5 Changes between 0.9.7i and 0.9.7j [XX xxx XXXX]
7 *) Add new Windows build target VC-32-GMAKE for VC++. This uses GNU make
8 from a Windows bash shell such as MSYS. It is autodetected from the
9 "config" script when run from a VC++ environment. Modify standard VC++
10 build to use fipscanister.o from the GNU make build.
13 Changes between 0.9.7h and 0.9.7i [14 Oct 2005]
15 *) Wrapped the definition of EVP_MAX_MD_SIZE in a #ifdef OPENSSL_FIPS.
16 The value now differs depending on if you build for FIPS or not.
17 BEWARE! A program linked with a shared FIPSed libcrypto can't be
18 safely run with a non-FIPSed libcrypto, as it may crash because of
19 the difference induced by this change.
22 Changes between 0.9.7g and 0.9.7h [11 Oct 2005]
24 *) Remove the functionality of SSL_OP_MSIE_SSLV2_RSA_PADDING
25 (part of SSL_OP_ALL). This option used to disable the
26 countermeasure against man-in-the-middle protocol-version
27 rollback in the SSL 2.0 server implementation, which is a bad
30 [Bodo Moeller; problem pointed out by Yutaka Oiwa (Research Center
31 for Information Security, National Institute of Advanced Industrial
32 Science and Technology [AIST], Japan)]
34 *) Minimal support for X9.31 signatures and PSS padding modes. This is
35 mainly for FIPS compliance and not fully integrated at this stage.
38 *) For DSA signing, unless DSA_FLAG_NO_EXP_CONSTTIME is set, perform
39 the exponentiation using a fixed-length exponent. (Otherwise,
40 the information leaked through timing could expose the secret key
41 after many signatures; cf. Bleichenbacher's attack on DSA with
45 *) Make a new fixed-window mod_exp implementation the default for
46 RSA, DSA, and DH private-key operations so that the sequence of
47 squares and multiplies and the memory access pattern are
48 independent of the particular secret key. This will mitigate
49 cache-timing and potential related attacks.
51 BN_mod_exp_mont_consttime() is the new exponentiation implementation,
52 and this is automatically used by BN_mod_exp_mont() if the new flag
53 BN_FLG_EXP_CONSTTIME is set for the exponent. RSA, DSA, and DH
54 will use this BN flag for private exponents unless the flag
55 RSA_FLAG_NO_EXP_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME, or
56 DH_FLAG_NO_EXP_CONSTTIME, respectively, is set.
58 [Matthew D Wood (Intel Corp), with some changes by Bodo Moeller]
60 *) Change the client implementation for SSLv23_method() and
61 SSLv23_client_method() so that is uses the SSL 3.0/TLS 1.0
62 Client Hello message format if the SSL_OP_NO_SSLv2 option is set.
63 (Previously, the SSL 2.0 backwards compatible Client Hello
64 message format would be used even with SSL_OP_NO_SSLv2.)
67 *) Add support for smime-type MIME parameter in S/MIME messages which some
71 *) New function BN_MONT_CTX_set_locked() to set montgomery parameters in
72 a threadsafe manner. Modify rsa code to use new function and add calls
73 to dsa and dh code (which had race conditions before).
76 *) Include the fixed error library code in the C error file definitions
77 instead of fixing them up at runtime. This keeps the error code
81 Changes between 0.9.7f and 0.9.7g [11 Apr 2005]
83 [NB: OpenSSL 0.9.7h and later 0.9.7 patch levels were released after
86 *) Fixes for newer kerberos headers. NB: the casts are needed because
87 the 'length' field is signed on one version and unsigned on another
88 with no (?) obvious way to tell the difference, without these VC++
89 complains. Also the "definition" of FAR (blank) is no longer included
90 nor is the error ENOMEM. KRB5_PRIVATE has to be set to 1 to pick up
91 some needed definitions.
94 *) Undo Cygwin change.
97 *) Added support for proxy certificates according to RFC 3820.
98 Because they may be a security thread to unaware applications,
99 they must be explicitely allowed in run-time. See
100 docs/HOWTO/proxy_certificates.txt for further information.
103 Changes between 0.9.7e and 0.9.7f [22 Mar 2005]
105 *) Use (SSL_RANDOM_VALUE - 4) bytes of pseudo random data when generating
106 server and client random values. Previously
107 (SSL_RANDOM_VALUE - sizeof(time_t)) would be used which would result in
108 less random data when sizeof(time_t) > 4 (some 64 bit platforms).
110 This change has negligible security impact because:
112 1. Server and client random values still have 24 bytes of pseudo random
115 2. Server and client random values are sent in the clear in the initial
118 3. The master secret is derived using the premaster secret (48 bytes in
119 size for static RSA ciphersuites) as well as client server and random
122 The OpenSSL team would like to thank the UK NISCC for bringing this issue
125 [Stephen Henson, reported by UK NISCC]
127 *) Use Windows randomness collection on Cygwin.
130 *) Fix hang in EGD/PRNGD query when communication socket is closed
131 prematurely by EGD/PRNGD.
132 [Darren Tucker <dtucker@zip.com.au> via Lutz Jänicke, resolves #1014]
134 *) Prompt for pass phrases when appropriate for PKCS12 input format.
137 *) Back-port of selected performance improvements from development
138 branch, as well as improved support for PowerPC platforms.
141 *) Add lots of checks for memory allocation failure, error codes to indicate
142 failure and freeing up memory if a failure occurs.
143 [Nauticus Networks SSL Team <openssl@nauticusnet.com>, Steve Henson]
145 *) Add new -passin argument to dgst.
148 *) Perform some character comparisons of different types in X509_NAME_cmp:
149 this is needed for some certificates that reencode DNs into UTF8Strings
150 (in violation of RFC3280) and can't or wont issue name rollover
154 *) Make an explicit check during certificate validation to see that
155 the CA setting in each certificate on the chain is correct. As a
156 side effect always do the following basic checks on extensions,
157 not just when there's an associated purpose to the check:
159 - if there is an unhandled critical extension (unless the user
160 has chosen to ignore this fault)
161 - if the path length has been exceeded (if one is set at all)
162 - that certain extensions fit the associated purpose (if one has
166 Changes between 0.9.7d and 0.9.7e [25 Oct 2004]
168 *) Avoid a race condition when CRLs are checked in a multi threaded
169 environment. This would happen due to the reordering of the revoked
170 entries during signature checking and serial number lookup. Now the
171 encoding is cached and the serial number sort performed under a lock.
172 Add new STACK function sk_is_sorted().
175 *) Add Delta CRL to the extension code.
178 *) Various fixes to s3_pkt.c so alerts are sent properly.
179 [David Holmes <d.holmes@f5.com>]
181 *) Reduce the chances of duplicate issuer name and serial numbers (in
182 violation of RFC3280) using the OpenSSL certificate creation utilities.
183 This is done by creating a random 64 bit value for the initial serial
184 number when a serial number file is created or when a self signed
185 certificate is created using 'openssl req -x509'. The initial serial
186 number file is created using 'openssl x509 -next_serial' in CA.pl
187 rather than being initialized to 1.
190 Changes between 0.9.7c and 0.9.7d [17 Mar 2004]
192 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
193 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
194 [Joe Orton, Steve Henson]
196 *) Fix flaw in SSL/TLS handshaking when using Kerberos ciphersuites
198 [Joe Orton, Steve Henson]
200 *) Make it possible to have multiple active certificates with the same
201 subject in the CA index file. This is done only if the keyword
202 'unique_subject' is set to 'no' in the main CA section (default
203 if 'CA_default') of the configuration file. The value is saved
204 with the database itself in a separate index attribute file,
205 named like the index file with '.attr' appended to the name.
208 *) X509 verify fixes. Disable broken certificate workarounds when
209 X509_V_FLAGS_X509_STRICT is set. Check CRL issuer has cRLSign set if
210 keyUsage extension present. Don't accept CRLs with unhandled critical
211 extensions: since verify currently doesn't process CRL extensions this
212 rejects a CRL with *any* critical extensions. Add new verify error codes
216 *) When creating an OCSP nonce use an OCTET STRING inside the extnValue.
217 A clarification of RFC2560 will require the use of OCTET STRINGs and
218 some implementations cannot handle the current raw format. Since OpenSSL
219 copies and compares OCSP nonces as opaque blobs without any attempt at
220 parsing them this should not create any compatibility issues.
223 *) New md flag EVP_MD_CTX_FLAG_REUSE this allows md_data to be reused when
224 calling EVP_MD_CTX_copy_ex() to avoid calling OPENSSL_malloc(). Without
225 this HMAC (and other) operations are several times slower than OpenSSL
229 *) Print out GeneralizedTime and UTCTime in ASN1_STRING_print_ex().
230 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
232 *) Use the correct content when signing type "other".
235 Changes between 0.9.7b and 0.9.7c [30 Sep 2003]
237 *) Fix various bugs revealed by running the NISCC test suite:
239 Stop out of bounds reads in the ASN1 code when presented with
240 invalid tags (CVE-2003-0543 and CVE-2003-0544).
242 Free up ASN1_TYPE correctly if ANY type is invalid (CVE-2003-0545).
244 If verify callback ignores invalid public key errors don't try to check
245 certificate signature with the NULL public key.
249 *) New -ignore_err option in ocsp application to stop the server
250 exiting on the first error in a request.
253 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
254 if the server requested one: as stated in TLS 1.0 and SSL 3.0
258 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
259 extra data after the compression methods not only for TLS 1.0
260 but also for SSL 3.0 (as required by the specification).
261 [Bodo Moeller; problem pointed out by Matthias Loepfe]
263 *) Change X509_certificate_type() to mark the key as exported/exportable
264 when it's 512 *bits* long, not 512 bytes.
267 *) Change AES_cbc_encrypt() so it outputs exact multiple of
268 blocks during encryption.
271 *) Various fixes to base64 BIO and non blocking I/O. On write
272 flushes were not handled properly if the BIO retried. On read
273 data was not being buffered properly and had various logic bugs.
274 This also affects blocking I/O when the data being decoded is a
278 *) Various S/MIME bugfixes and compatibility changes:
279 output correct application/pkcs7 MIME type if
280 PKCS7_NOOLDMIMETYPE is set. Tolerate some broken signatures.
281 Output CR+LF for EOL if PKCS7_CRLFEOL is set (this makes opening
282 of files as .eml work). Correctly handle very long lines in MIME
286 Changes between 0.9.7a and 0.9.7b [10 Apr 2003]
288 *) Countermeasure against the Klima-Pokorny-Rosa extension of
289 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
290 a protocol version number mismatch like a decryption error
291 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
294 *) Turn on RSA blinding by default in the default implementation
295 to avoid a timing attack. Applications that don't want it can call
296 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
297 They would be ill-advised to do so in most cases.
298 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
300 *) Change RSA blinding code so that it works when the PRNG is not
301 seeded (in this case, the secret RSA exponent is abused as
302 an unpredictable seed -- if it is not unpredictable, there
303 is no point in blinding anyway). Make RSA blinding thread-safe
304 by remembering the creator's thread ID in rsa->blinding and
305 having all other threads use local one-time blinding factors
306 (this requires more computation than sharing rsa->blinding, but
307 avoids excessive locking; and if an RSA object is not shared
308 between threads, blinding will still be very fast).
311 *) Fixed a typo bug that would cause ENGINE_set_default() to set an
312 ENGINE as defaults for all supported algorithms irrespective of
313 the 'flags' parameter. 'flags' is now honoured, so applications
314 should make sure they are passing it correctly.
317 *) Target "mingw" now allows native Windows code to be generated in
318 the Cygwin environment as well as with the MinGW compiler.
321 Changes between 0.9.7 and 0.9.7a [19 Feb 2003]
323 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
324 via timing by performing a MAC computation even if incorrrect
325 block cipher padding has been found. This is a countermeasure
326 against active attacks where the attacker has to distinguish
327 between bad padding and a MAC verification error. (CVE-2003-0078)
329 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
330 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
331 Martin Vuagnoux (EPFL, Ilion)]
333 *) Make the no-err option work as intended. The intention with no-err
334 is not to have the whole error stack handling routines removed from
335 libcrypto, it's only intended to remove all the function name and
336 reason texts, thereby removing some of the footprint that may not
337 be interesting if those errors aren't displayed anyway.
339 NOTE: it's still possible for any application or module to have it's
340 own set of error texts inserted. The routines are there, just not
341 used by default when no-err is given.
344 *) Add support for FreeBSD on IA64.
345 [dirk.meyer@dinoex.sub.org via Richard Levitte, resolves #454]
347 *) Adjust DES_cbc_cksum() so it returns the same value as the MIT
348 Kerberos function mit_des_cbc_cksum(). Before this change,
349 the value returned by DES_cbc_cksum() was like the one from
350 mit_des_cbc_cksum(), except the bytes were swapped.
351 [Kevin Greaney <Kevin.Greaney@hp.com> and Richard Levitte]
353 *) Allow an application to disable the automatic SSL chain building.
354 Before this a rather primitive chain build was always performed in
355 ssl3_output_cert_chain(): an application had no way to send the
356 correct chain if the automatic operation produced an incorrect result.
358 Now the chain builder is disabled if either:
360 1. Extra certificates are added via SSL_CTX_add_extra_chain_cert().
362 2. The mode flag SSL_MODE_NO_AUTO_CHAIN is set.
364 The reasoning behind this is that an application would not want the
365 auto chain building to take place if extra chain certificates are
366 present and it might also want a means of sending no additional
367 certificates (for example the chain has two certificates and the
371 *) Add the possibility to build without the ENGINE framework.
372 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
374 *) Under Win32 gmtime() can return NULL: check return value in
375 OPENSSL_gmtime(). Add error code for case where gmtime() fails.
378 *) DSA routines: under certain error conditions uninitialized BN objects
379 could be freed. Solution: make sure initialization is performed early
380 enough. (Reported and fix supplied by Ivan D Nestlerode <nestler@MIT.EDU>,
381 Nils Larsch <nla@trustcenter.de> via PR#459)
384 *) Another fix for SSLv2 session ID handling: the session ID was incorrectly
385 checked on reconnect on the client side, therefore session resumption
386 could still fail with a "ssl session id is different" error. This
387 behaviour is masked when SSL_OP_ALL is used due to
388 SSL_OP_MICROSOFT_SESS_ID_BUG being set.
389 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
393 *) IA-32 assembler support enhancements: unified ELF targets, support
394 for SCO/Caldera platforms, fix for Cygwin shared build.
397 *) Add support for FreeBSD on sparc64. As a consequence, support for
398 FreeBSD on non-x86 processors is separate from x86 processors on
399 the config script, much like the NetBSD support.
400 [Richard Levitte & Kris Kennaway <kris@obsecurity.org>]
402 Changes between 0.9.6h and 0.9.7 [31 Dec 2002]
404 [NB: OpenSSL 0.9.6i and later 0.9.6 patch levels were released after
407 *) Fix session ID handling in SSLv2 client code: the SERVER FINISHED
408 code (06) was taken as the first octet of the session ID and the last
409 octet was ignored consequently. As a result SSLv2 client side session
410 caching could not have worked due to the session ID mismatch between
412 Behaviour observed by Crispin Flowerday <crispin@flowerday.cx> as
416 *) Change the declaration of needed Kerberos libraries to use EX_LIBS
417 instead of the special (and badly supported) LIBKRB5. LIBKRB5 is
421 *) The hw_ncipher.c engine requires dynamic locks. Unfortunately, it
422 seems that in spite of existing for more than a year, many application
423 author have done nothing to provide the necessary callbacks, which
424 means that this particular engine will not work properly anywhere.
425 This is a very unfortunate situation which forces us, in the name
426 of usability, to give the hw_ncipher.c a static lock, which is part
428 NOTE: This is for the 0.9.7 series ONLY. This hack will never
429 appear in 0.9.8 or later. We EXPECT application authors to have
430 dealt properly with this when 0.9.8 is released (unless we actually
431 make such changes in the libcrypto locking code that changes will
432 have to be made anyway).
435 *) In asn1_d2i_read_bio() repeatedly call BIO_read() until all content
436 octets have been read, EOF or an error occurs. Without this change
437 some truncated ASN1 structures will not produce an error.
440 *) Disable Heimdal support, since it hasn't been fully implemented.
441 Still give the possibility to force the use of Heimdal, but with
442 warnings and a request that patches get sent to openssl-dev.
445 *) Add the VC-CE target, introduce the WINCE sysname, and add
446 INSTALL.WCE and appropriate conditionals to make it build.
447 [Steven Reddie <smr@essemer.com.au> via Richard Levitte]
449 *) Change the DLL names for Cygwin to cygcrypto-x.y.z.dll and
450 cygssl-x.y.z.dll, where x, y and z are the major, minor and
451 edit numbers of the version.
452 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
454 *) Introduce safe string copy and catenation functions
455 (BUF_strlcpy() and BUF_strlcat()).
456 [Ben Laurie (CHATS) and Richard Levitte]
458 *) Avoid using fixed-size buffers for one-line DNs.
461 *) Add BUF_MEM_grow_clean() to avoid information leakage when
462 resizing buffers containing secrets, and use where appropriate.
465 *) Avoid using fixed size buffers for configuration file location.
468 *) Avoid filename truncation for various CA files.
471 *) Use sizeof in preference to magic numbers.
474 *) Avoid filename truncation in cert requests.
477 *) Add assertions to check for (supposedly impossible) buffer
481 *) Don't cache truncated DNS entries in the local cache (this could
482 potentially lead to a spoofing attack).
485 *) Fix various buffers to be large enough for hex/decimal
486 representations in a platform independent manner.
489 *) Add CRYPTO_realloc_clean() to avoid information leakage when
490 resizing buffers containing secrets, and use where appropriate.
493 *) Add BIO_indent() to avoid much slightly worrying code to do
497 *) Convert sprintf()/BIO_puts() to BIO_printf().
500 *) buffer_gets() could terminate with the buffer only half
504 *) Add assertions to prevent user-supplied crypto functions from
505 overflowing internal buffers by having large block sizes, etc.
508 *) New OPENSSL_assert() macro (similar to assert(), but enabled
512 *) Eliminate unused copy of key in RC4.
515 *) Eliminate unused and incorrectly sized buffers for IV in pem.h.
518 *) Fix off-by-one error in EGD path.
521 *) If RANDFILE path is too long, ignore instead of truncating.
524 *) Eliminate unused and incorrectly sized X.509 structure
528 *) Eliminate unused and dangerous function knumber().
531 *) Eliminate unused and dangerous structure, KSSL_ERR.
534 *) Protect against overlong session ID context length in an encoded
535 session object. Since these are local, this does not appear to be
539 *) Change from security patch (see 0.9.6e below) that did not affect
540 the 0.9.6 release series:
542 Remote buffer overflow in SSL3 protocol - an attacker could
543 supply an oversized master key in Kerberos-enabled versions.
547 *) Change the SSL kerb5 codes to match RFC 2712.
550 *) Make -nameopt work fully for req and add -reqopt switch.
551 [Michael Bell <michael.bell@rz.hu-berlin.de>, Steve Henson]
553 *) The "block size" for block ciphers in CFB and OFB mode should be 1.
554 [Steve Henson, reported by Yngve Nysaeter Pettersen <yngve@opera.com>]
556 *) Make sure tests can be performed even if the corresponding algorithms
557 have been removed entirely. This was also the last step to make
558 OpenSSL compilable with DJGPP under all reasonable conditions.
559 [Richard Levitte, Doug Kaufman <dkaufman@rahul.net>]
561 *) Add cipher selection rules COMPLEMENTOFALL and COMPLEMENTOFDEFAULT
562 to allow version independent disabling of normally unselected ciphers,
563 which may be activated as a side-effect of selecting a single cipher.
565 (E.g., cipher list string "RSA" enables ciphersuites that are left
566 out of "ALL" because they do not provide symmetric encryption.
567 "RSA:!COMPLEMEMENTOFALL" avoids these unsafe ciphersuites.)
568 [Lutz Jaenicke, Bodo Moeller]
570 *) Add appropriate support for separate platform-dependent build
571 directories. The recommended way to make a platform-dependent
572 build directory is the following (tested on Linux), maybe with
575 # Place yourself outside of the OpenSSL source tree. In
576 # this example, the environment variable OPENSSL_SOURCE
577 # is assumed to contain the absolute OpenSSL source directory.
578 mkdir -p objtree/"`uname -s`-`uname -r`-`uname -m`"
579 cd objtree/"`uname -s`-`uname -r`-`uname -m`"
580 (cd $OPENSSL_SOURCE; find . -type f) | while read F; do
581 mkdir -p `dirname $F`
582 ln -s $OPENSSL_SOURCE/$F $F
585 To be absolutely sure not to disturb the source tree, a "make clean"
586 is a good thing. If it isn't successfull, don't worry about it,
587 it probably means the source directory is very clean.
590 *) Make sure any ENGINE control commands make local copies of string
591 pointers passed to them whenever necessary. Otherwise it is possible
592 the caller may have overwritten (or deallocated) the original string
593 data when a later ENGINE operation tries to use the stored values.
594 [Götz Babin-Ebell <babinebell@trustcenter.de>]
596 *) Improve diagnostics in file reading and command-line digests.
597 [Ben Laurie aided and abetted by Solar Designer <solar@openwall.com>]
599 *) Add AES modes CFB and OFB to the object database. Correct an
600 error in AES-CFB decryption.
603 *) Remove most calls to EVP_CIPHER_CTX_cleanup() in evp_enc.c, this
604 allows existing EVP_CIPHER_CTX structures to be reused after
605 calling EVP_*Final(). This behaviour is used by encryption
606 BIOs and some applications. This has the side effect that
607 applications must explicitly clean up cipher contexts with
608 EVP_CIPHER_CTX_cleanup() or they will leak memory.
611 *) Check the values of dna and dnb in bn_mul_recursive before calling
612 bn_mul_comba (a non zero value means the a or b arrays do not contain
613 n2 elements) and fallback to bn_mul_normal if either is not zero.
616 *) Fix escaping of non-ASCII characters when using the -subj option
617 of the "openssl req" command line tool. (Robert Joop <joop@fokus.gmd.de>)
620 *) Make object definitions compliant to LDAP (RFC2256): SN is the short
621 form for "surname", serialNumber has no short form.
622 Use "mail" as the short name for "rfc822Mailbox" according to RFC2798;
623 therefore remove "mail" short name for "internet 7".
624 The OID for unique identifiers in X509 certificates is
625 x500UniqueIdentifier, not uniqueIdentifier.
626 Some more OID additions. (Michael Bell <michael.bell@rz.hu-berlin.de>)
629 *) Add an "init" command to the ENGINE config module and auto initialize
630 ENGINEs. Without any "init" command the ENGINE will be initialized
631 after all ctrl commands have been executed on it. If init=1 the
632 ENGINE is initailized at that point (ctrls before that point are run
633 on the uninitialized ENGINE and after on the initialized one). If
634 init=0 then the ENGINE will not be iniatialized at all.
637 *) Fix the 'app_verify_callback' interface so that the user-defined
638 argument is actually passed to the callback: In the
639 SSL_CTX_set_cert_verify_callback() prototype, the callback
640 declaration has been changed from
643 int (*cb)(X509_STORE_CTX *,void *);
644 in ssl_verify_cert_chain (ssl/ssl_cert.c), the call
645 i=s->ctx->app_verify_callback(&ctx)
646 has been changed into
647 i=s->ctx->app_verify_callback(&ctx, s->ctx->app_verify_arg).
649 To update applications using SSL_CTX_set_cert_verify_callback(),
650 a dummy argument can be added to their callback functions.
651 [D. K. Smetters <smetters@parc.xerox.com>]
653 *) Added the '4758cca' ENGINE to support IBM 4758 cards.
654 [Maurice Gittens <maurice@gittens.nl>, touchups by Geoff Thorpe]
656 *) Add and OPENSSL_LOAD_CONF define which will cause
657 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
658 This allows older applications to transparently support certain
659 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
660 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
661 load the config file and OPENSSL_add_all_algorithms_conf() which will
662 always load it have also been added.
665 *) Add the OFB, CFB and CTR (all with 128 bit feedback) to AES.
666 Adjust NIDs and EVP layer.
667 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
669 *) Config modules support in openssl utility.
671 Most commands now load modules from the config file,
672 though in a few (such as version) this isn't done
673 because it couldn't be used for anything.
675 In the case of ca and req the config file used is
676 the same as the utility itself: that is the -config
677 command line option can be used to specify an
681 *) Move default behaviour from OPENSSL_config(). If appname is NULL
682 use "openssl_conf" if filename is NULL use default openssl config file.
685 *) Add an argument to OPENSSL_config() to allow the use of an alternative
686 config section name. Add a new flag to tolerate a missing config file
687 and move code to CONF_modules_load_file().
690 *) Support for crypto accelerator cards from Accelerated Encryption
691 Processing, www.aep.ie. (Use engine 'aep')
692 The support was copied from 0.9.6c [engine] and adapted/corrected
693 to work with the new engine framework.
694 [AEP Inc. and Richard Levitte]
696 *) Support for SureWare crypto accelerator cards from Baltimore
697 Technologies. (Use engine 'sureware')
698 The support was copied from 0.9.6c [engine] and adapted
699 to work with the new engine framework.
702 *) Have the CHIL engine fork-safe (as defined by nCipher) and actually
703 make the newer ENGINE framework commands for the CHIL engine work.
704 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
706 *) Make it possible to produce shared libraries on ReliantUNIX.
707 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
709 *) Add the configuration target debug-linux-ppro.
710 Make 'openssl rsa' use the general key loading routines
711 implemented in apps.c, and make those routines able to
712 handle the key format FORMAT_NETSCAPE and the variant
714 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
716 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
717 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
719 *) Add -keyform to rsautl, and document -engine.
720 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
722 *) Change BIO_new_file (crypto/bio/bss_file.c) to use new
723 BIO_R_NO_SUCH_FILE error code rather than the generic
724 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
729 ERR_peek_last_error_line
730 ERR_peek_last_error_line_data.
734 ERR_peek_error_line_data,
735 but report on the latest error recorded rather than the first one
736 still in the error queue.
737 [Ben Laurie, Bodo Moeller]
739 *) default_algorithms option in ENGINE config module. This allows things
741 default_algorithms = ALL
742 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
745 *) Prelminary ENGINE config module.
748 *) New experimental application configuration code.
751 *) Change the AES code to follow the same name structure as all other
752 symmetric ciphers, and behave the same way. Move everything to
753 the directory crypto/aes, thereby obsoleting crypto/rijndael.
754 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
756 *) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
757 [Ben Laurie and Theo de Raadt]
759 *) Add option to output public keys in req command.
760 [Massimiliano Pala madwolf@openca.org]
762 *) Use wNAFs in EC_POINTs_mul() for improved efficiency
763 (up to about 10% better than before for P-192 and P-224).
766 *) New functions/macros
768 SSL_CTX_set_msg_callback(ctx, cb)
769 SSL_CTX_set_msg_callback_arg(ctx, arg)
770 SSL_set_msg_callback(ssl, cb)
771 SSL_set_msg_callback_arg(ssl, arg)
773 to request calling a callback function
775 void cb(int write_p, int version, int content_type,
776 const void *buf, size_t len, SSL *ssl, void *arg)
778 whenever a protocol message has been completely received
779 (write_p == 0) or sent (write_p == 1). Here 'version' is the
780 protocol version according to which the SSL library interprets
781 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
782 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
783 the content type as defined in the SSL 3.0/TLS 1.0 protocol
784 specification (change_cipher_spec(20), alert(21), handshake(22)).
785 'buf' and 'len' point to the actual message, 'ssl' to the
786 SSL object, and 'arg' is the application-defined value set by
787 SSL[_CTX]_set_msg_callback_arg().
789 'openssl s_client' and 'openssl s_server' have new '-msg' options
790 to enable a callback that displays all protocol messages.
793 *) Change the shared library support so shared libraries are built as
794 soon as the corresponding static library is finished, and thereby get
795 openssl and the test programs linked against the shared library.
796 This still only happens when the keyword "shard" has been given to
797 the configuration scripts.
799 NOTE: shared library support is still an experimental thing, and
800 backward binary compatibility is still not guaranteed.
801 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
803 *) Add support for Subject Information Access extension.
804 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
806 *) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
807 additional bytes when new memory had to be allocated, not just
808 when reusing an existing buffer.
811 *) New command line and configuration option 'utf8' for the req command.
812 This allows field values to be specified as UTF8 strings.
815 *) Add -multi and -mr options to "openssl speed" - giving multiple parallel
816 runs for the former and machine-readable output for the latter.
