from command line arguments using the
       '-pre' and '-post' switches. '-post' is only used if '-t' is
       specified and the ENGINE is successfully initialised. The syntax for
       the individual commands are colon-separated, for example;
         openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
     [Geoff]

  +) New dynamic control command support for ENGINEs. ENGINEs can now
     declare their own commands (numbers), names (strings), descriptions,
     and input types for run-time discovery by calling applications. A
     subset of these commands are implicitly classed as "executable"
     depending on their input type, and only these can be invoked through
     the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
     can be based on user input, config files, etc). The distinction is
     that "executable" commands cannot return anything other than a boolean
     result and can only support numeric or string input, whereas some
     discoverable commands may only be for direct use through
     ENGINE_ctrl(), eg. supporting the exchange of binary data, function
     pointers, or other custom uses. The "executable" commands are to
     support parameterisations of ENGINE behaviour that can be
     unambiguously defined by ENGINEs and used consistently across any
     OpenSSL-based application. Commands have been added to all the
     existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
     control over shared-library paths without source code alterations.
     [Geoff]

  +) Changed all ENGINE implementations to dynamically allocate their
     ENGINEs rather than declaring them statically. Apart from this being
     necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
     this also allows the implementations to compile without using the
     internal engine_int.h header.
     [Geoff]

  +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
     'const' value. Any code that should be able to modify a RAND_METHOD
     should already have non-const pointers to it (ie. they should only
     modify their own ones).
     [Geoff]

  +) Made a variety of little tweaks to the ENGINE code.
     - "atalla" and "ubsec" string definitions were moved from header files
       to C code. "nuron" string definitions were placed in variables
       rather than hard-coded - allowing parameterisation of these values
       later on via ctrl() commands.
     - Removed unused "#if 0"'d code.
     - Fixed engine list iteration code so it uses ENGINE_free() to release
       structural references.
     - Constified the RAND_METHOD element of ENGINE structures.
     - Constified various get/set functions as appropriate and added
       missing functions (including a catch-all ENGINE_cpy that duplicates
       all ENGINE values onto a new ENGINE except reference counts/state).
     - Removed NULL parameter checks in get/set functions. Setting a method
       or function to NULL is a way of cancelling out a previously set
       value.  Passing a NULL ENGINE parameter is just plain stupid anyway
       and doesn't justify the extra error symbols and code.
     - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
       flags from engine_int.h to engine.h.
     - Changed prototypes for ENGINE handler functions (init(), finish(),
       ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
     [Geoff]

  *) Move 'if (!initialized) RAND_poll()' into regions protected by
     CRYPTO_LOCK_RAND.  This is not strictly necessary, but avoids
     having multiple threads call RAND_poll() concurrently.
     [Bodo Moeller]

  *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
     combination of a flag and a thread ID variable.
     Otherwise while one thread is in ssleay_rand_bytes (which sets the
     flag), *other* threads can enter ssleay_add_bytes without obeying
     the CRYPTO_LOCK_RAND lock (and may even illegaly release the lock
     that they do not hold after the first thread unsets add_do_not_lock).
     [Bodo Moeller]

  +) Implement binary inversion algorithm for BN_mod_inverse in addition
     to the algorithm using long divison.  The binary algorithm can be
     used only if the modulus is odd.  On 32-bit systems, it is faster
     only for relatively small moduli (roughly 20-30% for 128-bit moduli,
     roughly 5-15% for 256-bit moduli), so we use it only for moduli
     up to 450 bits.  In 64-bit environments, the binary algorithm
     appears to be advantageous for much longer moduli; here we use it
     for moduli up to 2048 bits.
     [Bodo Moeller]

  *) Change bctest again: '-x' expressions are not available in all
     versions of 'test'.
     [Bodo Moeller]

  -) OpenSSL 0.9.6a released [5 Apr 2001]

  *) Fix a couple of memory leaks in PKCS7_dataDecode()
     [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]

  *) Change Configure and Makefiles to provide EXE_EXT, which will contain
     the default extension for executables, if any.  Also, make the perl
     scripts that use symlink() to test if it really exists and use "cp"
     if it doesn't.  All this made OpenSSL compilable and installable in
     CygWin.
     [Richard Levitte]

  +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
     could not support the combine flag in choice fields.
     [Steve Henson]

  -) Fix for asn1_GetSequence() for indefinite length constructed data.
     If SEQUENCE is length is indefinite just set c->slen to the total
     amount of data available.
     [Steve Henson, reported by shige@FreeBSD.org]
     [This change does not apply to 0.9.7.]

  *) Change bctest to avoid here-documents inside command substitution
     (workaround for FreeBSD /bin/sh bug).
     For compatibility with Ultrix, avoid shell functions (introduced
     in the bctest version that searches along $PATH).
     [Bodo Moeller]

  *) Rename 'des_encrypt' to 'des_encrypt1'.  This avoids the clashes
     with des_encrypt() defined on some operating systems, like Solaris
     and UnixWare.
     [Richard Levitte]

  *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
     On the Importance of Eliminating Errors in Cryptographic
     Computations, J. Cryptology 14 (2001) 2, 101-119,
     http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
     [Ulf Moeller]
  
  *) MIPS assembler BIGNUM division bug fix. 
     [Andy Polyakov]

  *) Disabled incorrect Alpha assembler code.
     [Richard Levitte]

  -) Fix PKCS#7 decode routines so they correctly update the length
     after reading an EOC for the EXPLICIT tag.
     [Steve Henson]
     [This change does not apply to 0.9.7.]

  *) Fix bug in PKCS#12 key generation routines. This was triggered
     if a 3DES key was generated with a 0 initial byte. Include
     PKCS12_BROKEN_KEYGEN compilation option to retain the old
     (but broken) behaviour.
     [Steve Henson]

  *) Enhance bctest to search for a working bc along $PATH and print
     it when found.
     [Tim Rice <tim@multitalents.net> via Richard Levitte]

  +) Add a 'copy_extensions' option to the 'ca' utility. This copies
     extensions from a certificate request to the certificate.
     [Steve Henson]

  +) Allow multiple 'certopt' and 'nameopt' options to be separated
     by commas. Add 'namopt' and 'certopt' options to the 'ca' config
     file: this allows the display of the certificate about to be
     signed to be customised, to allow certain fields to be included
     or excluded and extension details. The old system didn't display
     multicharacter strings properly, omitted fields not in the policy
     and couldn't display additional details such as extensions.
     [Steve Henson]

  *) Fix memory leaks in err.c: free err_data string if necessary;
     don't write to the wrong index in ERR_set_error_data.
     [Bodo Moeller]

  +) Function EC_POINTs_mul for simultaneous scalar multiplication
     of an arbitrary number of elliptic curve points, optionally
     including the generator defined for the EC_GROUP.
     EC_POINT_mul is a simple wrapper function for the typical case
     that the point list has just one item (besides the optional
     generator).
     [Bodo Moeller]

  +) First EC_METHODs for curves over GF(p):

     EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
     operations and provides various method functions that can also
     operate with faster implementations of modular arithmetic.     

     EC_GFp_mont_method() reuses most functions that are part of
     EC_GFp_simple_method, but uses Montgomery arithmetic.

     [Bodo Moeller; point addition and point doubling
     implementation directly derived from source code provided by
     Lenka Fibikova <fibikova@exp-math.uni-essen.de>]

  +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
     crypto/ec/ec_lib.c):

     Curves are EC_GROUP objects (with an optional group generator)
     based on EC_METHODs that are built into the library.

     Points are EC_POINT objects based on EC_GROUP objects.

     Most of the framework would be able to handle curves over arbitrary
     finite fields, but as there are no obvious types for fields other
     than GF(p), some functions are limited to that for now.
     [Bodo Moeller]

  +) Add the -HTTP option to s_server.  It is similar to -WWW, but requires
     that the file contains a complete HTTP response.
     [Richard Levitte]

  +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
     change the def and num file printf format specifier from "%-40sXXX"
     to "%-39s XXX". The latter will always guarantee a space after the
     field while the former will cause them to run together if the field
     is 40 of more characters long.
     [Steve Henson]

  +) Constify the cipher and digest 'method' functions and structures
     and modify related functions to take constant EVP_MD and EVP_CIPHER
     pointers.
     [Steve Henson]

  *) Implement ssl23_peek (analogous to ssl23_read), which previously
     did not exist.
     [Bodo Moeller]

  *) Replace rdtsc with _emit statements for VC++ version 5.
     [Jeremy Cooper <jeremy@baymoo.org>]

  +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
     in <openssl/bn.h>.  Also further increase BN_CTX_NUM to 32.
     [Bodo Moeller]

  +) Modify EVP_Digest*() routines so they now return values. Although the
     internal software routines can never fail additional hardware versions
     might.
     [Steve Henson]

  +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:

     Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
     (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.

     ASN1 error codes
          ERR_R_NESTED_ASN1_ERROR
          ...
          ERR_R_MISSING_ASN1_EOS
     were 4 .. 9, conflicting with
          ERR_LIB_RSA (= ERR_R_RSA_LIB)
          ...
          ERR_LIB_PEM (= ERR_R_PEM_LIB).
     They are now 58 .. 63 (i.e., just below ERR_R_FATAL).

     Add new error code 'ERR_R_INTERNAL_ERROR'.
     [Bodo Moeller]

  +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
     suffices.
     [Bodo Moeller]

  +) New option '-subj arg' for 'openssl req' and 'openssl ca'.  This
     sets the subject name for a new request or supersedes the
     subject name in a given request. Formats that can be parsed are
          'CN=Some Name, OU=myOU, C=IT'
     and
          'CN=Some Name/OU=myOU/C=IT'.

     Add options '-batch' and '-verbose' to 'openssl req'.
     [Massimiliano Pala <madwolf@hackmasters.net>]

  *) Make it possible to reuse SSLv2 sessions.
     [Richard Levitte]

  +) Introduce the possibility to access global variables through
     functions on platform were that's the best way to handle exporting
     global variables in shared libraries.  To enable this functionality,
     one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
     "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
     is normally done by Configure or something similar).

     To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
     in the source file (foo.c) like this:

        OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
        OPENSSL_IMPLEMENT_GLOBAL(double,bar);

     To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
     and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:

        OPENSSL_DECLARE_GLOBAL(int,foo);
        #define foo OPENSSL_GLOBAL_REF(foo)
        OPENSSL_DECLARE_GLOBAL(double,bar);
        #define bar OPENSSL_GLOBAL_REF(bar)

     The #defines are very important, and therefore so is including the
     header file everywere where the defined globals are used.

     The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
     of ASN.1 items, but that structure is a bt different.

     The largest change is in util/mkdef.pl which has been enhanced with
     better and easier to understand logic to choose which symbols should
     go into the Windows .def files as well as a number of fixes and code
     cleanup (among others, algorithm keywords are now sorted
     lexicographically to avoid constant rewrites).
     [Richard Levitte]

  *) In copy_email() check for >= 0 as a return value for
     X509_NAME_get_index_by_NID() since 0 is a valid index.
     [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]

  +) In BN_div() keep a copy of the sign of 'num' before writing the
     result to 'rm' because if rm==num the value will be overwritten
     and produce the wrong result if 'num' is negative: this caused
     problems with BN_mod() and BN_nnmod().
     [Steve Henson]

  +) Function OCSP_request_verify(). This checks the signature on an
     OCSP request and verifies the signer certificate. The signer
     certificate is just checked for a generic purpose and OCSP request
     trust settings.
     [Steve Henson]

  +) Add OCSP_check_validity() function to check the validity of OCSP
     responses. OCSP responses are prepared in real time and may only
     be a few seconds old. Simply checking that the current time lies
     between thisUpdate and nextUpdate max reject otherwise valid responses
     caused by either OCSP responder or client clock innacuracy. Instead
     we allow thisUpdate and nextUpdate to fall within a certain period of
     the current time. The age of the response can also optionally be
     checked. Two new options -validity_period and -status_age added to
     ocsp utility.
     [Steve Henson]

  +) If signature or public key algorithm is unrecognized print out its
     OID rather that just UNKOWN.
     [Steve Henson]

  *) Avoid coredump with unsupported or invalid public keys by checking if
     X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
     PKCS7_verify() fails with non detached data.
     [Steve Henson]

  +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
     OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
     ID to be generated from the issuer certificate alone which can then be
     passed to OCSP_id_issuer_cmp().
     [Steve Henson]

  +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
     ASN1 modules to export functions returning ASN1_ITEM pointers
     instead of the ASN1_ITEM structures themselves. This adds several
     new macros which allow the underlying ASN1 function/structure to
     be accessed transparently. As a result code should not use ASN1_ITEM
     references directly (such as &X509_it) but instead use the relevant
     macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
     use of the new ASN1 code on platforms where exporting structures
     is problematical (for example in shared libraries) but exporting
     functions returning pointers to structures is not.
     [Steve Henson]

  +) Add support for overriding the generation of SSL/TLS session IDs.
     These callbacks can be registered either in an SSL_CTX or per SSL.
     The purpose of this is to allow applications to control, if they wish,
     the arbitrary values chosen for use as session IDs, particularly as it
     can be useful for session caching in multiple-server environments. A
     command-line switch for testing this (and any client code that wishes
     to use such a feature) has been added to "s_server".
     [Geoff Thorpe, Lutz Jaenicke]

  +) Modify mkdef.pl to recognise and parse prprocessor conditionals
     of the form '#if defined(...) || defined(...) || ...' and
     '#if !defined(...) && !defined(...) && ...'.  This also avoids
     the growing number of special cases it was previously handling.
     [Richard Levitte]

  *) Don't use getenv in library functions when run as setuid/setgid.
     New function OPENSSL_issetugid().
     [Ulf Moeller]

  +) Make all configuration macros available for application by making
     sure they are available in opensslconf.h, by giving them names starting
     with "OPENSSL_" to avoid conflicts with other packages and by making
     sure e_os2.h will cover all platform-specific cases together with
     opensslconf.h.
     Additionally, it is now possible to define configuration/platform-
     specific names (called "system identities").  In the C code, these
     are prefixed with "OPENSSL_SYSNAME_".  e_os2.h will create another
     macro with the name beginning with "OPENSSL_SYS_", which is determined
     from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
     what is available.
     [Richard Levitte]

  +) New option -set_serial to 'req' and 'x509' this allows the serial
     number to use to be specified on the command line. Previously self
     signed certificates were hard coded with serial number 0 and the 
     CA options of 'x509' had to use a serial number in a file which was
     auto incremented.
     [Steve Henson]

  *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
     due to incorrect handling of multi-threading:

     1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().

