CHANGES

     [Ulf Moeller]

  *) 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]

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

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

  *) 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]

  *) 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]

  *) 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]

  *) 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 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]

  *) 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]

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

  *) 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>]

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

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

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

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

  *) rand_win.c fix for Borland C.
     [Ulf Möller]

  *) BN_rshift bugfix for n == 0.
     [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]

  *) 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]

  *) 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]

  *) 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 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]

  *) 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 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 disabled 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 presenting 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 outputting
     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 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 instead. 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, meaning
     you effectively 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>]