819 *) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
820 of the e-mail address in the DN (i.e., it will go into a certificate
821 extension only). The new configuration file option 'email_in_dn = no'
823 [Massimiliano Pala madwolf@openca.org]
825 *) Change all functions with names starting with des_ to be starting
826 with DES_ instead. Add wrappers that are compatible with libdes,
827 but are named _ossl_old_des_*. Finally, add macros that map the
828 des_* symbols to the corresponding _ossl_old_des_* if libdes
829 compatibility is desired. If OpenSSL 0.9.6c compatibility is
830 desired, the des_* symbols will be mapped to DES_*, with one
833 Since we provide two compatibility mappings, the user needs to
834 define the macro OPENSSL_DES_LIBDES_COMPATIBILITY if libdes
835 compatibility is desired. The default (i.e., when that macro
836 isn't defined) is OpenSSL 0.9.6c compatibility.
838 There are also macros that enable and disable the support of old
839 des functions altogether. Those are OPENSSL_ENABLE_OLD_DES_SUPPORT
840 and OPENSSL_DISABLE_OLD_DES_SUPPORT. If none or both of those
841 are defined, the default will apply: to support the old des routines.
843 In either case, one must include openssl/des.h to get the correct
844 definitions. Do not try to just include openssl/des_old.h, that
847 NOTE: This is a major break of an old API into a new one. Software
848 authors are encouraged to switch to the DES_ style functions. Some
849 time in the future, des_old.h and the libdes compatibility functions
850 will be disable (i.e. OPENSSL_DISABLE_OLD_DES_SUPPORT will be the
851 default), and then completely removed.
854 *) Test for certificates which contain unsupported critical extensions.
855 If such a certificate is found during a verify operation it is
856 rejected by default: this behaviour can be overridden by either
857 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
858 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
859 X509_supported_extension() has also been added which returns 1 if a
860 particular extension is supported.
863 *) Modify the behaviour of EVP cipher functions in similar way to digests
864 to retain compatibility with existing code.
867 *) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
868 compatibility with existing code. In particular the 'ctx' parameter does
869 not have to be to be initialized before the call to EVP_DigestInit() and
870 it is tidied up after a call to EVP_DigestFinal(). New function
871 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
872 EVP_MD_CTX_copy() changed to not require the destination to be
873 initialized valid and new function EVP_MD_CTX_copy_ex() added which
874 requires the destination to be valid.
876 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
877 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
880 *) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
881 so that complete 'Handshake' protocol structures are kept in memory
882 instead of overwriting 'msg_type' and 'length' with 'body' data.
885 *) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
886 [Massimo Santin via Richard Levitte]
888 *) Major restructuring to the underlying ENGINE code. This includes
889 reduction of linker bloat, separation of pure "ENGINE" manipulation
890 (initialisation, etc) from functionality dealing with implementations
891 of specific crypto iterfaces. This change also introduces integrated
892 support for symmetric ciphers and digest implementations - so ENGINEs
893 can now accelerate these by providing EVP_CIPHER and EVP_MD
894 implementations of their own. This is detailed in crypto/engine/README
895 as it couldn't be adequately described here. However, there are a few
896 API changes worth noting - some RSA, DSA, DH, and RAND functions that
897 were changed in the original introduction of ENGINE code have now
898 reverted back - the hooking from this code to ENGINE is now a good
899 deal more passive and at run-time, operations deal directly with
900 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
901 dereferencing through an ENGINE pointer any more. Also, the ENGINE
902 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
903 they were not being used by the framework as there is no concept of a
904 BIGNUM_METHOD and they could not be generalised to the new
905 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
906 ENGINE_cpy() has been removed as it cannot be consistently defined in
910 *) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
913 *) Change mkdef.pl to sort symbols that get the same entry number,
914 and make sure the automatically generated functions ERR_load_*
915 become part of libeay.num as well.
918 *) New function SSL_renegotiate_pending(). This returns true once
919 renegotiation has been requested (either SSL_renegotiate() call
920 or HelloRequest/ClientHello receveived from the peer) and becomes
921 false once a handshake has been completed.
922 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
923 sends a HelloRequest, but does not ensure that a handshake takes
924 place. SSL_renegotiate_pending() is useful for checking if the
925 client has followed the request.)
928 *) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
929 By default, clients may request session resumption even during
930 renegotiation (if session ID contexts permit); with this option,
931 session resumption is possible only in the first handshake.
933 SSL_OP_ALL is now 0x00000FFFL instead of 0x000FFFFFL. This makes
934 more bits available for options that should not be part of
935 SSL_OP_ALL (such as SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION).
938 *) Add some demos for certificate and certificate request creation.
941 *) Make maximum certificate chain size accepted from the peer application
942 settable (SSL*_get/set_max_cert_list()), as proposed by
943 "Douglas E. Engert" <deengert@anl.gov>.
946 *) Add support for shared libraries for Unixware-7
947 (Boyd Lynn Gerber <gerberb@zenez.com>).
950 *) Add a "destroy" handler to ENGINEs that allows structural cleanup to
951 be done prior to destruction. Use this to unload error strings from
952 ENGINEs that load their own error strings. NB: This adds two new API
953 functions to "get" and "set" this destroy handler in an ENGINE.
956 *) Alter all existing ENGINE implementations (except "openssl" and
957 "openbsd") to dynamically instantiate their own error strings. This
958 makes them more flexible to be built both as statically-linked ENGINEs
959 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
960 Also, add stub code to each that makes building them as self-contained
961 shared-libraries easier (see README.ENGINE).
964 *) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
965 implementations into applications that are completely implemented in
966 self-contained shared-libraries. The "dynamic" ENGINE exposes control
967 commands that can be used to configure what shared-library to load and
968 to control aspects of the way it is handled. Also, made an update to
969 the README.ENGINE file that brings its information up-to-date and
970 provides some information and instructions on the "dynamic" ENGINE
971 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
974 *) Make it possible to unload ranges of ERR strings with a new
975 "ERR_unload_strings" function.
978 *) Add a copy() function to EVP_MD.
981 *) Make EVP_MD routines take a context pointer instead of just the
982 md_data void pointer.
985 *) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
986 that the digest can only process a single chunk of data
987 (typically because it is provided by a piece of
988 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
989 is only going to provide a single chunk of data, and hence the
990 framework needn't accumulate the data for oneshot drivers.
993 *) As with "ERR", make it possible to replace the underlying "ex_data"
994 functions. This change also alters the storage and management of global
995 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
996 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
997 index counters. The API functions that use this state have been changed
998 to take a "class_index" rather than pointers to the class's local STACK
999 and counter, and there is now an API function to dynamically create new
1000 classes. This centralisation allows us to (a) plug a lot of the
1001 thread-safety problems that existed, and (b) makes it possible to clean
1002 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
1003 such data would previously have always leaked in application code and
1004 workarounds were in place to make the memory debugging turn a blind eye
1005 to it. Application code that doesn't use this new function will still
1006 leak as before, but their memory debugging output will announce it now
1007 rather than letting it slide.
1009 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
1010 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
1011 has a return value to indicate success or failure.
1014 *) Make it possible to replace the underlying "ERR" functions such that the
1015 global state (2 LHASH tables and 2 locks) is only used by the "default"
1016 implementation. This change also adds two functions to "get" and "set"
1017 the implementation prior to it being automatically set the first time
1018 any other ERR function takes place. Ie. an application can call "get",
1019 pass the return value to a module it has just loaded, and that module
1020 can call its own "set" function using that value. This means the
1021 module's "ERR" operations will use (and modify) the error state in the
1022 application and not in its own statically linked copy of OpenSSL code.
1025 *) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
1026 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
1027 the operation, and provides a more encapsulated way for external code
1028 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
1029 to use these functions rather than manually incrementing the counts.
1031 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
1034 *) Add EVP test program.
1037 *) Add symmetric cipher support to ENGINE. Expect the API to change!
1040 *) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
1041 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
1042 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
1043 These allow a CRL to be built without having to access X509_CRL fields
1044 directly. Modify 'ca' application to use new functions.
1047 *) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
1048 bug workarounds. Rollback attack detection is a security feature.
1049 The problem will only arise on OpenSSL servers when TLSv1 is not
1050 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
1051 Software authors not wanting to support TLSv1 will have special reasons
1052 for their choice and can explicitly enable this option.
1053 [Bodo Moeller, Lutz Jaenicke]
1055 *) Rationalise EVP so it can be extended: don't include a union of
1056 cipher/digest structures, add init/cleanup functions for EVP_MD_CTX
1057 (similar to those existing for EVP_CIPHER_CTX).
1062 EVP_MD_CTX_init(&md); /* new function call */
1063 EVP_DigestInit(&md, EVP_sha1());
1064 EVP_DigestUpdate(&md, in, len);
1065 EVP_DigestFinal(&md, out, NULL);
1066 EVP_MD_CTX_cleanup(&md); /* new function call */
1070 *) Make DES key schedule conform to the usual scheme, as well as
1071 correcting its structure. This means that calls to DES functions
1072 now have to pass a pointer to a des_key_schedule instead of a
1073 plain des_key_schedule (which was actually always a pointer
1076 des_key_schedule ks;
1078 des_set_key_checked(..., &ks);
1079 des_ncbc_encrypt(..., &ks, ...);
1081 (Note that a later change renames 'des_...' into 'DES_...'.)
1084 *) Initial reduction of linker bloat: the use of some functions, such as
1085 PEM causes large amounts of unused functions to be linked in due to
1086 poor organisation. For example pem_all.c contains every PEM function
1087 which has a knock on effect of linking in large amounts of (unused)
1088 ASN1 code. Grouping together similar functions and splitting unrelated
1089 functions prevents this.
1092 *) Cleanup of EVP macros.
1095 *) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
1096 correct _ecb suffix.
1099 *) Add initial OCSP responder support to ocsp application. The
1100 revocation information is handled using the text based index
1101 use by the ca application. The responder can either handle
1102 requests generated internally, supplied in files (for example
1103 via a CGI script) or using an internal minimal server.
1106 *) Add configuration choices to get zlib compression for TLS.
1109 *) Changes to Kerberos SSL for RFC 2712 compliance:
1110 1. Implemented real KerberosWrapper, instead of just using
1111 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
1112 2. Implemented optional authenticator field of KerberosWrapper.
1114 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
1115 and authenticator structs; see crypto/krb5/.
1117 Generalized Kerberos calls to support multiple Kerberos libraries.
1118 [Vern Staats <staatsvr@asc.hpc.mil>,
1119 Jeffrey Altman <jaltman@columbia.edu>
1120 via Richard Levitte]
1122 *) Cause 'openssl speed' to use fully hard-coded DSA keys as it
1123 already does with RSA. testdsa.h now has 'priv_key/pub_key'
1124 values for each of the key sizes rather than having just
1125 parameters (and 'speed' generating keys each time).
1128 *) Speed up EVP routines.
1131 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
1132 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
1133 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
1134 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
1136 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
1137 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
1138 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
1141 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
1143 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
1146 *) Added the OS2-EMX target.
1147 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
1149 *) Rewrite apps to use NCONF routines instead of the old CONF. New functions
1150 to support NCONF routines in extension code. New function CONF_set_nconf()
1151 to allow functions which take an NCONF to also handle the old LHASH
1152 structure: this means that the old CONF compatible routines can be
1153 retained (in particular wrt extensions) without having to duplicate the
1154 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
1157 *) Enhance the general user interface with mechanisms for inner control
1158 and with possibilities to have yes/no kind of prompts.
1161 *) Change all calls to low level digest routines in the library and
1162 applications to use EVP. Add missing calls to HMAC_cleanup() and
1163 don't assume HMAC_CTX can be copied using memcpy().
1164 [Verdon Walker <VWalker@novell.com>, Steve Henson]
1166 *) Add the possibility to control engines through control names but with
1167 arbitrary arguments instead of just a string.
1168 Change the key loaders to take a UI_METHOD instead of a callback
1169 function pointer. NOTE: this breaks binary compatibility with earlier
1170 versions of OpenSSL [engine].
1171 Adapt the nCipher code for these new conditions and add a card insertion
1175 *) Enhance the general user interface with mechanisms to better support
1176 dialog box interfaces, application-defined prompts, the possibility
1177 to use defaults (for example default passwords from somewhere else)
1178 and interrupts/cancellations.
1181 *) Tidy up PKCS#12 attribute handling. Add support for the CSP name
1182 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
1185 *) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
1186 tidy up some unnecessarily weird code in 'sk_new()').
1187 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
1189 *) Change the key loading routines for ENGINEs to use the same kind
1190 callback (pem_password_cb) as all other routines that need this
1194 *) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
1195 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
1196 than this minimum value is recommended.
1199 *) New random seeder for OpenVMS, using the system process statistics
1200 that are easily reachable.
1203 *) Windows apparently can't transparently handle global
1204 variables defined in DLLs. Initialisations such as:
1206 const ASN1_ITEM *it = &ASN1_INTEGER_it;
1208 wont compile. This is used by the any applications that need to
1209 declare their own ASN1 modules. This was fixed by adding the option
1210 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
1211 needed for static libraries under Win32.
1214 *) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
1215 setting of purpose and trust fields. New X509_STORE trust and
1216 purpose functions and tidy up setting in other SSL functions.
1219 *) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
1220 structure. These are inherited by X509_STORE_CTX when it is
1221 initialised. This allows various defaults to be set in the
1222 X509_STORE structure (such as flags for CRL checking and custom
1223 purpose or trust settings) for functions which only use X509_STORE_CTX
1224 internally such as S/MIME.
1226 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
1227 trust settings if they are not set in X509_STORE. This allows X509_STORE
1228 purposes and trust (in S/MIME for example) to override any set by default.
1230 Add command line options for CRL checking to smime, s_client and s_server
1234 *) Initial CRL based revocation checking. If the CRL checking flag(s)
1235 are set then the CRL is looked up in the X509_STORE structure and
1236 its validity and signature checked, then if the certificate is found
1237 in the CRL the verify fails with a revoked error.
1239 Various new CRL related callbacks added to X509_STORE_CTX structure.
1241 Command line options added to 'verify' application to support this.
1243 This needs some additional work, such as being able to handle multiple
1244 CRLs with different times, extension based lookup (rather than just
1245 by subject name) and ultimately more complete V2 CRL extension
1249 *) Add a general user interface API (crypto/ui/). This is designed
1250 to replace things like des_read_password and friends (backward
1251 compatibility functions using this new API are provided).
1252 The purpose is to remove prompting functions from the DES code
1253 section as well as provide for prompting through dialog boxes in
1254 a window system and the like.
1257 *) Add "ex_data" support to ENGINE so implementations can add state at a
1258 per-structure level rather than having to store it globally.
1261 *) Make it possible for ENGINE structures to be copied when retrieved by
1262 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1263 This causes the "original" ENGINE structure to act like a template,
1264 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1265 operational state can be localised to each ENGINE structure, despite the
1266 fact they all share the same "methods". New ENGINE structures returned in
1267 this case have no functional references and the return value is the single
1268 structural reference. This matches the single structural reference returned
1269 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1273 *) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1274 needs to match any other type at all we need to manually clear the
1278 *) Changes to the "openssl engine" utility to include;
1279 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1280 about an ENGINE's available control commands.
1281 - executing control commands from command line arguments using the
1282 '-pre' and '-post' switches. '-post' is only used if '-t' is
1283 specified and the ENGINE is successfully initialised. The syntax for
1284 the individual commands are colon-separated, for example;
1285 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1288 *) New dynamic control command support for ENGINEs. ENGINEs can now
1289 declare their own commands (numbers), names (strings), descriptions,
1290 and input types for run-time discovery by calling applications. A
1291 subset of these commands are implicitly classed as "executable"
1292 depending on their input type, and only these can be invoked through
1293 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1294 can be based on user input, config files, etc). The distinction is
1295 that "executable" commands cannot return anything other than a boolean
1296 result and can only support numeric or string input, whereas some
1297 discoverable commands may only be for direct use through
1298 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1299 pointers, or other custom uses. The "executable" commands are to
1300 support parameterisations of ENGINE behaviour that can be
1301 unambiguously defined by ENGINEs and used consistently across any
1302 OpenSSL-based application. Commands have been added to all the
1303 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1304 control over shared-library paths without source code alterations.
1307 *) Changed all ENGINE implementations to dynamically allocate their
1308 ENGINEs rather than declaring them statically. Apart from this being
1309 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1310 this also allows the implementations to compile without using the
1311 internal engine_int.h header.
1314 *) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1315 'const' value. Any code that should be able to modify a RAND_METHOD
1316 should already have non-const pointers to it (ie. they should only
1317 modify their own ones).
1320 *) Made a variety of little tweaks to the ENGINE code.
1321 - "atalla" and "ubsec" string definitions were moved from header files
1322 to C code. "nuron" string definitions were placed in variables
1323 rather than hard-coded - allowing parameterisation of these values
1324 later on via ctrl() commands.
1325 - Removed unused "#if 0"'d code.
1326 - Fixed engine list iteration code so it uses ENGINE_free() to release
1327 structural references.
1328 - Constified the RAND_METHOD element of ENGINE structures.
1329 - Constified various get/set functions as appropriate and added
1330 missing functions (including a catch-all ENGINE_cpy that duplicates
1331 all ENGINE values onto a new ENGINE except reference counts/state).
1332 - Removed NULL parameter checks in get/set functions. Setting a method
1333 or function to NULL is a way of cancelling out a previously set
1334 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1335 and doesn't justify the extra error symbols and code.
1336 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1337 flags from engine_int.h to engine.h.
1338 - Changed prototypes for ENGINE handler functions (init(), finish(),
1339 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1342 *) Implement binary inversion algorithm for BN_mod_inverse in addition
1343 to the algorithm using long division. The binary algorithm can be
1344 used only if the modulus is odd. On 32-bit systems, it is faster
1345 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1346 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1347 up to 450 bits. In 64-bit environments, the binary algorithm
1348 appears to be advantageous for much longer moduli; here we use it
1349 for moduli up to 2048 bits.
1352 *) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1353 could not support the combine flag in choice fields.
1356 *) Add a 'copy_extensions' option to the 'ca' utility. This copies
1357 extensions from a certificate request to the certificate.
1360 *) Allow multiple 'certopt' and 'nameopt' options to be separated
1361 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1362 file: this allows the display of the certificate about to be
1363 signed to be customised, to allow certain fields to be included
1364 or excluded and extension details. The old system didn't display
1365 multicharacter strings properly, omitted fields not in the policy
1366 and couldn't display additional details such as extensions.
1369 *) Function EC_POINTs_mul for multiple scalar multiplication
1370 of an arbitrary number of elliptic curve points
1371 \sum scalars[i]*points[i],
1372 optionally including the generator defined for the EC_GROUP:
1373 scalar*generator + \sum scalars[i]*points[i].
1375 EC_POINT_mul is a simple wrapper function for the typical case
1376 that the point list has just one item (besides the optional
1380 *) First EC_METHODs for curves over GF(p):
1382 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1383 operations and provides various method functions that can also
1384 operate with faster implementations of modular arithmetic.
1386 EC_GFp_mont_method() reuses most functions that are part of
1387 EC_GFp_simple_method, but uses Montgomery arithmetic.
1389 [Bodo Moeller; point addition and point doubling
1390 implementation directly derived from source code provided by
1391 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1393 *) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1394 crypto/ec/ec_lib.c):
1396 Curves are EC_GROUP objects (with an optional group generator)
1397 based on EC_METHODs that are built into the library.
1399 Points are EC_POINT objects based on EC_GROUP objects.
1401 Most of the framework would be able to handle curves over arbitrary
1402 finite fields, but as there are no obvious types for fields other
1403 than GF(p), some functions are limited to that for now.
1406 *) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1407 that the file contains a complete HTTP response.
1410 *) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1411 change the def and num file printf format specifier from "%-40sXXX"
1412 to "%-39s XXX". The latter will always guarantee a space after the
1413 field while the former will cause them to run together if the field
1414 is 40 of more characters long.
1417 *) Constify the cipher and digest 'method' functions and structures
1418 and modify related functions to take constant EVP_MD and EVP_CIPHER
1422 *) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1423 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1426 *) Modify EVP_Digest*() routines so they now return values. Although the
1427 internal software routines can never fail additional hardware versions
1431 *) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1433 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1434 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1437 ERR_R_NESTED_ASN1_ERROR
1439 ERR_R_MISSING_ASN1_EOS
1440 were 4 .. 9, conflicting with
1441 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1443 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1444 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1446 Add new error code 'ERR_R_INTERNAL_ERROR'.
1449 *) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1453 *) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1454 sets the subject name for a new request or supersedes the
1455 subject name in a given request. Formats that can be parsed are
1456 'CN=Some Name, OU=myOU, C=IT'
1458 'CN=Some Name/OU=myOU/C=IT'.
1460 Add options '-batch' and '-verbose' to 'openssl req'.
1461 [Massimiliano Pala <madwolf@hackmasters.net>]
1463 *) Introduce the possibility to access global variables through
1464 functions on platform were that's the best way to handle exporting
1465 global variables in shared libraries. To enable this functionality,
1466 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1467 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1468 is normally done by Configure or something similar).
1470 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1471 in the source file (foo.c) like this:
1473 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1474 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1476 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1477 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1479 OPENSSL_DECLARE_GLOBAL(int,foo);
1480 #define foo OPENSSL_GLOBAL_REF(foo)
1481 OPENSSL_DECLARE_GLOBAL(double,bar);
1482 #define bar OPENSSL_GLOBAL_REF(bar)
1484 The #defines are very important, and therefore so is including the
1485 header file everywhere where the defined globals are used.
1487 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1488 of ASN.1 items, but that structure is a bit different.
1490 The largest change is in util/mkdef.pl which has been enhanced with
1491 better and easier to understand logic to choose which symbols should
1492 go into the Windows .def files as well as a number of fixes and code
1493 cleanup (among others, algorithm keywords are now sorted
1494 lexicographically to avoid constant rewrites).
1497 *) In BN_div() keep a copy of the sign of 'num' before writing the
1498 result to 'rm' because if rm==num the value will be overwritten
1499 and produce the wrong result if 'num' is negative: this caused
1500 problems with BN_mod() and BN_nnmod().
1503 *) Function OCSP_request_verify(). This checks the signature on an
1504 OCSP request and verifies the signer certificate. The signer
1505 certificate is just checked for a generic purpose and OCSP request
1509 *) Add OCSP_check_validity() function to check the validity of OCSP
1510 responses. OCSP responses are prepared in real time and may only
1511 be a few seconds old. Simply checking that the current time lies
1512 between thisUpdate and nextUpdate max reject otherwise valid responses
1513 caused by either OCSP responder or client clock inaccuracy. Instead
1514 we allow thisUpdate and nextUpdate to fall within a certain period of
1515 the current time. The age of the response can also optionally be
1516 checked. Two new options -validity_period and -status_age added to
1520 *) If signature or public key algorithm is unrecognized print out its
1521 OID rather that just UNKNOWN.
1524 *) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1525 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1526 ID to be generated from the issuer certificate alone which can then be
1527 passed to OCSP_id_issuer_cmp().
1530 *) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1531 ASN1 modules to export functions returning ASN1_ITEM pointers
1532 instead of the ASN1_ITEM structures themselves. This adds several
1533 new macros which allow the underlying ASN1 function/structure to
1534 be accessed transparently. As a result code should not use ASN1_ITEM
1535 references directly (such as &X509_it) but instead use the relevant
1536 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1537 use of the new ASN1 code on platforms where exporting structures
1538 is problematical (for example in shared libraries) but exporting
1539 functions returning pointers to structures is not.
1542 *) Add support for overriding the generation of SSL/TLS session IDs.
1543 These callbacks can be registered either in an SSL_CTX or per SSL.
1544 The purpose of this is to allow applications to control, if they wish,
1545 the arbitrary values chosen for use as session IDs, particularly as it
1546 can be useful for session caching in multiple-server environments. A
1547 command-line switch for testing this (and any client code that wishes
1548 to use such a feature) has been added to "s_server".
1549 [Geoff Thorpe, Lutz Jaenicke]
1551 *) Modify mkdef.pl to recognise and parse preprocessor conditionals
1552 of the form '#if defined(...) || defined(...) || ...' and
1553 '#if !defined(...) && !defined(...) && ...'. This also avoids
1554 the growing number of special cases it was previously handling.
1557 *) Make all configuration macros available for application by making
1558 sure they are available in opensslconf.h, by giving them names starting
1559 with "OPENSSL_" to avoid conflicts with other packages and by making
1560 sure e_os2.h will cover all platform-specific cases together with
1562 Additionally, it is now possible to define configuration/platform-
1563 specific names (called "system identities"). In the C code, these
1564 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1565 macro with the name beginning with "OPENSSL_SYS_", which is determined
1566 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1570 *) New option -set_serial to 'req' and 'x509' this allows the serial
1571 number to use to be specified on the command line. Previously self
1572 signed certificates were hard coded with serial number 0 and the
1573 CA options of 'x509' had to use a serial number in a file which was
1577 *) New options to 'ca' utility to support V2 CRL entry extensions.
1578 Currently CRL reason, invalidity date and hold instruction are
1579 supported. Add new CRL extensions to V3 code and some new objects.
1582 *) New function EVP_CIPHER_CTX_set_padding() this is used to
1583 disable standard block padding (aka PKCS#5 padding) in the EVP
1584 API, which was previously mandatory. This means that the data is
1585 not padded in any way and so the total length much be a multiple
1586 of the block size, otherwise an error occurs.
1589 *) Initial (incomplete) OCSP SSL support.
1592 *) New function OCSP_parse_url(). This splits up a URL into its host,
1593 port and path components: primarily to parse OCSP URLs. New -url
1594 option to ocsp utility.
1597 *) New nonce behavior. The return value of OCSP_check_nonce() now
1598 reflects the various checks performed. Applications can decide
1599 whether to tolerate certain situations such as an absent nonce
1600 in a response when one was present in a request: the ocsp application
1601 just prints out a warning. New function OCSP_add1_basic_nonce()
1602 this is to allow responders to include a nonce in a response even if
1603 the request is nonce-less.
1606 *) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1607 skipped when using openssl x509 multiple times on a single input file,
1608 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1611 *) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1612 set string type: to handle setting ASN1_TIME structures. Fix ca
1613 utility to correctly initialize revocation date of CRLs.
1616 *) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1617 the clients preferred ciphersuites and rather use its own preferences.
1618 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1619 Internet Explorer by ensuring unchanged hash method during stepup.
1620 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1623 *) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1624 to aes and add a new 'exist' option to print out symbols that don't
1628 *) Additional options to ocsp utility to allow flags to be set and
1629 additional certificates supplied.
1632 *) Add the option -VAfile to 'openssl ocsp', so the user can give the
1633 OCSP client a number of certificate to only verify the response
1637 *) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1638 handle the new API. Currently only ECB, CBC modes supported. Add new
1641 Add TLS AES ciphersuites as described in RFC3268, "Advanced
1642 Encryption Standard (AES) Ciphersuites for Transport Layer
1643 Security (TLS)". (In beta versions of OpenSSL 0.9.7, these were
1644 not enabled by default and were not part of the "ALL" ciphersuite
1645 alias because they were not yet official; they could be
1646 explicitly requested by specifying the "AESdraft" ciphersuite
1647 group alias. In the final release of OpenSSL 0.9.7, the group
1648 alias is called "AES" and is part of "ALL".)
1649 [Ben Laurie, Steve Henson, Bodo Moeller]
1651 *) New function OCSP_copy_nonce() to copy nonce value (if present) from
1652 request to response.
1655 *) Functions for OCSP responders. OCSP_request_onereq_count(),
1656 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1657 extract information from a certificate request. OCSP_response_create()
1658 creates a response and optionally adds a basic response structure.
1659 OCSP_basic_add1_status() adds a complete single response to a basic
1660 response and returns the OCSP_SINGLERESP structure just added (to allow
1661 extensions to be included for example). OCSP_basic_add1_cert() adds a
1662 certificate to a basic response and OCSP_basic_sign() signs a basic
1663 response with various flags. New helper functions ASN1_TIME_check()
1664 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1665 (converts ASN1_TIME to GeneralizedTime).
1668 *) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1669 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1670 structure from a certificate. X509_pubkey_digest() digests the public_key
1671 contents: this is used in various key identifiers.
1674 *) Make sk_sort() tolerate a NULL argument.
1675 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1677 *) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1678 passed by the function are trusted implicitly. If any of them signed the
1679 response then it is assumed to be valid and is not verified.
1682 *) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1683 to data. This was previously part of the PKCS7 ASN1 code. This
1684 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1685 [Steve Henson, reported by Kenneth R. Robinette
1686 <support@securenetterm.com>]
1688 *) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1689 routines: without these tracing memory leaks is very painful.
1690 Fix leaks in PKCS12 and PKCS7 routines.
1693 *) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1694 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1695 effectively meant GeneralizedTime would never be used. Now it
1696 is initialised to -1 but X509_time_adj() now has to check the value
1697 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1698 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1699 [Steve Henson, reported by Kenneth R. Robinette
1700 <support@securenetterm.com>]
1702 *) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1703 result in a zero length in the ASN1_INTEGER structure which was
1704 not consistent with the structure when d2i_ASN1_INTEGER() was used
1705 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1706 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1707 where it did not print out a minus for negative ASN1_INTEGER.