     2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().

     3. Count how many times MemCheck_off() has been called so that
        nested use can be treated correctly.  This also avoids 
        inband-signalling in the previous code (which relied on the
        assumption that thread ID 0 is impossible).
     [Bodo Moeller]

  +) New options to 'ca' utility to support V2 CRL entry extensions.
     Currently CRL reason, invalidity date and hold instruction are
     supported. Add new CRL extensions to V3 code and some new objects.
     [Steve Henson]

  *) Add "-rand" option also to s_client and s_server.
     [Lutz Jaenicke]

  +) New function EVP_CIPHER_CTX_set_padding() this is used to
     disable standard block padding (aka PKCS#5 padding) in the EVP
     API, which was previously mandatory. This means that the data is
     not padded in any way and so the total length much be a multiple
     of the block size, otherwise an error occurs.
     [Steve Henson]

  +) Initial (incomplete) OCSP SSL support.
     [Steve Henson]

  *) Fix CPU detection on Irix 6.x.
     [Kurt Hockenbury <khockenb@stevens-tech.edu> and
      "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]

  +) New function OCSP_parse_url(). This splits up a URL into its host,
     port and path components: primarily to parse OCSP URLs. New -url
     option to ocsp utility.
     [Steve Henson]

  +) New nonce behavior. The return value of OCSP_check_nonce() now 
     reflects the various checks performed. Applications can decide
     whether to tolerate certain situations such as an absent nonce
     in a response when one was present in a request: the ocsp application
     just prints out a warning. New function OCSP_add1_basic_nonce()
     this is to allow responders to include a nonce in a response even if
     the request is nonce-less.
     [Steve Henson]

  *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
     was empty.
     [Steve Henson]
     [This change does not apply to 0.9.7.]

  *) Use the cached encoding of an X509_NAME structure rather than
     copying it. This is apparently the reason for the libsafe "errors"
     but the code is actually correct.
     [Steve Henson]

  +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
     skipped when using openssl x509 multiple times on a single input file,
     e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
     [Bodo Moeller]

  +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
     set string type: to handle setting ASN1_TIME structures. Fix ca
     utility to correctly initialize revocation date of CRLs.
     [Steve Henson]

  +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
     the clients preferred ciphersuites and rather use its own preferences.
     Should help to work around M$ SGC (Server Gated Cryptography) bug in
     Internet Explorer by ensuring unchanged hash method during stepup.
     (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
     [Lutz Jaenicke]

  +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
     to aes and add a new 'exist' option to print out symbols that don't
     appear to exist.
     [Steve Henson]

  +) Additional options to ocsp utility to allow flags to be set and
     additional certificates supplied.
     [Steve Henson]

  +) Add the option -VAfile to 'openssl ocsp', so the user can give the
     OCSP client a number of certificate to only verify the response
     signature against.
     [Richard Levitte]

  *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
     Bleichenbacher's DSA attack.
     Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
     to be set and top=0 forces the highest bit to be set; top=-1 is new
     and leaves the highest bit random.
     [Ulf Moeller, Bodo Moeller]

  +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
     handle the new API. Currently only ECB, CBC modes supported. Add new
     AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
     for TLS" draft-ietf-tls-ciphersuite-03.txt.
     [Ben Laurie, Steve Henson]

  *) In the NCONF_...-based implementations for CONF_... queries
     (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
     a temporary CONF structure with the data component set to NULL
     (which gives segmentation faults in lh_retrieve).
     Instead, use NULL for the CONF pointer in CONF_get_string and
     CONF_get_number (which may use environment variables) and directly
     return NULL from CONF_get_section.
     [Bodo Moeller]

  *) Fix potential buffer overrun for EBCDIC.
     [Ulf Moeller]

  +) New function OCSP_copy_nonce() to copy nonce value (if present) from
     request to response.
     [Steve Henson]

  +) Functions for OCSP responders. OCSP_request_onereq_count(),
     OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
     extract information from a certificate request. OCSP_response_create()
     creates a response and optionally adds a basic response structure.
     OCSP_basic_add1_status() adds a complete single response to a basic
     reponse and returns the OCSP_SINGLERESP structure just added (to allow
     extensions to be included for example). OCSP_basic_add1_cert() adds a
     certificate to a basic response and OCSP_basic_sign() signs a basic
     response with various flags. New helper functions ASN1_TIME_check()
     (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
     (converts ASN1_TIME to GeneralizedTime).
     [Steve Henson]

  +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
     in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
     structure from a certificate. X509_pubkey_digest() digests tha public_key
     contents: this is used in various key identifiers. 
     [Steve Henson]

  *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
     keyUsage if basicConstraints absent for a CA.
     [Steve Henson]

  *) Make SMIME_write_PKCS7() write mail header values with a format that
     is more generally accepted (no spaces before the semicolon), since
     some programs can't parse those values properly otherwise.  Also make
     sure BIO's that break lines after each write do not create invalid
     headers.
     [Richard Levitte]

  +) Make sk_sort() tolerate a NULL argument.
     [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]

  +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
     passed by the function are trusted implicitly. If any of them signed the
     reponse then it is assumed to be valid and is not verified.
     [Steve Henson]

  -) Make the CRL encoding routines work with empty SEQUENCE OF. The
     macros previously used would not encode an empty SEQUENCE OF
     and break the signature.
     [Steve Henson]
     [This change does not apply to 0.9.7.]

  *) Zero the premaster secret after deriving the master secret in
     DH ciphersuites.
     [Steve Henson]

  +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
     to data. This was previously part of the PKCS7 ASN1 code. This
     was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
     [Steve Henson, reported by Kenneth R. Robinette
                                <support@securenetterm.com>]

  +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
     routines: without these tracing memory leaks is very painful.
     Fix leaks in PKCS12 and PKCS7 routines.
     [Steve Henson]

  *) Add some EVP_add_digest_alias registrations (as found in
     OpenSSL_add_all_digests()) to SSL_library_init()
     aka OpenSSL_add_ssl_algorithms().  This provides improved
     compatibility with peers using X.509 certificates
     with unconventional AlgorithmIdentifier OIDs.
     [Bodo Moeller]

  *) Fix for Irix with NO_ASM.
     ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]

  *) ./config script fixes.
     [Ulf Moeller, Richard Levitte]

  +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
     Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
     effectively meant GeneralizedTime would never be used. Now it
     is initialised to -1 but X509_time_adj() now has to check the value
     and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
     V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
     [Steve Henson, reported by Kenneth R. Robinette
                                <support@securenetterm.com>]

  +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
     result in a zero length in the ASN1_INTEGER structure which was
     not consistent with the structure when d2i_ASN1_INTEGER() was used
     and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
     to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
     where it did not print out a minus for negative ASN1_INTEGER.
     [Steve Henson]

  *) Fix 'openssl passwd -1'.
     [Bodo Moeller]

  +) Add summary printout to ocsp utility. The various functions which
     convert status values to strings have been renamed to:
     OCSP_response_status_str(), OCSP_cert_status_str() and
     OCSP_crl_reason_str() and are no longer static. New options
     to verify nonce values and to disable verification. OCSP response
     printout format cleaned up.
     [Steve Henson]

  +) Add additional OCSP certificate checks. These are those specified
     in RFC2560. This consists of two separate checks: the CA of the
     certificate being checked must either be the OCSP signer certificate
     or the issuer of the OCSP signer certificate. In the latter case the
     OCSP signer certificate must contain the OCSP signing extended key
     usage. This check is performed by attempting to match the OCSP
     signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
     in the OCSP_CERTID structures of the response.
     [Steve Henson]

  +) Initial OCSP certificate verification added to OCSP_basic_verify()
     and related routines. This uses the standard OpenSSL certificate
     verify routines to perform initial checks (just CA validity) and
     to obtain the certificate chain. Then additional checks will be
     performed on the chain. Currently the root CA is checked to see
     if it is explicitly trusted for OCSP signing. This is used to set
     a root CA as a global signing root: that is any certificate that
     chains to that CA is an acceptable OCSP signing certificate.
     [Steve Henson]

  +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
     extensions from a separate configuration file.
     As when reading extensions from the main configuration file,
     the '-extensions ...' option may be used for specifying the
     section to use.
     [Massimiliano Pala <madwolf@comune.modena.it>]

  *) Change PKCS12_key_gen_asc() so it can cope with non null
     terminated strings whose length is passed in the passlen
     parameter, for example from PEM callbacks. This was done
     by adding an extra length parameter to asc2uni().
     [Steve Henson, reported by <oddissey@samsung.co.kr>]

  +) New OCSP utility. Allows OCSP requests to be generated or
     read. The request can be sent to a responder and the output
     parsed, outputed or printed in text form. Not complete yet:
     still needs to check the OCSP response validity.
     [Steve Henson]

  +) New subcommands for 'openssl ca':
     'openssl ca -status <serial>' prints the status of the cert with
     the given serial number (according to the index file).
     'openssl ca -updatedb' updates the expiry status of certificates
     in the index file.
     [Massimiliano Pala <madwolf@comune.modena.it>]

  +) New '-newreq-nodes' command option to CA.pl.  This is like
     '-newreq', but calls 'openssl req' with the '-nodes' option
     so that the resulting key is not encrypted.
     [Damien Miller <djm@mindrot.org>]

  +) New configuration for the GNU Hurd.
     [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]

  +) Initial code to implement OCSP basic response verify. This
     is currently incomplete. Currently just finds the signer's
     certificate and verifies the signature on the response.
     [Steve Henson]

  +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
     value of OPENSSLDIR.  This is available via the new '-d' option
     to 'openssl version', and is also included in 'openssl version -a'.
     [Bodo Moeller]

  *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
     call failed, free the DSA structure.
     [Bodo Moeller]

  +) Allowing defining memory allocation callbacks that will be given
     file name and line number information in additional arguments
     (a const char* and an int).  The basic functionality remains, as
     well as the original possibility to just replace malloc(),
     realloc() and free() by functions that do not know about these
     additional arguments.  To register and find out the current
     settings for extended allocation functions, the following
     functions are provided:

        CRYPTO_set_mem_ex_functions
        CRYPTO_set_locked_mem_ex_functions
        CRYPTO_get_mem_ex_functions
        CRYPTO_get_locked_mem_ex_functions

     These work the same way as CRYPTO_set_mem_functions and friends.
     CRYPTO_get_[locked_]mem_functions now writes 0 where such an
     extended allocation function is enabled.
     Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
     a conventional allocation function is enabled.
     [Richard Levitte, Bodo Moeller]

  *) Fix to uni2asc() to cope with zero length Unicode strings.
     These are present in some PKCS#12 files.
     [Steve Henson]

  +) Finish off removing the remaining LHASH function pointer casts.
     There should no longer be any prototype-casting required when using
     the LHASH abstraction, and any casts that remain are "bugs". See
     the callback types and macros at the head of lhash.h for details
     (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
     [Geoff Thorpe]

  +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
     If /dev/[u]random devices are not available or do not return enough
     entropy, EGD style sockets (served by EGD or PRNGD) will automatically
     be queried.
     The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
     /etc/entropy will be queried once each in this sequence, quering stops
     when enough entropy was collected without querying more sockets.
     [Lutz Jaenicke]

  +) Change the Unix RAND_poll() variant to be able to poll several
     random devices, as specified by DEVRANDOM, until a sufficient amount
     of data has been collected.   We spend at most 10 ms on each file
     (select timeout) and read in non-blocking mode.  DEVRANDOM now
     defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
     (previously it was just the string "/dev/urandom"), so on typical
     platforms the 10 ms delay will never occur.
     Also separate out the Unix variant to its own file, rand_unix.c.
     For VMS, there's a currently-empty rand_vms.c.
     [Richard Levitte]

  +) Move OCSP client related routines to ocsp_cl.c. These
     provide utility functions which an application needing
     to issue a request to an OCSP responder and analyse the
     response will typically need: as opposed to those which an
     OCSP responder itself would need which will be added later.

     OCSP_request_sign() signs an OCSP request with an API similar
     to PKCS7_sign(). OCSP_response_status() returns status of OCSP
     response. OCSP_response_get1_basic() extracts basic response
     from response. OCSP_resp_find_status(): finds and extracts status
     information from an OCSP_CERTID structure (which will be created
     when the request structure is built). These are built from lower
     level functions which work on OCSP_SINGLERESP structures but
     wont normally be used unless the application wishes to examine
     extensions in the OCSP response for example.

