CHANGES

 OpenSSL CHANGES
 _______________

 Changes between 0.9.3a and 0.9.4  [xx Jul/Aug/...? 1999]

  *) 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>, with 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.

  *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
     (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
     problems not only on Windows, but also on some Unix platforms.
     To avoid problematic command lines, these definitions are now in an
     auto-generated file, which also defines DATE and thus replaces
     crypto/date.h.  Standard "make" builds use crypto/buildinf.h,
     mk1mf.pl builds use crypto/mk1mfinf.h (created by mk1mf.pl).
     [Bodo Moeller]

  *) MIPS III/IV assembler module is reimplemented.
     [Andy Polyakov]

  *) More DES library cleanups: remove references to srand/rand and
     delete an unused file.
     [Ulf Möller]

  *) Add support for the the free Netwide assembler (NASM) under Win32,
     since not many people have MASM (ml) and it can be hard to obtain.
     This is currently experimental but it seems to work OK and pass all
     the tests. Check out INSTALL.W32 for info.
     [Steve Henson]

  *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
     without temporary keys kept an extra copy of the server key,
     and connections with temporary keys did not free everything in case
     of an error.
     [Bodo Moeller]

  *) New function RSA_check_key and new openssl rsa option -check
     for verifying the consistency of RSA keys.
     [Ulf Moeller, Bodo Moeller]

  *) Various changes to make Win32 compile work: 
     1. Casts to avoid "loss of data" warnings in p5_crpt2.c
     2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
        comparison" warnings.
     3. Add sk_<TYPE>_sort to DEF file generator and do make update.
     [Steve Henson]

  *) Add a debugging option to PKCS#5 v2 key generation function: when
     you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
     derived keys are printed to stderr.
     [Steve Henson]

  *) Copy the flags in ASN1_STRING_dup().
     [Roman E. Pavlov <pre@mo.msk.ru>]

  *) The x509 application mishandled signing requests containing DSA
     keys when the signing key was also DSA and the parameters didn't match.

     It was supposed to omit the parameters when they matched the signing key:
     the verifying software was then supposed to automatically use the CA's
     parameters if they were absent from the end user certificate.

     Omitting parameters is no longer recommended. The test was also
     the wrong way round! This was probably due to unusual behaviour in
     EVP_cmp_parameters() which returns 1 if the parameters match. 
     This meant that parameters were omitted when they *didn't* match and
     the certificate was useless. Certificates signed with 'ca' didn't have
     this bug.
     [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]

  *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
     The interface is as follows:
     Applications can use
         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
     "off" is now the default.
     The library internally uses
         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
         CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
     to disable memory-checking temporarily.

     Some inconsistent states that previously were possible (and were
     even the default) are now avoided.

     -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
     with each memory chunk allocated; this is occasionally more helpful
     than just having a counter.

     -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.

     -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
     extensions.
     [Bodo Moeller]

  *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
     which largely parallels "options", but is for changing API behaviour,
     whereas "options" are about protocol behaviour.
     Initial "mode" flags are:

     SSL_MODE_ENABLE_PARTIAL_WRITE   Allow SSL_write to report success when
                                     a single record has been written.
     SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER  Don't insist that SSL_write
                                     retries use the same buffer location.
                                     (But all of the contents must be
                                     copied!)
     [Bodo Moeller]

  *) Bugfix: SSL_set_mode ignored its parameter, only SSL_CTX_set_mode
     worked.

  *) Fix problems with no-hmac etc.
     [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]

  *) New functions RSA_get_default_method(), RSA_set_method() and
     RSA_get_method(). These allows replacement of RSA_METHODs without having
     to mess around with the internals of an RSA structure.
     [Steve Henson]

  *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
     Also really enable memory leak checks in openssl.c and in some
     test programs.
     [Chad C. Mulligan, Bodo Moeller]

  *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
     up the length of negative integers. This has now been simplified to just
     store the length when it is first determined and use it later, rather
     than trying to keep track of where data is copied and updating it to
     point to the end.
     [Steve Henson, reported by Brien Wheeler
      <bwheeler@authentica-security.com>]

  *) Add a new function PKCS7_signatureVerify. This allows the verification
     of a PKCS#7 signature but with the signing certificate passed to the
     function itself. This contrasts with PKCS7_dataVerify which assumes the
     certificate is present in the PKCS#7 structure. This isn't always the
     case: certificates can be omitted from a PKCS#7 structure and be
     distributed by "out of band" means (such as a certificate database).
     [Steve Henson]

  *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
     function prototypes in pem.h, also change util/mkdef.pl to add the
     necessary function names. 
     [Steve Henson]

  *) mk1mf.pl (used by Windows builds) did not properly read the
     options set by Configure in the top level Makefile, and Configure
     was not even able to write more than one option correctly.
     Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
     [Bodo Moeller]

  *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
     file to be loaded from a BIO or FILE pointer. The BIO version will
     for example allow memory BIOs to contain config info.
     [Steve Henson]

  *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
     Whoever hopes to achieve shared-library compatibility across versions
     must use this, not the compile-time macro.
     (Exercise 0.9.4: Which is the minimum library version required by
     such programs?)
     Note: All this applies only to multi-threaded programs, others don't
     need locks.
     [Bodo Moeller]