1710 *) Add summary printout to ocsp utility. The various functions which
1711 convert status values to strings have been renamed to:
1712 OCSP_response_status_str(), OCSP_cert_status_str() and
1713 OCSP_crl_reason_str() and are no longer static. New options
1714 to verify nonce values and to disable verification. OCSP response
1715 printout format cleaned up.
1718 *) Add additional OCSP certificate checks. These are those specified
1719 in RFC2560. This consists of two separate checks: the CA of the
1720 certificate being checked must either be the OCSP signer certificate
1721 or the issuer of the OCSP signer certificate. In the latter case the
1722 OCSP signer certificate must contain the OCSP signing extended key
1723 usage. This check is performed by attempting to match the OCSP
1724 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1725 in the OCSP_CERTID structures of the response.
1728 *) Initial OCSP certificate verification added to OCSP_basic_verify()
1729 and related routines. This uses the standard OpenSSL certificate
1730 verify routines to perform initial checks (just CA validity) and
1731 to obtain the certificate chain. Then additional checks will be
1732 performed on the chain. Currently the root CA is checked to see
1733 if it is explicitly trusted for OCSP signing. This is used to set
1734 a root CA as a global signing root: that is any certificate that
1735 chains to that CA is an acceptable OCSP signing certificate.
1738 *) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1739 extensions from a separate configuration file.
1740 As when reading extensions from the main configuration file,
1741 the '-extensions ...' option may be used for specifying the
1743 [Massimiliano Pala <madwolf@comune.modena.it>]
1745 *) New OCSP utility. Allows OCSP requests to be generated or
1746 read. The request can be sent to a responder and the output
1747 parsed, outputed or printed in text form. Not complete yet:
1748 still needs to check the OCSP response validity.
1751 *) New subcommands for 'openssl ca':
1752 'openssl ca -status <serial>' prints the status of the cert with
1753 the given serial number (according to the index file).
1754 'openssl ca -updatedb' updates the expiry status of certificates
1756 [Massimiliano Pala <madwolf@comune.modena.it>]
1758 *) New '-newreq-nodes' command option to CA.pl. This is like
1759 '-newreq', but calls 'openssl req' with the '-nodes' option
1760 so that the resulting key is not encrypted.
1761 [Damien Miller <djm@mindrot.org>]
1763 *) New configuration for the GNU Hurd.
1764 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1766 *) Initial code to implement OCSP basic response verify. This
1767 is currently incomplete. Currently just finds the signer's
1768 certificate and verifies the signature on the response.
1771 *) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1772 value of OPENSSLDIR. This is available via the new '-d' option
1773 to 'openssl version', and is also included in 'openssl version -a'.
1776 *) Allowing defining memory allocation callbacks that will be given
1777 file name and line number information in additional arguments
1778 (a const char* and an int). The basic functionality remains, as
1779 well as the original possibility to just replace malloc(),
1780 realloc() and free() by functions that do not know about these
1781 additional arguments. To register and find out the current
1782 settings for extended allocation functions, the following
1783 functions are provided:
1785 CRYPTO_set_mem_ex_functions
1786 CRYPTO_set_locked_mem_ex_functions
1787 CRYPTO_get_mem_ex_functions
1788 CRYPTO_get_locked_mem_ex_functions
1790 These work the same way as CRYPTO_set_mem_functions and friends.
1791 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1792 extended allocation function is enabled.
1793 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1794 a conventional allocation function is enabled.
1795 [Richard Levitte, Bodo Moeller]
1797 *) Finish off removing the remaining LHASH function pointer casts.
1798 There should no longer be any prototype-casting required when using
1799 the LHASH abstraction, and any casts that remain are "bugs". See
1800 the callback types and macros at the head of lhash.h for details
1801 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1804 *) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1805 If /dev/[u]random devices are not available or do not return enough
1806 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1808 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1809 /etc/entropy will be queried once each in this sequence, quering stops
1810 when enough entropy was collected without querying more sockets.
1813 *) Change the Unix RAND_poll() variant to be able to poll several
1814 random devices, as specified by DEVRANDOM, until a sufficient amount
1815 of data has been collected. We spend at most 10 ms on each file
1816 (select timeout) and read in non-blocking mode. DEVRANDOM now
1817 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1818 (previously it was just the string "/dev/urandom"), so on typical
1819 platforms the 10 ms delay will never occur.
1820 Also separate out the Unix variant to its own file, rand_unix.c.
1821 For VMS, there's a currently-empty rand_vms.c.
1824 *) Move OCSP client related routines to ocsp_cl.c. These
1825 provide utility functions which an application needing
1826 to issue a request to an OCSP responder and analyse the
1827 response will typically need: as opposed to those which an
1828 OCSP responder itself would need which will be added later.
1830 OCSP_request_sign() signs an OCSP request with an API similar
1831 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1832 response. OCSP_response_get1_basic() extracts basic response
1833 from response. OCSP_resp_find_status(): finds and extracts status
1834 information from an OCSP_CERTID structure (which will be created
1835 when the request structure is built). These are built from lower
1836 level functions which work on OCSP_SINGLERESP structures but
1837 wont normally be used unless the application wishes to examine
1838 extensions in the OCSP response for example.
1840 Replace nonce routines with a pair of functions.
1841 OCSP_request_add1_nonce() adds a nonce value and optionally
1842 generates a random value. OCSP_check_nonce() checks the
1843 validity of the nonce in an OCSP response.
1846 *) Change function OCSP_request_add() to OCSP_request_add0_id().
1847 This doesn't copy the supplied OCSP_CERTID and avoids the
1848 need to free up the newly created id. Change return type
1849 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1850 This can then be used to add extensions to the request.
1851 Deleted OCSP_request_new(), since most of its functionality
1852 is now in OCSP_REQUEST_new() (and the case insensitive name
1853 clash) apart from the ability to set the request name which
1854 will be added elsewhere.
1857 *) Update OCSP API. Remove obsolete extensions argument from
1858 various functions. Extensions are now handled using the new
1859 OCSP extension code. New simple OCSP HTTP function which
1860 can be used to send requests and parse the response.
1863 *) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1864 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1865 uses the special reorder version of SET OF to sort the attributes
1866 and reorder them to match the encoded order. This resolves a long
1867 standing problem: a verify on a PKCS7 structure just after signing
1868 it used to fail because the attribute order did not match the
1869 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1870 it uses the received order. This is necessary to tolerate some broken
1871 software that does not order SET OF. This is handled by encoding
1872 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1873 to produce the required SET OF.
1876 *) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1877 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1878 files to get correct declarations of the ASN.1 item variables.
1881 *) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1882 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1883 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1884 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1885 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1886 ASN1_ITEM and no wrapper functions.
1889 *) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1890 replace the old function pointer based I/O routines. Change most of
1891 the *_d2i_bio() and *_d2i_fp() functions to use these.
1894 *) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1895 lines, recognice more "algorithms" that can be deselected, and make
1896 it complain about algorithm deselection that isn't recognised.
1899 *) New ASN1 functions to handle dup, sign, verify, digest, pack and
1900 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1901 to use new functions. Add NO_ASN1_OLD which can be set to remove
1902 some old style ASN1 functions: this can be used to determine if old
1903 code will still work when these eventually go away.
1906 *) New extension functions for OCSP structures, these follow the
1907 same conventions as certificates and CRLs.
1910 *) New function X509V3_add1_i2d(). This automatically encodes and
1911 adds an extension. Its behaviour can be customised with various
1912 flags to append, replace or delete. Various wrappers added for
1913 certifcates and CRLs.
1916 *) Fix to avoid calling the underlying ASN1 print routine when
1917 an extension cannot be parsed. Correct a typo in the
1918 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1921 *) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1922 entries for variables.
1925 *) Add functionality to apps/openssl.c for detecting locking
1926 problems: As the program is single-threaded, all we have
1927 to do is register a locking callback using an array for
1928 storing which locks are currently held by the program.
1931 *) Use a lock around the call to CRYPTO_get_ex_new_index() in
1932 SSL_get_ex_data_X509_STORE_idx(), which is used in
1933 ssl_verify_cert_chain() and thus can be called at any time
1934 during TLS/SSL handshakes so that thread-safety is essential.
1935 Unfortunately, the ex_data design is not at all suited
1936 for multi-threaded use, so it probably should be abolished.
1939 *) Added Broadcom "ubsec" ENGINE to OpenSSL.
1940 [Broadcom, tweaked and integrated by Geoff Thorpe]
1942 *) Move common extension printing code to new function
1943 X509V3_print_extensions(). Reorganise OCSP print routines and
1944 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1947 *) New function X509_signature_print() to remove duplication in some
1951 *) Add a special meaning when SET OF and SEQUENCE OF flags are both
1952 set (this was treated exactly the same as SET OF previously). This
1953 is used to reorder the STACK representing the structure to match the
1954 encoding. This will be used to get round a problem where a PKCS7
1955 structure which was signed could not be verified because the STACK
1956 order did not reflect the encoded order.
1959 *) Reimplement the OCSP ASN1 module using the new code.
1962 *) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1963 for its ASN1 operations. The old style function pointers still exist
1964 for now but they will eventually go away.
1967 *) Merge in replacement ASN1 code from the ASN1 branch. This almost
1968 completely replaces the old ASN1 functionality with a table driven
1969 encoder and decoder which interprets an ASN1_ITEM structure describing
1970 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1971 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1972 has also been converted to the new form.
1975 *) Change BN_mod_exp_recp so that negative moduli are tolerated
1976 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1977 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1978 for negative moduli.
1981 *) Fix BN_uadd and BN_usub: Always return non-negative results instead
1982 of not touching the result's sign bit.
1985 *) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1989 *) Changed the LHASH code to use prototypes for callbacks, and created
1990 macros to declare and implement thin (optionally static) functions
1991 that provide type-safety and avoid function pointer casting for the
1992 type-specific callbacks.
1995 *) Added Kerberos Cipher Suites to be used with TLS, as written in
1997 [Veers Staats <staatsvr@asc.hpc.mil>,
1998 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
2000 *) Reformat the FAQ so the different questions and answers can be divided
2001 in sections depending on the subject.
2004 *) Have the zlib compression code load ZLIB.DLL dynamically under
2008 *) New function BN_mod_sqrt for computing square roots modulo a prime
2009 (using the probabilistic Tonelli-Shanks algorithm unless
2010 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
2011 be handled deterministically).
2012 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2014 *) Make BN_mod_inverse faster by explicitly handling small quotients
2015 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
2016 512 bits], about 30% for larger ones [1024 or 2048 bits].)
2019 *) New function BN_kronecker.
2022 *) Fix BN_gcd so that it works on negative inputs; the result is
2023 positive unless both parameters are zero.
2024 Previously something reasonably close to an infinite loop was
2025 possible because numbers could be growing instead of shrinking
2026 in the implementation of Euclid's algorithm.
2029 *) Fix BN_is_word() and BN_is_one() macros to take into account the
2030 sign of the number in question.
2032 Fix BN_is_word(a,w) to work correctly for w == 0.
2034 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
2035 because its test if the absolute value of 'a' equals 'w'.
2036 Note that BN_abs_is_word does *not* handle w == 0 reliably;
2037 it exists mostly for use in the implementations of BN_is_zero(),
2038 BN_is_one(), and BN_is_word().
2041 *) New function BN_swap.
2044 *) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
2045 the exponentiation functions are more likely to produce reasonable
2046 results on negative inputs.
2049 *) Change BN_mod_mul so that the result is always non-negative.
2050 Previously, it could be negative if one of the factors was negative;
2051 I don't think anyone really wanted that behaviour.
2054 *) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
2055 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
2056 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
2057 and add new functions:
2066 BN_mod_lshift1_quick
2070 These functions always generate non-negative results.
2072 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2073 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2075 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2076 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2077 be reduced modulo m.
2078 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2081 The following entry accidentily appeared in the CHANGES file
2082 distributed with OpenSSL 0.9.7. The modifications described in
2083 it do *not* apply to OpenSSL 0.9.7.
2085 *) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2086 was actually never needed) and in BN_mul(). The removal in BN_mul()
2087 required a small change in bn_mul_part_recursive() and the addition
2088 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2089 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2090 bn_sub_words() and bn_add_words() except they take arrays with
2095 *) In 'openssl passwd', verify passwords read from the terminal
2096 unless the '-salt' option is used (which usually means that
2097 verification would just waste user's time since the resulting
2098 hash is going to be compared with some given password hash)
2099 or the new '-noverify' option is used.
2101 This is an incompatible change, but it does not affect
2102 non-interactive use of 'openssl passwd' (passwords on the command
2103 line, '-stdin' option, '-in ...' option) and thus should not
2107 *) Remove all references to RSAref, since there's no more need for it.
2110 *) Make DSO load along a path given through an environment variable
2111 (SHLIB_PATH) with shl_load().
2114 *) Constify the ENGINE code as a result of BIGNUM constification.
2115 Also constify the RSA code and most things related to it. In a
2116 few places, most notable in the depth of the ASN.1 code, ugly
2117 casts back to non-const were required (to be solved at a later
2121 *) Make it so the openssl application has all engines loaded by default.
2124 *) Constify the BIGNUM routines a little more.
2127 *) Add the following functions:
2129 ENGINE_load_cswift()
2131 ENGINE_load_atalla()
2133 ENGINE_load_builtin_engines()
2135 That way, an application can itself choose if external engines that
2136 are built-in in OpenSSL shall ever be used or not. The benefit is
2137 that applications won't have to be linked with libdl or other dso
2138 libraries unless it's really needed.
2140 Changed 'openssl engine' to load all engines on demand.
2141 Changed the engine header files to avoid the duplication of some
2142 declarations (they differed!).
2145 *) 'openssl engine' can now list capabilities.
2148 *) Better error reporting in 'openssl engine'.
2151 *) Never call load_dh_param(NULL) in s_server.
2154 *) Add engine application. It can currently list engines by name and
2155 identity, and test if they are actually available.
2158 *) Improve RPM specification file by forcing symbolic linking and making
2159 sure the installed documentation is also owned by root.root.
2160 [Damien Miller <djm@mindrot.org>]
2162 *) Give the OpenSSL applications more possibilities to make use of
2163 keys (public as well as private) handled by engines.
2166 *) Add OCSP code that comes from CertCo.
2169 *) Add VMS support for the Rijndael code.
2172 *) Added untested support for Nuron crypto accelerator.
2175 *) Add support for external cryptographic devices. This code was
2176 previously distributed separately as the "engine" branch.
2177 [Geoff Thorpe, Richard Levitte]
2179 *) Rework the filename-translation in the DSO code. It is now possible to
2180 have far greater control over how a "name" is turned into a filename
2181 depending on the operating environment and any oddities about the
2182 different shared library filenames on each system.
2185 *) Support threads on FreeBSD-elf in Configure.
2188 *) Fix for SHA1 assembly problem with MASM: it produces
2189 warnings about corrupt line number information when assembling
2190 with debugging information. This is caused by the overlapping
2192 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2195 NCONF_get_number() has no error checking at all. As a replacement,
2196 NCONF_get_number_e() is defined (_e for "error checking") and is
2197 promoted strongly. The old NCONF_get_number is kept around for
2198 binary backward compatibility.
2199 Make it possible for methods to load from something other than a BIO,
2200 by providing a function pointer that is given a name instead of a BIO.
2201 For example, this could be used to load configuration data from an
2205 *) Fix for non blocking accept BIOs. Added new I/O special reason
2206 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2207 with non blocking I/O was not possible because no retry code was
2208 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2212 *) Added the beginnings of Rijndael support.
2215 *) Fix for bug in DirectoryString mask setting. Add support for
2216 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2217 to allow certificate printing to more controllable, additional
2218 'certopt' option to 'x509' to allow new printing options to be
2222 *) Clean old EAY MD5 hack from e_os.h.
2225 Changes between 0.9.6l and 0.9.6m [17 Mar 2004]
2227 *) Fix null-pointer assignment in do_change_cipher_spec() revealed
2228 by using the Codenomicon TLS Test Tool (CVE-2004-0079)
2229 [Joe Orton, Steve Henson]
2231 Changes between 0.9.6k and 0.9.6l [04 Nov 2003]
2233 *) Fix additional bug revealed by the NISCC test suite:
2235 Stop bug triggering large recursion when presented with
2236 certain ASN.1 tags (CVE-2003-0851)
2239 Changes between 0.9.6j and 0.9.6k [30 Sep 2003]
2241 *) Fix various bugs revealed by running the NISCC test suite:
2243 Stop out of bounds reads in the ASN1 code when presented with
2244 invalid tags (CVE-2003-0543 and CVE-2003-0544).
2246 If verify callback ignores invalid public key errors don't try to check
2247 certificate signature with the NULL public key.
2251 *) In ssl3_accept() (ssl/s3_srvr.c) only accept a client certificate
2252 if the server requested one: as stated in TLS 1.0 and SSL 3.0
2256 *) In ssl3_get_client_hello() (ssl/s3_srvr.c), tolerate additional
2257 extra data after the compression methods not only for TLS 1.0
2258 but also for SSL 3.0 (as required by the specification).
2259 [Bodo Moeller; problem pointed out by Matthias Loepfe]
2261 *) Change X509_certificate_type() to mark the key as exported/exportable
2262 when it's 512 *bits* long, not 512 bytes.
2265 Changes between 0.9.6i and 0.9.6j [10 Apr 2003]
2267 *) Countermeasure against the Klima-Pokorny-Rosa extension of
2268 Bleichbacher's attack on PKCS #1 v1.5 padding: treat
2269 a protocol version number mismatch like a decryption error
2270 in ssl3_get_client_key_exchange (ssl/s3_srvr.c).
2273 *) Turn on RSA blinding by default in the default implementation
2274 to avoid a timing attack. Applications that don't want it can call
2275 RSA_blinding_off() or use the new flag RSA_FLAG_NO_BLINDING.
2276 They would be ill-advised to do so in most cases.
2277 [Ben Laurie, Steve Henson, Geoff Thorpe, Bodo Moeller]
2279 *) Change RSA blinding code so that it works when the PRNG is not
2280 seeded (in this case, the secret RSA exponent is abused as
2281 an unpredictable seed -- if it is not unpredictable, there
2282 is no point in blinding anyway). Make RSA blinding thread-safe
2283 by remembering the creator's thread ID in rsa->blinding and
2284 having all other threads use local one-time blinding factors
2285 (this requires more computation than sharing rsa->blinding, but
2286 avoids excessive locking; and if an RSA object is not shared
2287 between threads, blinding will still be very fast).
2290 Changes between 0.9.6h and 0.9.6i [19 Feb 2003]
2292 *) In ssl3_get_record (ssl/s3_pkt.c), minimize information leaked
2293 via timing by performing a MAC computation even if incorrrect
2294 block cipher padding has been found. This is a countermeasure
2295 against active attacks where the attacker has to distinguish
2296 between bad padding and a MAC verification error. (CVE-2003-0078)
2298 [Bodo Moeller; problem pointed out by Brice Canvel (EPFL),
2299 Alain Hiltgen (UBS), Serge Vaudenay (EPFL), and
2300 Martin Vuagnoux (EPFL, Ilion)]
2302 Changes between 0.9.6g and 0.9.6h [5 Dec 2002]
2304 *) New function OPENSSL_cleanse(), which is used to cleanse a section of
2305 memory from it's contents. This is done with a counter that will
2306 place alternating values in each byte. This can be used to solve
2307 two issues: 1) the removal of calls to memset() by highly optimizing
2308 compilers, and 2) cleansing with other values than 0, since those can
2309 be read through on certain media, for example a swap space on disk.
2312 *) Bugfix: client side session caching did not work with external caching,
2313 because the session->cipher setting was not restored when reloading
2314 from the external cache. This problem was masked, when
2315 SSL_OP_NETSCAPE_REUSE_CIPHER_CHANGE_BUG (part of SSL_OP_ALL) was set.
2316 (Found by Steve Haslam <steve@araqnid.ddts.net>.)
2319 *) Fix client_certificate (ssl/s2_clnt.c): The permissible total
2320 length of the REQUEST-CERTIFICATE message is 18 .. 34, not 17 .. 33.
2321 [Zeev Lieber <zeev-l@yahoo.com>]
2323 *) Undo an undocumented change introduced in 0.9.6e which caused
2324 repeated calls to OpenSSL_add_all_ciphers() and
2325 OpenSSL_add_all_digests() to be ignored, even after calling
2329 *) Change the default configuration reader to deal with last line not
2330 being properly terminated.
2333 *) Change X509_NAME_cmp() so it applies the special rules on handling
2334 DN values that are of type PrintableString, as well as RDNs of type
2335 emailAddress where the value has the type ia5String.
2336 [stefank@valicert.com via Richard Levitte]
2338 *) Add a SSL_SESS_CACHE_NO_INTERNAL_STORE flag to take over half
2339 the job SSL_SESS_CACHE_NO_INTERNAL_LOOKUP was inconsistently
2340 doing, define a new flag (SSL_SESS_CACHE_NO_INTERNAL) to be
2341 the bitwise-OR of the two for use by the majority of applications
2342 wanting this behaviour, and update the docs. The documented
2343 behaviour and actual behaviour were inconsistent and had been
2344 changing anyway, so this is more a bug-fix than a behavioural
2346 [Geoff Thorpe, diagnosed by Nadav Har'El]
2348 *) Don't impose a 16-byte length minimum on session IDs in ssl/s3_clnt.c
2349 (the SSL 3.0 and TLS 1.0 specifications allow any length up to 32 bytes).
2352 *) Fix initialization code race conditions in
2353 SSLv23_method(), SSLv23_client_method(), SSLv23_server_method(),
2354 SSLv2_method(), SSLv2_client_method(), SSLv2_server_method(),
2355 SSLv3_method(), SSLv3_client_method(), SSLv3_server_method(),
2356 TLSv1_method(), TLSv1_client_method(), TLSv1_server_method(),
2357 ssl2_get_cipher_by_char(),
2358 ssl3_get_cipher_by_char().
2359 [Patrick McCormick <patrick@tellme.com>, Bodo Moeller]
2361 *) Reorder cleanup sequence in SSL_CTX_free(): only remove the ex_data after
2362 the cached sessions are flushed, as the remove_cb() might use ex_data
2363 contents. Bug found by Sam Varshavchik <mrsam@courier-mta.com>
2364 (see [openssl.org #212]).
2365 [Geoff Thorpe, Lutz Jaenicke]
2367 *) Fix typo in OBJ_txt2obj which incorrectly passed the content
2368 length, instead of the encoding length to d2i_ASN1_OBJECT.
2371 Changes between 0.9.6f and 0.9.6g [9 Aug 2002]
2373 *) [In 0.9.6g-engine release:]
2374 Fix crypto/engine/vendor_defns/cswift.h for WIN32 (use '_stdcall').
2375 [Lynn Gazis <lgazis@rainbow.com>]
2377 Changes between 0.9.6e and 0.9.6f [8 Aug 2002]
2379 *) Fix ASN1 checks. Check for overflow by comparing with LONG_MAX
2380 and get fix the header length calculation.
2381 [Florian Weimer <Weimer@CERT.Uni-Stuttgart.DE>,
2382 Alon Kantor <alonk@checkpoint.com> (and others),
2385 *) Use proper error handling instead of 'assertions' in buffer
2386 overflow checks added in 0.9.6e. This prevents DoS (the
2387 assertions could call abort()).
2388 [Arne Ansper <arne@ats.cyber.ee>, Bodo Moeller]
2390 Changes between 0.9.6d and 0.9.6e [30 Jul 2002]
2392 *) Add various sanity checks to asn1_get_length() to reject
2393 the ASN1 length bytes if they exceed sizeof(long), will appear
2394 negative or the content length exceeds the length of the
2396 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2398 *) Fix cipher selection routines: ciphers without encryption had no flags
2399 for the cipher strength set and where therefore not handled correctly
2400 by the selection routines (PR #130).
2403 *) Fix EVP_dsa_sha macro.
2407 SSL_OP_DONT_INSERT_EMPTY_FRAGMENTS
2408 for disabling the SSL 3.0/TLS 1.0 CBC vulnerability countermeasure
2409 that was added in OpenSSL 0.9.6d.
2411 As the countermeasure turned out to be incompatible with some
2412 broken SSL implementations, the new option is part of SSL_OP_ALL.
2413 SSL_OP_ALL is usually employed when compatibility with weird SSL
2414 implementations is desired (e.g. '-bugs' option to 's_client' and
2415 's_server'), so the new option is automatically set in many
2419 *) Changes in security patch:
2421 Changes marked "(CHATS)" were sponsored by the Defense Advanced
2422 Research Projects Agency (DARPA) and Air Force Research Laboratory,
2423 Air Force Materiel Command, USAF, under agreement number
2426 *) Add various sanity checks to asn1_get_length() to reject
2427 the ASN1 length bytes if they exceed sizeof(long), will appear
2428 negative or the content length exceeds the length of the
2429 supplied buffer. (CVE-2002-0659)
2430 [Steve Henson, Adi Stav <stav@mercury.co.il>, James Yonan <jim@ntlp.com>]
2432 *) Assertions for various potential buffer overflows, not known to
2434 [Ben Laurie (CHATS)]
2436 *) Various temporary buffers to hold ASCII versions of integers were
2437 too small for 64 bit platforms. (CVE-2002-0655)
2438 [Matthew Byng-Maddick <mbm@aldigital.co.uk> and Ben Laurie (CHATS)>
2440 *) Remote buffer overflow in SSL3 protocol - an attacker could
2441 supply an oversized session ID to a client. (CVE-2002-0656)
2442 [Ben Laurie (CHATS)]
2444 *) Remote buffer overflow in SSL2 protocol - an attacker could
2445 supply an oversized client master key. (CVE-2002-0656)
2446 [Ben Laurie (CHATS)]
2448 Changes between 0.9.6c and 0.9.6d [9 May 2002]
2450 *) Fix crypto/asn1/a_sign.c so that 'parameters' is omitted (not
2451 encoded as NULL) with id-dsa-with-sha1.
2452 [Nils Larsch <nla@trustcenter.de>; problem pointed out by Bodo Moeller]
2454 *) Check various X509_...() return values in apps/req.c.
2455 [Nils Larsch <nla@trustcenter.de>]
2457 *) Fix BASE64 decode (EVP_DecodeUpdate) for data with CR/LF ended lines:
2458 an end-of-file condition would erronously be flagged, when the CRLF
2459 was just at the end of a processed block. The bug was discovered when
2460 processing data through a buffering memory BIO handing the data to a
2461 BASE64-decoding BIO. Bug fund and patch submitted by Pavel Tsekov
2462 <ptsekov@syntrex.com> and Nedelcho Stanev.
2465 *) Implement a countermeasure against a vulnerability recently found
2466 in CBC ciphersuites in SSL 3.0/TLS 1.0: Send an empty fragment
2467 before application data chunks to avoid the use of known IVs
2468 with data potentially chosen by the attacker.
2471 *) Fix length checks in ssl3_get_client_hello().
2474 *) TLS/SSL library bugfix: use s->s3->in_read_app_data differently
2475 to prevent ssl3_read_internal() from incorrectly assuming that
2476 ssl3_read_bytes() found application data while handshake
2477 processing was enabled when in fact s->s3->in_read_app_data was
2478 merely automatically cleared during the initial handshake.
2479 [Bodo Moeller; problem pointed out by Arne Ansper <arne@ats.cyber.ee>]
2481 *) Fix object definitions for Private and Enterprise: they were not
2482 recognized in their shortname (=lowercase) representation. Extend
2483 obj_dat.pl to issue an error when using undefined keywords instead
2484 of silently ignoring the problem (Svenning Sorensen
2485 <sss@sss.dnsalias.net>).
2488 *) Fix DH_generate_parameters() so that it works for 'non-standard'
2489 generators, i.e. generators other than 2 and 5. (Previously, the
2490 code did not properly initialise the 'add' and 'rem' values to
2491 BN_generate_prime().)
2493 In the new general case, we do not insist that 'generator' is
2494 actually a primitive root: This requirement is rather pointless;
2495 a generator of the order-q subgroup is just as good, if not
2499 *) Map new X509 verification errors to alerts. Discovered and submitted by
2500 Tom Wu <tom@arcot.com>.
2503 *) Fix ssl3_pending() (ssl/s3_lib.c) to prevent SSL_pending() from
2504 returning non-zero before the data has been completely received
2505 when using non-blocking I/O.
2506 [Bodo Moeller; problem pointed out by John Hughes]
2508 *) Some of the ciphers missed the strength entry (SSL_LOW etc).
2509 [Ben Laurie, Lutz Jaenicke]
2511 *) Fix bug in SSL_clear(): bad sessions were not removed (found by
2512 Yoram Zahavi <YoramZ@gilian.com>).
2515 *) Add information about CygWin 1.3 and on, and preserve proper
2516 configuration for the versions before that.