     Replace nonce routines with a pair of functions.
     OCSP_request_add1_nonce() adds a nonce value and optionally
     generates a random value. OCSP_check_nonce() checks the
     validity of the nonce in an OCSP response.
     [Steve Henson]

  +) Change function OCSP_request_add() to OCSP_request_add0_id().
     This doesn't copy the supplied OCSP_CERTID and avoids the
     need to free up the newly created id. Change return type
     to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
     This can then be used to add extensions to the request.
     Deleted OCSP_request_new(), since most of its functionality
     is now in OCSP_REQUEST_new() (and the case insensitive name
     clash) apart from the ability to set the request name which
     will be added elsewhere.
     [Steve Henson]

  +) Update OCSP API. Remove obsolete extensions argument from
     various functions. Extensions are now handled using the new
     OCSP extension code. New simple OCSP HTTP function which 
     can be used to send requests and parse the response.
     [Steve Henson]

  +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
     ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
     uses the special reorder version of SET OF to sort the attributes
     and reorder them to match the encoded order. This resolves a long
     standing problem: a verify on a PKCS7 structure just after signing
     it used to fail because the attribute order did not match the
     encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
     it uses the received order. This is necessary to tolerate some broken
     software that does not order SET OF. This is handled by encoding
     as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
     to produce the required SET OF.
     [Steve Henson]

  +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
     OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
     files to get correct declarations of the ASN.1 item variables.
     [Richard Levitte]

  +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
     PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
     asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
     NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
     New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
     ASN1_ITEM and no wrapper functions.
     [Steve Henson]

  +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
     replace the old function pointer based I/O routines. Change most of
     the *_d2i_bio() and *_d2i_fp() functions to use these.
     [Steve Henson]

  +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
     lines, recognice more "algorithms" that can be deselected, and make
     it complain about algorithm deselection that isn't recognised.
     [Richard Levitte]

  +) New ASN1 functions to handle dup, sign, verify, digest, pack and
     unpack operations in terms of ASN1_ITEM. Modify existing wrappers
     to use new functions. Add NO_ASN1_OLD which can be set to remove
     some old style ASN1 functions: this can be used to determine if old
     code will still work when these eventually go away.
     [Steve Henson]

  +) New extension functions for OCSP structures, these follow the
     same conventions as certificates and CRLs.
     [Steve Henson]

  +) New function X509V3_add1_i2d(). This automatically encodes and
     adds an extension. Its behaviour can be customised with various
     flags to append, replace or delete. Various wrappers added for
     certifcates and CRLs.
     [Steve Henson]

  +) Fix to avoid calling the underlying ASN1 print routine when
     an extension cannot be parsed. Correct a typo in the
     OCSP_SERVICELOC extension. Tidy up print OCSP format.
     [Steve Henson]

  *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
     Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
     when writing a 32767 byte record.
     [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]

  *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
     obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.

     (RSA objects have a reference count access to which is protected
     by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
     so they are meant to be shared between threads.)
     [Bodo Moeller, Geoff Thorpe; original patch submitted by
     "Reddie, Steven" <Steven.Reddie@ca.com>]

  +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
     entries for variables.
     [Steve Henson]

  *) Fix a deadlock in CRYPTO_mem_leaks().
     [Bodo Moeller]

  +) Add functionality to apps/openssl.c for detecting locking
     problems: As the program is single-threaded, all we have
     to do is register a locking callback using an array for
     storing which locks are currently held by the program.
     [Bodo Moeller]

  +) Use a lock around the call to CRYPTO_get_ex_new_index() in
     SSL_get_ex_data_X509_STORE_idx(), which is used in
     ssl_verify_cert_chain() and thus can be called at any time
     during TLS/SSL handshakes so that thread-safety is essential.
     Unfortunately, the ex_data design is not at all suited
     for multi-threaded use, so it probably should be abolished.
     [Bodo Moeller]

  +) Added Broadcom "ubsec" ENGINE to OpenSSL.
     [Broadcom, tweaked and integrated by Geoff Thorpe]

  +) Move common extension printing code to new function
     X509V3_print_extensions(). Reorganise OCSP print routines and
     implement some needed OCSP ASN1 functions. Add OCSP extensions.
     [Steve Henson]

  +) New function X509_signature_print() to remove duplication in some
     print routines.
     [Steve Henson]

  +) Add a special meaning when SET OF and SEQUENCE OF flags are both
     set (this was treated exactly the same as SET OF previously). This
     is used to reorder the STACK representing the structure to match the
     encoding. This will be used to get round a problem where a PKCS7
     structure which was signed could not be verified because the STACK
     order did not reflect the encoded order.
     [Steve Henson]

  +) Reimplement the OCSP ASN1 module using the new code.
     [Steve Henson]

  +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
     for its ASN1 operations. The old style function pointers still exist
     for now but they will eventually go away.
     [Steve Henson]

  +) Merge in replacement ASN1 code from the ASN1 branch. This almost
     completely replaces the old ASN1 functionality with a table driven
     encoder and decoder which interprets an ASN1_ITEM structure describing
     the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
     largely maintained. Almost all of the old asn1_mac.h macro based ASN1
     has also been converted to the new form.
     [Steve Henson]

  +) Change BN_mod_exp_recp so that negative moduli are tolerated
     (the sign is ignored).  Similarly, ignore the sign in BN_MONT_CTX_set
     so that BN_mod_exp_mont and BN_mod_exp_mont_word work
     for negative moduli.
     [Bodo Moeller]

  +) Fix BN_uadd and BN_usub: Always return non-negative results instead
     of not touching the result's sign bit.
     [Bodo Moeller]

  +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
     set.
     [Bodo Moeller]

  +) Changed the LHASH code to use prototypes for callbacks, and created
     macros to declare and implement thin (optionally static) functions
     that provide type-safety and avoid function pointer casting for the
     type-specific callbacks.
     [Geoff Thorpe]

  *) Use better test patterns in bntest.
     [Ulf Möller]

  +) Added Kerberos Cipher Suites to be used with TLS, as written in
     RFC 2712.
     [Veers Staats <staatsvr@asc.hpc.mil>,
      Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]

  *) rand_win.c fix for Borland C.
     [Ulf Möller]
 
  *) BN_rshift bugfix for n == 0.
     [Bodo Moeller]

  +) Reformat the FAQ so the different questions and answers can be divided
     in sections depending on the subject.
     [Richard Levitte]

  +) Have the zlib compression code load ZLIB.DLL dynamically under
     Windows.
     [Richard Levitte]

  +) New function BN_mod_sqrt for computing square roots modulo a prime
     (using the probabilistic Tonelli-Shanks algorithm unless
     p == 3 (mod 4)  or  p == 5 (mod 8),  which are cases that can
     be handled deterministically).
     [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]

  *) Add a 'bctest' script that checks for some known 'bc' bugs
     so that 'make test' does not abort just because 'bc' is broken.
     [Bodo Moeller]

  *) Store verify_result within SSL_SESSION also for client side to
     avoid potential security hole. (Re-used sessions on the client side
     always resulted in verify_result==X509_V_OK, not using the original
     result of the server certificate verification.)
     [Lutz Jaenicke]

  +) Make BN_mod_inverse faster by explicitly handling small quotients
     in the Euclid loop. (Speed gain about 20% for small moduli [256 or
     512 bits], about 30% for larger ones [1024 or 2048 bits].)
     [Bodo Moeller]

  *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
     SSL3_RT_APPLICATION_DATA, return 0.
     Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
     [Bodo Moeller]

  *) Fix SSL_peek:
     Both ssl2_peek and ssl3_peek, which were totally broken in earlier
     releases, have been re-implemented by renaming the previous
     implementations of ssl2_read and ssl3_read to ssl2_read_internal
     and ssl3_read_internal, respectively, and adding 'peek' parameters
     to them.  The new ssl[23]_{read,peek} functions are calls to
     ssl[23]_read_internal with the 'peek' flag set appropriately.
     A 'peek' parameter has also been added to ssl3_read_bytes, which
     does the actual work for ssl3_read_internal.
     [Bodo Moeller]

  +) New function BN_kronecker.
     [Bodo Moeller]

  +) Fix BN_gcd so that it works on negative inputs; the result is
     positive unless both parameters are zero.
     Previously something reasonably close to an infinite loop was
     possible because numbers could be growing instead of shrinking
     in the implementation of Euclid's algorithm.
     [Bodo Moeller]

  +) Fix BN_is_word() and BN_is_one() macros to take into account the
     sign of the number in question.

     Fix BN_is_word(a,w) to work correctly for w == 0.

     The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
     because its test if the absolute value of 'a' equals 'w'.
     Note that BN_abs_is_word does *not* handle w == 0 reliably;
     it exists mostly for use in the implementations of BN_is_zero(),
     BN_is_one(), and BN_is_word().
     [Bodo Moeller]

  *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
     the method-specific "init()" handler. Also clean up ex_data after
     calling the method-specific "finish()" handler. Previously, this was
     happening the other way round.
     [Geoff Thorpe]

  +) New function BN_swap.
     [Bodo Moeller]

  +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
     the exponentiation functions are more likely to produce reasonable
     results on negative inputs.
     [Bodo Moeller]

  +) Change BN_mod_mul so that the result is always non-negative.
     Previously, it could be negative if one of the factors was negative;
     I don't think anyone really wanted that behaviour.
     [Bodo Moeller]

  +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
     (except for exponentiation, which stays in crypto/bn/bn_exp.c,
     and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
     and add new functions:

          BN_nnmod
          BN_mod_sqr
          BN_mod_add
          BN_mod_add_quick
          BN_mod_sub
          BN_mod_sub_quick
          BN_mod_lshift1
          BN_mod_lshift1_quick
          BN_mod_lshift
          BN_mod_lshift_quick

     These functions always generate non-negative results.

     BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder  r
     such that  |m| < r < 0,  BN_nnmod will output  rem + |m|  instead).

     BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
     BN_mod_XXX(r, a, [b,] m, ctx), but requires that  a  [and  b]
     be reduced modulo  m.
     [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]

  +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
     was actually never needed) and in BN_mul().  The removal in BN_mul()
     required a small change in bn_mul_part_recursive() and the addition
     of the functions bn_cmp_part_words(), bn_sub_part_words() and
     bn_add_part_words(), which do the same thing as bn_cmp_words(),
     bn_sub_words() and bn_add_words() except they take arrays with
     differing sizes.
     [Richard Levitte]

  +) In 'openssl passwd', verify passwords read from the terminal
     unless the '-salt' option is used (which usually means that
     verification would just waste user's time since the resulting
     hash is going to be compared with some given password hash)
     or the new '-noverify' option is used.

     This is an incompatible change, but it does not affect
     non-interactive use of 'openssl passwd' (passwords on the command
     line, '-stdin' option, '-in ...' option) and thus should not
     cause any problems.
     [Bodo Moeller]

  +) Remove all references to RSAref, since there's no more need for it.
     [Richard Levitte]

  *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
     The previous value, 12, was not always sufficient for BN_mod_exp().
     [Bodo Moeller]

  +) Make DSO load along a path given through an environment variable
     (SHLIB_PATH) with shl_load().
     [Richard Levitte]

  +) Constify the ENGINE code as a result of BIGNUM constification.
     Also constify the RSA code and most things related to it.  In a
     few places, most notable in the depth of the ASN.1 code, ugly
     casts back to non-const were required (to be solved at a later
     time)
     [Richard Levitte]

  +) Make it so the openssl application has all engines loaded by default.
     [Richard Levitte]

  +) Constify the BIGNUM routines a little more.
     [Richard Levitte]

  *) Make sure that shared libraries get the internal name engine with
     the full version number and not just 0.  This should mark the
     shared libraries as not backward compatible.  Of course, this should
     be changed again when we can guarantee backward binary compatibility.
     [Richard Levitte]

  +) Add the following functions:

        ENGINE_load_cswift()
        ENGINE_load_chil()
        ENGINE_load_atalla()
        ENGINE_load_nuron()
        ENGINE_load_builtin_engines()

     That way, an application can itself choose if external engines that
     are built-in in OpenSSL shall ever be used or not.  The benefit is
     that applications won't have to be linked with libdl or other dso
     libraries unless it's really needed.

     Changed 'openssl engine' to load all engines on demand.
     Changed the engine header files to avoid the duplication of some
     declarations (they differed!).
     [Richard Levitte]

  +) 'openssl engine' can now list capabilities.
     [Richard Levitte]

  +) Better error reporting in 'openssl engine'.
     [Richard Levitte]

  +) Never call load_dh_param(NULL) in s_server.
     [Bodo Moeller]

  +) Add engine application.  It can currently list engines by name and
     identity, and test if they are actually available.
     [Richard Levitte]

  +) Add support for shared libraries under Irix.
     [Albert Chin-A-Young <china@thewrittenword.com>]

  +) Improve RPM specification file by forcing symbolic linking and making
     sure the installed documentation is also owned by root.root.
     [Damien Miller <djm@mindrot.org>]

  +) Add configuration option to build on Linux on both big-endian and
     little-endian MIPS.
     [Ralf Baechle <ralf@uni-koblenz.de>]

  +) Give the OpenSSL applications more possibilities to make use of
     keys (public as well as private) handled by engines.
     [Richard Levitte]

  +) Add OCSP code that comes from CertCo.
     [Richard Levitte]

  +) Add VMS support for the Rijndael code.
     [Richard Levitte]

  +) Added untested support for Nuron crypto accelerator.
     [Ben Laurie]

  +) Add support for external cryptographic devices.  This code was
     previously distributed separately as the "engine" branch.
     [Geoff Thorpe, Richard Levitte]

  +) Rework the filename-translation in the DSO code. It is now possible to
     have far greater control over how a "name" is turned into a filename
     depending on the operating environment and any oddities about the
     different shared library filenames on each system.
     [Geoff Thorpe]

  +) Support threads on FreeBSD-elf in Configure.
     [Richard Levitte]

  +) Add the possibility to create shared libraries on HP-UX
     [Richard Levitte]

  +) Fix for SHA1 assembly problem with MASM: it produces
     warnings about corrupt line number information when assembling
     with debugging information. This is caused by the overlapping
     of two sections.
     [Bernd Matthes <mainbug@celocom.de>, Steve Henson]

  +) NCONF changes.
     NCONF_get_number() has no error checking at all.  As a replacement,
     NCONF_get_number_e() is defined (_e for "error checking") and is
     promoted strongly.  The old NCONF_get_number is kept around for
     binary backward compatibility.
     Make it possible for methods to load from something other than a BIO,
     by providing a function pointer that is given a name instead of a BIO.
     For example, this could be used to load configuration data from an
     LDAP server.
     [Richard Levitte]

  *) Fix typo in get_cert_by_subject() in by_dir.c
     [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]

  *) Rework the system to generate shared libraries:

     - Make note of the expected extension for the shared libraries and
       if there is a need for symbolic links from for example libcrypto.so.0
       to libcrypto.so.0.9.7.  There is extended info in Configure for
       that.

     - Make as few rebuilds of the shared libraries as possible.

     - Still avoid linking the OpenSSL programs with the shared libraries.