2517 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
2519 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
2520 check whether we deal with a copy of a session and do not delete from
2521 the cache in this case. Problem reported by "Izhar Shoshani Levi"
2522 <izhar@checkpoint.com>.
2525 *) Do not store session data into the internal session cache, if it
2526 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
2527 flag is set). Proposed by Aslam <aslam@funk.com>.
2530 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
2534 *) [In 0.9.6d-engine release:]
2535 Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
2536 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
2538 *) Add the configuration target linux-s390x.
2539 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
2541 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
2542 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
2543 variable as an indication that a ClientHello message has been
2544 received. As the flag value will be lost between multiple
2545 invocations of ssl3_accept when using non-blocking I/O, the
2546 function may not be aware that a handshake has actually taken
2547 place, thus preventing a new session from being added to the
2550 To avoid this problem, we now set s->new_session to 2 instead of
2551 using a local variable.
2552 [Lutz Jaenicke, Bodo Moeller]
2554 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
2555 if the SSL_R_LENGTH_MISMATCH error is detected.
2556 [Geoff Thorpe, Bodo Moeller]
2558 *) New 'shared_ldflag' column in Configure platform table.
2561 *) Fix EVP_CIPHER_mode macro.
2562 ["Dan S. Camper" <dan@bti.net>]
2564 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
2565 type, we must throw them away by setting rr->length to 0.
2566 [D P Chang <dpc@qualys.com>]
2568 Changes between 0.9.6b and 0.9.6c [21 dec 2001]
2570 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
2571 <Dominikus.Scherkl@biodata.com>. (The previous implementation
2572 worked incorrectly for those cases where range = 10..._2 and
2573 3*range is two bits longer than range.)
2576 *) Only add signing time to PKCS7 structures if it is not already
2580 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
2581 OBJ_ld_ce should be OBJ_id_ce.
2582 Also some ip-pda OIDs in crypto/objects/objects.txt were
2583 incorrect (cf. RFC 3039).
2584 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
2586 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
2587 returns early because it has nothing to do.
2588 [Andy Schneider <andy.schneider@bjss.co.uk>]
2590 *) [In 0.9.6c-engine release:]
2591 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
2592 [Andy Schneider <andy.schneider@bjss.co.uk>]
2594 *) [In 0.9.6c-engine release:]
2595 Add support for Cryptographic Appliance's keyserver technology.
2596 (Use engine 'keyclient')
2597 [Cryptographic Appliances and Geoff Thorpe]
2599 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
2600 is called via tools/c89.sh because arguments have to be
2601 rearranged (all '-L' options must appear before the first object
2603 [Richard Shapiro <rshapiro@abinitio.com>]
2605 *) [In 0.9.6c-engine release:]
2606 Add support for Broadcom crypto accelerator cards, backported
2608 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
2610 *) [In 0.9.6c-engine release:]
2611 Add support for SureWare crypto accelerator cards from
2612 Baltimore Technologies. (Use engine 'sureware')
2613 [Baltimore Technologies and Mark Cox]
2615 *) [In 0.9.6c-engine release:]
2616 Add support for crypto accelerator cards from Accelerated
2617 Encryption Processing, www.aep.ie. (Use engine 'aep')
2618 [AEP Inc. and Mark Cox]
2620 *) Add a configuration entry for gcc on UnixWare.
2621 [Gary Benson <gbenson@redhat.com>]
2623 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
2624 messages are stored in a single piece (fixed-length part and
2625 variable-length part combined) and fix various bugs found on the way.
2628 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
2629 instead. BIO_gethostbyname() does not know what timeouts are
2630 appropriate, so entries would stay in cache even when they have
2632 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
2634 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
2635 faced with a pathologically small ClientHello fragment that does
2636 not contain client_version: Instead of aborting with an error,
2637 simply choose the highest available protocol version (i.e.,
2638 TLS 1.0 unless it is disabled). In practice, ClientHello
2639 messages are never sent like this, but this change gives us
2640 strictly correct behaviour at least for TLS.
2643 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
2644 never resets s->method to s->ctx->method when called from within
2645 one of the SSL handshake functions.
2646 [Bodo Moeller; problem pointed out by Niko Baric]
2648 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
2649 (sent using the client's version number) if client_version is
2650 smaller than the protocol version in use. Also change
2651 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
2652 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
2653 the client will at least see that alert.
2656 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
2660 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
2661 client receives HelloRequest while in a handshake.
2662 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
2664 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
2665 should end in 'break', not 'goto end' which circuments various
2666 cleanups done in state SSL_ST_OK. But session related stuff
2667 must be disabled for SSL_ST_OK in the case that we just sent a
2670 Also avoid some overhead by not calling ssl_init_wbio_buffer()
2671 before just sending a HelloRequest.
2672 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
2674 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
2675 reveal whether illegal block cipher padding was found or a MAC
2676 verification error occured. (Neither SSLerr() codes nor alerts
2677 are directly visible to potential attackers, but the information
2678 may leak via logfiles.)
2680 Similar changes are not required for the SSL 2.0 implementation
2681 because the number of padding bytes is sent in clear for SSL 2.0,
2682 and the extra bytes are just ignored. However ssl/s2_pkt.c
2683 failed to verify that the purported number of padding bytes is in
2687 *) Add OpenUNIX-8 support including shared libraries
2688 (Boyd Lynn Gerber <gerberb@zenez.com>).
2691 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
2692 'wristwatch attack' using huge encoding parameters (cf.
2693 James H. Manger's CRYPTO 2001 paper). Note that the
2694 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
2695 encoding parameters and hence was not vulnerable.
2698 *) BN_sqr() bug fix.
2699 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
2701 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
2702 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
2703 followed by modular reduction.
2704 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
2706 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
2707 equivalent based on BN_pseudo_rand() instead of BN_rand().
2710 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
2711 This function was broken, as the check for a new client hello message
2712 to handle SGC did not allow these large messages.
2713 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
2716 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
2719 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
2720 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
2723 *) Rework the configuration and shared library support for Tru64 Unix.
2724 The configuration part makes use of modern compiler features and
2725 still retains old compiler behavior for those that run older versions
2726 of the OS. The shared library support part includes a variant that
2727 uses the RPATH feature, and is available through the special
2728 configuration target "alpha-cc-rpath", which will never be selected
2730 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
2732 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
2733 with the same message size as in ssl3_get_certificate_request().
2734 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
2735 messages might inadvertently be reject as too long.
2736 [Petr Lampa <lampa@fee.vutbr.cz>]
2738 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
2741 *) Modified SSL library such that the verify_callback that has been set
2742 specificly for an SSL object with SSL_set_verify() is actually being
2743 used. Before the change, a verify_callback set with this function was
2744 ignored and the verify_callback() set in the SSL_CTX at the time of
2745 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
2746 to allow the necessary settings.
2749 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
2750 explicitly to NULL, as at least on Solaris 8 this seems not always to be
2751 done automatically (in contradiction to the requirements of the C
2752 standard). This made problems when used from OpenSSH.
2755 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
2756 dh->length and always used
2758 BN_rand_range(priv_key, dh->p).
2760 BN_rand_range() is not necessary for Diffie-Hellman, and this
2761 specific range makes Diffie-Hellman unnecessarily inefficient if
2762 dh->length (recommended exponent length) is much smaller than the
2763 length of dh->p. We could use BN_rand_range() if the order of
2764 the subgroup was stored in the DH structure, but we only have
2769 BN_rand(priv_key, l, ...)
2771 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
2777 RSA_eay_public_encrypt
2778 RSA_eay_private_decrypt
2779 RSA_eay_private_encrypt (signing)
2780 RSA_eay_public_decrypt (signature verification)
2782 (default implementations for RSA_public_encrypt,
2783 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
2784 always reject numbers >= n.
2787 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
2788 to synchronize access to 'locking_thread'. This is necessary on
2789 systems where access to 'locking_thread' (an 'unsigned long'
2790 variable) is not atomic.
2793 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
2794 *before* setting the 'crypto_lock_rand' flag. The previous code had
2795 a race condition if 0 is a valid thread ID.
2796 [Travis Vitek <vitek@roguewave.com>]
2798 *) Add support for shared libraries under Irix.
2799 [Albert Chin-A-Young <china@thewrittenword.com>]
2801 *) Add configuration option to build on Linux on both big-endian and
2803 [Ralf Baechle <ralf@uni-koblenz.de>]
2805 *) Add the possibility to create shared libraries on HP-UX.
2808 Changes between 0.9.6a and 0.9.6b [9 Jul 2001]
2810 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
2811 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
2812 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
2813 PRNG state recovery was possible based on the output of
2814 one PRNG request appropriately sized to gain knowledge on
2815 'md' followed by enough consecutive 1-byte PRNG requests
2816 to traverse all of 'state'.
2818 1. When updating 'md_local' (the current thread's copy of 'md')
2819 during PRNG output generation, hash all of the previous
2820 'md_local' value, not just the half used for PRNG output.
2822 2. Make the number of bytes from 'state' included into the hash
2823 independent from the number of PRNG bytes requested.
2825 The first measure alone would be sufficient to avoid
2826 Markku-Juhani's attack. (Actually it had never occurred
2827 to me that the half of 'md_local' used for chaining was the
2828 half from which PRNG output bytes were taken -- I had always
2829 assumed that the secret half would be used.) The second
2830 measure makes sure that additional data from 'state' is never
2831 mixed into 'md_local' in small portions; this heuristically
2832 further strengthens the PRNG.
2835 *) Fix crypto/bn/asm/mips3.s.
2838 *) When only the key is given to "enc", the IV is undefined. Print out
2839 an error message in this case.
2842 *) Handle special case when X509_NAME is empty in X509 printing routines.
2845 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
2846 positive and less than q.
2849 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
2850 used: it isn't thread safe and the add_lock_callback should handle
2852 [Paul Rose <Paul.Rose@bridge.com>]
2854 *) Verify that incoming data obeys the block size in
2855 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
2859 [Ulf Möller, Bodo Möller]
2861 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
2862 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
2863 when fixing the server behaviour for backwards-compatible 'client
2864 hello' messages. (Note that the attack is impractical against
2865 SSL 3.0 and TLS 1.0 anyway because length and version checking
2866 means that the probability of guessing a valid ciphertext is
2867 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
2870 Before 0.9.5, the countermeasure (hide the error by generating a
2871 random 'decryption result') did not work properly because
2872 ERR_clear_error() was missing, meaning that SSL_get_error() would
2873 detect the supposedly ignored error.
2875 Both problems are now fixed.
2878 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
2879 (previously it was 1024).
2882 *) Fix for compatibility mode trust settings: ignore trust settings
2883 unless some valid trust or reject settings are present.
2886 *) Fix for blowfish EVP: its a variable length cipher.
2889 *) Fix various bugs related to DSA S/MIME verification. Handle missing
2890 parameters in DSA public key structures and return an error in the
2891 DSA routines if parameters are absent.
2894 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
2895 in the current directory if neither $RANDFILE nor $HOME was set.
2896 RAND_file_name() in 0.9.6a returned NULL in this case. This has
2897 caused some confusion to Windows users who haven't defined $HOME.
2898 Thus RAND_file_name() is changed again: e_os.h can define a
2899 DEFAULT_HOME, which will be used if $HOME is not set.
2900 For Windows, we use "C:"; on other platforms, we still require
2901 environment variables.
2903 *) Move 'if (!initialized) RAND_poll()' into regions protected by
2904 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
2905 having multiple threads call RAND_poll() concurrently.
2908 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
2909 combination of a flag and a thread ID variable.
2910 Otherwise while one thread is in ssleay_rand_bytes (which sets the
2911 flag), *other* threads can enter ssleay_add_bytes without obeying
2912 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
2913 that they do not hold after the first thread unsets add_do_not_lock).
2916 *) Change bctest again: '-x' expressions are not available in all
2920 Changes between 0.9.6 and 0.9.6a [5 Apr 2001]
2922 *) Fix a couple of memory leaks in PKCS7_dataDecode()
2923 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
2925 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
2926 the default extension for executables, if any. Also, make the perl
2927 scripts that use symlink() to test if it really exists and use "cp"
2928 if it doesn't. All this made OpenSSL compilable and installable in
2932 *) Fix for asn1_GetSequence() for indefinite length constructed data.
2933 If SEQUENCE is length is indefinite just set c->slen to the total
2934 amount of data available.
2935 [Steve Henson, reported by shige@FreeBSD.org]
2936 [This change does not apply to 0.9.7.]
2938 *) Change bctest to avoid here-documents inside command substitution
2939 (workaround for FreeBSD /bin/sh bug).
2940 For compatibility with Ultrix, avoid shell functions (introduced
2941 in the bctest version that searches along $PATH).
2944 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
2945 with des_encrypt() defined on some operating systems, like Solaris
2949 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
2950 On the Importance of Eliminating Errors in Cryptographic
2951 Computations, J. Cryptology 14 (2001) 2, 101-119,
2952 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
2955 *) MIPS assembler BIGNUM division bug fix.
2958 *) Disabled incorrect Alpha assembler code.
2961 *) Fix PKCS#7 decode routines so they correctly update the length
2962 after reading an EOC for the EXPLICIT tag.
2964 [This change does not apply to 0.9.7.]
2966 *) Fix bug in PKCS#12 key generation routines. This was triggered
2967 if a 3DES key was generated with a 0 initial byte. Include
2968 PKCS12_BROKEN_KEYGEN compilation option to retain the old
2969 (but broken) behaviour.
2972 *) Enhance bctest to search for a working bc along $PATH and print
2974 [Tim Rice <tim@multitalents.net> via Richard Levitte]
2976 *) Fix memory leaks in err.c: free err_data string if necessary;
2977 don't write to the wrong index in ERR_set_error_data.
2980 *) Implement ssl23_peek (analogous to ssl23_read), which previously
2984 *) Replace rdtsc with _emit statements for VC++ version 5.
2985 [Jeremy Cooper <jeremy@baymoo.org>]
2987 *) Make it possible to reuse SSLv2 sessions.
2990 *) In copy_email() check for >= 0 as a return value for
2991 X509_NAME_get_index_by_NID() since 0 is a valid index.
2992 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
2994 *) Avoid coredump with unsupported or invalid public keys by checking if
2995 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
2996 PKCS7_verify() fails with non detached data.
2999 *) Don't use getenv in library functions when run as setuid/setgid.
3000 New function OPENSSL_issetugid().
3003 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
3004 due to incorrect handling of multi-threading:
3006 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
3008 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
3010 3. Count how many times MemCheck_off() has been called so that
3011 nested use can be treated correctly. This also avoids
3012 inband-signalling in the previous code (which relied on the
3013 assumption that thread ID 0 is impossible).
3016 *) Add "-rand" option also to s_client and s_server.
3019 *) Fix CPU detection on Irix 6.x.
3020 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
3021 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3023 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
3026 [This change does not apply to 0.9.7.]
3028 *) Use the cached encoding of an X509_NAME structure rather than
3029 copying it. This is apparently the reason for the libsafe "errors"
3030 but the code is actually correct.
3033 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
3034 Bleichenbacher's DSA attack.
3035 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
3036 to be set and top=0 forces the highest bit to be set; top=-1 is new
3037 and leaves the highest bit random.
3038 [Ulf Moeller, Bodo Moeller]
3040 *) In the NCONF_...-based implementations for CONF_... queries
3041 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
3042 a temporary CONF structure with the data component set to NULL
3043 (which gives segmentation faults in lh_retrieve).
3044 Instead, use NULL for the CONF pointer in CONF_get_string and
3045 CONF_get_number (which may use environment variables) and directly
3046 return NULL from CONF_get_section.
3049 *) Fix potential buffer overrun for EBCDIC.
3052 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
3053 keyUsage if basicConstraints absent for a CA.
3056 *) Make SMIME_write_PKCS7() write mail header values with a format that
3057 is more generally accepted (no spaces before the semicolon), since
3058 some programs can't parse those values properly otherwise. Also make
3059 sure BIO's that break lines after each write do not create invalid
3063 *) Make the CRL encoding routines work with empty SEQUENCE OF. The
3064 macros previously used would not encode an empty SEQUENCE OF
3065 and break the signature.
3067 [This change does not apply to 0.9.7.]
3069 *) Zero the premaster secret after deriving the master secret in
3073 *) Add some EVP_add_digest_alias registrations (as found in
3074 OpenSSL_add_all_digests()) to SSL_library_init()
3075 aka OpenSSL_add_ssl_algorithms(). This provides improved
3076 compatibility with peers using X.509 certificates
3077 with unconventional AlgorithmIdentifier OIDs.
3080 *) Fix for Irix with NO_ASM.
3081 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
3083 *) ./config script fixes.
3084 [Ulf Moeller, Richard Levitte]
3086 *) Fix 'openssl passwd -1'.
3089 *) Change PKCS12_key_gen_asc() so it can cope with non null
3090 terminated strings whose length is passed in the passlen
3091 parameter, for example from PEM callbacks. This was done
3092 by adding an extra length parameter to asc2uni().
3093 [Steve Henson, reported by <oddissey@samsung.co.kr>]
3095 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
3096 call failed, free the DSA structure.
3099 *) Fix to uni2asc() to cope with zero length Unicode strings.
3100 These are present in some PKCS#12 files.
3103 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
3104 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
3105 when writing a 32767 byte record.
3106 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
3108 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
3109 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
3111 (RSA objects have a reference count access to which is protected
3112 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
3113 so they are meant to be shared between threads.)
3114 [Bodo Moeller, Geoff Thorpe; original patch submitted by
3115 "Reddie, Steven" <Steven.Reddie@ca.com>]
3117 *) Fix a deadlock in CRYPTO_mem_leaks().
3120 *) Use better test patterns in bntest.
3123 *) rand_win.c fix for Borland C.
3126 *) BN_rshift bugfix for n == 0.
3129 *) Add a 'bctest' script that checks for some known 'bc' bugs
3130 so that 'make test' does not abort just because 'bc' is broken.
3133 *) Store verify_result within SSL_SESSION also for client side to
3134 avoid potential security hole. (Re-used sessions on the client side
3135 always resulted in verify_result==X509_V_OK, not using the original
3136 result of the server certificate verification.)
3139 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
3140 SSL3_RT_APPLICATION_DATA, return 0.
3141 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
3145 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
3146 releases, have been re-implemented by renaming the previous
3147 implementations of ssl2_read and ssl3_read to ssl2_read_internal
3148 and ssl3_read_internal, respectively, and adding 'peek' parameters
3149 to them. The new ssl[23]_{read,peek} functions are calls to
3150 ssl[23]_read_internal with the 'peek' flag set appropriately.
3151 A 'peek' parameter has also been added to ssl3_read_bytes, which
3152 does the actual work for ssl3_read_internal.
3155 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
3156 the method-specific "init()" handler. Also clean up ex_data after
3157 calling the method-specific "finish()" handler. Previously, this was
3158 happening the other way round.
3161 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
3162 The previous value, 12, was not always sufficient for BN_mod_exp().
3165 *) Make sure that shared libraries get the internal name engine with
3166 the full version number and not just 0. This should mark the
3167 shared libraries as not backward compatible. Of course, this should
3168 be changed again when we can guarantee backward binary compatibility.
3171 *) Fix typo in get_cert_by_subject() in by_dir.c
3172 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
3174 *) Rework the system to generate shared libraries:
3176 - Make note of the expected extension for the shared libraries and
3177 if there is a need for symbolic links from for example libcrypto.so.0
3178 to libcrypto.so.0.9.7. There is extended info in Configure for
3181 - Make as few rebuilds of the shared libraries as possible.
3183 - Still avoid linking the OpenSSL programs with the shared libraries.
3185 - When installing, install the shared libraries separately from the
3189 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
3191 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
3192 and not in SSL_clear because the latter is also used by the
3193 accept/connect functions; previously, the settings made by
3194 SSL_set_read_ahead would be lost during the handshake.
3195 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
3197 *) Correct util/mkdef.pl to be selective about disabled algorithms.
3198 Previously, it would create entries for disableed algorithms no
3202 *) Added several new manual pages for SSL_* function.
3205 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
3207 *) In ssl23_get_client_hello, generate an error message when faced
3208 with an initial SSL 3.0/TLS record that is too small to contain the
3209 first two bytes of the ClientHello message, i.e. client_version.
3210 (Note that this is a pathologic case that probably has never happened
3211 in real life.) The previous approach was to use the version number
3212 from the record header as a substitute; but our protocol choice
3213 should not depend on that one because it is not authenticated
3214 by the Finished messages.
3217 *) More robust randomness gathering functions for Windows.
3218 [Jeffrey Altman <jaltman@columbia.edu>]
3220 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
3221 not set then we don't setup the error code for issuer check errors
3222 to avoid possibly overwriting other errors which the callback does
3223 handle. If an application does set the flag then we assume it knows
3224 what it is doing and can handle the new informational codes
3228 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
3229 a general "ANY" type, as such it should be able to decode anything
3230 including tagged types. However it didn't check the class so it would
3231 wrongly interpret tagged types in the same way as their universal
3232 counterpart and unknown types were just rejected. Changed so that the
3233 tagged and unknown types are handled in the same way as a SEQUENCE:
3234 that is the encoding is stored intact. There is also a new type
3235 "V_ASN1_OTHER" which is used when the class is not universal, in this
3236 case we have no idea what the actual type is so we just lump them all
3240 *) On VMS, stdout may very well lead to a file that is written to
3241 in a record-oriented fashion. That means that every write() will
3242 write a separate record, which will be read separately by the
3243 programs trying to read from it. This can be very confusing.
3245 The solution is to put a BIO filter in the way that will buffer
3246 text until a linefeed is reached, and then write everything a
3247 line at a time, so every record written will be an actual line,
3248 not chunks of lines and not (usually doesn't happen, but I've
3249 seen it once) several lines in one record. BIO_f_linebuffer() is
3252 Currently, it's a VMS-only method, because that's where it has
3253 been tested well enough.
3256 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
3257 it can return incorrect results.
3258 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
3259 but it was in 0.9.6-beta[12].)
3262 *) Disable the check for content being present when verifying detached
3263 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
3264 include zero length content when signing messages.
3267 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
3268 BIO_ctrl (for BIO pairs).
3271 *) Add DSO method for VMS.
3274 *) Bug fix: Montgomery multiplication could produce results with the
3278 *) Add RPM specification openssl.spec and modify it to build three
3279 packages. The default package contains applications, application
3280 documentation and run-time libraries. The devel package contains
3281 include files, static libraries and function documentation. The
3282 doc package contains the contents of the doc directory. The original
3283 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
3286 *) Add a large number of documentation files for many SSL routines.
3287 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3289 *) Add a configuration entry for Sony News 4.
3290 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
3292 *) Don't set the two most significant bits to one when generating a
3293 random number < q in the DSA library.
3296 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
3297 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
3298 the underlying transport is blocking) if a handshake took place.
3299 (The default behaviour is needed by applications such as s_client
3300 and s_server that use select() to determine when to use SSL_read;
3301 but for applications that know in advance when to expect data, it
3302 just makes things more complicated.)
3305 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
3309 *) Add a few more EBCDIC conditionals that make `req' and `x509'
3310 work better on such systems.
3311 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
3313 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
3314 Update PKCS12_parse() so it copies the friendlyName and the
3315 keyid to the certificates aux info.
3318 *) Fix bug in PKCS7_verify() which caused an infinite loop
3319 if there was more than one signature.
3320 [Sven Uszpelkat <su@celocom.de>]
3322 *) Major change in util/mkdef.pl to include extra information
3323 about each symbol, as well as presentig variables as well
3324 as functions. This change means that there's n more need
3325 to rebuild the .num files when some algorithms are excluded.
3328 *) Allow the verify time to be set by an application,
3329 rather than always using the current time.
3332 *) Phase 2 verify code reorganisation. The certificate
3333 verify code now looks up an issuer certificate by a
3334 number of criteria: subject name, authority key id
3335 and key usage. It also verifies self signed certificates
3336 by the same criteria. The main comparison function is
3337 X509_check_issued() which performs these checks.
3339 Lot of changes were necessary in order to support this
3340 without completely rewriting the lookup code.
3342 Authority and subject key identifier are now cached.
3344 The LHASH 'certs' is X509_STORE has now been replaced
3345 by a STACK_OF(X509_OBJECT). This is mainly because an
3346 LHASH can't store or retrieve multiple objects with
3347 the same hash value.
3349 As a result various functions (which were all internal
3350 use only) have changed to handle the new X509_STORE
3351 structure. This will break anything that messed round
3352 with X509_STORE internally.
3354 The functions X509_STORE_add_cert() now checks for an
3355 exact match, rather than just subject name.
3357 The X509_STORE API doesn't directly support the retrieval
3358 of multiple certificates matching a given criteria, however
3359 this can be worked round by performing a lookup first
3360 (which will fill the cache with candidate certificates)
3361 and then examining the cache for matches. This is probably
3362 the best we can do without throwing out X509_LOOKUP
3363 entirely (maybe later...).
3365 The X509_VERIFY_CTX structure has been enhanced considerably.
3367 All certificate lookup operations now go via a get_issuer()
3368 callback. Although this currently uses an X509_STORE it
3369 can be replaced by custom lookups. This is a simple way
3370 to bypass the X509_STORE hackery necessary to make this
3371 work and makes it possible to use more efficient techniques
3372 in future. A very simple version which uses a simple
3373 STACK for its trusted certificate store is also provided
3374 using X509_STORE_CTX_trusted_stack().
3376 The verify_cb() and verify() callbacks now have equivalents
3377 in the X509_STORE_CTX structure.
3379 X509_STORE_CTX also has a 'flags' field which can be used
3380 to customise the verify behaviour.
3383 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
3384 excludes S/MIME capabilities.
3387 *) When a certificate request is read in keep a copy of the
3388 original encoding of the signed data and use it when outputing
3389 again. Signatures then use the original encoding rather than
3390 a decoded, encoded version which may cause problems if the
3391 request is improperly encoded.
3394 *) For consistency with other BIO_puts implementations, call
3395 buffer_write(b, ...) directly in buffer_puts instead of calling
3398 In BIO_puts, increment b->num_write as in BIO_write.
3399 [Peter.Sylvester@EdelWeb.fr]
3401 *) Fix BN_mul_word for the case where the word is 0. (We have to use
3402 BN_zero, we may not return a BIGNUM with an array consisting of
3406 *) Avoid calling abort() from within the library when problems are
3407 detected, except if preprocessor symbols have been defined
3408 (such as REF_CHECK, BN_DEBUG etc.).
3411 *) New openssl application 'rsautl'. This utility can be
3412 used for low level RSA operations. DER public key
3413 BIO/fp routines also added.
3416 *) New Configure entry and patches for compiling on QNX 4.
3417 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
3419 *) A demo state-machine implementation was sponsored by
3420 Nuron (http://www.nuron.com/) and is now available in
3421 demos/state_machine.
3424 *) New options added to the 'dgst' utility for signature
3425 generation and verification.
3428 *) Unrecognized PKCS#7 content types are now handled via a
3429 catch all ASN1_TYPE structure. This allows unsupported
3430 types to be stored as a "blob" and an application can
3431 encode and decode it manually.
3434 *) Fix various signed/unsigned issues to make a_strex.c
3436 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
3438 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
3439 length if passed a buffer. ASN1_INTEGER_to_BN failed
3440 if passed a NULL BN and its argument was negative.
3441 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
3443 *) Modification to PKCS#7 encoding routines to output definite
3444 length encoding. Since currently the whole structures are in
3445 memory there's not real point in using indefinite length
3446 constructed encoding. However if OpenSSL is compiled with
3447 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
3450 *) Added BIO_vprintf() and BIO_vsnprintf().
3453 *) Added more prefixes to parse for in the the strings written
3454 through a logging bio, to cover all the levels that are available
3455 through syslog. The prefixes are now:
3457 PANIC, EMERG, EMR => LOG_EMERG
3458 ALERT, ALR => LOG_ALERT
3459 CRIT, CRI => LOG_CRIT
3460 ERROR, ERR => LOG_ERR
3461 WARNING, WARN, WAR => LOG_WARNING
3462 NOTICE, NOTE, NOT => LOG_NOTICE
3463 INFO, INF => LOG_INFO
3464 DEBUG, DBG => LOG_DEBUG
3466 and as before, if none of those prefixes are present at the
3467 beginning of the string, LOG_ERR is chosen.
3469 On Win32, the LOG_* levels are mapped according to this:
3471 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
3472 LOG_WARNING => EVENTLOG_WARNING_TYPE
3473 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
3477 *) Made it possible to reconfigure with just the configuration
3478 argument "reconf" or "reconfigure". The command line arguments
3479 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
3480 and are retrieved from there when reconfiguring.