     - When installing, install the shared libraries separately from the
       static ones.
     [Richard Levitte]

  +) Fix for non blocking accept BIOs. Added new I/O special reason
     BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
     with non blocking I/O was not possible because no retry code was
     implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
     this case.
     [Steve Henson]

  +) Added the beginnings of Rijndael support.
     [Ben Laurie]

  +) Fix for bug in DirectoryString mask setting. Add support for
     X509_NAME_print_ex() in 'req' and X509_print_ex() function
     to allow certificate printing to more controllable, additional
     'certopt' option to 'x509' to allow new printing options to be
     set.
     [Steve Henson]

  +) Clean old EAY MD5 hack from e_os.h.
     [Richard Levitte]

  *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.

     Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
     and not in SSL_clear because the latter is also used by the
     accept/connect functions; previously, the settings made by
     SSL_set_read_ahead would be lost during the handshake.
     [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]     

  *) Correct util/mkdef.pl to be selective about disabled algorithms.
     Previously, it would create entries for disableed algorithms no
     matter what.
     [Richard Levitte]

  *) Added several new manual pages for SSL_* function.
     [Lutz Jaenicke]

 Changes between 0.9.5a and 0.9.6  [24 Sep 2000]

  *) In ssl23_get_client_hello, generate an error message when faced
     with an initial SSL 3.0/TLS record that is too small to contain the
     first two bytes of the ClientHello message, i.e. client_version.
     (Note that this is a pathologic case that probably has never happened
     in real life.)  The previous approach was to use the version number
     from the record header as a substitute; but our protocol choice
     should not depend on that one because it is not authenticated
     by the Finished messages.
     [Bodo Moeller]

  *) More robust randomness gathering functions for Windows.
     [Jeffrey Altman <jaltman@columbia.edu>]

  *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
     not set then we don't setup the error code for issuer check errors
     to avoid possibly overwriting other errors which the callback does
     handle. If an application does set the flag then we assume it knows
     what it is doing and can handle the new informational codes
     appropriately.
     [Steve Henson]

  *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
     a general "ANY" type, as such it should be able to decode anything
     including tagged types. However it didn't check the class so it would
     wrongly interpret tagged types in the same way as their universal
     counterpart and unknown types were just rejected. Changed so that the
     tagged and unknown types are handled in the same way as a SEQUENCE:
     that is the encoding is stored intact. There is also a new type
     "V_ASN1_OTHER" which is used when the class is not universal, in this
     case we have no idea what the actual type is so we just lump them all
     together.
     [Steve Henson]

  *) On VMS, stdout may very well lead to a file that is written to
     in a record-oriented fashion.  That means that every write() will
     write a separate record, which will be read separately by the
     programs trying to read from it.  This can be very confusing.

     The solution is to put a BIO filter in the way that will buffer
     text until a linefeed is reached, and then write everything a
     line at a time, so every record written will be an actual line,
     not chunks of lines and not (usually doesn't happen, but I've
     seen it once) several lines in one record.  BIO_f_linebuffer() is
     the answer.

     Currently, it's a VMS-only method, because that's where it has
     been tested well enough.
     [Richard Levitte]

  *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
     it can return incorrect results.
     (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
     but it was in 0.9.6-beta[12].)
     [Bodo Moeller]

  *) Disable the check for content being present when verifying detached
     signatures in pk7_smime.c. Some versions of Netscape (wrongly)
     include zero length content when signing messages.
     [Steve Henson]

  *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
     BIO_ctrl (for BIO pairs).
     [Bodo Möller]

  *) Add DSO method for VMS.
     [Richard Levitte]

  *) Bug fix: Montgomery multiplication could produce results with the
     wrong sign.
     [Ulf Möller]

  *) Add RPM specification openssl.spec and modify it to build three
     packages.  The default package contains applications, application
     documentation and run-time libraries.  The devel package contains
     include files, static libraries and function documentation.  The
     doc package contains the contents of the doc directory.  The original
     openssl.spec was provided by Damien Miller <djm@mindrot.org>.
     [Richard Levitte]
     
  *) Add a large number of documentation files for many SSL routines.
     [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]

  *) Add a configuration entry for Sony News 4.
     [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]

  *) Don't set the two most significant bits to one when generating a
     random number < q in the DSA library.
     [Ulf Möller]

  *) New SSL API mode 'SSL_MODE_AUTO_RETRY'.  This disables the default
     behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
     the underlying transport is blocking) if a handshake took place.
     (The default behaviour is needed by applications such as s_client
     and s_server that use select() to determine when to use SSL_read;
     but for applications that know in advance when to expect data, it
     just makes things more complicated.)
     [Bodo Moeller]

  *) Add RAND_egd_bytes(), which gives control over the number of bytes read
     from EGD.
     [Ben Laurie]

  *) Add a few more EBCDIC conditionals that make `req' and `x509'
     work better on such systems.
     [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]

  *) Add two demo programs for PKCS12_parse() and PKCS12_create().
     Update PKCS12_parse() so it copies the friendlyName and the
     keyid to the certificates aux info.
     [Steve Henson]

  *) Fix bug in PKCS7_verify() which caused an infinite loop
     if there was more than one signature.
     [Sven Uszpelkat <su@celocom.de>]

  *) Major change in util/mkdef.pl to include extra information
     about each symbol, as well as presentig variables as well
     as functions.  This change means that there's n more need
     to rebuild the .num files when some algorithms are excluded.
     [Richard Levitte]

  *) Allow the verify time to be set by an application,
     rather than always using the current time.
     [Steve Henson]
  
  *) Phase 2 verify code reorganisation. The certificate
     verify code now looks up an issuer certificate by a
     number of criteria: subject name, authority key id
     and key usage. It also verifies self signed certificates
     by the same criteria. The main comparison function is
     X509_check_issued() which performs these checks.
 
     Lot of changes were necessary in order to support this
     without completely rewriting the lookup code.
 
     Authority and subject key identifier are now cached.
 
     The LHASH 'certs' is X509_STORE has now been replaced
     by a STACK_OF(X509_OBJECT). This is mainly because an
     LHASH can't store or retrieve multiple objects with
     the same hash value.

     As a result various functions (which were all internal
     use only) have changed to handle the new X509_STORE
     structure. This will break anything that messed round
     with X509_STORE internally.
 
     The functions X509_STORE_add_cert() now checks for an
     exact match, rather than just subject name.
 
     The X509_STORE API doesn't directly support the retrieval
     of multiple certificates matching a given criteria, however
     this can be worked round by performing a lookup first
     (which will fill the cache with candidate certificates)
     and then examining the cache for matches. This is probably
     the best we can do without throwing out X509_LOOKUP
     entirely (maybe later...).
 
     The X509_VERIFY_CTX structure has been enhanced considerably.
 
     All certificate lookup operations now go via a get_issuer()
     callback. Although this currently uses an X509_STORE it
     can be replaced by custom lookups. This is a simple way
     to bypass the X509_STORE hackery necessary to make this
     work and makes it possible to use more efficient techniques
     in future. A very simple version which uses a simple
     STACK for its trusted certificate store is also provided
     using X509_STORE_CTX_trusted_stack().
 
     The verify_cb() and verify() callbacks now have equivalents
     in the X509_STORE_CTX structure.
 
     X509_STORE_CTX also has a 'flags' field which can be used
     to customise the verify behaviour.
     [Steve Henson]
 
  *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which 
     excludes S/MIME capabilities.
     [Steve Henson]

  *) When a certificate request is read in keep a copy of the
     original encoding of the signed data and use it when outputing
     again. Signatures then use the original encoding rather than
     a decoded, encoded version which may cause problems if the
     request is improperly encoded.
     [Steve Henson]

  *) For consistency with other BIO_puts implementations, call
     buffer_write(b, ...) directly in buffer_puts instead of calling
     BIO_write(b, ...).

     In BIO_puts, increment b->num_write as in BIO_write.
     [Peter.Sylvester@EdelWeb.fr]

  *) Fix BN_mul_word for the case where the word is 0. (We have to use
     BN_zero, we may not return a BIGNUM with an array consisting of
     words set to zero.)
     [Bodo Moeller]

  *) Avoid calling abort() from within the library when problems are
     detected, except if preprocessor symbols have been defined
     (such as REF_CHECK, BN_DEBUG etc.).
     [Bodo Moeller]

  *) New openssl application 'rsautl'. This utility can be
     used for low level RSA operations. DER public key
     BIO/fp routines also added.
     [Steve Henson]

  *) New Configure entry and patches for compiling on QNX 4.
     [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]

  *) A demo state-machine implementation was sponsored by
     Nuron (http://www.nuron.com/) and is now available in
     demos/state_machine.
     [Ben Laurie]

  *) New options added to the 'dgst' utility for signature
     generation and verification.
     [Steve Henson]

  *) Unrecognized PKCS#7 content types are now handled via a
     catch all ASN1_TYPE structure. This allows unsupported
     types to be stored as a "blob" and an application can
     encode and decode it manually.
     [Steve Henson]

  *) Fix various signed/unsigned issues to make a_strex.c
     compile under VC++.
     [Oscar Jacobsson <oscar.jacobsson@celocom.com>]

  *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
     length if passed a buffer. ASN1_INTEGER_to_BN failed
     if passed a NULL BN and its argument was negative.
     [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]

  *) Modification to PKCS#7 encoding routines to output definite
     length encoding. Since currently the whole structures are in
     memory there's not real point in using indefinite length 
     constructed encoding. However if OpenSSL is compiled with
     the flag PKCS7_INDEFINITE_ENCODING the old form is used.
     [Steve Henson]

  *) Added BIO_vprintf() and BIO_vsnprintf().
     [Richard Levitte]

  *) Added more prefixes to parse for in the the strings written
     through a logging bio, to cover all the levels that are available
     through syslog.  The prefixes are now:

        PANIC, EMERG, EMR       =>      LOG_EMERG
        ALERT, ALR              =>      LOG_ALERT
        CRIT, CRI               =>      LOG_CRIT
        ERROR, ERR              =>      LOG_ERR
        WARNING, WARN, WAR      =>      LOG_WARNING
        NOTICE, NOTE, NOT       =>      LOG_NOTICE
        INFO, INF               =>      LOG_INFO
        DEBUG, DBG              =>      LOG_DEBUG

     and as before, if none of those prefixes are present at the
     beginning of the string, LOG_ERR is chosen.

     On Win32, the LOG_* levels are mapped according to this:

        LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
        LOG_WARNING                             => EVENTLOG_WARNING_TYPE
        LOG_NOTICE, LOG_INFO, LOG_DEBUG         => EVENTLOG_INFORMATION_TYPE

     [Richard Levitte]

  *) Made it possible to reconfigure with just the configuration
     argument "reconf" or "reconfigure".  The command line arguments
     are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
     and are retrieved from there when reconfiguring.
     [Richard Levitte]

  *) MD4 implemented.
     [Assar Westerlund <assar@sics.se>, Richard Levitte]

  *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
     [Richard Levitte]

  *) The obj_dat.pl script was messing up the sorting of object
     names. The reason was that it compared the quoted version
     of strings as a result "OCSP" > "OCSP Signing" because
     " > SPACE. Changed script to store unquoted versions of
     names and add quotes on output. It was also omitting some
     names from the lookup table if they were given a default
     value (that is if SN is missing it is given the same
     value as LN and vice versa), these are now added on the
     grounds that if an object has a name we should be able to
     look it up. Finally added warning output when duplicate
     short or long names are found.
     [Steve Henson]

  *) Changes needed for Tandem NSK.
     [Scott Uroff <scott@xypro.com>]

  *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
     RSA_padding_check_SSLv23(), special padding was never detected
     and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
     version rollback attacks was not effective.

     In s23_clnt.c, don't use special rollback-attack detection padding
     (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
     client; similarly, in s23_srvr.c, don't do the rollback check if
     SSL 2.0 is the only protocol enabled in the server.
     [Bodo Moeller]

  *) Make it possible to get hexdumps of unprintable data with 'openssl
     asn1parse'.  By implication, the functions ASN1_parse_dump() and
     BIO_dump_indent() are added.
     [Richard Levitte]

  *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
     these print out strings and name structures based on various
     flags including RFC2253 support and proper handling of
     multibyte characters. Added options to the 'x509' utility 
     to allow the various flags to be set.
     [Steve Henson]

  *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
     Also change the functions X509_cmp_current_time() and
     X509_gmtime_adj() work with an ASN1_TIME structure,
     this will enable certificates using GeneralizedTime in validity
     dates to be checked.
     [Steve Henson]

  *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
     negative public key encodings) on by default,
     NO_NEG_PUBKEY_BUG can be set to disable it.
     [Steve Henson]

  *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
     content octets. An i2c_ASN1_OBJECT is unnecessary because
     the encoding can be trivially obtained from the structure.
     [Steve Henson]

  *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
     not read locks (CRYPTO_r_[un]lock).
     [Bodo Moeller]

  *) A first attempt at creating official support for shared
     libraries through configuration.  I've kept it so the
     default is static libraries only, and the OpenSSL programs
     are always statically linked for now, but there are
     preparations for dynamic linking in place.
     This has been tested on Linux and Tru64.
     [Richard Levitte]

  *) Randomness polling function for Win9x, as described in:
     Peter Gutmann, Software Generation of Practically Strong
     Random Numbers.
     [Ulf Möller]

  *) Fix so PRNG is seeded in req if using an already existing
     DSA key.
     [Steve Henson]

  *) New options to smime application. -inform and -outform
     allow alternative formats for the S/MIME message including
     PEM and DER. The -content option allows the content to be
     specified separately. This should allow things like Netscape
     form signing output easier to verify.
     [Steve Henson]

  *) Fix the ASN1 encoding of tags using the 'long form'.
     [Steve Henson]

  *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
     STRING types. These convert content octets to and from the
     underlying type. The actual tag and length octets are
     already assumed to have been read in and checked. These
     are needed because all other string types have virtually
     identical handling apart from the tag. By having versions
     of the ASN1 functions that just operate on content octets
     IMPLICIT tagging can be handled properly. It also allows
     the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
     and ASN1_INTEGER are identical apart from the tag.
     [Steve Henson]

  *) Change the handling of OID objects as follows:

     - New object identifiers are inserted in objects.txt, following
       the syntax given in objects.README.
     - objects.pl is used to process obj_mac.num and create a new
       obj_mac.h.
     - obj_dat.pl is used to create a new obj_dat.h, using the data in
       obj_mac.h.

     This is currently kind of a hack, and the perl code in objects.pl
     isn't very elegant, but it works as I intended.  The simplest way
     to check that it worked correctly is to look in obj_dat.h and
     check the array nid_objs and make sure the objects haven't moved
     around (this is important!).  Additions are OK, as well as
     consistent name changes. 
     [Richard Levitte]

  *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
     [Bodo Moeller]

  *) Addition of the command line parameter '-rand file' to 'openssl req'.
     The given file adds to whatever has already been seeded into the
     random pool through the RANDFILE configuration file option or
     environment variable, or the default random state file.
     [Richard Levitte]

  *) mkstack.pl now sorts each macro group into lexical order.
     Previously the output order depended on the order the files
     appeared in the directory, resulting in needless rewriting
     of safestack.h .
     [Steve Henson]

  *) Patches to make OpenSSL compile under Win32 again. Mostly
     work arounds for the VC++ problem that it treats func() as
     func(void). Also stripped out the parts of mkdef.pl that
     added extra typesafe functions: these no longer exist.
     [Steve Henson]

  *) Reorganisation of the stack code. The macros are now all 
     collected in safestack.h . Each macro is defined in terms of
     a "stack macro" of the form SKM_<name>(type, a, b). The 
     DEBUG_SAFESTACK is now handled in terms of function casts,
     this has the advantage of retaining type safety without the
     use of additional functions. If DEBUG_SAFESTACK is not defined
     then the non typesafe macros are used instead. Also modified the
     mkstack.pl script to handle the new form. Needs testing to see
     if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
     the default if no major problems. Similar behaviour for ASN1_SET_OF
     and PKCS12_STACK_OF.
     [Steve Henson]

  *) When some versions of IIS use the 'NET' form of private key the
     key derivation algorithm is different. Normally MD5(password) is
     used as a 128 bit RC4 key. In the modified case
     MD5(MD5(password) + "SGCKEYSALT")  is used insted. Added some
     new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
     as the old Netscape_RSA functions except they have an additional
     'sgckey' parameter which uses the modified algorithm. Also added
     an -sgckey command line option to the rsa utility. Thanks to 
     Adrian Peck <bertie@ncipher.com> for posting details of the modified
     algorithm to openssl-dev.
     [Steve Henson]

  *) The evp_local.h macros were using 'c.##kname' which resulted in
     invalid expansion on some systems (SCO 5.0.5 for example).
     Corrected to 'c.kname'.
     [Phillip Porch <root@theporch.com>]

  *) New X509_get1_email() and X509_REQ_get1_email() functions that return
     a STACK of email addresses from a certificate or request, these look
     in the subject name and the subject alternative name extensions and 
     omit any duplicate addresses.
     [Steve Henson]

  *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
     This makes DSA verification about 2 % faster.
     [Bodo Moeller]

  *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
     (meaning that now 2^5 values will be precomputed, which is only 4 KB
     plus overhead for 1024 bit moduli).
     This makes exponentiations about 0.5 % faster for 1024 bit
     exponents (as measured by "openssl speed rsa2048").
     [Bodo Moeller]

  *) Rename memory handling macros to avoid conflicts with other
     software:
          Malloc         =>  OPENSSL_malloc
          Malloc_locked  =>  OPENSSL_malloc_locked
          Realloc        =>  OPENSSL_realloc
          Free           =>  OPENSSL_free
     [Richard Levitte]

  *) New function BN_mod_exp_mont_word for small bases (roughly 15%
     faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
     [Bodo Moeller]

  *) CygWin32 support.
     [John Jarvie <jjarvie@newsguy.com>]

  *) The type-safe stack code has been rejigged. It is now only compiled
     in when OpenSSL is configured with the DEBUG_SAFESTACK option and
     by default all type-specific stack functions are "#define"d back to
     standard stack functions. This results in more streamlined output
     but retains the type-safety checking possibilities of the original
     approach.
     [Geoff Thorpe]

  *) The STACK code has been cleaned up, and certain type declarations
     that didn't make a lot of sense have been brought in line. This has
     also involved a cleanup of sorts in safestack.h to more correctly
     map type-safe stack functions onto their plain stack counterparts.
     This work has also resulted in a variety of "const"ifications of
     lots of the code, especially "_cmp" operations which should normally
     be prototyped with "const" parameters anyway.
     [Geoff Thorpe]

  *) When generating bytes for the first time in md_rand.c, 'stir the pool'
     by seeding with STATE_SIZE dummy bytes (with zero entropy count).
     (The PRNG state consists of two parts, the large pool 'state' and 'md',
     where all of 'md' is used each time the PRNG is used, but 'state'
     is used only indexed by a cyclic counter. As entropy may not be
     well distributed from the beginning, 'md' is important as a
     chaining variable. However, the output function chains only half
     of 'md', i.e. 80 bits.  ssleay_rand_add, on the other hand, chains
     all of 'md', and seeding with STATE_SIZE dummy bytes will result
     in all of 'state' being rewritten, with the new values depending
     on virtually all of 'md'.  This overcomes the 80 bit limitation.)
     [Bodo Moeller]

  *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
     the handshake is continued after ssl_verify_cert_chain();
     otherwise, if SSL_VERIFY_NONE is set, remaining error codes
     can lead to 'unexplainable' connection aborts later.
     [Bodo Moeller; problem tracked down by Lutz Jaenicke]

  *) Major EVP API cipher revision.
     Add hooks for extra EVP features. This allows various cipher
     parameters to be set in the EVP interface. Support added for variable
     key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
     setting of RC2 and RC5 parameters.

     Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
     ciphers.

     Remove lots of duplicated code from the EVP library. For example *every*
     cipher init() function handles the 'iv' in the same way according to the
     cipher mode. They also all do nothing if the 'key' parameter is NULL and
     for CFB and OFB modes they zero ctx->num.

     New functionality allows removal of S/MIME code RC2 hack.

     Most of the routines have the same form and so can be declared in terms
     of macros.

     By shifting this to the top level EVP_CipherInit() it can be removed from
     all individual ciphers. If the cipher wants to handle IVs or keys
     differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
     flags.

     Change lots of functions like EVP_EncryptUpdate() to now return a
     value: although software versions of the algorithms cannot fail
     any installed hardware versions can.
     [Steve Henson]

  *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
     this option is set, tolerate broken clients that send the negotiated
     protocol version number instead of the requested protocol version
     number.
     [Bodo Moeller]

  *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
     i.e. non-zero for export ciphersuites, zero otherwise.
     Previous versions had this flag inverted, inconsistent with
     rsa_tmp_cb (..._TMP_RSA_CB).
     [Bodo Moeller; problem reported by Amit Chopra]

  *) Add missing DSA library text string. Work around for some IIS
     key files with invalid SEQUENCE encoding.
     [Steve Henson]

  *) Add a document (doc/standards.txt) that list all kinds of standards
     and so on that are implemented in OpenSSL.
     [Richard Levitte]

  *) Enhance c_rehash script. Old version would mishandle certificates
     with the same subject name hash and wouldn't handle CRLs at all.
     Added -fingerprint option to crl utility, to support new c_rehash
     features.
     [Steve Henson]

  *) Eliminate non-ANSI declarations in crypto.h and stack.h.
     [Ulf Möller]

  *) Fix for SSL server purpose checking. Server checking was
     rejecting certificates which had extended key usage present
     but no ssl client purpose.
     [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]

  *) Make PKCS#12 code work with no password. The PKCS#12 spec
     is a little unclear about how a blank password is handled.
     Since the password in encoded as a BMPString with terminating
     double NULL a zero length password would end up as just the
     double NULL. However no password at all is different and is
     handled differently in the PKCS#12 key generation code. NS
     treats a blank password as zero length. MSIE treats it as no
     password on export: but it will try both on import. We now do
     the same: PKCS12_parse() tries zero length and no password if
     the password is set to "" or NULL (NULL is now a valid password:
     it wasn't before) as does the pkcs12 application.
     [Steve Henson]

  *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
     perror when PEM_read_bio_X509_REQ fails, the error message must
     be obtained from the error queue.
     [Bodo Moeller]

  *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
     it in ERR_remove_state if appropriate, and change ERR_get_state
     accordingly to avoid race conditions (this is necessary because
     thread_hash is no longer constant once set).
     [Bodo Moeller]

  *) Bugfix for linux-elf makefile.one.
     [Ulf Möller]

  *) RSA_get_default_method() will now cause a default
     RSA_METHOD to be chosen if one doesn't exist already.
     Previously this was only set during a call to RSA_new()
     or RSA_new_method(NULL) meaning it was possible for
     RSA_get_default_method() to return NULL.
     [Geoff Thorpe]

  *) Added native name translation to the existing DSO code
     that will convert (if the flag to do so is set) filenames
     that are sufficiently small and have no path information
     into a canonical native form. Eg. "blah" converted to
     "libblah.so" or "blah.dll" etc.
     [Geoff Thorpe]

  *) New function ERR_error_string_n(e, buf, len) which is like
     ERR_error_string(e, buf), but writes at most 'len' bytes
     including the 0 terminator.  For ERR_error_string_n, 'buf'
     may not be NULL.
     [Damien Miller <djm@mindrot.org>, Bodo Moeller]

  *) CONF library reworked to become more general.  A new CONF
     configuration file reader "class" is implemented as well as a
     new functions (NCONF_*, for "New CONF") to handle it.  The now
     old CONF_* functions are still there, but are reimplemented to
     work in terms of the new functions.  Also, a set of functions
     to handle the internal storage of the configuration data is
     provided to make it easier to write new configuration file
     reader "classes" (I can definitely see something reading a
     configuration file in XML format, for example), called _CONF_*,
     or "the configuration storage API"...

     The new configuration file reading functions are:

        NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
        NCONF_get_section, NCONF_get_string, NCONF_get_numbre

        NCONF_default, NCONF_WIN32

        NCONF_dump_fp, NCONF_dump_bio

     NCONF_default and NCONF_WIN32 are method (or "class") choosers,
     NCONF_new creates a new CONF object.  This works in the same way
     as other interfaces in OpenSSL, like the BIO interface.
     NCONF_dump_* dump the internal storage of the configuration file,
     which is useful for debugging.  All other functions take the same
     arguments as the old CONF_* functions wth the exception of the
     first that must be a `CONF *' instead of a `LHASH *'.

     To make it easer to use the new classes with the old CONF_* functions,
     the function CONF_set_default_method is provided.
     [Richard Levitte]

  *) Add '-tls1' option to 'openssl ciphers', which was already
     mentioned in the documentation but had not been implemented.
     (This option is not yet really useful because even the additional
     experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
     [Bodo Moeller]

  *) Initial DSO code added into libcrypto for letting OpenSSL (and
     OpenSSL-based applications) load shared libraries and bind to
     them in a portable way.
     [Geoff Thorpe, with contributions from Richard Levitte]

 Changes between 0.9.5 and 0.9.5a  [1 Apr 2000]

  *) Make sure _lrotl and _lrotr are only used with MSVC.

  *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
     (the default implementation of RAND_status).

  *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
     to '-clrext' (= clear extensions), as intended and documented.
     [Bodo Moeller; inconsistency pointed out by Michael Attili
     <attili@amaxo.com>]

  *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
     was larger than the MD block size.      
     [Steve Henson, pointed out by Yost William <YostW@tce.com>]

  *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
     fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
     using the passed key: if the passed key was a private key the result
     of X509_print(), for example, would be to print out all the private key
     components.
     [Steve Henson]

  *) des_quad_cksum() byte order bug fix.
     [Ulf Möller, using the problem description in krb4-0.9.7, where
      the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]

  *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
     discouraged.
     [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]

  *) For easily testing in shell scripts whether some command
     'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
     returns with exit code 0 iff no command of the given name is available.
     'no-XXX' is printed in this case, 'XXX' otherwise.  In both cases,
     the output goes to stdout and nothing is printed to stderr.
     Additional arguments are always ignored.

     Since for each cipher there is a command of the same name,
     the 'no-cipher' compilation switches can be tested this way.

     ('openssl no-XXX' is not able to detect pseudo-commands such
     as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
     [Bodo Moeller]

  *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
     [Bodo Moeller]

  *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
     is set; it will be thrown away anyway because each handshake creates
     its own key.
     ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
     to parameters -- in previous versions (since OpenSSL 0.9.3) the
     'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
     you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
     [Bodo Moeller]

  *) New s_client option -ign_eof: EOF at stdin is ignored, and
     'Q' and 'R' lose their special meanings (quit/renegotiate).
     This is part of what -quiet does; unlike -quiet, -ign_eof
     does not suppress any output.
     [Richard Levitte]

  *) Add compatibility options to the purpose and trust code. The
     purpose X509_PURPOSE_ANY is "any purpose" which automatically
     accepts a certificate or CA, this was the previous behaviour,
     with all the associated security issues.