3484 [Assar Westerlund <assar@sics.se>, Richard Levitte]
3486 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
3489 *) The obj_dat.pl script was messing up the sorting of object
3490 names. The reason was that it compared the quoted version
3491 of strings as a result "OCSP" > "OCSP Signing" because
3492 " > SPACE. Changed script to store unquoted versions of
3493 names and add quotes on output. It was also omitting some
3494 names from the lookup table if they were given a default
3495 value (that is if SN is missing it is given the same
3496 value as LN and vice versa), these are now added on the
3497 grounds that if an object has a name we should be able to
3498 look it up. Finally added warning output when duplicate
3499 short or long names are found.
3502 *) Changes needed for Tandem NSK.
3503 [Scott Uroff <scott@xypro.com>]
3505 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
3506 RSA_padding_check_SSLv23(), special padding was never detected
3507 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
3508 version rollback attacks was not effective.
3510 In s23_clnt.c, don't use special rollback-attack detection padding
3511 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
3512 client; similarly, in s23_srvr.c, don't do the rollback check if
3513 SSL 2.0 is the only protocol enabled in the server.
3516 *) Make it possible to get hexdumps of unprintable data with 'openssl
3517 asn1parse'. By implication, the functions ASN1_parse_dump() and
3518 BIO_dump_indent() are added.
3521 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
3522 these print out strings and name structures based on various
3523 flags including RFC2253 support and proper handling of
3524 multibyte characters. Added options to the 'x509' utility
3525 to allow the various flags to be set.
3528 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
3529 Also change the functions X509_cmp_current_time() and
3530 X509_gmtime_adj() work with an ASN1_TIME structure,
3531 this will enable certificates using GeneralizedTime in validity
3532 dates to be checked.
3535 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
3536 negative public key encodings) on by default,
3537 NO_NEG_PUBKEY_BUG can be set to disable it.
3540 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
3541 content octets. An i2c_ASN1_OBJECT is unnecessary because
3542 the encoding can be trivially obtained from the structure.
3545 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
3546 not read locks (CRYPTO_r_[un]lock).
3549 *) A first attempt at creating official support for shared
3550 libraries through configuration. I've kept it so the
3551 default is static libraries only, and the OpenSSL programs
3552 are always statically linked for now, but there are
3553 preparations for dynamic linking in place.
3554 This has been tested on Linux and Tru64.
3557 *) Randomness polling function for Win9x, as described in:
3558 Peter Gutmann, Software Generation of Practically Strong
3562 *) Fix so PRNG is seeded in req if using an already existing
3566 *) New options to smime application. -inform and -outform
3567 allow alternative formats for the S/MIME message including
3568 PEM and DER. The -content option allows the content to be
3569 specified separately. This should allow things like Netscape
3570 form signing output easier to verify.
3573 *) Fix the ASN1 encoding of tags using the 'long form'.
3576 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
3577 STRING types. These convert content octets to and from the
3578 underlying type. The actual tag and length octets are
3579 already assumed to have been read in and checked. These
3580 are needed because all other string types have virtually
3581 identical handling apart from the tag. By having versions
3582 of the ASN1 functions that just operate on content octets
3583 IMPLICIT tagging can be handled properly. It also allows
3584 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
3585 and ASN1_INTEGER are identical apart from the tag.
3588 *) Change the handling of OID objects as follows:
3590 - New object identifiers are inserted in objects.txt, following
3591 the syntax given in objects.README.
3592 - objects.pl is used to process obj_mac.num and create a new
3594 - obj_dat.pl is used to create a new obj_dat.h, using the data in
3597 This is currently kind of a hack, and the perl code in objects.pl
3598 isn't very elegant, but it works as I intended. The simplest way
3599 to check that it worked correctly is to look in obj_dat.h and
3600 check the array nid_objs and make sure the objects haven't moved
3601 around (this is important!). Additions are OK, as well as
3602 consistent name changes.
3605 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
3608 *) Addition of the command line parameter '-rand file' to 'openssl req'.
3609 The given file adds to whatever has already been seeded into the
3610 random pool through the RANDFILE configuration file option or
3611 environment variable, or the default random state file.
3614 *) mkstack.pl now sorts each macro group into lexical order.
3615 Previously the output order depended on the order the files
3616 appeared in the directory, resulting in needless rewriting
3620 *) Patches to make OpenSSL compile under Win32 again. Mostly
3621 work arounds for the VC++ problem that it treats func() as
3622 func(void). Also stripped out the parts of mkdef.pl that
3623 added extra typesafe functions: these no longer exist.
3626 *) Reorganisation of the stack code. The macros are now all
3627 collected in safestack.h . Each macro is defined in terms of
3628 a "stack macro" of the form SKM_<name>(type, a, b). The
3629 DEBUG_SAFESTACK is now handled in terms of function casts,
3630 this has the advantage of retaining type safety without the
3631 use of additional functions. If DEBUG_SAFESTACK is not defined
3632 then the non typesafe macros are used instead. Also modified the
3633 mkstack.pl script to handle the new form. Needs testing to see
3634 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
3635 the default if no major problems. Similar behaviour for ASN1_SET_OF
3636 and PKCS12_STACK_OF.
3639 *) When some versions of IIS use the 'NET' form of private key the
3640 key derivation algorithm is different. Normally MD5(password) is
3641 used as a 128 bit RC4 key. In the modified case
3642 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
3643 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
3644 as the old Netscape_RSA functions except they have an additional
3645 'sgckey' parameter which uses the modified algorithm. Also added
3646 an -sgckey command line option to the rsa utility. Thanks to
3647 Adrian Peck <bertie@ncipher.com> for posting details of the modified
3648 algorithm to openssl-dev.
3651 *) The evp_local.h macros were using 'c.##kname' which resulted in
3652 invalid expansion on some systems (SCO 5.0.5 for example).
3653 Corrected to 'c.kname'.
3654 [Phillip Porch <root@theporch.com>]
3656 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
3657 a STACK of email addresses from a certificate or request, these look
3658 in the subject name and the subject alternative name extensions and
3659 omit any duplicate addresses.
3662 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
3663 This makes DSA verification about 2 % faster.
3666 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
3667 (meaning that now 2^5 values will be precomputed, which is only 4 KB
3668 plus overhead for 1024 bit moduli).
3669 This makes exponentiations about 0.5 % faster for 1024 bit
3670 exponents (as measured by "openssl speed rsa2048").
3673 *) Rename memory handling macros to avoid conflicts with other
3675 Malloc => OPENSSL_malloc
3676 Malloc_locked => OPENSSL_malloc_locked
3677 Realloc => OPENSSL_realloc
3678 Free => OPENSSL_free
3681 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
3682 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
3685 *) CygWin32 support.
3686 [John Jarvie <jjarvie@newsguy.com>]
3688 *) The type-safe stack code has been rejigged. It is now only compiled
3689 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
3690 by default all type-specific stack functions are "#define"d back to
3691 standard stack functions. This results in more streamlined output
3692 but retains the type-safety checking possibilities of the original
3696 *) The STACK code has been cleaned up, and certain type declarations
3697 that didn't make a lot of sense have been brought in line. This has
3698 also involved a cleanup of sorts in safestack.h to more correctly
3699 map type-safe stack functions onto their plain stack counterparts.
3700 This work has also resulted in a variety of "const"ifications of
3701 lots of the code, especially "_cmp" operations which should normally
3702 be prototyped with "const" parameters anyway.
3705 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
3706 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
3707 (The PRNG state consists of two parts, the large pool 'state' and 'md',
3708 where all of 'md' is used each time the PRNG is used, but 'state'
3709 is used only indexed by a cyclic counter. As entropy may not be
3710 well distributed from the beginning, 'md' is important as a
3711 chaining variable. However, the output function chains only half
3712 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
3713 all of 'md', and seeding with STATE_SIZE dummy bytes will result
3714 in all of 'state' being rewritten, with the new values depending
3715 on virtually all of 'md'. This overcomes the 80 bit limitation.)
3718 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
3719 the handshake is continued after ssl_verify_cert_chain();
3720 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
3721 can lead to 'unexplainable' connection aborts later.
3722 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
3724 *) Major EVP API cipher revision.
3725 Add hooks for extra EVP features. This allows various cipher
3726 parameters to be set in the EVP interface. Support added for variable
3727 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
3728 setting of RC2 and RC5 parameters.
3730 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
3733 Remove lots of duplicated code from the EVP library. For example *every*
3734 cipher init() function handles the 'iv' in the same way according to the
3735 cipher mode. They also all do nothing if the 'key' parameter is NULL and
3736 for CFB and OFB modes they zero ctx->num.
3738 New functionality allows removal of S/MIME code RC2 hack.
3740 Most of the routines have the same form and so can be declared in terms
3743 By shifting this to the top level EVP_CipherInit() it can be removed from
3744 all individual ciphers. If the cipher wants to handle IVs or keys
3745 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
3748 Change lots of functions like EVP_EncryptUpdate() to now return a
3749 value: although software versions of the algorithms cannot fail
3750 any installed hardware versions can.
3753 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
3754 this option is set, tolerate broken clients that send the negotiated
3755 protocol version number instead of the requested protocol version
3759 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
3760 i.e. non-zero for export ciphersuites, zero otherwise.
3761 Previous versions had this flag inverted, inconsistent with
3762 rsa_tmp_cb (..._TMP_RSA_CB).
3763 [Bodo Moeller; problem reported by Amit Chopra]
3765 *) Add missing DSA library text string. Work around for some IIS
3766 key files with invalid SEQUENCE encoding.
3769 *) Add a document (doc/standards.txt) that list all kinds of standards
3770 and so on that are implemented in OpenSSL.
3773 *) Enhance c_rehash script. Old version would mishandle certificates
3774 with the same subject name hash and wouldn't handle CRLs at all.
3775 Added -fingerprint option to crl utility, to support new c_rehash
3779 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
3782 *) Fix for SSL server purpose checking. Server checking was
3783 rejecting certificates which had extended key usage present
3784 but no ssl client purpose.
3785 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
3787 *) Make PKCS#12 code work with no password. The PKCS#12 spec
3788 is a little unclear about how a blank password is handled.
3789 Since the password in encoded as a BMPString with terminating
3790 double NULL a zero length password would end up as just the
3791 double NULL. However no password at all is different and is
3792 handled differently in the PKCS#12 key generation code. NS
3793 treats a blank password as zero length. MSIE treats it as no
3794 password on export: but it will try both on import. We now do
3795 the same: PKCS12_parse() tries zero length and no password if
3796 the password is set to "" or NULL (NULL is now a valid password:
3797 it wasn't before) as does the pkcs12 application.
3800 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
3801 perror when PEM_read_bio_X509_REQ fails, the error message must
3802 be obtained from the error queue.
3805 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
3806 it in ERR_remove_state if appropriate, and change ERR_get_state
3807 accordingly to avoid race conditions (this is necessary because
3808 thread_hash is no longer constant once set).
3811 *) Bugfix for linux-elf makefile.one.
3814 *) RSA_get_default_method() will now cause a default
3815 RSA_METHOD to be chosen if one doesn't exist already.
3816 Previously this was only set during a call to RSA_new()
3817 or RSA_new_method(NULL) meaning it was possible for
3818 RSA_get_default_method() to return NULL.
3821 *) Added native name translation to the existing DSO code
3822 that will convert (if the flag to do so is set) filenames
3823 that are sufficiently small and have no path information
3824 into a canonical native form. Eg. "blah" converted to
3825 "libblah.so" or "blah.dll" etc.
3828 *) New function ERR_error_string_n(e, buf, len) which is like
3829 ERR_error_string(e, buf), but writes at most 'len' bytes
3830 including the 0 terminator. For ERR_error_string_n, 'buf'
3832 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
3834 *) CONF library reworked to become more general. A new CONF
3835 configuration file reader "class" is implemented as well as a
3836 new functions (NCONF_*, for "New CONF") to handle it. The now
3837 old CONF_* functions are still there, but are reimplemented to
3838 work in terms of the new functions. Also, a set of functions
3839 to handle the internal storage of the configuration data is
3840 provided to make it easier to write new configuration file
3841 reader "classes" (I can definitely see something reading a
3842 configuration file in XML format, for example), called _CONF_*,
3843 or "the configuration storage API"...
3845 The new configuration file reading functions are:
3847 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
3848 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
3850 NCONF_default, NCONF_WIN32
3852 NCONF_dump_fp, NCONF_dump_bio
3854 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
3855 NCONF_new creates a new CONF object. This works in the same way
3856 as other interfaces in OpenSSL, like the BIO interface.
3857 NCONF_dump_* dump the internal storage of the configuration file,
3858 which is useful for debugging. All other functions take the same
3859 arguments as the old CONF_* functions wth the exception of the
3860 first that must be a `CONF *' instead of a `LHASH *'.
3862 To make it easer to use the new classes with the old CONF_* functions,
3863 the function CONF_set_default_method is provided.
3866 *) Add '-tls1' option to 'openssl ciphers', which was already
3867 mentioned in the documentation but had not been implemented.
3868 (This option is not yet really useful because even the additional
3869 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
3872 *) Initial DSO code added into libcrypto for letting OpenSSL (and
3873 OpenSSL-based applications) load shared libraries and bind to
3874 them in a portable way.
3875 [Geoff Thorpe, with contributions from Richard Levitte]
3877 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
3879 *) Make sure _lrotl and _lrotr are only used with MSVC.
3881 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
3882 (the default implementation of RAND_status).
3884 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
3885 to '-clrext' (= clear extensions), as intended and documented.
3886 [Bodo Moeller; inconsistency pointed out by Michael Attili
3889 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
3890 was larger than the MD block size.
3891 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
3893 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
3894 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
3895 using the passed key: if the passed key was a private key the result
3896 of X509_print(), for example, would be to print out all the private key
3900 *) des_quad_cksum() byte order bug fix.
3901 [Ulf Möller, using the problem description in krb4-0.9.7, where
3902 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
3904 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
3906 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
3908 *) For easily testing in shell scripts whether some command
3909 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
3910 returns with exit code 0 iff no command of the given name is available.
3911 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
3912 the output goes to stdout and nothing is printed to stderr.
3913 Additional arguments are always ignored.
3915 Since for each cipher there is a command of the same name,
3916 the 'no-cipher' compilation switches can be tested this way.
3918 ('openssl no-XXX' is not able to detect pseudo-commands such
3919 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
3922 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
3925 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
3926 is set; it will be thrown away anyway because each handshake creates
3928 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
3929 to parameters -- in previous versions (since OpenSSL 0.9.3) the
3930 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
3931 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
3934 *) New s_client option -ign_eof: EOF at stdin is ignored, and
3935 'Q' and 'R' lose their special meanings (quit/renegotiate).
3936 This is part of what -quiet does; unlike -quiet, -ign_eof
3937 does not suppress any output.
3940 *) Add compatibility options to the purpose and trust code. The
3941 purpose X509_PURPOSE_ANY is "any purpose" which automatically
3942 accepts a certificate or CA, this was the previous behaviour,
3943 with all the associated security issues.
3945 X509_TRUST_COMPAT is the old trust behaviour: only and
3946 automatically trust self signed roots in certificate store. A
3947 new trust setting X509_TRUST_DEFAULT is used to specify that
3948 a purpose has no associated trust setting and it should instead
3949 use the value in the default purpose.
3952 *) Fix the PKCS#8 DSA private key code so it decodes keys again
3953 and fix a memory leak.
3956 *) In util/mkerr.pl (which implements 'make errors'), preserve
3957 reason strings from the previous version of the .c file, as
3958 the default to have only downcase letters (and digits) in
3959 automatically generated reasons codes is not always appropriate.
3962 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
3963 using strerror. Previously, ERR_reason_error_string() returned
3964 library names as reason strings for SYSerr; but SYSerr is a special
3965 case where small numbers are errno values, not library numbers.
3968 *) Add '-dsaparam' option to 'openssl dhparam' application. This
3969 converts DSA parameters into DH parameters. (When creating parameters,
3970 DSA_generate_parameters is used.)
3973 *) Include 'length' (recommended exponent length) in C code generated
3974 by 'openssl dhparam -C'.
3977 *) The second argument to set_label in perlasm was already being used
3978 so couldn't be used as a "file scope" flag. Moved to third argument
3982 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
3983 instead of RAND_bytes for encryption IVs and salts.
3986 *) Include RAND_status() into RAND_METHOD instead of implementing
3987 it only for md_rand.c Otherwise replacing the PRNG by calling
3988 RAND_set_rand_method would be impossible.
3991 *) Don't let DSA_generate_key() enter an infinite loop if the random
3992 number generation fails.
3995 *) New 'rand' application for creating pseudo-random output.
3998 *) Added configuration support for Linux/IA64
3999 [Rolf Haberrecker <rolf@suse.de>]
4001 *) Assembler module support for Mingw32.
4004 *) Shared library support for HPUX (in shlib/).
4005 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
4007 *) Shared library support for Solaris gcc.
4008 [Lutz Behnke <behnke@trustcenter.de>]
4010 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
4012 *) PKCS7_encrypt() was adding text MIME headers twice because they
4013 were added manually and by SMIME_crlf_copy().
4016 *) In bntest.c don't call BN_rand with zero bits argument.
4017 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
4019 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
4020 case was implemented. This caused BN_div_recp() to fail occasionally.
4023 *) Add an optional second argument to the set_label() in the perl
4024 assembly language builder. If this argument exists and is set
4025 to 1 it signals that the assembler should use a symbol whose
4026 scope is the entire file, not just the current function. This
4027 is needed with MASM which uses the format label:: for this scope.
4028 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
4030 *) Change the ASN1 types so they are typedefs by default. Before
4031 almost all types were #define'd to ASN1_STRING which was causing
4032 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
4036 *) Change names of new functions to the new get1/get0 naming
4037 convention: After 'get1', the caller owns a reference count
4038 and has to call ..._free; 'get0' returns a pointer to some
4039 data structure without incrementing reference counters.
4040 (Some of the existing 'get' functions increment a reference
4041 counter, some don't.)
4042 Similarly, 'set1' and 'add1' functions increase reference
4043 counters or duplicate objects.
4046 *) Allow for the possibility of temp RSA key generation failure:
4047 the code used to assume it always worked and crashed on failure.
4050 *) Fix potential buffer overrun problem in BIO_printf().
4051 [Ulf Möller, using public domain code by Patrick Powell; problem
4052 pointed out by David Sacerdote <das33@cornell.edu>]
4054 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
4055 RAND_egd() and RAND_status(). In the command line application,
4056 the EGD socket can be specified like a seed file using RANDFILE
4060 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
4061 Some CAs (e.g. Verisign) distribute certificates in this form.
4064 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
4065 list to exclude them. This means that no special compilation option
4066 is needed to use anonymous DH: it just needs to be included in the
4070 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
4071 EVP_MD_type. The old functionality is available in a new macro called
4072 EVP_MD_md(). Change code that uses it and update docs.
4075 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
4076 where the 'void *' argument is replaced by a function pointer argument.
4077 Previously 'void *' was abused to point to functions, which works on
4078 many platforms, but is not correct. As these functions are usually
4079 called by macros defined in OpenSSL header files, most source code
4080 should work without changes.
4083 *) <openssl/opensslconf.h> (which is created by Configure) now contains
4084 sections with information on -D... compiler switches used for
4085 compiling the library so that applications can see them. To enable
4086 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
4087 must be defined. E.g.,
4088 #define OPENSSL_ALGORITHM_DEFINES
4089 #include <openssl/opensslconf.h>
4090 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
4091 [Richard Levitte, Ulf and Bodo Möller]
4093 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
4097 *) Change the 'other' type in certificate aux info to a STACK_OF
4098 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
4099 the required ASN1 format: arbitrary types determined by an OID.
4102 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
4103 argument to 'req'. This is not because the function is newer or
4104 better than others it just uses the work 'NEW' in the certificate
4105 request header lines. Some software needs this.
4108 *) Reorganise password command line arguments: now passwords can be
4109 obtained from various sources. Delete the PEM_cb function and make
4110 it the default behaviour: i.e. if the callback is NULL and the
4111 usrdata argument is not NULL interpret it as a null terminated pass
4112 phrase. If usrdata and the callback are NULL then the pass phrase
4113 is prompted for as usual.
4116 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
4117 the support is automatically enabled. The resulting binaries will
4118 autodetect the card and use it if present.
4119 [Ben Laurie and Compaq Inc.]
4121 *) Work around for Netscape hang bug. This sends certificate request
4122 and server done in one record. Since this is perfectly legal in the
4123 SSL/TLS protocol it isn't a "bug" option and is on by default. See
4124 the bugs/SSLv3 entry for more info.
4127 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
4130 *) Add -rand argument to smime and pkcs12 applications and read/write
4134 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
4137 *) Add command line password options to the remaining applications.
4140 *) Bug fix for BN_div_recp() for numerators with an even number of
4144 *) More tests in bntest.c, and changed test_bn output.
4147 *) ./config recognizes MacOS X now.
4150 *) Bug fix for BN_div() when the first words of num and divsor are
4151 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
4154 *) Add support for various broken PKCS#8 formats, and command line
4155 options to produce them.
4158 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
4159 get temporary BIGNUMs from a BN_CTX.
4162 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
4166 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
4167 include a #define from the old name to the new. The original intent
4168 was that statically linked binaries could for example just call
4169 SSLeay_add_all_ciphers() to just add ciphers to the table and not
4170 link with digests. This never worked becayse SSLeay_add_all_digests()
4171 and SSLeay_add_all_ciphers() were in the same source file so calling
4172 one would link with the other. They are now in separate source files.
4175 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
4178 *) Use a less unusual form of the Miller-Rabin primality test (it used
4179 a binary algorithm for exponentiation integrated into the Miller-Rabin
4180 loop, our standard modexp algorithms are faster).
4183 *) Support for the EBCDIC character set completed.
4184 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
4186 *) Source code cleanups: use const where appropriate, eliminate casts,
4187 use void * instead of char * in lhash.
4190 *) Bugfix: ssl3_send_server_key_exchange was not restartable
4191 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
4192 this the server could overwrite ephemeral keys that the client
4196 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
4197 using 50 iterations of the Rabin-Miller test.
4199 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
4200 iterations of the Rabin-Miller test as required by the appendix
4201 to FIPS PUB 186[-1]) instead of DSA_is_prime.
4202 As BN_is_prime_fasttest includes trial division, DSA parameter
4203 generation becomes much faster.
4205 This implies a change for the callback functions in DSA_is_prime
4206 and DSA_generate_parameters: The callback function is called once
4207 for each positive witness in the Rabin-Miller test, not just
4208 occasionally in the inner loop; and the parameters to the
4209 callback function now provide an iteration count for the outer
4210 loop rather than for the current invocation of the inner loop.
4211 DSA_generate_parameters additionally can call the callback
4212 function with an 'iteration count' of -1, meaning that a
4213 candidate has passed the trial division test (when q is generated
4214 from an application-provided seed, trial division is skipped).
4217 *) New function BN_is_prime_fasttest that optionally does trial
4218 division before starting the Rabin-Miller test and has
4219 an additional BN_CTX * argument (whereas BN_is_prime always
4220 has to allocate at least one BN_CTX).
4221 'callback(1, -1, cb_arg)' is called when a number has passed the
4222 trial division stage.
4225 *) Fix for bug in CRL encoding. The validity dates weren't being handled
4229 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
4232 *) New function BN_pseudo_rand().
4235 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
4236 bignum version of BN_from_montgomery() with the working code from
4237 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
4241 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
4242 made it impossible to use the same SSL_SESSION data structure in
4243 SSL2 clients in multiple threads.
4246 *) The return value of RAND_load_file() no longer counts bytes obtained
4247 by stat(). RAND_load_file(..., -1) is new and uses the complete file
4248 to seed the PRNG (previously an explicit byte count was required).
4249 [Ulf Möller, Bodo Möller]
4251 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
4252 used (char *) instead of (void *) and had casts all over the place.
4255 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
4258 *) Retain source code compatibility for BN_prime_checks macro:
4259 BN_is_prime(..., BN_prime_checks, ...) now uses
4260 BN_prime_checks_for_size to determine the appropriate number of
4261 Rabin-Miller iterations.
4264 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
4265 DH_CHECK_P_NOT_SAFE_PRIME.
4266 (Check if this is true? OpenPGP calls them "strong".)
4269 *) Merge the functionality of "dh" and "gendh" programs into a new program
4270 "dhparam". The old programs are retained for now but will handle DH keys
4271 (instead of parameters) in future.
4274 *) Make the ciphers, s_server and s_client programs check the return values
4275 when a new cipher list is set.
4278 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
4279 ciphers. Before when the 56bit ciphers were enabled the sorting was
4282 The syntax for the cipher sorting has been extended to support sorting by
4283 cipher-strength (using the strength_bits hard coded in the tables).
4284 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
4286 Fix a bug in the cipher-command parser: when supplying a cipher command
4287 string with an "undefined" symbol (neither command nor alphanumeric
4288 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
4289 an error is flagged.
4291 Due to the strength-sorting extension, the code of the
4292 ssl_create_cipher_list() function was completely rearranged. I hope that
4293 the readability was also increased :-)
4294 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
4296 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
4297 for the first serial number and places 2 in the serial number file. This
4298 avoids problems when the root CA is created with serial number zero and
4299 the first user certificate has the same issuer name and serial number
4303 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
4304 the new code. Add documentation for this stuff.
4307 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
4308 X509_*() to X509at_*() on the grounds that they don't handle X509
4309 structures and behave in an analagous way to the X509v3 functions:
4310 they shouldn't be called directly but wrapper functions should be used
4313 So we also now have some wrapper functions that call the X509at functions
4314 when passed certificate requests. (TO DO: similar things can be done with
4315 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
4316 things. Some of these need some d2i or i2d and print functionality
4317 because they handle more complex structures.)
4320 *) Add missing #ifndefs that caused missing symbols when building libssl
4321 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
4322 NO_RSA in ssl/s2*.c.
4323 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
4325 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
4326 has a return value which indicates the quality of the random data
4327 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
4328 error queue. New function RAND_pseudo_bytes() generates output that is
4329 guaranteed to be unique but not unpredictable. RAND_add is like
4330 RAND_seed, but takes an extra argument for an entropy estimate
4331 (RAND_seed always assumes full entropy).
4334 *) Do more iterations of Rabin-Miller probable prime test (specifically,
4335 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
4336 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
4337 in crypto/bn/bn_prime.c for the complete table). This guarantees a
4338 false-positive rate of at most 2^-80 for random input.
4341 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
4344 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
4345 in the 0.9.5 release), this returns the chain
4346 from an X509_CTX structure with a dup of the stack and all
4347 the X509 reference counts upped: so the stack will exist
4348 after X509_CTX_cleanup() has been called. Modify pkcs12.c
4351 Also make SSL_SESSION_print() print out the verify return
4355 *) Add manpage for the pkcs12 command. Also change the default
4356 behaviour so MAC iteration counts are used unless the new
4357 -nomaciter option is used. This improves file security and
4358 only older versions of MSIE (4.0 for example) need it.
4361 *) Honor the no-xxx Configure options when creating .DEF files.
4364 *) Add PKCS#10 attributes to field table: challengePassword,
4365 unstructuredName and unstructuredAddress. These are taken from
4366 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
4367 international characters are used.
4369 More changes to X509_ATTRIBUTE code: allow the setting of types
4370 based on strings. Remove the 'loc' parameter when adding
4371 attributes because these will be a SET OF encoding which is sorted
4375 *) Initial changes to the 'req' utility to allow request generation
4376 automation. This will allow an application to just generate a template
4377 file containing all the field values and have req construct the
4380 Initial support for X509_ATTRIBUTE handling. Stacks of these are
4381 used all over the place including certificate requests and PKCS#7
4382 structures. They are currently handled manually where necessary with
4383 some primitive wrappers for PKCS#7. The new functions behave in a
4384 manner analogous to the X509 extension functions: they allow
4385 attributes to be looked up by NID and added.
4387 Later something similar to the X509V3 code would be desirable to
4388 automatically handle the encoding, decoding and printing of the
4389 more complex types. The string types like challengePassword can
4390 be handled by the string table functions.
4392 Also modified the multi byte string table handling. Now there is
4393 a 'global mask' which masks out certain types. The table itself
4394 can use the flag STABLE_NO_MASK to ignore the mask setting: this
4395 is useful when for example there is only one permissible type
4396 (as in countryName) and using the mask might result in no valid
4400 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
4401 SSL_get_peer_finished to allow applications to obtain the latest
4402 Finished messages sent to the peer or expected from the peer,
4403 respectively. (SSL_get_peer_finished is usually the Finished message
4404 actually received from the peer, otherwise the protocol will be aborted.)
4406 As the Finished message are message digests of the complete handshake
4407 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
4408 be used for external authentication procedures when the authentication
4409 provided by SSL/TLS is not desired or is not enough.
4412 *) Enhanced support for Alpha Linux is added. Now ./config checks if
4413 the host supports BWX extension and if Compaq C is present on the
4414 $PATH. Just exploiting of the BWX extension results in 20-30%
4415 performance kick for some algorithms, e.g. DES and RC4 to mention
4416 a couple. Compaq C in turn generates ~20% faster code for MD5 and
4420 *) Add support for MS "fast SGC". This is arguably a violation of the
4421 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
4422 weak crypto and after checking the certificate is SGC a second one
4423 with strong crypto. MS SGC stops the first handshake after receiving
4424 the server certificate message and sends a second client hello. Since
4425 a server will typically do all the time consuming operations before
4426 expecting any further messages from the client (server key exchange
4427 is the most expensive) there is little difference between the two.