     X509_TRUST_COMPAT is the old trust behaviour: only and
     automatically trust self signed roots in certificate store. A
     new trust setting X509_TRUST_DEFAULT is used to specify that
     a purpose has no associated trust setting and it should instead
     use the value in the default purpose.
     [Steve Henson]

  *) Fix the PKCS#8 DSA private key code so it decodes keys again
     and fix a memory leak.
     [Steve Henson]

  *) In util/mkerr.pl (which implements 'make errors'), preserve
     reason strings from the previous version of the .c file, as
     the default to have only downcase letters (and digits) in
     automatically generated reasons codes is not always appropriate.
     [Bodo Moeller]

  *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
     using strerror.  Previously, ERR_reason_error_string() returned
     library names as reason strings for SYSerr; but SYSerr is a special
     case where small numbers are errno values, not library numbers.
     [Bodo Moeller]

  *) Add '-dsaparam' option to 'openssl dhparam' application.  This
     converts DSA parameters into DH parameters. (When creating parameters,
     DSA_generate_parameters is used.)
     [Bodo Moeller]

  *) Include 'length' (recommended exponent length) in C code generated
     by 'openssl dhparam -C'.
     [Bodo Moeller]

  *) The second argument to set_label in perlasm was already being used
     so couldn't be used as a "file scope" flag. Moved to third argument
     which was free.
     [Steve Henson]

  *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
     instead of RAND_bytes for encryption IVs and salts.
     [Bodo Moeller]

  *) Include RAND_status() into RAND_METHOD instead of implementing
     it only for md_rand.c  Otherwise replacing the PRNG by calling
     RAND_set_rand_method would be impossible.
     [Bodo Moeller]

  *) Don't let DSA_generate_key() enter an infinite loop if the random
     number generation fails.
     [Bodo Moeller]

  *) New 'rand' application for creating pseudo-random output.
     [Bodo Moeller]

  *) Added configuration support for Linux/IA64
     [Rolf Haberrecker <rolf@suse.de>]

  *) Assembler module support for Mingw32.
     [Ulf Möller]

  *) Shared library support for HPUX (in shlib/).
     [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]

  *) Shared library support for Solaris gcc.
     [Lutz Behnke <behnke@trustcenter.de>]

 Changes between 0.9.4 and 0.9.5  [28 Feb 2000]

  *) PKCS7_encrypt() was adding text MIME headers twice because they
     were added manually and by SMIME_crlf_copy().
     [Steve Henson]

  *) In bntest.c don't call BN_rand with zero bits argument.
     [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]

  *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
     case was implemented. This caused BN_div_recp() to fail occasionally.
     [Ulf Möller]

  *) Add an optional second argument to the set_label() in the perl
     assembly language builder. If this argument exists and is set
     to 1 it signals that the assembler should use a symbol whose 
     scope is the entire file, not just the current function. This
     is needed with MASM which uses the format label:: for this scope.
     [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]

  *) Change the ASN1 types so they are typedefs by default. Before
     almost all types were #define'd to ASN1_STRING which was causing
     STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
     for example.
     [Steve Henson]

  *) Change names of new functions to the new get1/get0 naming
     convention: After 'get1', the caller owns a reference count
     and has to call ..._free; 'get0' returns a pointer to some
     data structure without incrementing reference counters.
     (Some of the existing 'get' functions increment a reference
     counter, some don't.)
     Similarly, 'set1' and 'add1' functions increase reference
     counters or duplicate objects.
     [Steve Henson]

  *) Allow for the possibility of temp RSA key generation failure:
     the code used to assume it always worked and crashed on failure.
     [Steve Henson]

  *) Fix potential buffer overrun problem in BIO_printf().
     [Ulf Möller, using public domain code by Patrick Powell; problem
      pointed out by David Sacerdote <das33@cornell.edu>]

  *) Support EGD <http://www.lothar.com/tech/crypto/>.  New functions
     RAND_egd() and RAND_status().  In the command line application,
     the EGD socket can be specified like a seed file using RANDFILE
     or -rand.
     [Ulf Möller]

  *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
     Some CAs (e.g. Verisign) distribute certificates in this form.
     [Steve Henson]

  *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
     list to exclude them. This means that no special compilation option
     is needed to use anonymous DH: it just needs to be included in the
     cipher list.
     [Steve Henson]

  *) Change the EVP_MD_CTX_type macro so its meaning consistent with
     EVP_MD_type. The old functionality is available in a new macro called
     EVP_MD_md(). Change code that uses it and update docs.
     [Steve Henson]

  *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
     where the 'void *' argument is replaced by a function pointer argument.
     Previously 'void *' was abused to point to functions, which works on
     many platforms, but is not correct.  As these functions are usually
     called by macros defined in OpenSSL header files, most source code
     should work without changes.
     [Richard Levitte]

  *) <openssl/opensslconf.h> (which is created by Configure) now contains
     sections with information on -D... compiler switches used for
     compiling the library so that applications can see them.  To enable
     one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
     must be defined.  E.g.,
        #define OPENSSL_ALGORITHM_DEFINES
        #include <openssl/opensslconf.h>
     defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
     [Richard Levitte, Ulf and Bodo Möller]

  *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
     record layer.
     [Bodo Moeller]

  *) Change the 'other' type in certificate aux info to a STACK_OF
     X509_ALGOR. Although not an AlgorithmIdentifier as such it has
     the required ASN1 format: arbitrary types determined by an OID.
     [Steve Henson]

  *) Add some PEM_write_X509_REQ_NEW() functions and a command line
     argument to 'req'. This is not because the function is newer or
     better than others it just uses the work 'NEW' in the certificate
     request header lines. Some software needs this.
     [Steve Henson]

  *) Reorganise password command line arguments: now passwords can be
     obtained from various sources. Delete the PEM_cb function and make
     it the default behaviour: i.e. if the callback is NULL and the
     usrdata argument is not NULL interpret it as a null terminated pass
     phrase. If usrdata and the callback are NULL then the pass phrase
     is prompted for as usual.
     [Steve Henson]

  *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
     the support is automatically enabled. The resulting binaries will
     autodetect the card and use it if present.
     [Ben Laurie and Compaq Inc.]

  *) Work around for Netscape hang bug. This sends certificate request
     and server done in one record. Since this is perfectly legal in the
     SSL/TLS protocol it isn't a "bug" option and is on by default. See
     the bugs/SSLv3 entry for more info.
     [Steve Henson]

  *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
     [Andy Polyakov]

  *) Add -rand argument to smime and pkcs12 applications and read/write
     of seed file.
     [Steve Henson]

  *) New 'passwd' tool for crypt(3) and apr1 password hashes.
     [Bodo Moeller]

  *) Add command line password options to the remaining applications.
     [Steve Henson]

  *) Bug fix for BN_div_recp() for numerators with an even number of
     bits.
     [Ulf Möller]

  *) More tests in bntest.c, and changed test_bn output.
     [Ulf Möller]

  *) ./config recognizes MacOS X now.
     [Andy Polyakov]

  *) Bug fix for BN_div() when the first words of num and divsor are
     equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
     [Ulf Möller]

  *) Add support for various broken PKCS#8 formats, and command line
     options to produce them.
     [Steve Henson]

  *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
     get temporary BIGNUMs from a BN_CTX.
     [Ulf Möller]

  *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
     for p == 0.
     [Ulf Möller]

  *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
     include a #define from the old name to the new. The original intent
     was that statically linked binaries could for example just call
     SSLeay_add_all_ciphers() to just add ciphers to the table and not
     link with digests. This never worked becayse SSLeay_add_all_digests()
     and SSLeay_add_all_ciphers() were in the same source file so calling
     one would link with the other. They are now in separate source files.
     [Steve Henson]

  *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
     [Steve Henson]

  *) Use a less unusual form of the Miller-Rabin primality test (it used
     a binary algorithm for exponentiation integrated into the Miller-Rabin
     loop, our standard modexp algorithms are faster).
     [Bodo Moeller]

  *) Support for the EBCDIC character set completed.
     [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]

  *) Source code cleanups: use const where appropriate, eliminate casts,
     use void * instead of char * in lhash.
     [Ulf Möller] 

  *) Bugfix: ssl3_send_server_key_exchange was not restartable
     (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
     this the server could overwrite ephemeral keys that the client
     has already seen).
     [Bodo Moeller]

  *) Turn DSA_is_prime into a macro that calls BN_is_prime,
     using 50 iterations of the Rabin-Miller test.

     DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
     iterations of the Rabin-Miller test as required by the appendix
     to FIPS PUB 186[-1]) instead of DSA_is_prime.
     As BN_is_prime_fasttest includes trial division, DSA parameter
     generation becomes much faster.

     This implies a change for the callback functions in DSA_is_prime
     and DSA_generate_parameters: The callback function is called once
     for each positive witness in the Rabin-Miller test, not just
     occasionally in the inner loop; and the parameters to the
     callback function now provide an iteration count for the outer
     loop rather than for the current invocation of the inner loop.
     DSA_generate_parameters additionally can call the callback
     function with an 'iteration count' of -1, meaning that a
     candidate has passed the trial division test (when q is generated 
     from an application-provided seed, trial division is skipped).
     [Bodo Moeller]

  *) New function BN_is_prime_fasttest that optionally does trial
     division before starting the Rabin-Miller test and has
     an additional BN_CTX * argument (whereas BN_is_prime always
     has to allocate at least one BN_CTX).
     'callback(1, -1, cb_arg)' is called when a number has passed the
     trial division stage.
     [Bodo Moeller]

  *) Fix for bug in CRL encoding. The validity dates weren't being handled
     as ASN1_TIME.
     [Steve Henson]

  *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
     [Steve Henson]

  *) New function BN_pseudo_rand().
     [Ulf Möller]

  *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
     bignum version of BN_from_montgomery() with the working code from
     SSLeay 0.9.0 (the word based version is faster anyway), and clean up
     the comments.
     [Ulf Möller]

  *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
     made it impossible to use the same SSL_SESSION data structure in
     SSL2 clients in multiple threads.
     [Bodo Moeller]

  *) The return value of RAND_load_file() no longer counts bytes obtained
     by stat().  RAND_load_file(..., -1) is new and uses the complete file
     to seed the PRNG (previously an explicit byte count was required).
     [Ulf Möller, Bodo Möller]

  *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
     used (char *) instead of (void *) and had casts all over the place.
     [Steve Henson]

  *) Make BN_generate_prime() return NULL on error if ret!=NULL.
     [Ulf Möller]

  *) Retain source code compatibility for BN_prime_checks macro:
     BN_is_prime(..., BN_prime_checks, ...) now uses
     BN_prime_checks_for_size to determine the appropriate number of
     Rabin-Miller iterations.
     [Ulf Möller]

  *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
     DH_CHECK_P_NOT_SAFE_PRIME.
     (Check if this is true? OpenPGP calls them "strong".)
     [Ulf Möller]

  *) Merge the functionality of "dh" and "gendh" programs into a new program
     "dhparam". The old programs are retained for now but will handle DH keys
     (instead of parameters) in future.
     [Steve Henson]

  *) Make the ciphers, s_server and s_client programs check the return values
     when a new cipher list is set.
     [Steve Henson]

  *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
     ciphers. Before when the 56bit ciphers were enabled the sorting was
     wrong.

     The syntax for the cipher sorting has been extended to support sorting by
     cipher-strength (using the strength_bits hard coded in the tables).
     The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).

     Fix a bug in the cipher-command parser: when supplying a cipher command
     string with an "undefined" symbol (neither command nor alphanumeric
     [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
     an error is flagged.

     Due to the strength-sorting extension, the code of the
     ssl_create_cipher_list() function was completely rearranged. I hope that
     the readability was also increased :-)
     [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]

  *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
     for the first serial number and places 2 in the serial number file. This
     avoids problems when the root CA is created with serial number zero and
     the first user certificate has the same issuer name and serial number
     as the root CA.
     [Steve Henson]

  *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
     the new code. Add documentation for this stuff.
     [Steve Henson]

  *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
     X509_*() to X509at_*() on the grounds that they don't handle X509
     structures and behave in an analagous way to the X509v3 functions:
     they shouldn't be called directly but wrapper functions should be used
     instead.

     So we also now have some wrapper functions that call the X509at functions
     when passed certificate requests. (TO DO: similar things can be done with
     PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
     things. Some of these need some d2i or i2d and print functionality
     because they handle more complex structures.)
     [Steve Henson]

  *) Add missing #ifndefs that caused missing symbols when building libssl
     as a shared library without RSA.  Use #ifndef NO_SSL2 instead of
     NO_RSA in ssl/s2*.c. 
     [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]

  *) Precautions against using the PRNG uninitialized: RAND_bytes() now
     has a return value which indicates the quality of the random data
     (1 = ok, 0 = not seeded).  Also an error is recorded on the thread's
     error queue. New function RAND_pseudo_bytes() generates output that is
     guaranteed to be unique but not unpredictable. RAND_add is like
     RAND_seed, but takes an extra argument for an entropy estimate
     (RAND_seed always assumes full entropy).
     [Ulf Möller]

  *) Do more iterations of Rabin-Miller probable prime test (specifically,
     3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
     instead of only 2 for all lengths; see BN_prime_checks_for_size definition
     in crypto/bn/bn_prime.c for the complete table).  This guarantees a
     false-positive rate of at most 2^-80 for random input.
     [Bodo Moeller]

  *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
     [Bodo Moeller]

  *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
     in the 0.9.5 release), this returns the chain
     from an X509_CTX structure with a dup of the stack and all
     the X509 reference counts upped: so the stack will exist
     after X509_CTX_cleanup() has been called. Modify pkcs12.c
     to use this.

     Also make SSL_SESSION_print() print out the verify return
     code.
     [Steve Henson]

  *) Add manpage for the pkcs12 command. Also change the default
     behaviour so MAC iteration counts are used unless the new
     -nomaciter option is used. This improves file security and
     only older versions of MSIE (4.0 for example) need it.
     [Steve Henson]

  *) Honor the no-xxx Configure options when creating .DEF files.
     [Ulf Möller]

  *) Add PKCS#10 attributes to field table: challengePassword, 
     unstructuredName and unstructuredAddress. These are taken from
     draft PKCS#9 v2.0 but are compatible with v1.2 provided no 
     international characters are used.

     More changes to X509_ATTRIBUTE code: allow the setting of types
     based on strings. Remove the 'loc' parameter when adding
     attributes because these will be a SET OF encoding which is sorted
     in ASN1 order.
     [Steve Henson]

  *) Initial changes to the 'req' utility to allow request generation
     automation. This will allow an application to just generate a template
     file containing all the field values and have req construct the
     request.

     Initial support for X509_ATTRIBUTE handling. Stacks of these are
     used all over the place including certificate requests and PKCS#7
     structures. They are currently handled manually where necessary with
     some primitive wrappers for PKCS#7. The new functions behave in a
     manner analogous to the X509 extension functions: they allow
     attributes to be looked up by NID and added.

     Later something similar to the X509V3 code would be desirable to
     automatically handle the encoding, decoding and printing of the
     more complex types. The string types like challengePassword can
     be handled by the string table functions.