4429 To get OpenSSL to support MS SGC we have to permit a second client
4430 hello message after we have sent server done. In addition we have to
4431 reset the MAC if we do get this second client hello.
4434 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
4435 if a DER encoded private key is RSA or DSA traditional format. Changed
4436 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
4437 format DER encoded private key. Newer code should use PKCS#8 format which
4438 has the key type encoded in the ASN1 structure. Added DER private key
4439 support to pkcs8 application.
4442 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
4443 ciphersuites has been selected (as required by the SSL 3/TLS 1
4444 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
4445 is set, we interpret this as a request to violate the specification
4446 (the worst that can happen is a handshake failure, and 'correct'
4447 behaviour would result in a handshake failure anyway).
4450 *) In SSL_CTX_add_session, take into account that there might be multiple
4451 SSL_SESSION structures with the same session ID (e.g. when two threads
4452 concurrently obtain them from an external cache).
4453 The internal cache can handle only one SSL_SESSION with a given ID,
4454 so if there's a conflict, we now throw out the old one to achieve
4458 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
4459 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
4460 some routines that use cipher OIDs: some ciphers do not have OIDs
4461 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
4465 *) Simplify the trust setting structure and code. Now we just have
4466 two sequences of OIDs for trusted and rejected settings. These will
4467 typically have values the same as the extended key usage extension
4468 and any application specific purposes.
4470 The trust checking code now has a default behaviour: it will just
4471 check for an object with the same NID as the passed id. Functions can
4472 be provided to override either the default behaviour or the behaviour
4473 for a given id. SSL client, server and email already have functions
4474 in place for compatibility: they check the NID and also return "trusted"
4475 if the certificate is self signed.
4478 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
4479 traditional format into an EVP_PKEY structure.
4482 *) Add a password callback function PEM_cb() which either prompts for
4483 a password if usr_data is NULL or otherwise assumes it is a null
4484 terminated password. Allow passwords to be passed on command line
4485 environment or config files in a few more utilities.
4488 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
4489 keys. Add some short names for PKCS#8 PBE algorithms and allow them
4490 to be specified on the command line for the pkcs8 and pkcs12 utilities.
4491 Update documentation.
4494 *) Support for ASN1 "NULL" type. This could be handled before by using
4495 ASN1_TYPE but there wasn't any function that would try to read a NULL
4496 and produce an error if it couldn't. For compatibility we also have
4497 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
4498 don't allocate anything because they don't need to.
4501 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
4503 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
4505 *) Rebuild of the memory allocation routines used by OpenSSL code and
4506 possibly others as well. The purpose is to make an interface that
4507 provide hooks so anyone can build a separate set of allocation and
4508 deallocation routines to be used by OpenSSL, for example memory
4509 pool implementations, or something else, which was previously hard
4510 since Malloc(), Realloc() and Free() were defined as macros having
4511 the values malloc, realloc and free, respectively (except for Win32
4512 compilations). The same is provided for memory debugging code.
4513 OpenSSL already comes with functionality to find memory leaks, but
4514 this gives people a chance to debug other memory problems.
4516 With these changes, a new set of functions and macros have appeared:
4518 CRYPTO_set_mem_debug_functions() [F]
4519 CRYPTO_get_mem_debug_functions() [F]
4520 CRYPTO_dbg_set_options() [F]
4521 CRYPTO_dbg_get_options() [F]
4522 CRYPTO_malloc_debug_init() [M]
4524 The memory debug functions are NULL by default, unless the library
4525 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
4526 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
4527 gives the standard debugging functions that come with OpenSSL) or
4528 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
4529 provided by the library user) must be used. When the standard
4530 debugging functions are used, CRYPTO_dbg_set_options can be used to
4531 request additional information:
4532 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
4533 the CRYPTO_MDEBUG_xxx macro when compiling the library.
4535 Also, things like CRYPTO_set_mem_functions will always give the
4536 expected result (the new set of functions is used for allocation
4537 and deallocation) at all times, regardless of platform and compiler
4540 To finish it up, some functions that were never use in any other
4541 way than through macros have a new API and new semantic:
4544 CRYPTO_dbg_realloc()
4547 All macros of value have retained their old syntax.
4548 [Richard Levitte and Bodo Moeller]
4550 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
4551 ordering of SMIMECapabilities wasn't in "strength order" and there
4552 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
4556 *) Some ASN1 types with illegal zero length encoding (INTEGER,
4557 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
4558 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
4560 *) Merge in my S/MIME library for OpenSSL. This provides a simple
4561 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
4562 functionality to handle multipart/signed properly) and a utility
4563 called 'smime' to call all this stuff. This is based on code I
4564 originally wrote for Celo who have kindly allowed it to be
4565 included in OpenSSL.
4568 *) Add variants des_set_key_checked and des_set_key_unchecked of
4569 des_set_key (aka des_key_sched). Global variable des_check_key
4570 decides which of these is called by des_set_key; this way
4571 des_check_key behaves as it always did, but applications and
4572 the library itself, which was buggy for des_check_key == 1,
4573 have a cleaner way to pick the version they need.
4576 *) New function PKCS12_newpass() which changes the password of a
4580 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
4581 dynamic mix. In both cases the ids can be used as an index into the
4582 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
4583 functions so they accept a list of the field values and the
4584 application doesn't need to directly manipulate the X509_TRUST
4588 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
4592 *) Modify the way the V3 extension code looks up extensions. This now
4593 works in a similar way to the object code: we have some "standard"
4594 extensions in a static table which is searched with OBJ_bsearch()
4595 and the application can add dynamic ones if needed. The file
4596 crypto/x509v3/ext_dat.h now has the info: this file needs to be
4597 updated whenever a new extension is added to the core code and kept
4598 in ext_nid order. There is a simple program 'tabtest.c' which checks
4599 this. New extensions are not added too often so this file can readily
4600 be maintained manually.
4602 There are two big advantages in doing things this way. The extensions
4603 can be looked up immediately and no longer need to be "added" using
4604 X509V3_add_standard_extensions(): this function now does nothing.
4605 [Side note: I get *lots* of email saying the extension code doesn't
4606 work because people forget to call this function]
4607 Also no dynamic allocation is done unless new extensions are added:
4608 so if we don't add custom extensions there is no need to call
4609 X509V3_EXT_cleanup().
4612 *) Modify enc utility's salting as follows: make salting the default. Add a
4613 magic header, so unsalted files fail gracefully instead of just decrypting
4614 to garbage. This is because not salting is a big security hole, so people
4615 should be discouraged from doing it.
4618 *) Fixes and enhancements to the 'x509' utility. It allowed a message
4619 digest to be passed on the command line but it only used this
4620 parameter when signing a certificate. Modified so all relevant
4621 operations are affected by the digest parameter including the
4622 -fingerprint and -x509toreq options. Also -x509toreq choked if a
4623 DSA key was used because it didn't fix the digest.
4626 *) Initial certificate chain verify code. Currently tests the untrusted
4627 certificates for consistency with the verify purpose (which is set
4628 when the X509_STORE_CTX structure is set up) and checks the pathlength.
4630 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
4631 this is because it will reject chains with invalid extensions whereas
4632 every previous version of OpenSSL and SSLeay made no checks at all.
4634 Trust code: checks the root CA for the relevant trust settings. Trust
4635 settings have an initial value consistent with the verify purpose: e.g.
4636 if the verify purpose is for SSL client use it expects the CA to be
4637 trusted for SSL client use. However the default value can be changed to
4638 permit custom trust settings: one example of this would be to only trust
4639 certificates from a specific "secure" set of CAs.
4641 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
4642 which should be used for version portability: especially since the
4643 verify structure is likely to change more often now.
4645 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
4646 to set them. If not set then assume SSL clients will verify SSL servers
4649 Two new options to the verify program: -untrusted allows a set of
4650 untrusted certificates to be passed in and -purpose which sets the
4651 intended purpose of the certificate. If a purpose is set then the
4652 new chain verify code is used to check extension consistency.
4655 *) Support for the authority information access extension.
4658 *) Modify RSA and DSA PEM read routines to transparently handle
4659 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
4660 public keys in a format compatible with certificate
4661 SubjectPublicKeyInfo structures. Unfortunately there were already
4662 functions called *_PublicKey_* which used various odd formats so
4663 these are retained for compatibility: however the DSA variants were
4664 never in a public release so they have been deleted. Changed dsa/rsa
4665 utilities to handle the new format: note no releases ever handled public
4666 keys so we should be OK.
4668 The primary motivation for this change is to avoid the same fiasco
4669 that dogs private keys: there are several incompatible private key
4670 formats some of which are standard and some OpenSSL specific and
4671 require various evil hacks to allow partial transparent handling and
4672 even then it doesn't work with DER formats. Given the option anything
4673 other than PKCS#8 should be dumped: but the other formats have to
4674 stay in the name of compatibility.
4676 With public keys and the benefit of hindsight one standard format
4677 is used which works with EVP_PKEY, RSA or DSA structures: though
4678 it clearly returns an error if you try to read the wrong kind of key.
4680 Added a -pubkey option to the 'x509' utility to output the public key.
4681 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
4682 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
4683 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
4684 that do the same as the EVP_PKEY_assign_*() except they up the
4685 reference count of the added key (they don't "swallow" the
4689 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
4690 CRLs would fail if the file contained no certificates or no CRLs:
4691 added a new function to read in both types and return the number
4692 read: this means that if none are read it will be an error. The
4693 DER versions of the certificate and CRL reader would always fail
4694 because it isn't possible to mix certificates and CRLs in DER format
4695 without choking one or the other routine. Changed this to just read
4696 a certificate: this is the best we can do. Also modified the code
4697 in apps/verify.c to take notice of return codes: it was previously
4698 attempting to read in certificates from NULL pointers and ignoring
4699 any errors: this is one reason why the cert and CRL reader seemed
4700 to work. It doesn't check return codes from the default certificate
4701 routines: these may well fail if the certificates aren't installed.
4704 *) Code to support otherName option in GeneralName.
4707 *) First update to verify code. Change the verify utility
4708 so it warns if it is passed a self signed certificate:
4709 for consistency with the normal behaviour. X509_verify
4710 has been modified to it will now verify a self signed
4711 certificate if *exactly* the same certificate appears
4712 in the store: it was previously impossible to trust a
4713 single self signed certificate. This means that:
4714 openssl verify ss.pem
4715 now gives a warning about a self signed certificate but
4716 openssl verify -CAfile ss.pem ss.pem
4720 *) For servers, store verify_result in SSL_SESSION data structure
4721 (and add it to external session representation).
4722 This is needed when client certificate verifications fails,
4723 but an application-provided verification callback (set by
4724 SSL_CTX_set_cert_verify_callback) allows accepting the session
4725 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
4726 but returns 1): When the session is reused, we have to set
4727 ssl->verify_result to the appropriate error code to avoid
4729 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
4731 *) Fix a bug in the new PKCS#7 code: it didn't consider the
4732 case in PKCS7_dataInit() where the signed PKCS7 structure
4733 didn't contain any existing data because it was being created.
4734 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
4736 *) Add a salt to the key derivation routines in enc.c. This
4737 forms the first 8 bytes of the encrypted file. Also add a
4738 -S option to allow a salt to be input on the command line.
4741 *) New function X509_cmp(). Oddly enough there wasn't a function
4742 to compare two certificates. We do this by working out the SHA1
4743 hash and comparing that. X509_cmp() will be needed by the trust
4747 *) SSL_get1_session() is like SSL_get_session(), but increments
4748 the reference count in the SSL_SESSION returned.
4749 [Geoff Thorpe <geoff@eu.c2.net>]
4751 *) Fix for 'req': it was adding a null to request attributes.
4752 Also change the X509_LOOKUP and X509_INFO code to handle
4753 certificate auxiliary information.
4756 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
4760 *) Add the possibility to add extra information to the memory leak
4761 detecting output, to form tracebacks, showing from where each
4762 allocation was originated: CRYPTO_push_info("constant string") adds
4763 the string plus current file name and line number to a per-thread
4764 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
4765 is like calling CYRPTO_pop_info() until the stack is empty.
4766 Also updated memory leak detection code to be multi-thread-safe.
4769 *) Add options -text and -noout to pkcs7 utility and delete the
4770 encryption options which never did anything. Update docs.
4773 *) Add options to some of the utilities to allow the pass phrase
4774 to be included on either the command line (not recommended on
4775 OSes like Unix) or read from the environment. Update the
4776 manpages and fix a few bugs.
4779 *) Add a few manpages for some of the openssl commands.
4782 *) Fix the -revoke option in ca. It was freeing up memory twice,
4783 leaking and not finding already revoked certificates.
4786 *) Extensive changes to support certificate auxiliary information.
4787 This involves the use of X509_CERT_AUX structure and X509_AUX
4788 functions. An X509_AUX function such as PEM_read_X509_AUX()
4789 can still read in a certificate file in the usual way but it
4790 will also read in any additional "auxiliary information". By
4791 doing things this way a fair degree of compatibility can be
4792 retained: existing certificates can have this information added
4793 using the new 'x509' options.
4795 Current auxiliary information includes an "alias" and some trust
4796 settings. The trust settings will ultimately be used in enhanced
4797 certificate chain verification routines: currently a certificate
4798 can only be trusted if it is self signed and then it is trusted
4802 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
4803 The problem was that one of the replacement routines had not been working
4804 since SSLeay releases. For now the offending routine has been replaced
4805 with non-optimised assembler. Even so, this now gives around 95%
4806 performance improvement for 1024 bit RSA signs.
4809 *) Hack to fix PKCS#7 decryption when used with some unorthodox RC2
4810 handling. Most clients have the effective key size in bits equal to
4811 the key length in bits: so a 40 bit RC2 key uses a 40 bit (5 byte) key.
4812 A few however don't do this and instead use the size of the decrypted key
4813 to determine the RC2 key length and the AlgorithmIdentifier to determine
4814 the effective key length. In this case the effective key length can still
4815 be 40 bits but the key length can be 168 bits for example. This is fixed
4816 by manually forcing an RC2 key into the EVP_PKEY structure because the
4817 EVP code can't currently handle unusual RC2 key sizes: it always assumes
4818 the key length and effective key length are equal.
4821 *) Add a bunch of functions that should simplify the creation of
4822 X509_NAME structures. Now you should be able to do:
4823 X509_NAME_add_entry_by_txt(nm, "CN", MBSTRING_ASC, "Steve", -1, -1, 0);
4824 and have it automatically work out the correct field type and fill in
4825 the structures. The more adventurous can try:
4826 X509_NAME_add_entry_by_txt(nm, field, MBSTRING_UTF8, str, -1, -1, 0);
4827 and it will (hopefully) work out the correct multibyte encoding.
4830 *) Change the 'req' utility to use the new field handling and multibyte
4831 copy routines. Before the DN field creation was handled in an ad hoc
4832 way in req, ca, and x509 which was rather broken and didn't support
4833 BMPStrings or UTF8Strings. Since some software doesn't implement
4834 BMPStrings or UTF8Strings yet, they can be enabled using the config file
4835 using the dirstring_type option. See the new comment in the default
4836 openssl.cnf for more info.
4839 *) Make crypto/rand/md_rand.c more robust:
4840 - Assure unique random numbers after fork().
4841 - Make sure that concurrent threads access the global counter and
4842 md serializably so that we never lose entropy in them
4843 or use exactly the same state in multiple threads.
4844 Access to the large state is not always serializable because
4845 the additional locking could be a performance killer, and
4846 md should be large enough anyway.
4849 *) New file apps/app_rand.c with commonly needed functionality
4850 for handling the random seed file.
4852 Use the random seed file in some applications that previously did not:
4854 dsaparam -genkey (which also ignored its '-rand' option),
4857 x509 (when signing).
4858 Except on systems with /dev/urandom, it is crucial to have a random
4859 seed file at least for key creation, DSA signing, and for DH exchanges;
4860 for RSA signatures we could do without one.
4862 gendh and gendsa (unlike genrsa) used to read only the first byte
4863 of each file listed in the '-rand' option. The function as previously
4864 found in genrsa is now in app_rand.c and is used by all programs
4865 that support '-rand'.
4868 *) In RAND_write_file, use mode 0600 for creating files;
4869 don't just chmod when it may be too late.
4872 *) Report an error from X509_STORE_load_locations
4873 when X509_LOOKUP_load_file or X509_LOOKUP_add_dir failed.
4876 *) New function ASN1_mbstring_copy() this copies a string in either
4877 ASCII, Unicode, Universal (4 bytes per character) or UTF8 format
4878 into an ASN1_STRING type. A mask of permissible types is passed
4879 and it chooses the "minimal" type to use or an error if not type
4883 *) Add function equivalents to the various macros in asn1.h. The old
4884 macros are retained with an M_ prefix. Code inside the library can
4885 use the M_ macros. External code (including the openssl utility)
4886 should *NOT* in order to be "shared library friendly".
4889 *) Add various functions that can check a certificate's extensions
4890 to see if it usable for various purposes such as SSL client,
4891 server or S/MIME and CAs of these types. This is currently
4892 VERY EXPERIMENTAL but will ultimately be used for certificate chain
4893 verification. Also added a -purpose flag to x509 utility to
4894 print out all the purposes.
4897 *) Add a CRYPTO_EX_DATA to X509 certificate structure and associated
4901 *) New X509V3_{X509,CRL,REVOKED}_get_d2i() functions. These will search
4902 for, obtain and decode and extension and obtain its critical flag.
4903 This allows all the necessary extension code to be handled in a
4904 single function call.
4907 *) RC4 tune-up featuring 30-40% performance improvement on most RISC
4908 platforms. See crypto/rc4/rc4_enc.c for further details.
4911 *) New -noout option to asn1parse. This causes no output to be produced
4912 its main use is when combined with -strparse and -out to extract data
4913 from a file (which may not be in ASN.1 format).
4916 *) Fix for pkcs12 program. It was hashing an invalid certificate pointer
4917 when producing the local key id.
4918 [Richard Levitte <levitte@stacken.kth.se>]
4920 *) New option -dhparam in s_server. This allows a DH parameter file to be
4921 stated explicitly. If it is not stated then it tries the first server
4922 certificate file. The previous behaviour hard coded the filename
4926 *) Add -pubin and -pubout options to the rsa and dsa commands. These allow
4927 a public key to be input or output. For example:
4928 openssl rsa -in key.pem -pubout -out pubkey.pem
4929 Also added necessary DSA public key functions to handle this.
4932 *) Fix so PKCS7_dataVerify() doesn't crash if no certificates are contained
4933 in the message. This was handled by allowing
4934 X509_find_by_issuer_and_serial() to tolerate a NULL passed to it.
4935 [Steve Henson, reported by Sampo Kellomaki <sampo@mail.neuronio.pt>]
4937 *) Fix for bug in d2i_ASN1_bytes(): other ASN1 functions add an extra null
4938 to the end of the strings whereas this didn't. This would cause problems
4939 if strings read with d2i_ASN1_bytes() were later modified.
4940 [Steve Henson, reported by Arne Ansper <arne@ats.cyber.ee>]
4942 *) Fix for base64 decode bug. When a base64 bio reads only one line of
4943 data and it contains EOF it will end up returning an error. This is
4944 caused by input 46 bytes long. The cause is due to the way base64
4945 BIOs find the start of base64 encoded data. They do this by trying a
4946 trial decode on each line until they find one that works. When they
4947 do a flag is set and it starts again knowing it can pass all the
4948 data directly through the decoder. Unfortunately it doesn't reset
4949 the context it uses. This means that if EOF is reached an attempt
4950 is made to pass two EOFs through the context and this causes the
4951 resulting error. This can also cause other problems as well. As is
4952 usual with these problems it takes *ages* to find and the fix is
4953 trivial: move one line.
4954 [Steve Henson, reported by ian@uns.ns.ac.yu (Ivan Nejgebauer) ]
4956 *) Ugly workaround to get s_client and s_server working under Windows. The
4957 old code wouldn't work because it needed to select() on sockets and the
4958 tty (for keypresses and to see if data could be written). Win32 only
4959 supports select() on sockets so we select() with a 1s timeout on the
4960 sockets and then see if any characters are waiting to be read, if none
4961 are present then we retry, we also assume we can always write data to
4962 the tty. This isn't nice because the code then blocks until we've
4963 received a complete line of data and it is effectively polling the
4964 keyboard at 1s intervals: however it's quite a bit better than not
4965 working at all :-) A dedicated Windows application might handle this
4966 with an event loop for example.
4969 *) Enhance RSA_METHOD structure. Now there are two extra methods, rsa_sign
4970 and rsa_verify. When the RSA_FLAGS_SIGN_VER option is set these functions
4971 will be called when RSA_sign() and RSA_verify() are used. This is useful
4972 if rsa_pub_dec() and rsa_priv_enc() equivalents are not available.
4973 For this to work properly RSA_public_decrypt() and RSA_private_encrypt()
4974 should *not* be used: RSA_sign() and RSA_verify() must be used instead.
4975 This necessitated the support of an extra signature type NID_md5_sha1
4976 for SSL signatures and modifications to the SSL library to use it instead
4977 of calling RSA_public_decrypt() and RSA_private_encrypt().
4980 *) Add new -verify -CAfile and -CApath options to the crl program, these
4981 will lookup a CRL issuers certificate and verify the signature in a
4982 similar way to the verify program. Tidy up the crl program so it
4983 no longer accesses structures directly. Make the ASN1 CRL parsing a bit
4984 less strict. It will now permit CRL extensions even if it is not
4985 a V2 CRL: this will allow it to tolerate some broken CRLs.
4988 *) Initialize all non-automatic variables each time one of the openssl
4989 sub-programs is started (this is necessary as they may be started
4990 multiple times from the "OpenSSL>" prompt).
4991 [Lennart Bang, Bodo Moeller]
4993 *) Preliminary compilation option RSA_NULL which disables RSA crypto without
4994 removing all other RSA functionality (this is what NO_RSA does). This
4995 is so (for example) those in the US can disable those operations covered
4996 by the RSA patent while allowing storage and parsing of RSA keys and RSA
5000 *) Non-copying interface to BIO pairs.
5001 (still largely untested)
5004 *) New function ANS1_tag2str() to convert an ASN1 tag to a descriptive
5005 ASCII string. This was handled independently in various places before.
5008 *) New functions UTF8_getc() and UTF8_putc() that parse and generate
5009 UTF8 strings a character at a time.
5012 *) Use client_version from client hello to select the protocol
5013 (s23_srvr.c) and for RSA client key exchange verification
5014 (s3_srvr.c), as required by the SSL 3.0/TLS 1.0 specifications.
5017 *) Add various utility functions to handle SPKACs, these were previously
5018 handled by poking round in the structure internals. Added new function
5019 NETSCAPE_SPKI_print() to print out SPKAC and a new utility 'spkac' to
5020 print, verify and generate SPKACs. Based on an original idea from
5021 Massimiliano Pala <madwolf@comune.modena.it> but extensively modified.
5024 *) RIPEMD160 is operational on all platforms and is back in 'make test'.
5027 *) Allow the config file extension section to be overwritten on the
5028 command line. Based on an original idea from Massimiliano Pala
5029 <madwolf@comune.modena.it>. The new option is called -extensions
5030 and can be applied to ca, req and x509. Also -reqexts to override
5031 the request extensions in req and -crlexts to override the crl extensions
5035 *) Add new feature to the SPKAC handling in ca. Now you can include
5036 the same field multiple times by preceding it by "XXXX." for example:
5039 this is the same syntax as used in the req config file.
5042 *) Allow certificate extensions to be added to certificate requests. These
5043 are specified in a 'req_extensions' option of the req section of the
5044 config file. They can be printed out with the -text option to req but
5045 are otherwise ignored at present.
5048 *) Fix a horrible bug in enc_read() in crypto/evp/bio_enc.c: if the first
5049 data read consists of only the final block it would not decrypted because
5050 EVP_CipherUpdate() would correctly report zero bytes had been decrypted.
5051 A misplaced 'break' also meant the decrypted final block might not be
5052 copied until the next read.
5055 *) Initial support for DH_METHOD. Again based on RSA_METHOD. Also added
5056 a few extra parameters to the DH structure: these will be useful if
5057 for example we want the value of 'q' or implement X9.42 DH.
5060 *) Initial support for DSA_METHOD. This is based on the RSA_METHOD and
5061 provides hooks that allow the default DSA functions or functions on a
5062 "per key" basis to be replaced. This allows hardware acceleration and
5063 hardware key storage to be handled without major modification to the
5064 library. Also added low level modexp hooks and CRYPTO_EX structure and
5065 associated functions.
5068 *) Add a new flag to memory BIOs, BIO_FLAG_MEM_RDONLY. This marks the BIO
5069 as "read only": it can't be written to and the buffer it points to will
5070 not be freed. Reading from a read only BIO is much more efficient than
5071 a normal memory BIO. This was added because there are several times when
5072 an area of memory needs to be read from a BIO. The previous method was
5073 to create a memory BIO and write the data to it, this results in two
5074 copies of the data and an O(n^2) reading algorithm. There is a new
5075 function BIO_new_mem_buf() which creates a read only memory BIO from
5076 an area of memory. Also modified the PKCS#7 routines to use read only
5080 *) Bugfix: ssl23_get_client_hello did not work properly when called in
5081 state SSL23_ST_SR_CLNT_HELLO_B, i.e. when the first 7 bytes of
5082 a SSLv2-compatible client hello for SSLv3 or TLSv1 could be read,
5083 but a retry condition occured while trying to read the rest.
5086 *) The PKCS7_ENC_CONTENT_new() function was setting the content type as
5087 NID_pkcs7_encrypted by default: this was wrong since this should almost
5088 always be NID_pkcs7_data. Also modified the PKCS7_set_type() to handle
5089 the encrypted data type: this is a more sensible place to put it and it
5090 allows the PKCS#12 code to be tidied up that duplicated this
5094 *) Changed obj_dat.pl script so it takes its input and output files on
5095 the command line. This should avoid shell escape redirection problems
5099 *) Initial support for certificate extension requests, these are included
5100 in things like Xenroll certificate requests. Included functions to allow
5101 extensions to be obtained and added.
5104 *) -crlf option to s_client and s_server for sending newlines as
5105 CRLF (as required by many protocols).
5108 Changes between 0.9.3a and 0.9.4 [09 Aug 1999]
5110 *) Install libRSAglue.a when OpenSSL is built with RSAref.
5111 [Ralf S. Engelschall]
5113 *) A few more ``#ifndef NO_FP_API / #endif'' pairs for consistency.
5114 [Andrija Antonijevic <TheAntony2@bigfoot.com>]
5116 *) Fix -startdate and -enddate (which was missing) arguments to 'ca'
5120 *) New function DSA_dup_DH, which duplicates DSA parameters/keys as
5121 DH parameters/keys (q is lost during that conversion, but the resulting
5122 DH parameters contain its length).
5124 For 1024-bit p, DSA_generate_parameters followed by DSA_dup_DH is
5125 much faster than DH_generate_parameters (which creates parameters
5126 where p = 2*q + 1), and also the smaller q makes DH computations
5127 much more efficient (160-bit exponentiation instead of 1024-bit
5128 exponentiation); so this provides a convenient way to support DHE
5129 ciphersuites in SSL/TLS servers (see ssl/ssltest.c). It is of
5130 utter importance to use
5131 SSL_CTX_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
5133 SSL_set_options(s_ctx, SSL_OP_SINGLE_DH_USE);
5134 when such DH parameters are used, because otherwise small subgroup
5135 attacks may become possible!
5138 *) Avoid memory leak in i2d_DHparams.
5141 *) Allow the -k option to be used more than once in the enc program:
5142 this allows the same encrypted message to be read by multiple recipients.
5145 *) New function OBJ_obj2txt(buf, buf_len, a, no_name), this converts
5146 an ASN1_OBJECT to a text string. If the "no_name" parameter is set then
5147 it will always use the numerical form of the OID, even if it has a short
5151 *) Added an extra RSA flag: RSA_FLAG_EXT_PKEY. Previously the rsa_mod_exp
5152 method only got called if p,q,dmp1,dmq1,iqmp components were present,
5153 otherwise bn_mod_exp was called. In the case of hardware keys for example
5154 no private key components need be present and it might store extra data
5155 in the RSA structure, which cannot be accessed from bn_mod_exp.
5156 By setting RSA_FLAG_EXT_PKEY rsa_mod_exp will always be called for
5157 private key operations.
5160 *) Added support for SPARC Linux.
5163 *) pem_password_cb function type incompatibly changed from
5164 typedef int pem_password_cb(char *buf, int size, int rwflag);
5166 ....(char *buf, int size, int rwflag, void *userdata);
5167 so that applications can pass data to their callbacks:
5168 The PEM[_ASN1]_{read,write}... functions and macros now take an
5169 additional void * argument, which is just handed through whenever
5170 the password callback is called.
5171 [Damien Miller <dmiller@ilogic.com.au>; tiny changes by Bodo Moeller]
5173 New function SSL_CTX_set_default_passwd_cb_userdata.
5175 Compatibility note: As many C implementations push function arguments
5176 onto the stack in reverse order, the new library version is likely to
5177 interoperate with programs that have been compiled with the old
5178 pem_password_cb definition (PEM_whatever takes some data that
5179 happens to be on the stack as its last argument, and the callback
5180 just ignores this garbage); but there is no guarantee whatsoever that
5183 *) The -DPLATFORM="\"$(PLATFORM)\"" definition and the similar -DCFLAGS=...
5184 (both in crypto/Makefile.ssl for use by crypto/cversion.c) caused
5185 problems not only on Windows, but also on some Unix platforms.
5186 To avoid problematic command lines, these definitions are now in an
5187 auto-generated file crypto/buildinf.h (created by crypto/Makefile.ssl
5188 for standard "make" builds, by util/mk1mf.pl for "mk1mf" builds).
5191 *) MIPS III/IV assembler module is reimplemented.
5194 *) More DES library cleanups: remove references to srand/rand and
5195 delete an unused file.
5198 *) Add support for the the free Netwide assembler (NASM) under Win32,
5199 since not many people have MASM (ml) and it can be hard to obtain.
5200 This is currently experimental but it seems to work OK and pass all
5201 the tests. Check out INSTALL.W32 for info.
5204 *) Fix memory leaks in s3_clnt.c: All non-anonymous SSL3/TLS1 connections
5205 without temporary keys kept an extra copy of the server key,
5206 and connections with temporary keys did not free everything in case
5210 *) New function RSA_check_key and new openssl rsa option -check
5211 for verifying the consistency of RSA keys.
5212 [Ulf Moeller, Bodo Moeller]
5214 *) Various changes to make Win32 compile work:
5215 1. Casts to avoid "loss of data" warnings in p5_crpt2.c
5216 2. Change unsigned int to int in b_dump.c to avoid "signed/unsigned
5217 comparison" warnings.
5218 3. Add sk_<TYPE>_sort to DEF file generator and do make update.
5221 *) Add a debugging option to PKCS#5 v2 key generation function: when
5222 you #define DEBUG_PKCS5V2 passwords, salts, iteration counts and
5223 derived keys are printed to stderr.
5226 *) Copy the flags in ASN1_STRING_dup().
5227 [Roman E. Pavlov <pre@mo.msk.ru>]
5229 *) The x509 application mishandled signing requests containing DSA
5230 keys when the signing key was also DSA and the parameters didn't match.
5232 It was supposed to omit the parameters when they matched the signing key:
5233 the verifying software was then supposed to automatically use the CA's
5234 parameters if they were absent from the end user certificate.
5236 Omitting parameters is no longer recommended. The test was also
5237 the wrong way round! This was probably due to unusual behaviour in
5238 EVP_cmp_parameters() which returns 1 if the parameters match.
5239 This meant that parameters were omitted when they *didn't* match and
5240 the certificate was useless. Certificates signed with 'ca' didn't have
5242 [Steve Henson, reported by Doug Erickson <Doug.Erickson@Part.NET>]
5244 *) Memory leak checking (-DCRYPTO_MDEBUG) had some problems.
5245 The interface is as follows:
5246 Applications can use
5247 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ON) aka MemCheck_start(),
5248 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_OFF) aka MemCheck_stop();
5249 "off" is now the default.
5250 The library internally uses
5251 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_DISABLE) aka MemCheck_off(),
5252 CRYPTO_mem_ctrl(CRYPTO_MEM_CHECK_ENABLE) aka MemCheck_on()
5253 to disable memory-checking temporarily.
5255 Some inconsistent states that previously were possible (and were
5256 even the default) are now avoided.
5258 -DCRYPTO_MDEBUG_TIME is new and additionally stores the current time
5259 with each memory chunk allocated; this is occasionally more helpful
5260 than just having a counter.
5262 -DCRYPTO_MDEBUG_THREAD is also new and adds the thread ID.
5264 -DCRYPTO_MDEBUG_ALL enables all of the above, plus any future
5268 *) Introduce "mode" for SSL structures (with defaults in SSL_CTX),
5269 which largely parallels "options", but is for changing API behaviour,
5270 whereas "options" are about protocol behaviour.
5271 Initial "mode" flags are:
5273 SSL_MODE_ENABLE_PARTIAL_WRITE Allow SSL_write to report success when
5274 a single record has been written.
5275 SSL_MODE_ACCEPT_MOVING_WRITE_BUFFER Don't insist that SSL_write
5276 retries use the same buffer location.
5277 (But all of the contents must be
5281 *) Bugfix: SSL_set_options ignored its parameter, only SSL_CTX_set_options
5284 *) Fix problems with no-hmac etc.
5285 [Ulf Möller, pointed out by Brian Wellington <bwelling@tislabs.com>]
5287 *) New functions RSA_get_default_method(), RSA_set_method() and
5288 RSA_get_method(). These allows replacement of RSA_METHODs without having
5289 to mess around with the internals of an RSA structure.
5292 *) Fix memory leaks in DSA_do_sign and DSA_is_prime.
5293 Also really enable memory leak checks in openssl.c and in some
5295 [Chad C. Mulligan, Bodo Moeller]
5297 *) Fix a bug in d2i_ASN1_INTEGER() and i2d_ASN1_INTEGER() which can mess
5298 up the length of negative integers. This has now been simplified to just
5299 store the length when it is first determined and use it later, rather
5300 than trying to keep track of where data is copied and updating it to
5302 [Steve Henson, reported by Brien Wheeler
5303 <bwheeler@authentica-security.com>]
5305 *) Add a new function PKCS7_signatureVerify. This allows the verification
5306 of a PKCS#7 signature but with the signing certificate passed to the
5307 function itself. This contrasts with PKCS7_dataVerify which assumes the
5308 certificate is present in the PKCS#7 structure. This isn't always the
5309 case: certificates can be omitted from a PKCS#7 structure and be
5310 distributed by "out of band" means (such as a certificate database).
5313 *) Complete the PEM_* macros with DECLARE_PEM versions to replace the
5314 function prototypes in pem.h, also change util/mkdef.pl to add the
5315 necessary function names.
5318 *) mk1mf.pl (used by Windows builds) did not properly read the
5319 options set by Configure in the top level Makefile, and Configure
5320 was not even able to write more than one option correctly.
5321 Fixed, now "no-idea no-rc5 -DCRYPTO_MDEBUG" etc. works as intended.
5324 *) New functions CONF_load_bio() and CONF_load_fp() to allow a config
5325 file to be loaded from a BIO or FILE pointer. The BIO version will
5326 for example allow memory BIOs to contain config info.
5329 *) New function "CRYPTO_num_locks" that returns CRYPTO_NUM_LOCKS.
5330 Whoever hopes to achieve shared-library compatibility across versions
5331 must use this, not the compile-time macro.
5332 (Exercise 0.9.4: Which is the minimum library version required by
5334 Note: All this applies only to multi-threaded programs, others don't
5338 *) Add missing case to s3_clnt.c state machine -- one of the new SSL tests
5339 through a BIO pair triggered the default case, i.e.
5340 SSLerr(...,SSL_R_UNKNOWN_STATE).
5343 *) New "BIO pair" concept (crypto/bio/bss_bio.c) so that applications
5344 can use the SSL library even if none of the specific BIOs is
5348 *) Fix a bug in i2d_DSAPublicKey() which meant it returned the wrong value
5349 for the encoded length.
5350 [Jeon KyoungHo <khjeon@sds.samsung.co.kr>]
5352 *) Add initial documentation of the X509V3 functions.
5355 *) Add a new pair of functions PEM_write_PKCS8PrivateKey() and
5356 PEM_write_bio_PKCS8PrivateKey() that are equivalent to
5357 PEM_write_PrivateKey() and PEM_write_bio_PrivateKey() but use the more
5358 secure PKCS#8 private key format with a high iteration count.
5361 *) Fix determination of Perl interpreter: A perl or perl5
5362 _directory_ in $PATH was also accepted as the interpreter.
5363 [Ralf S. Engelschall]
5365 *) Fix demos/sign/sign.c: well there wasn't anything strictly speaking
5366 wrong with it but it was very old and did things like calling
5367 PEM_ASN1_read() directly and used MD5 for the hash not to mention some
5371 *) Fix demos/selfsign.c: it used obsolete and deleted functions, changed
5372 to use the new extension code.
5375 *) Implement the PEM_read/PEM_write functions in crypto/pem/pem_all.c
5376 with macros. This should make it easier to change their form, add extra
5377 arguments etc. Fix a few PEM prototypes which didn't have cipher as a
5381 *) Add to configuration table a new entry that can specify an alternative
5382 name for unistd.h (for pre-POSIX systems); we need this for NeXTstep,
5383 according to Mark Crispin <MRC@Panda.COM>.
5387 *) DES CBC did not update the IV. Weird.
5390 des_cbc_encrypt does not update the IV, but des_ncbc_encrypt does.
5391 Changing the behaviour of the former might break existing programs --
5392 where IV updating is needed, des_ncbc_encrypt can be used.
5395 *) When bntest is run from "make test" it drives bc to check its
5396 calculations, as well as internally checking them. If an internal check
5397 fails, it needs to cause bc to give a non-zero result or make test carries
5398 on without noticing the failure. Fixed.
5401 *) DES library cleanups.
5404 *) Add support for PKCS#5 v2.0 PBE algorithms. This will permit PKCS#8 to be
5405 used with any cipher unlike PKCS#5 v1.5 which can at most handle 64 bit
5406 ciphers. NOTE: although the key derivation function has been verified
5407 against some published test vectors it has not been extensively tested
5408 yet. Added a -v2 "cipher" option to pkcs8 application to allow the use
5412 *) Instead of "mkdir -p", which is not fully portable, use new
5413 Perl script "util/mkdir-p.pl".
5416 *) Rewrite the way password based encryption (PBE) is handled. It used to
5417 assume that the ASN1 AlgorithmIdentifier parameter was a PBEParameter
5418 structure. This was true for the PKCS#5 v1.5 and PKCS#12 PBE algorithms
5419 but doesn't apply to PKCS#5 v2.0 where it can be something else. Now
5420 the 'parameter' field of the AlgorithmIdentifier is passed to the
5421 underlying key generation function so it must do its own ASN1 parsing.
5422 This has also changed the EVP_PBE_CipherInit() function which now has a
5423 'parameter' argument instead of literal salt and iteration count values
5424 and the function EVP_PBE_ALGOR_CipherInit() has been deleted.
5427 *) Support for PKCS#5 v1.5 compatible password based encryption algorithms
5428 and PKCS#8 functionality. New 'pkcs8' application linked to openssl.
5429 Needed to change the PEM_STRING_EVP_PKEY value which was just "PRIVATE
5430 KEY" because this clashed with PKCS#8 unencrypted string. Since this
5431 value was just used as a "magic string" and not used directly its
5432 value doesn't matter.
5435 *) Introduce some semblance of const correctness to BN. Shame C doesn't
5439 *) "linux-sparc64" configuration (ultrapenguin).
5440 [Ray Miller <ray.miller@oucs.ox.ac.uk>]
5441 "linux-sparc" configuration.
5442 [Christian Forster <fo@hawo.stw.uni-erlangen.de>]
5444 *) config now generates no-xxx options for missing ciphers.
5447 *) Support the EBCDIC character set (work in progress).
5448 File ebcdic.c not yet included because it has a different license.
5449 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
5451 *) Support BS2000/OSD-POSIX.
5452 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
5454 *) Make callbacks for key generation use void * instead of char *.
5457 *) Make S/MIME samples compile (not yet tested).
5460 *) Additional typesafe stacks.
5463 *) New configuration variants "bsdi-elf-gcc" (BSD/OS 4.x).
5467 Changes between 0.9.3 and 0.9.3a [29 May 1999]
5469 *) New configuration variant "sco5-gcc".
5471 *) Updated some demos.
5472 [Sean O Riordain, Wade Scholine]
5474 *) Add missing BIO_free at exit of pkcs12 application.
5477 *) Fix memory leak in conf.c.
5480 *) Updates for Win32 to assembler version of MD5.
5483 *) Set #! path to perl in apps/der_chop to where we found it
5484 instead of using a fixed path.
5487 *) SHA library changes for irix64-mips4-cc.
5490 *) Improvements for VMS support.
5494 Changes between 0.9.2b and 0.9.3 [24 May 1999]
5496 *) Bignum library bug fix. IRIX 6 passes "make test" now!
5497 This also avoids the problems with SC4.2 and unpatched SC5.
5498 [Andy Polyakov <appro@fy.chalmers.se>]
5500 *) New functions sk_num, sk_value and sk_set to replace the previous macros.
5501 These are required because of the typesafe stack would otherwise break
5502 existing code. If old code used a structure member which used to be STACK
5503 and is now STACK_OF (for example cert in a PKCS7_SIGNED structure) with
5504 sk_num or sk_value it would produce an error because the num, data members
5505 are not present in STACK_OF. Now it just produces a warning. sk_set
5506 replaces the old method of assigning a value to sk_value
5507 (e.g. sk_value(x, i) = y) which the library used in a few cases. Any code
5508 that does this will no longer work (and should use sk_set instead) but
5509 this could be regarded as a "questionable" behaviour anyway.
5512 *) Fix most of the other PKCS#7 bugs. The "experimental" code can now
5513 correctly handle encrypted S/MIME data.
5516 *) Change type of various DES function arguments from des_cblock
5517 (which means, in function argument declarations, pointer to char)
5518 to des_cblock * (meaning pointer to array with 8 char elements),
5519 which allows the compiler to do more typechecking; it was like
5520 that back in SSLeay, but with lots of ugly casts.
5522 Introduce new type const_des_cblock.
5525 *) Reorganise the PKCS#7 library and get rid of some of the more obvious
5526 problems: find RecipientInfo structure that matches recipient certificate
5527 and initialise the ASN1 structures properly based on passed cipher.
5530 *) Belatedly make the BN tests actually check the results.
5533 *) Fix the encoding and decoding of negative ASN1 INTEGERS and conversion
5534 to and from BNs: it was completely broken. New compilation option
5535 NEG_PUBKEY_BUG to allow for some broken certificates that encode public
5536 key elements as negative integers.
5539 *) Reorganize and speed up MD5.
5540 [Andy Polyakov <appro@fy.chalmers.se>]
5543 [Richard Levitte <richard@levitte.org>]
5545 *) New option -out to asn1parse to allow the parsed structure to be
5546 output to a file. This is most useful when combined with the -strparse
5547 option to examine the output of things like OCTET STRINGS.
5550 *) Make SSL library a little more fool-proof by not requiring any longer
5551 that SSL_set_{accept,connect}_state be called before
5552 SSL_{accept,connect} may be used (SSL_set_..._state is omitted
5553 in many applications because usually everything *appeared* to work as
5554 intended anyway -- now it really works as intended).
5557 *) Move openssl.cnf out of lib/.
5560 *) Fix various things to let OpenSSL even pass ``egcc -pipe -O2 -Wall
5561 -Wshadow -Wpointer-arith -Wcast-align -Wmissing-prototypes
5562 -Wmissing-declarations -Wnested-externs -Winline'' with EGCS 1.1.2+
5563 [Ralf S. Engelschall]
5565 *) Various fixes to the EVP and PKCS#7 code. It may now be able to
5566 handle PKCS#7 enveloped data properly.
5567 [Sebastian Akerman <sak@parallelconsulting.com>, modified by Steve]
5569 *) Create a duplicate of the SSL_CTX's CERT in SSL_new instead of
5570 copying pointers. The cert_st handling is changed by this in
5571 various ways (and thus what used to be known as ctx->default_cert
5572 is now called ctx->cert, since we don't resort to s->ctx->[default_]cert
5573 any longer when s->cert does not give us what we need).
5574 ssl_cert_instantiate becomes obsolete by this change.
5575 As soon as we've got the new code right (possibly it already is?),
5576 we have solved a couple of bugs of the earlier code where s->cert
5577 was used as if it could not have been shared with other SSL structures.
5579 Note that using the SSL API in certain dirty ways now will result
5580 in different behaviour than observed with earlier library versions:
5581 Changing settings for an SSL_CTX *ctx after having done s = SSL_new(ctx)
5582 does not influence s as it used to.
5584 In order to clean up things more thoroughly, inside SSL_SESSION
5585 we don't use CERT any longer, but a new structure SESS_CERT
5586 that holds per-session data (if available); currently, this is
5587 the peer's certificate chain and, for clients, the server's certificate
5588 and temporary key. CERT holds only those values that can have
5589 meaningful defaults in an SSL_CTX.
5592 *) New function X509V3_EXT_i2d() to create an X509_EXTENSION structure
5593 from the internal representation. Various PKCS#7 fixes: remove some
5594 evil casts and set the enc_dig_alg field properly based on the signing
5598 *) Allow PKCS#12 password to be set from the command line or the
5599 environment. Let 'ca' get its config file name from the environment
5600 variables "OPENSSL_CONF" or "SSLEAY_CONF" (for consistency with 'req'
5604 *) Allow certificate policies extension to use an IA5STRING for the
5605 organization field. This is contrary to the PKIX definition but
5606 VeriSign uses it and IE5 only recognises this form. Document 'x509'
5610 *) Add PEDANTIC compiler flag to allow compilation with gcc -pedantic,
5611 without disallowing inline assembler and the like for non-pedantic builds.
5614 *) Support Borland C++ builder.
5615 [Janez Jere <jj@void.si>, modified by Ulf Möller]
5620 *) SHA-1 cleanups and performance enhancements.
5621 [Andy Polyakov <appro@fy.chalmers.se>]
5623 *) Sparc v8plus assembler for the bignum library.
5624 [Andy Polyakov <appro@fy.chalmers.se>]
5626 *) Accept any -xxx and +xxx compiler options in Configure.
5629 *) Update HPUX configuration.
5632 *) Add missing sk_<type>_unshift() function to safestack.h
5633 [Ralf S. Engelschall]
5635 *) New function SSL_CTX_use_certificate_chain_file that sets the
5636 "extra_cert"s in addition to the certificate. (This makes sense
5637 only for "PEM" format files, as chains as a whole are not
5641 *) Support verify_depth from the SSL API.
5642 x509_vfy.c had what can be considered an off-by-one-error:
5643 Its depth (which was not part of the external interface)
5644 was actually counting the number of certificates in a chain;
5645 now it really counts the depth.
5648 *) Bugfix in crypto/x509/x509_cmp.c: The SSLerr macro was used
5649 instead of X509err, which often resulted in confusing error
5650 messages since the error codes are not globally unique
5651 (e.g. an alleged error in ssl3_accept when a certificate
5652 didn't match the private key).
5654 *) New function SSL_CTX_set_session_id_context that allows to set a default
5655 value (so that you don't need SSL_set_session_id_context for each
5656 connection using the SSL_CTX).
5659 *) OAEP decoding bug fix.
5662 *) Support INSTALL_PREFIX for package builders, as proposed by
5666 *) New Configure options "threads" and "no-threads". For systems
5667 where the proper compiler options are known (currently Solaris
5668 and Linux), "threads" is the default.
5671 *) New script util/mklink.pl as a faster substitute for util/mklink.sh.
5674 *) Install various scripts to $(OPENSSLDIR)/misc, not to
5675 $(INSTALLTOP)/bin -- they shouldn't clutter directories
5676 such as /usr/local/bin.
5679 *) "make linux-shared" to build shared libraries.
5680 [Niels Poppe <niels@netbox.org>]
5682 *) New Configure option no-<cipher> (rsa, idea, rc5, ...).
5685 *) Add the PKCS#12 API documentation to openssl.txt. Preliminary support for
5686 extension adding in x509 utility.
5689 *) Remove NOPROTO sections and error code comments.
5692 *) Partial rewrite of the DEF file generator to now parse the ANSI
5696 *) New Configure options --prefix=DIR and --openssldir=DIR.
5699 *) Complete rewrite of the error code script(s). It is all now handled
5700 by one script at the top level which handles error code gathering,
5701 header rewriting and C source file generation. It should be much better
5702 than the old method: it now uses a modified version of Ulf's parser to
5703 read the ANSI prototypes in all header files (thus the old K&R definitions
5704 aren't needed for error creation any more) and do a better job of
5705 translating function codes into names. The old 'ASN1 error code imbedded
5706 in a comment' is no longer necessary and it doesn't use .err files which
5707 have now been deleted. Also the error code call doesn't have to appear all
5708 on one line (which resulted in some large lines...).
5711 *) Change #include filenames from <foo.h> to <openssl/foo.h>.
5714 *) Change behaviour of ssl2_read when facing length-0 packets: Don't return
5715 0 (which usually indicates a closed connection), but continue reading.
5718 *) Fix some race conditions.
5721 *) Add support for CRL distribution points extension. Add Certificate
5722 Policies and CRL distribution points documentation.
5725 *) Move the autogenerated header file parts to crypto/opensslconf.h.
5728 *) Fix new 56-bit DES export ciphersuites: they were using 7 bytes instead of
5729 8 of keying material. Merlin has also confirmed interop with this fix
5730 between OpenSSL and Baltimore C/SSL 2.0 and J/SSL 2.0.
5731 [Merlin Hughes <merlin@baltimore.ie>]
5733 *) Fix lots of warnings.
5734 [Richard Levitte <levitte@stacken.kth.se>]
5736 *) In add_cert_dir() in crypto/x509/by_dir.c, break out of the loop if
5737 the directory spec didn't end with a LIST_SEPARATOR_CHAR.
5738 [Richard Levitte <levitte@stacken.kth.se>]
5740 *) Fix problems with sizeof(long) == 8.
5741 [Andy Polyakov <appro@fy.chalmers.se>]
5743 *) Change functions to ANSI C.
5746 *) Fix typos in error codes.
5747 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>, Ulf Möller]
5749 *) Remove defunct assembler files from Configure.
5752 *) SPARC v8 assembler BIGNUM implementation.
5753 [Andy Polyakov <appro@fy.chalmers.se>]
5755 *) Support for Certificate Policies extension: both print and set.
5756 Various additions to support the r2i method this uses.
5759 *) A lot of constification, and fix a bug in X509_NAME_oneline() that could
5760 return a const string when you are expecting an allocated buffer.
5763 *) Add support for ASN1 types UTF8String and VISIBLESTRING, also the CHOICE
5764 types DirectoryString and DisplayText.
5767 *) Add code to allow r2i extensions to access the configuration database,
5768 add an LHASH database driver and add several ctx helper functions.
5771 *) Fix an evil bug in bn_expand2() which caused various BN functions to
5772 fail when they extended the size of a BIGNUM.
5775 *) Various utility functions to handle SXNet extension. Modify mkdef.pl to
5776 support typesafe stack.
5779 *) Fix typo in SSL_[gs]et_options().
5780 [Nils Frostberg <nils@medcom.se>]
5782 *) Delete various functions and files that belonged to the (now obsolete)
5783 old X509V3 handling code.
5786 *) New Configure option "rsaref".
5789 *) Don't auto-generate pem.h.
5792 *) Introduce type-safe ASN.1 SETs.
5795 *) Convert various additional casted stacks to type-safe STACK_OF() variants.
5796 [Ben Laurie, Ralf S. Engelschall, Steve Henson]
5798 *) Introduce type-safe STACKs. This will almost certainly break lots of code
5799 that links with OpenSSL (well at least cause lots of warnings), but fear
5800 not: the conversion is trivial, and it eliminates loads of evil casts. A
5801 few STACKed things have been converted already. Feel free to convert more.
5802 In the fullness of time, I'll do away with the STACK type altogether.
5805 *) Add `openssl ca -revoke <certfile>' facility which revokes a certificate
5806 specified in <certfile> by updating the entry in the index.txt file.
5807 This way one no longer has to edit the index.txt file manually for
5808 revoking a certificate. The -revoke option does the gory details now.
5809 [Massimiliano Pala <madwolf@openca.org>, Ralf S. Engelschall]
5811 *) Fix `openssl crl -noout -text' combination where `-noout' killed the
5812 `-text' option at all and this way the `-noout -text' combination was
5813 inconsistent in `openssl crl' with the friends in `openssl x509|rsa|dsa'.
5814 [Ralf S. Engelschall]
5816 *) Make sure a corresponding plain text error message exists for the
5817 X509_V_ERR_CERT_REVOKED/23 error number which can occur when a
5818 verify callback function determined that a certificate was revoked.
5819 [Ralf S. Engelschall]
5821 *) Bugfix: In test/testenc, don't test "openssl <cipher>" for
5822 ciphers that were excluded, e.g. by -DNO_IDEA. Also, test
5823 all available cipers including rc5, which was forgotten until now.
5824 In order to let the testing shell script know which algorithms
5825 are available, a new (up to now undocumented) command
5826 "openssl list-cipher-commands" is used.
5829 *) Bugfix: s_client occasionally would sleep in select() when
5830 it should have checked SSL_pending() first.
5833 *) New functions DSA_do_sign and DSA_do_verify to provide access to
5834 the raw DSA values prior to ASN.1 encoding.
5837 *) Tweaks to Configure
5838 [Niels Poppe <niels@netbox.org>]
5840 *) Add support for PKCS#5 v2.0 ASN1 PBES2 structures. No other support,
5844 *) New variables $(RANLIB) and $(PERL) in the Makefiles.
5847 *) New config option to avoid instructions that are illegal on the 80386.
5848 The default code is faster, but requires at least a 486.
5851 *) Got rid of old SSL2_CLIENT_VERSION (inconsistently used) and
5852 SSL2_SERVER_VERSION (not used at all) macros, which are now the
5853 same as SSL2_VERSION anyway.
5856 *) New "-showcerts" option for s_client.
5859 *) Still more PKCS#12 integration. Add pkcs12 application to openssl
5860 application. Various cleanups and fixes.
5863 *) More PKCS#12 integration. Add new pkcs12 directory with Makefile.ssl and
5864 modify error routines to work internally. Add error codes and PBE init
5865 to library startup routines.
5868 *) Further PKCS#12 integration. Added password based encryption, PKCS#8 and
5869 packing functions to asn1 and evp. Changed function names and error
5870 codes along the way.
5873 *) PKCS12 integration: and so it begins... First of several patches to
5874 slowly integrate PKCS#12 functionality into OpenSSL. Add PKCS#12
5875 objects to objects.h
5878 *) Add a new 'indent' option to some X509V3 extension code. Initial ASN1
5879 and display support for Thawte strong extranet extension.
5882 *) Add LinuxPPC support.
5883 [Jeff Dubrule <igor@pobox.org>]
5885 *) Get rid of redundant BN file bn_mulw.c, and rename bn_div64 to
5886 bn_div_words in alpha.s.
5887 [Hannes Reinecke <H.Reinecke@hw.ac.uk> and Ben Laurie]
5889 *) Make sure the RSA OAEP test is skipped under -DRSAref because
5890 OAEP isn't supported when OpenSSL is built with RSAref.
5891 [Ulf Moeller <ulf@fitug.de>]
5893 *) Move definitions of IS_SET/IS_SEQUENCE inside crypto/asn1/asn1.h
5894 so they no longer are missing under -DNOPROTO.
5895 [Soren S. Jorvang <soren@t.dk>]
5898 Changes between 0.9.1c and 0.9.2b [22 Mar 1999]
5900 *) Make SSL_get_peer_cert_chain() work in servers. Unfortunately, it still
5901 doesn't work when the session is reused. Coming soon!
5904 *) Fix a security hole, that allows sessions to be reused in the wrong
5905 context thus bypassing client cert protection! All software that uses
5906 client certs and session caches in multiple contexts NEEDS PATCHING to
5907 allow session reuse! A fuller solution is in the works.
5908 [Ben Laurie, problem pointed out by Holger Reif, Bodo Moeller (and ???)]
5910 *) Some more source tree cleanups (removed obsolete files
5911 crypto/bf/asm/bf586.pl, test/test.txt and crypto/sha/asm/f.s; changed
5912 permission on "config" script to be executable) and a fix for the INSTALL
5914 [Ulf Moeller <ulf@fitug.de>]
5916 *) Remove some legacy and erroneous uses of malloc, free instead of
5918 [Lennart Bang <lob@netstream.se>, with minor changes by Steve]
5920 *) Make rsa_oaep_test return non-zero on error.
5921 [Ulf Moeller <ulf@fitug.de>]
5923 *) Add support for native Solaris shared libraries. Configure
5924 solaris-sparc-sc4-pic, make, then run shlib/solaris-sc4.sh. It'd be nice
5925 if someone would make that last step automatic.