     Also modified the multi byte string table handling. Now there is
     a 'global mask' which masks out certain types. The table itself
     can use the flag STABLE_NO_MASK to ignore the mask setting: this
     is useful when for example there is only one permissible type
     (as in countryName) and using the mask might result in no valid
     types at all.
     [Steve Henson]

  *) Clean up 'Finished' handling, and add functions SSL_get_finished and
     SSL_get_peer_finished to allow applications to obtain the latest
     Finished messages sent to the peer or expected from the peer,
     respectively.  (SSL_get_peer_finished is usually the Finished message
     actually received from the peer, otherwise the protocol will be aborted.)

     As the Finished message are message digests of the complete handshake
     (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
     be used for external authentication procedures when the authentication
     provided by SSL/TLS is not desired or is not enough.
     [Bodo Moeller]

  *) Enhanced support for Alpha Linux is added. Now ./config checks if
     the host supports BWX extension and if Compaq C is present on the
     $PATH. Just exploiting of the BWX extension results in 20-30%
     performance kick for some algorithms, e.g. DES and RC4 to mention
     a couple. Compaq C in turn generates ~20% faster code for MD5 and
     SHA1.
     [Andy Polyakov]

  *) Add support for MS "fast SGC". This is arguably a violation of the
     SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
     weak crypto and after checking the certificate is SGC a second one
     with strong crypto. MS SGC stops the first handshake after receiving
     the server certificate message and sends a second client hello. Since
     a server will typically do all the time consuming operations before
     expecting any further messages from the client (server key exchange
     is the most expensive) there is little difference between the two.

     To get OpenSSL to support MS SGC we have to permit a second client
     hello message after we have sent server done. In addition we have to
     reset the MAC if we do get this second client hello.
     [Steve Henson]

  *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
     if a DER encoded private key is RSA or DSA traditional format. Changed
     d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
     format DER encoded private key. Newer code should use PKCS#8 format which
     has the key type encoded in the ASN1 structure. Added DER private key
     support to pkcs8 application.
     [Steve Henson]

  *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
     ciphersuites has been selected (as required by the SSL 3/TLS 1
     specifications).  Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
     is set, we interpret this as a request to violate the specification
     (the worst that can happen is a handshake failure, and 'correct'
     behaviour would result in a handshake failure anyway).
     [Bodo Moeller]

  *) In SSL_CTX_add_session, take into account that there might be multiple
     SSL_SESSION structures with the same session ID (e.g. when two threads
     concurrently obtain them from an external cache).
     The internal cache can handle only one SSL_SESSION with a given ID,
     so if there's a conflict, we now throw out the old one to achieve
     consistency.
     [Bodo Moeller]

  *) Add OIDs for idea and blowfish in CBC mode. This will allow both
     to be used in PKCS#5 v2.0 and S/MIME.  Also add checking to
     some routines that use cipher OIDs: some ciphers do not have OIDs
     defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
     example.
     [Steve Henson]

  *) Simplify the trust setting structure and code. Now we just have
     two sequences of OIDs for trusted and rejected settings. These will
     typically have values the same as the extended key usage extension
     and any application specific purposes.

     The trust checking code now has a default behaviour: it will just
     check for an object with the same NID as the passed id. Functions can
     be provided to override either the default behaviour or the behaviour
     for a given id. SSL client, server and email already have functions
     in place for compatibility: they check the NID and also return "trusted"
     if the certificate is self signed.
     [Steve Henson]

  *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
     traditional format into an EVP_PKEY structure.
     [Steve Henson]

  *) Add a password callback function PEM_cb() which either prompts for
     a password if usr_data is NULL or otherwise assumes it is a null
     terminated password. Allow passwords to be passed on command line
     environment or config files in a few more utilities.
     [Steve Henson]

  *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
     keys. Add some short names for PKCS#8 PBE algorithms and allow them
     to be specified on the command line for the pkcs8 and pkcs12 utilities.
     Update documentation.
     [Steve Henson]

  *) Support for ASN1 "NULL" type. This could be handled before by using
     ASN1_TYPE but there wasn't any function that would try to read a NULL
     and produce an error if it couldn't. For compatibility we also have
     ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
     don't allocate anything because they don't need to.
     [Steve Henson]

  *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
     for details.
     [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]

  *) Rebuild of the memory allocation routines used by OpenSSL code and
     possibly others as well.  The purpose is to make an interface that
     provide hooks so anyone can build a separate set of allocation and
     deallocation routines to be used by OpenSSL, for example memory
     pool implementations, or something else, which was previously hard
     since Malloc(), Realloc() and Free() were defined as macros having
     the values malloc, realloc and free, respectively (except for Win32
     compilations).  The same is provided for memory debugging code.
     OpenSSL already comes with functionality to find memory leaks, but
     this gives people a chance to debug other memory problems.

     With these changes, a new set of functions and macros have appeared:

       CRYPTO_set_mem_debug_functions()         [F]
       CRYPTO_get_mem_debug_functions()         [F]
       CRYPTO_dbg_set_options()                 [F]
       CRYPTO_dbg_get_options()                 [F]
       CRYPTO_malloc_debug_init()               [M]

     The memory debug functions are NULL by default, unless the library
     is compiled with CRYPTO_MDEBUG or friends is defined.  If someone
     wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
     gives the standard debugging functions that come with OpenSSL) or
     CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
     provided by the library user) must be used.  When the standard
     debugging functions are used, CRYPTO_dbg_set_options can be used to
     request additional information:
     CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
     the CRYPTO_MDEBUG_xxx macro when compiling the library.   

     Also, things like CRYPTO_set_mem_functions will always give the
     expected result (the new set of functions is used for allocation
     and deallocation) at all times, regardless of platform and compiler
     options.

     To finish it up, some functions that were never use in any other
     way than through macros have a new API and new semantic:

       CRYPTO_dbg_malloc()
       CRYPTO_dbg_realloc()
       CRYPTO_dbg_free()

     All macros of value have retained their old syntax.
     [Richard Levitte and Bodo Moeller]

  *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
     ordering of SMIMECapabilities wasn't in "strength order" and there
     was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
     algorithm.
     [Steve Henson]

  *) Some ASN1 types with illegal zero length encoding (INTEGER,
     ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
     [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]

  *) Merge in my S/MIME library for OpenSSL. This provides a simple
     S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
     functionality to handle multipart/signed properly) and a utility
     called 'smime' to call all this stuff. This is based on code I
     originally wrote for Celo who have kindly allowed it to be
     included in OpenSSL.
     [Steve Henson]

  *) Add variants des_set_key_checked and des_set_key_unchecked of
     des_set_key (aka des_key_sched).  Global variable des_check_key
     decides which of these is called by des_set_key; this way
     des_check_key behaves as it always did, but applications and
     the library itself, which was buggy for des_check_key == 1,
     have a cleaner way to pick the version they need.
     [Bodo Moeller]

  *) New function PKCS12_newpass() which changes the password of a
     PKCS12 structure.
     [Steve Henson]

  *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
     dynamic mix. In both cases the ids can be used as an index into the
     table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
     functions so they accept a list of the field values and the
     application doesn't need to directly manipulate the X509_TRUST
     structure.
     [Steve Henson]

  *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
     need initialising.
     [Steve Henson]

  *) Modify the way the V3 extension code looks up extensions. This now
     works in a similar way to the object code: we have some "standard"
     extensions in a static table which is searched with OBJ_bsearch()
     and the application can add dynamic ones if needed. The file
     crypto/x509v3/ext_dat.h now has the info: this file needs to be
     updated whenever a new extension is added to the core code and kept
     in ext_nid order. There is a simple program 'tabtest.c' which checks
     this. New extensions are not added too often so this file can readily
     be maintained manually.

     There are two big advantages in doing things this way. The extensions
     can be looked up immediately and no longer need to be "added" using
     X509V3_add_standard_extensions(): this function now does nothing.
     [Side note: I get *lots* of email saying the extension code doesn't
      work because people forget to call this function]
     Also no dynamic allocation is done unless new extensions are added:
     so if we don't add custom extensions there is no need to call
     X509V3_EXT_cleanup().
     [Steve Henson]

  *) Modify enc utility's salting as follows: make salting the default. Add a
     magic header, so unsalted files fail gracefully instead of just decrypting
     to garbage. This is because not salting is a big security hole, so people
     should be discouraged from doing it.
     [Ben Laurie]

  *) Fixes and enhancements to the 'x509' utility. It allowed a message
     digest to be passed on the command line but it only used this
     parameter when signing a certificate. Modified so all relevant
     operations are affected by the digest parameter including the
     -fingerprint and -x509toreq options. Also -x509toreq choked if a
     DSA key was used because it didn't fix the digest.
     [Steve Henson]

  *) Initial certificate chain verify code. Currently tests the untrusted
     certificates for consistency with the verify purpose (which is set
     when the X509_STORE_CTX structure is set up) and checks the pathlength.

     There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
     this is because it will reject chains with invalid extensions whereas
     every previous version of OpenSSL and SSLeay made no checks at all.

     Trust code: checks the root CA for the relevant trust settings. Trust
     settings have an initial value consistent with the verify purpose: e.g.
     if the verify purpose is for SSL client use it expects the CA to be
     trusted for SSL client use. However the default value can be changed to
     permit custom trust settings: one example of this would be to only trust
     certificates from a specific "secure" set of CAs.

     Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
     which should be used for version portability: especially since the
     verify structure is likely to change more often now.

     SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
     to set them. If not set then assume SSL clients will verify SSL servers
     and vice versa.

     Two new options to the verify program: -untrusted allows a set of
     untrusted certificates to be passed in and -purpose which sets the
     intended purpose of the certificate. If a purpose is set then the
     new chain verify code is used to check extension consistency.
     [Steve Henson]

  *) Support for the authority information access extension.
     [Steve Henson]

  *) Modify RSA and DSA PEM read routines to transparently handle
     PKCS#8 format private keys. New *_PUBKEY_* functions that handle
     public keys in a format compatible with certificate
     SubjectPublicKeyInfo structures. Unfortunately there were already
     functions called *_PublicKey_* which used various odd formats so
     these are retained for compatibility: however the DSA variants were
     never in a public release so they have been deleted. Changed dsa/rsa
     utilities to handle the new format: note no releases ever handled public
     keys so we should be OK.

     The primary motivation for this change is to avoid the same fiasco
     that dogs private keys: there are several incompatible private key
     formats some of which are standard and some OpenSSL specific and
     require various evil hacks to allow partial transparent handling and
     even then it doesn't work with DER formats. Given the option anything
     other than PKCS#8 should be dumped: but the other formats have to
     stay in the name of compatibility.

     With public keys and the benefit of hindsight one standard format 
     is used which works with EVP_PKEY, RSA or DSA structures: though
     it clearly returns an error if you try to read the wrong kind of key.

     Added a -pubkey option to the 'x509' utility to output the public key.
     Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
     (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
     EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
     that do the same as the EVP_PKEY_assign_*() except they up the
     reference count of the added key (they don't "swallow" the
     supplied key).
     [Steve Henson]

  *) Fixes to crypto/x509/by_file.c the code to read in certificates and
     CRLs would fail if the file contained no certificates or no CRLs:
     added a new function to read in both types and return the number
     read: this means that if none are read it will be an error. The
     DER versions of the certificate and CRL reader would always fail
     because it isn't possible to mix certificates and CRLs in DER format
     without choking one or the other routine. Changed this to just read
     a certificate: this is the best we can do. Also modified the code
     in apps/verify.c to take notice of return codes: it was previously
     attempting to read in certificates from NULL pointers and ignoring
     any errors: this is one reason why the cert and CRL reader seemed
     to work. It doesn't check return codes from the default certificate
     routines: these may well fail if the certificates aren't installed.
     [Steve Henson]

  *) Code to support otherName option in GeneralName.
     [Steve Henson]

  *) First update to verify code. Change the verify utility
     so it warns if it is passed a self signed certificate:
     for consistency with the normal behaviour. X509_verify
     has been modified to it will now verify a self signed
     certificate if *exactly* the same certificate appears
     in the store: it was previously impossible to trust a
     single self signed certificate. This means that:
     openssl verify ss.pem
     now gives a warning about a self signed certificate but
     openssl verify -CAfile ss.pem ss.pem
     is OK.
     [Steve Henson]

  *) For servers, store verify_result in SSL_SESSION data structure
     (and add it to external session representation).
     This is needed when client certificate verifications fails,
     but an application-provided verification callback (set by
     SSL_CTX_set_cert_verify_callback) allows accepting the session
     anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
     but returns 1): When the session is reused, we have to set
     ssl->verify_result to the appropriate error code to avoid
     security holes.
     [Bodo Moeller, problem pointed out by Lutz Jaenicke]

  *) Fix a bug in the new PKCS#7 code: it didn't consider the
     case in PKCS7_dataInit() where the signed PKCS7 structure
     didn't contain any existing data because it was being created.
     [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]

  *) Add a salt to the key derivation routines in enc.c. This
     forms the first 8 bytes of the encrypted file. Also add a
     -S option to allow a salt to be input on the command line.
     [Steve Henson]

  *) New function X509_cmp(). Oddly enough there wasn't a function
     to compare two certificates. We do this by working out the SHA1
     hash and comparing that. X509_cmp() will be needed by the trust
     code.
     [Steve Henson]

  *) SSL_get1_session() is like SSL_get_session(), but increments
     the reference count in the SSL_SESSION returned.
     [Geoff Thorpe <geoff@eu.c2.net>]

  *) Fix for 'req': it was adding a null to request attributes.
     Also change the X509_LOOKUP and X509_INFO code to handle
     certificate auxiliary information.
     [Steve Henson]

  *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
     the 'enc' command.
     [Steve Henson]

  *) Add the possibility to add extra information to the memory leak
     detecting output, to form tracebacks, showing from where each
     allocation was originated: CRYPTO_push_info("constant string") adds
     the string plus current file name and line number to a per-thread
     stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
     is like calling CYRPTO_pop_info() until the stack is empty.
     Also updated memory leak detection code to be multi-thread-safe.
     [Richard Levitte]

  *) Add options -text and -noout to pkcs7 utility and delete the
     encryption options which never did anything. Update docs.
     [Steve Henson]

  *) Add options to some of the utilities to allow the pass phrase
     to be included on either the command line (not recommended on
     OSes like Unix) or read from the environment. Update the
     manpages and fix a few bugs.
     [Steve Henson]

  *) Add a few manpages for some of the openssl commands.
     [Steve Henson]

  *) Fix the -revoke option in ca. It was freeing up memory twice,
     leaking and not finding already revoked certificates.
     [Steve Henson]

  *) Extensive changes to support certificate auxiliary information.
     This involves the use of X509_CERT_AUX structure and X509_AUX
     functions. An X509_AUX function such as PEM_read_X509_AUX()
     can still read in a certificate file in the usual way but it
     will also read in any additional "auxiliary information". By
     doing things this way a fair degree of compatibility can be
     retained: existing certificates can have this information added
     using the new 'x509' options. 

     Current auxiliary information includes an "alias" and some trust
     settings. The trust settings will ultimately be used in enhanced
     certificate chain verification routines: currently a certificate
     can only be trusted if it is self signed and then it is trusted
     for all purposes.
     [Steve Henson]

  *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
     The problem was that one of the replacement routines had not been working
     since SSLeay releases.  For now the offending routine has been replaced
     with non-optimised assembler.  Even so, this now gives around 95%
     performance improvement for 1024 bit RSA signs.
     [Mark Cox]

  *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2 
     handling. Most clients have the effective key size in bits equal to
     the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
     A few however don't do this and instead use the size of the decrypted key
     to determine the RC2 key length and the AlgorithmIdentifier to determine
     the effective key length. In this case the effective key length can still
     be 40 bits but the key length can be 168 bits for example. This is fixed
     by manually forcing an RC2 key into the EVP_PKEY structure because the
     EVP code can't currently handle unusual RC2 key sizes: it always assumes
     the key length and effective key length are equal.
     [Steve Henson]

  *) Add a bunch of functions that should simplify the creation of 
     X509_NAME structures. Now you should be able to do:
     X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
     and have it automatically work out the correct field type and fill in
     the structures. The more adventurous can try:
     X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
     and it will (hopefully) work out the correct multibyte encoding.
     [Steve Henson]

  *) Change the 'req' utility to use the new field handling and multibyte
     copy routines. Before the DN field creation was handled in an ad hoc
     way in req, ca, and x509 which was rather broken and didn't support
     BMPStrings or UTF8Strings. Since some software doesn't implement
     BMPStrings or UTF8Strings yet, they can be enabled using the config file
     using the dirstring_type option. See the new comment in the default
     openssl.cnf for more info.
     [Steve Henson]

  *) Make crypto/rand/md_rand.c more robust:
     - Assure unique random numbers after fork().
     - Make sure that concurrent threads access the global counter and
       md serializably so that we never lose entropy in them
       or use exactly the same state in multiple threads.
       Access to the large state is not always serializable because
       the additional locking could be a performance killer, and
       md should be large enough anyway.
     [Bodo Moeller]

  *) New file apps/app_rand.c with commonly needed functionality
     for handling the random seed file.

     Use the random seed file in some applications that previously did not:
          ca,
          dsaparam -genkey (which also ignored its '-rand' option), 
          s_client,
          s_server,
          x509 (when signing).
     Except on systems with /dev/urandom, it is crucial to have a random
     seed file at least for key creation, DSA signing, and for DH exchanges;
     for RSA signatures we could do without one.

     gendh and gendsa (unlike genrsa) used to read only the first byte
     of each file listed in the '-rand' option.  The function as previously
     found in genrsa is now in app_rand.c and is used by all programs
     that support '-rand'.
     [Bodo Moeller]

  *) In RAND_write_file, use mode 0600 for creating files;
     don't just chmod when it may be too late.
     [Bodo Moeller]

  *) Report an error from X509_STORE_load_locations
     when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
     [Bill Perry]

  *) New function ASN1_mbstring_copy() this copies a string in either
     ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
     into an ASN1_STRING type. A mask of permissible types is passed
     and it chooses the "minimal" type to use or an error if not type
     is suitable.
     [Steve Henson]

  *) Add function equivalents to the various macros in asn1.h. The old
     macros are retained with an M_ prefix. Code inside the library can
     use the M_ macros. External code (including the openssl utility)
     should *NOT* in order to be "shared library friendly".
     [Steve Henson]

  *) Add various functions that can check a certificate's extensions
     to see if it usable for various purposes such as SSL client,
     server or S/MIME and CAs of these types. This is currently 
     VERY EXPERIMENTAL but will ultimately be used for certificate chain
     verification. Also added a -purpose flag to x509 utility to
     print out all the purposes.
     [Steve Henson]

  *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
     functions.
     [Steve Henson]

  *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
     for, obtain and decode and extension and obtain its critical flag.
     This allows all the necessary extension code to be handled in a
     single function call.
     [Steve Henson]

  *) RC4 tune-up featuring 30-40% performance improvement on most RISC
     platforms. See crypto/rc4/rc4_enc.c for further details.
     [Andy Polyakov]

  *) New -noout option to asn1parse. This causes no output to be produced
     its main use is when combined with -strparse and -out to extract data
     from a file (which may not be in ASN.1 format).
     [Steve Henson]

  *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
     when producing the local key id.
     [Richard Levitte <levitte@stacken.kth.se>]

  *) New option -dhparam in s_server. This allows a DH parameter file to be
     stated explicitly. If it is not stated then it tries the first server
     certificate file. The previous behaviour hard coded the filename
     "server.pem".
     [Steve Henson]

  *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
     a public key to be input or output. For example:
     openssl rsa -in key.pem -pubout -out pubkey.pem
     Also added necessary DSA public key functions to handle this.
     [Steve Henson]

  *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
     in the message. This was handled by allowing
     X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
     [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]

  *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
     to the end of the strings whereas this didn't. This would cause problems
     if strings read with d2i_ASN1_bytes() were later modified.
     [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]

  *) Fix for base64 decode bug. When a base64 bio reads only one line of
     data and it contains EOF it will end up returning an error. This is
     caused by input 46 bytes long. The cause is due to the way base64
     BIOs find the start of base64 encoded data. They do this by trying a
     trial decode on each line until they find one that works. When they
     do a flag is set and it starts again knowing it can pass all the
     data directly through the decoder. Unfortunately it doesn't reset
     the context it uses. This means that if EOF is reached an attempt
     is made to pass two EOFs through the context and this causes the
     resulting error. This can also cause other problems as well. As is
     usual with these problems it takes *ages* to find and the fix is
     trivial: move one line.
     [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]

  *) Ugly workaround to get s_client and s_server working under Windows. The
     old code wouldn't work because it needed to select() on sockets and the
     tty (for keypresses and to see if data could be written). Win32 only
     supports select() on sockets so we select() with a 1s timeout on the
     sockets and then see if any characters are waiting to be read, if none
     are present then we retry, we also assume we can always write data to
     the tty. This isn't nice because the code then blocks until we've
     received a complete line of data and it is effectively polling the
     keyboard at 1s intervals: however it's quite a bit better than not
     working at all :-) A dedicated Windows application might handle this
     with an event loop for example.
     [Steve Henson]

  *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
     and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
     will be called when RSA_sign() and RSA_verify() are used. This is useful
     if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
     For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
     should *not* be used: RSA_sign() and RSA_verify() must be used instead.
     This necessitated the support of an extra signature type NID_md5_sha1
     for SSL signatures and modifications to the SSL library to use it instead
     of calling RSA_public_decrypt() and RSA_private_encrypt().
     [Steve Henson]

  *) Add new -verify -CAfile and -CApath options to the crl program, these
     will lookup a CRL issuers certificate and verify the signature in a
     similar way to the verify program. Tidy up the crl program so it
     no longer accesses structures directly. Make the ASN1 CRL parsing a bit
     less strict. It will now permit CRL extensions even if it is not
     a V2 CRL: this will allow it to tolerate some broken CRLs.
     [Steve Henson]

  *) Initialize all non-automatic variables each time one of the openssl
     sub-programs is started (this is necessary as they may be started
     multiple times from the "OpenSSL>" prompt).
     [Lennart Bang, Bodo Moeller]

  *) Preliminary compilation option RSA_NULL which disables RSA crypto without
     removing all other RSA functionality (this is what NO_RSA does). This
     is so (for example) those in the US can disable those operations covered
     by the RSA patent while allowing storage and parsing of RSA keys and RSA
     key generation.
     [Steve Henson]

  *) Non-copying interface to BIO pairs.
     (still largely untested)
     [Bodo Moeller]

  *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
     ASCII string. This was handled independently in various places before.
     [Steve Henson]

  *) New functions UTF8_getc() and UTF8_putc() that parse and generate
     UTF8 strings a character at a time.
     [Steve Henson]

  *) Use client_version from client hello to select the protocol
     (s23_srvr.c) and for RSA client key exchange verification
     (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
     [Bodo Moeller]

  *) Add various utility functions to handle SPKACs, these were previously
     handled by poking round in the structure internals. Added new function
     NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
     print, verify and generate SPKACs. Based on an original idea from
     Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
     [Steve Henson]

  *) RIPEMD160 is operational on all platforms and is back in 'make test'.
     [Andy Polyakov]

  *) Allow the config file extension section to be overwritten on the
     command line. Based on an original idea from Massimiliano Pala
     <madwolf@comune.modena.it>. The new option is called -extensions
     and can be applied to ca, req and x509. Also -reqexts to override
     the request extensions in req and -crlexts to override the crl extensions
     in ca.
     [Steve Henson]

  *) Add new feature to the SPKAC handling in ca.  Now you can include
     the same field multiple times by preceding it by "XXXX." for example:
     1.OU="Unit name 1"
     2.OU="Unit name 2"
     this is the same syntax as used in the req config file.
     [Steve Henson]

  *) Allow certificate extensions to be added to certificate requests. These
     are specified in a 'req_extensions' option of the req section of the
     config file. They can be printed out with the -text option to req but
     are otherwise ignored at present.
     [Steve Henson]

  *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
     data read consists of only the final block it would not decrypted because
     EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
     A misplaced 'break' also meant the decrypted final block might not be
     copied until the next read.
     [Steve Henson]

  *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
     a few extra parameters to the DH structure: these will be useful if
     for example we want the value of 'q' or implement X9.42 DH.
     [Steve Henson]

  *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
     provides hooks that allow the default DSA functions or functions on a
     "per key" basis to be replaced. This allows hardware acceleration and
     hardware key storage to be handled without major modification to the
     library. Also added low level modexp hooks and CRYPTO_EX structure and 
     associated functions.
     [Steve Henson]

  *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
     as "read only": it can't be written to and the buffer it points to will
     not be freed. Reading from a read only BIO is much more efficient than
     a normal memory BIO. This was added because there are several times when
     an area of memory needs to be read from a BIO. The previous method was
     to create a memory BIO and write the data to it, this results in two
     copies of the data and an O(n^2) reading algorithm. There is a new
     function BIO_new_mem_buf() which creates a read only memory BIO from
     an area of memory. Also modified the PKCS#7 routines to use read only
     memory BIOs.
     [Steve Henson]

  *) Bugfix: ssl23_get_client_hello did not work properly when called in
     state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
     a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
     but a retry condition occured while trying to read the rest.
     [Bodo Moeller]

  *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
     NID_pkcs7_encrypted by default: this was wrong since this should almost
     always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
     the encrypted data type: this is a more sensible place to put it and it
     allows the PKCS#12 code to be tidied up that duplicated this
     functionality.
     [Steve Henson]

  *) Changed obj_dat.pl script so it takes its input and output files on
     the command line. This should avoid shell escape redirection problems
     under Win32.
     [Steve Henson]

  *) Initial support for certificate extension requests, these are included
     in things like Xenroll certificate requests. Included functions to allow
     extensions to be obtained and added.
     [Steve Henson]

  *) -crlf option to s_client and s_server for sending newlines as
     CRLF (as required by many protocols).
     [Bodo Moeller]

 Changes between 0.9.3a and 0.9.4  [09 Aug 1999]
  
  *) Install libRSAglue.a when OpenSSL is built with RSAref.
     [Ralf S. Engelschall]

  *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
     [Andrija Antonijevic <TheAntony2@bigfoot.com>]

  *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
     program.
     [Steve Henson]

  *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
     DH parameters/keys (q is lost during that conversion, but the resulting
     DH parameters contain its length).

     For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
     much faster than DH_generate_parameters (which creates parameters
     where p = 2*q + 1), and also the smaller q makes DH computations
     much more efficient (160-bit exponentiation instead of 1024-bit
     exponentiation); so this provides a convenient way to support DHE
     ciphersuites in SSL/TLS servers (see ssl/ssltest.c).  It is of
     utter importance to use
         SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
     or
         SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
     when such DH parameters are used, because otherwise small subgroup
     attacks may become possible!
     [Bodo Moeller]

  *) Avoid memory leak in i2d_DHparams.
     [Bodo Moeller]

  *) Allow the -k option to be used more than once in the enc program:
     this allows the same encrypted message to be read by multiple recipients.
     [Steve Henson]

  *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
     an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
     it will always use the numerical form of the OID, even if it has a short
     or long name.
     [Steve Henson]

  *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
     method only got called if p,q,dmp1,dmq1,iqmp components were present,
     otherwise bn_mod_exp was called. In the case of hardware keys for example
     no private key components need be present and it might store extra data
     in the RSA structure, which cannot be accessed from bn_mod_exp.
     By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
     private key operations.
     [Steve Henson]

  *) Added support for SPARC Linux.
     [Andy Polyakov]

  *) pem_password_cb function type incompatibly changed from
          typedef int pem_password_cb(char *buf, int size, int rwflag);
     to
          ....(char *buf, int size, int rwflag, void *userdata);
     so that applications can pass data to their callbacks:
     The PEM[_ASN1]_{read,write}... functions and macros now take an
     additional void * argument, which is just handed through whenever
     the password callback is called.
     [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]

     New function SSL_CTX_set_default_passwd_cb_userdata.

     Compatibility note: As many C implementations push function arguments
     onto the stack in reverse order, the new library version is likely to
     interoperate with programs that have been compiled with the old
     pem_password_cb definition (PEM_whatever takes some data that
     happens to be on the stack as its last argument, and the callback
     just ignores this garbage); but there is no guarantee whatsoever that
     this will work.