5926 [Matthias Loepfe <Matthias.Loepfe@AdNovum.CH>]
5928 *) ctx_size was not built with the right compiler during "make links". Fixed.
5931 *) Change the meaning of 'ALL' in the cipher list. It now means "everything
5932 except NULL ciphers". This means the default cipher list will no longer
5933 enable NULL ciphers. They need to be specifically enabled e.g. with
5934 the string "DEFAULT:eNULL".
5937 *) Fix to RSA private encryption routines: if p < q then it would
5938 occasionally produce an invalid result. This will only happen with
5939 externally generated keys because OpenSSL (and SSLeay) ensure p > q.
5942 *) Be less restrictive and allow also `perl util/perlpath.pl
5943 /path/to/bin/perl' in addition to `perl util/perlpath.pl /path/to/bin',
5944 because this way one can also use an interpreter named `perl5' (which is
5945 usually the name of Perl 5.xxx on platforms where an Perl 4.x is still
5946 installed as `perl').
5947 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
5949 *) Let util/clean-depend.pl work also with older Perl 5.00x versions.
5950 [Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
5952 *) Fix Makefile.org so CC,CFLAG etc are passed to 'make links' add
5953 advapi32.lib to Win32 build and change the pem test comparision
5954 to fc.exe (thanks to Ulrich Kroener <kroneru@yahoo.com> for the
5955 suggestion). Fix misplaced ASNI prototypes and declarations in evp.h
5956 and crypto/des/ede_cbcm_enc.c.
5959 *) DES quad checksum was broken on big-endian architectures. Fixed.
5962 *) Comment out two functions in bio.h that aren't implemented. Fix up the
5963 Win32 test batch file so it (might) work again. The Win32 test batch file
5964 is horrible: I feel ill....
5967 *) Move various #ifdefs around so NO_SYSLOG, NO_DIRENT etc are now selected
5968 in e_os.h. Audit of header files to check ANSI and non ANSI
5969 sections: 10 functions were absent from non ANSI section and not exported
5970 from Windows DLLs. Fixed up libeay.num for new functions.
5973 *) Make `openssl version' output lines consistent.
5974 [Ralf S. Engelschall]
5976 *) Fix Win32 symbol export lists for BIO functions: Added
5977 BIO_get_ex_new_index, BIO_get_ex_num, BIO_get_ex_data and BIO_set_ex_data
5978 to ms/libeay{16,32}.def.
5979 [Ralf S. Engelschall]
5981 *) Second round of fixing the OpenSSL perl/ stuff. It now at least compiled
5982 fine under Unix and passes some trivial tests I've now added. But the
5983 whole stuff is horribly incomplete, so a README.1ST with a disclaimer was
5984 added to make sure no one expects that this stuff really works in the
5985 OpenSSL 0.9.2 release. Additionally I've started to clean the XS sources
5986 up and fixed a few little bugs and inconsistencies in OpenSSL.{pm,xs} and
5988 [Ralf S. Engelschall]
5990 *) Fix the generation of two part addresses in perl.
5991 [Kenji Miyake <kenji@miyake.org>, integrated by Ben Laurie]
5993 *) Add config entry for Linux on MIPS.
5994 [John Tobey <jtobey@channel1.com>]
5996 *) Make links whenever Configure is run, unless we are on Windoze.
5999 *) Permit extensions to be added to CRLs using crl_section in openssl.cnf.
6000 Currently only issuerAltName and AuthorityKeyIdentifier make any sense
6004 *) Add a useful kludge to allow package maintainers to specify compiler and
6005 other platforms details on the command line without having to patch the
6006 Configure script everytime: One now can use ``perl Configure
6007 <id>:<details>'', i.e. platform ids are allowed to have details appended
6008 to them (seperated by colons). This is treated as there would be a static
6009 pre-configured entry in Configure's %table under key <id> with value
6010 <details> and ``perl Configure <id>'' is called. So, when you want to
6011 perform a quick test-compile under FreeBSD 3.1 with pgcc and without
6012 assembler stuff you can use ``perl Configure "FreeBSD-elf:pgcc:-O6:::"''
6013 now, which overrides the FreeBSD-elf entry on-the-fly.
6014 [Ralf S. Engelschall]
6016 *) Disable new TLS1 ciphersuites by default: they aren't official yet.
6019 *) Allow DSO flags like -fpic, -fPIC, -KPIC etc. to be specified
6020 on the `perl Configure ...' command line. This way one can compile
6021 OpenSSL libraries with Position Independent Code (PIC) which is needed
6022 for linking it into DSOs.
6023 [Ralf S. Engelschall]
6025 *) Remarkably, export ciphers were totally broken and no-one had noticed!
6029 *) Cleaned up the LICENSE document: The official contact for any license
6030 questions now is the OpenSSL core team under openssl-core@openssl.org.
6031 And add a paragraph about the dual-license situation to make sure people
6032 recognize that _BOTH_ the OpenSSL license _AND_ the SSLeay license apply
6033 to the OpenSSL toolkit.
6034 [Ralf S. Engelschall]
6036 *) General source tree makefile cleanups: Made `making xxx in yyy...'
6037 display consistent in the source tree and replaced `/bin/rm' by `rm'.
6038 Additonally cleaned up the `make links' target: Remove unnecessary
6039 semicolons, subsequent redundant removes, inline point.sh into mklink.sh
6040 to speed processing and no longer clutter the display with confusing
6041 stuff. Instead only the actually done links are displayed.
6042 [Ralf S. Engelschall]
6044 *) Permit null encryption ciphersuites, used for authentication only. It used
6045 to be necessary to set the preprocessor define SSL_ALLOW_ENULL to do this.
6046 It is now necessary to set SSL_FORBID_ENULL to prevent the use of null
6050 *) Add a bunch of fixes to the PKCS#7 stuff. It used to sometimes reorder
6051 signed attributes when verifying signatures (this would break them),
6052 the detached data encoding was wrong and public keys obtained using
6053 X509_get_pubkey() weren't freed.
6056 *) Add text documentation for the BUFFER functions. Also added a work around
6057 to a Win95 console bug. This was triggered by the password read stuff: the
6058 last character typed gets carried over to the next fread(). If you were
6059 generating a new cert request using 'req' for example then the last
6060 character of the passphrase would be CR which would then enter the first
6064 *) Added the new `Includes OpenSSL Cryptography Software' button as
6065 doc/openssl_button.{gif,html} which is similar in style to the old SSLeay
6066 button and can be used by applications based on OpenSSL to show the
6067 relationship to the OpenSSL project.
6068 [Ralf S. Engelschall]
6070 *) Remove confusing variables in function signatures in files
6071 ssl/ssl_lib.c and ssl/ssl.h.
6072 [Lennart Bong <lob@kulthea.stacken.kth.se>]
6074 *) Don't install bss_file.c under PREFIX/include/
6075 [Lennart Bong <lob@kulthea.stacken.kth.se>]
6077 *) Get the Win32 compile working again. Modify mkdef.pl so it can handle
6078 functions that return function pointers and has support for NT specific
6079 stuff. Fix mk1mf.pl and VC-32.pl to support NT differences also. Various
6080 #ifdef WIN32 and WINNTs sprinkled about the place and some changes from
6081 unsigned to signed types: this was killing the Win32 compile.
6084 *) Add new certificate file to stack functions,
6085 SSL_add_dir_cert_subjects_to_stack() and
6086 SSL_add_file_cert_subjects_to_stack(). These largely supplant
6087 SSL_load_client_CA_file(), and can be used to add multiple certs easily
6088 to a stack (usually this is then handed to SSL_CTX_set_client_CA_list()).
6089 This means that Apache-SSL and similar packages don't have to mess around
6090 to add as many CAs as they want to the preferred list.
6093 *) Experiment with doxygen documentation. Currently only partially applied to
6095 See http://www.stack.nl/~dimitri/doxygen/index.html, and run doxygen with
6096 openssl.doxy as the configuration file.
6099 *) Get rid of remaining C++-style comments which strict C compilers hate.
6100 [Ralf S. Engelschall, pointed out by Carlos Amengual]
6102 *) Changed BN_RECURSION in bn_mont.c to BN_RECURSION_MONT so it is not
6103 compiled in by default: it has problems with large keys.
6106 *) Add a bunch of SSL_xxx() functions for configuring the temporary RSA and
6107 DH private keys and/or callback functions which directly correspond to
6108 their SSL_CTX_xxx() counterparts but work on a per-connection basis. This
6109 is needed for applications which have to configure certificates on a
6110 per-connection basis (e.g. Apache+mod_ssl) instead of a per-context basis
6112 For the RSA certificate situation is makes no difference, but
6113 for the DSA certificate situation this fixes the "no shared cipher"
6114 problem where the OpenSSL cipher selection procedure failed because the
6115 temporary keys were not overtaken from the context and the API provided
6116 no way to reconfigure them.
6117 The new functions now let applications reconfigure the stuff and they
6118 are in detail: SSL_need_tmp_RSA, SSL_set_tmp_rsa, SSL_set_tmp_dh,
6119 SSL_set_tmp_rsa_callback and SSL_set_tmp_dh_callback. Additionally a new
6120 non-public-API function ssl_cert_instantiate() is used as a helper
6121 function and also to reduce code redundancy inside ssl_rsa.c.
6122 [Ralf S. Engelschall]
6124 *) Move s_server -dcert and -dkey options out of the undocumented feature
6125 area because they are useful for the DSA situation and should be
6126 recognized by the users.
6127 [Ralf S. Engelschall]
6129 *) Fix the cipher decision scheme for export ciphers: the export bits are
6130 *not* within SSL_MKEY_MASK or SSL_AUTH_MASK, they are within
6131 SSL_EXP_MASK. So, the original variable has to be used instead of the
6132 already masked variable.
6133 [Richard Levitte <levitte@stacken.kth.se>]
6135 *) Fix 'port' variable from `int' to `unsigned int' in crypto/bio/b_sock.c
6136 [Richard Levitte <levitte@stacken.kth.se>]
6138 *) Change type of another md_len variable in pk7_doit.c:PKCS7_dataFinal()
6139 from `int' to `unsigned int' because it's a length and initialized by
6140 EVP_DigestFinal() which expects an `unsigned int *'.
6141 [Richard Levitte <levitte@stacken.kth.se>]
6143 *) Don't hard-code path to Perl interpreter on shebang line of Configure
6144 script. Instead use the usual Shell->Perl transition trick.
6145 [Ralf S. Engelschall]
6147 *) Make `openssl x509 -noout -modulus' functional also for DSA certificates
6148 (in addition to RSA certificates) to match the behaviour of `openssl dsa
6149 -noout -modulus' as it's already the case for `openssl rsa -noout
6150 -modulus'. For RSA the -modulus is the real "modulus" while for DSA
6151 currently the public key is printed (a decision which was already done by
6152 `openssl dsa -modulus' in the past) which serves a similar purpose.
6153 Additionally the NO_RSA no longer completely removes the whole -modulus
6154 option; it now only avoids using the RSA stuff. Same applies to NO_DSA
6156 [Ralf S. Engelschall]
6158 *) Add Arne Ansper's reliable BIO - this is an encrypted, block-digested
6159 BIO. See the source (crypto/evp/bio_ok.c) for more info.
6160 [Arne Ansper <arne@ats.cyber.ee>]
6162 *) Dump the old yucky req code that tried (and failed) to allow raw OIDs
6163 to be added. Now both 'req' and 'ca' can use new objects defined in the
6167 *) Add cool BIO that does syslog (or event log on NT).
6168 [Arne Ansper <arne@ats.cyber.ee>, integrated by Ben Laurie]
6170 *) Add support for new TLS ciphersuites, TLS_RSA_EXPORT56_WITH_RC4_56_MD5,
6171 TLS_RSA_EXPORT56_WITH_RC2_CBC_56_MD5 and
6172 TLS_RSA_EXPORT56_WITH_DES_CBC_SHA, as specified in "56-bit Export Cipher
6173 Suites For TLS", draft-ietf-tls-56-bit-ciphersuites-00.txt.
6176 *) Add preliminary config info for new extension code.
6179 *) Make RSA_NO_PADDING really use no padding.
6180 [Ulf Moeller <ulf@fitug.de>]
6182 *) Generate errors when private/public key check is done.
6185 *) Overhaul for 'crl' utility. New function X509_CRL_print. Partial support
6186 for some CRL extensions and new objects added.
6189 *) Really fix the ASN1 IMPLICIT bug this time... Partial support for private
6190 key usage extension and fuller support for authority key id.
6193 *) Add OAEP encryption for the OpenSSL crypto library. OAEP is the improved
6194 padding method for RSA, which is recommended for new applications in PKCS
6195 #1 v2.0 (RFC 2437, October 1998).
6196 OAEP (Optimal Asymmetric Encryption Padding) has better theoretical
6197 foundations than the ad-hoc padding used in PKCS #1 v1.5. It is secure
6198 against Bleichbacher's attack on RSA.
6199 [Ulf Moeller <ulf@fitug.de>, reformatted, corrected and integrated by
6202 *) Updates to the new SSL compression code
6203 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
6205 *) Fix so that the version number in the master secret, when passed
6206 via RSA, checks that if TLS was proposed, but we roll back to SSLv3
6207 (because the server will not accept higher), that the version number
6208 is 0x03,0x01, not 0x03,0x00
6209 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
6211 *) Run extensive memory leak checks on SSL apps. Fixed *lots* of memory
6212 leaks in ssl/ relating to new X509_get_pubkey() behaviour. Also fixes
6213 in apps/ and an unrelated leak in crypto/dsa/dsa_vrf.c
6216 *) Support for RAW extensions where an arbitrary extension can be
6217 created by including its DER encoding. See apps/openssl.cnf for
6221 *) Make sure latest Perl versions don't interpret some generated C array
6222 code as Perl array code in the crypto/err/err_genc.pl script.
6223 [Lars Weber <3weber@informatik.uni-hamburg.de>]
6225 *) Modify ms/do_ms.bat to not generate assembly language makefiles since
6226 not many people have the assembler. Various Win32 compilation fixes and
6227 update to the INSTALL.W32 file with (hopefully) more accurate Win32
6231 *) Modify configure script 'Configure' to automatically create crypto/date.h
6232 file under Win32 and also build pem.h from pem.org. New script
6233 util/mkfiles.pl to create the MINFO file on environments that can't do a
6234 'make files': perl util/mkfiles.pl >MINFO should work.
6237 *) Major rework of DES function declarations, in the pursuit of correctness
6238 and purity. As a result, many evil casts evaporated, and some weirdness,
6239 too. You may find this causes warnings in your code. Zapping your evil
6240 casts will probably fix them. Mostly.
6243 *) Fix for a typo in asn1.h. Bug fix to object creation script
6244 obj_dat.pl. It considered a zero in an object definition to mean
6245 "end of object": none of the objects in objects.h have any zeros
6246 so it wasn't spotted.
6247 [Steve Henson, reported by Erwann ABALEA <eabalea@certplus.com>]
6249 *) Add support for Triple DES Cipher Block Chaining with Output Feedback
6250 Masking (CBCM). In the absence of test vectors, the best I have been able
6251 to do is check that the decrypt undoes the encrypt, so far. Send me test
6252 vectors if you have them.
6255 *) Correct calculation of key length for export ciphers (too much space was
6256 allocated for null ciphers). This has not been tested!
6259 *) Modifications to the mkdef.pl for Win32 DEF file creation. The usage
6260 message is now correct (it understands "crypto" and "ssl" on its
6261 command line). There is also now an "update" option. This will update
6262 the util/ssleay.num and util/libeay.num files with any new functions.
6264 perl util/mkdef.pl crypto ssl update
6265 it will update them.
6268 *) Overhauled the Perl interface (perl/*):
6269 - ported BN stuff to OpenSSL's different BN library
6270 - made the perl/ source tree CVS-aware
6271 - renamed the package from SSLeay to OpenSSL (the files still contain
6272 their history because I've copied them in the repository)
6273 - removed obsolete files (the test scripts will be replaced
6274 by better Test::Harness variants in the future)
6275 [Ralf S. Engelschall]
6277 *) First cut for a very conservative source tree cleanup:
6278 1. merge various obsolete readme texts into doc/ssleay.txt
6279 where we collect the old documents and readme texts.
6280 2. remove the first part of files where I'm already sure that we no
6281 longer need them because of three reasons: either they are just temporary
6282 files which were left by Eric or they are preserved original files where
6283 I've verified that the diff is also available in the CVS via "cvs diff
6284 -rSSLeay_0_8_1b" or they were renamed (as it was definitely the case for
6285 the crypto/md/ stuff).
6286 [Ralf S. Engelschall]
6288 *) More extension code. Incomplete support for subject and issuer alt
6289 name, issuer and authority key id. Change the i2v function parameters
6290 and add an extra 'crl' parameter in the X509V3_CTX structure: guess
6291 what that's for :-) Fix to ASN1 macro which messed up
6292 IMPLICIT tag and add f_enum.c which adds a2i, i2a for ENUMERATED.
6295 *) Preliminary support for ENUMERATED type. This is largely copied from the
6299 *) Add new function, EVP_MD_CTX_copy() to replace frequent use of memcpy.
6300 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
6302 *) Make sure `make rehash' target really finds the `openssl' program.
6303 [Ralf S. Engelschall, Matthias Loepfe <Matthias.Loepfe@adnovum.ch>]
6305 *) Squeeze another 7% of speed out of MD5 assembler, at least on a P2. I'd
6306 like to hear about it if this slows down other processors.
6309 *) Add CygWin32 platform information to Configure script.
6310 [Alan Batie <batie@aahz.jf.intel.com>]
6312 *) Fixed ms/32all.bat script: `no_asm' -> `no-asm'
6313 [Rainer W. Gerling <gerling@mpg-gv.mpg.de>]
6315 *) New program nseq to manipulate netscape certificate sequences
6318 *) Modify crl2pkcs7 so it supports multiple -certfile arguments. Fix a
6322 *) Fixes to BN code. Previously the default was to define BN_RECURSION
6323 but the BN code had some problems that would cause failures when
6324 doing certificate verification and some other functions.
6325 [Eric A. Young, (from changes to C2Net SSLeay, integrated by Mark Cox)]
6327 *) Add ASN1 and PEM code to support netscape certificate sequences.
6330 *) Add ASN1 and PEM code to support netscape certificate sequences.
6333 *) Add several PKIX and private extended key usage OIDs.
6336 *) Modify the 'ca' program to handle the new extension code. Modify
6337 openssl.cnf for new extension format, add comments.
6340 *) More X509 V3 changes. Fix typo in v3_bitstr.c. Add support to 'req'
6341 and add a sample to openssl.cnf so req -x509 now adds appropriate
6345 *) Continued X509 V3 changes. Add to other makefiles, integrate with the
6346 error code, add initial support to X509_print() and x509 application.
6349 *) Takes a deep breath and start addding X509 V3 extension support code. Add
6350 files in crypto/x509v3. Move original stuff to crypto/x509v3/old. All this
6351 stuff is currently isolated and isn't even compiled yet.
6354 *) Continuing patches for GeneralizedTime. Fix up certificate and CRL
6355 ASN1 to use ASN1_TIME and modify print routines to use ASN1_TIME_print.
6356 Removed the versions check from X509 routines when loading extensions:
6357 this allows certain broken certificates that don't set the version
6358 properly to be processed.
6361 *) Deal with irritating shit to do with dependencies, in YAAHW (Yet Another
6362 Ad Hoc Way) - Makefile.ssls now all contain local dependencies, which
6363 can still be regenerated with "make depend".
6366 *) Spelling mistake in C version of CAST-128.
6367 [Ben Laurie, reported by Jeremy Hylton <jeremy@cnri.reston.va.us>]
6369 *) Changes to the error generation code. The perl script err-code.pl
6370 now reads in the old error codes and retains the old numbers, only
6371 adding new ones if necessary. It also only changes the .err files if new
6372 codes are added. The makefiles have been modified to only insert errors
6373 when needed (to avoid needlessly modifying header files). This is done
6374 by only inserting errors if the .err file is newer than the auto generated
6375 C file. To rebuild all the error codes from scratch (the old behaviour)
6376 either modify crypto/Makefile.ssl to pass the -regen flag to err_code.pl
6377 or delete all the .err files.
6380 *) CAST-128 was incorrectly implemented for short keys. The C version has
6381 been fixed, but is untested. The assembler versions are also fixed, but
6382 new assembler HAS NOT BEEN GENERATED FOR WIN32 - the Makefile needs fixing
6383 to regenerate it if needed.
6384 [Ben Laurie, reported (with fix for C version) by Jun-ichiro itojun
6385 Hagino <itojun@kame.net>]
6387 *) File was opened incorrectly in randfile.c.
6388 [Ulf Möller <ulf@fitug.de>]
6390 *) Beginning of support for GeneralizedTime. d2i, i2d, check and print
6391 functions. Also ASN1_TIME suite which is a CHOICE of UTCTime or
6392 GeneralizedTime. ASN1_TIME is the proper type used in certificates et
6393 al: it's just almost always a UTCTime. Note this patch adds new error
6394 codes so do a "make errors" if there are problems.
6397 *) Correct Linux 1 recognition in config.
6398 [Ulf Möller <ulf@fitug.de>]
6400 *) Remove pointless MD5 hash when using DSA keys in ca.
6401 [Anonymous <nobody@replay.com>]
6403 *) Generate an error if given an empty string as a cert directory. Also
6404 generate an error if handed NULL (previously returned 0 to indicate an
6405 error, but didn't set one).
6406 [Ben Laurie, reported by Anonymous <nobody@replay.com>]
6408 *) Add prototypes to SSL methods. Make SSL_write's buffer const, at last.
6411 *) Fix the dummy function BN_ref_mod_exp() in rsaref.c to have the correct
6412 parameters. This was causing a warning which killed off the Win32 compile.
6415 *) Remove C++ style comments from crypto/bn/bn_local.h.
6416 [Neil Costigan <neil.costigan@celocom.com>]
6418 *) The function OBJ_txt2nid was broken. It was supposed to return a nid
6419 based on a text string, looking up short and long names and finally
6420 "dot" format. The "dot" format stuff didn't work. Added new function
6421 OBJ_txt2obj to do the same but return an ASN1_OBJECT and rewrote
6422 OBJ_txt2nid to use it. OBJ_txt2obj can also return objects even if the
6423 OID is not part of the table.
6426 *) Add prototypes to X509 lookup/verify methods, fixing a bug in
6427 X509_LOOKUP_by_alias().
6430 *) Sort openssl functions by name.
6433 *) Get the gendsa program working (hopefully) and add it to app list. Remove
6434 encryption from sample DSA keys (in case anyone is interested the password
6438 *) Make _all_ *_free functions accept a NULL pointer.
6439 [Frans Heymans <fheymans@isaserver.be>]
6441 *) If a DH key is generated in s3_srvr.c, don't blow it by trying to use
6443 [Anonymous <nobody@replay.com>]
6445 *) s_server should send the CAfile as acceptable CAs, not its own cert.
6446 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
6448 *) Don't blow it for numeric -newkey arguments to apps/req.
6449 [Bodo Moeller <3moeller@informatik.uni-hamburg.de>]
6451 *) Temp key "for export" tests were wrong in s3_srvr.c.
6452 [Anonymous <nobody@replay.com>]
6454 *) Add prototype for temp key callback functions
6455 SSL_CTX_set_tmp_{rsa,dh}_callback().
6458 *) Make DH_free() tolerate being passed a NULL pointer (like RSA_free() and
6459 DSA_free()). Make X509_PUBKEY_set() check for errors in d2i_PublicKey().
6462 *) X509_name_add_entry() freed the wrong thing after an error.
6463 [Arne Ansper <arne@ats.cyber.ee>]
6465 *) rsa_eay.c would attempt to free a NULL context.
6466 [Arne Ansper <arne@ats.cyber.ee>]
6468 *) BIO_s_socket() had a broken should_retry() on Windoze.
6469 [Arne Ansper <arne@ats.cyber.ee>]
6471 *) BIO_f_buffer() didn't pass on BIO_CTRL_FLUSH.
6472 [Arne Ansper <arne@ats.cyber.ee>]
6474 *) Make sure the already existing X509_STORE->depth variable is initialized
6475 in X509_STORE_new(), but document the fact that this variable is still
6476 unused in the certificate verification process.
6477 [Ralf S. Engelschall]
6479 *) Fix the various library and apps files to free up pkeys obtained from
6480 X509_PUBKEY_get() et al. Also allow x509.c to handle netscape extensions.
6483 *) Fix reference counting in X509_PUBKEY_get(). This makes
6484 demos/maurice/example2.c work, amongst others, probably.
6485 [Steve Henson and Ben Laurie]
6487 *) First cut of a cleanup for apps/. First the `ssleay' program is now named
6488 `openssl' and second, the shortcut symlinks for the `openssl <command>'
6489 are no longer created. This way we have a single and consistent command
6490 line interface `openssl <command>', similar to `cvs <command>'.
6491 [Ralf S. Engelschall, Paul Sutton and Ben Laurie]
6493 *) ca.c: move test for DSA keys inside #ifndef NO_DSA. Make pubkey
6494 BIT STRING wrapper always have zero unused bits.
6497 *) Add CA.pl, perl version of CA.sh, add extended key usage OID.
6500 *) Make the top-level INSTALL documentation easier to understand.
6503 *) Makefiles updated to exit if an error occurs in a sub-directory
6504 make (including if user presses ^C) [Paul Sutton]
6506 *) Make Montgomery context stuff explicit in RSA data structure.
6509 *) Fix build order of pem and err to allow for generated pem.h.
6512 *) Fix renumbering bug in X509_NAME_delete_entry().
6515 *) Enhanced the err-ins.pl script so it makes the error library number
6516 global and can add a library name. This is needed for external ASN1 and
6517 other error libraries.
6520 *) Fixed sk_insert which never worked properly.
6523 *) Fix ASN1 macros so they can handle indefinite length construted
6524 EXPLICIT tags. Some non standard certificates use these: they can now
6528 *) Merged the various old/obsolete SSLeay documentation files (doc/xxx.doc)
6529 into a single doc/ssleay.txt bundle. This way the information is still
6530 preserved but no longer messes up this directory. Now it's new room for
6531 the new set of documenation files.
6532 [Ralf S. Engelschall]
6534 *) SETs were incorrectly DER encoded. This was a major pain, because they
6535 shared code with SEQUENCEs, which aren't coded the same. This means that
6536 almost everything to do with SETs or SEQUENCEs has either changed name or
6537 number of arguments.
6538 [Ben Laurie, based on a partial fix by GP Jayan <gp@nsj.co.jp>]
6540 *) Fix test data to work with the above.
6543 *) Fix the RSA header declarations that hid a bug I fixed in 0.9.0b but
6544 was already fixed by Eric for 0.9.1 it seems.
6545 [Ben Laurie - pointed out by Ulf Möller <ulf@fitug.de>]
6547 *) Autodetect FreeBSD3.
6550 *) Fix various bugs in Configure. This affects the following platforms:
6554 unixware-2.0-pentium
6558 *) Eliminate generated files from CVS. Reorder tests to regenerate files
6559 before they are needed.
6562 *) Generate Makefile.ssl from Makefile.org (to keep CVS happy).
6566 Changes between 0.9.1b and 0.9.1c [23-Dec-1998]
6568 *) Added OPENSSL_VERSION_NUMBER to crypto/crypto.h and
6569 changed SSLeay to OpenSSL in version strings.
6570 [Ralf S. Engelschall]
6572 *) Some fixups to the top-level documents.
6575 *) Fixed the nasty bug where rsaref.h was not found under compile-time
6576 because the symlink to include/ was missing.
6577 [Ralf S. Engelschall]
6579 *) Incorporated the popular no-RSA/DSA-only patches
6580 which allow to compile a RSA-free SSLeay.
6581 [Andrew Cooke / Interrader Ldt., Ralf S. Engelschall]
6583 *) Fixed nasty rehash problem under `make -f Makefile.ssl links'
6584 when "ssleay" is still not found.
6585 [Ralf S. Engelschall]
6587 *) Added more platforms to Configure: Cray T3E, HPUX 11,
6588 [Ralf S. Engelschall, Beckmann <beckman@acl.lanl.gov>]
6590 *) Updated the README file.
6591 [Ralf S. Engelschall]
6593 *) Added various .cvsignore files in the CVS repository subdirs
6594 to make a "cvs update" really silent.
6595 [Ralf S. Engelschall]
6597 *) Recompiled the error-definition header files and added
6598 missing symbols to the Win32 linker tables.
6599 [Ralf S. Engelschall]
6601 *) Cleaned up the top-level documents;
6602 o new files: CHANGES and LICENSE
6603 o merged VERSION, HISTORY* and README* files a CHANGES.SSLeay
6604 o merged COPYRIGHT into LICENSE
6605 o removed obsolete TODO file
6606 o renamed MICROSOFT to INSTALL.W32
6607 [Ralf S. Engelschall]
6609 *) Removed dummy files from the 0.9.1b source tree:
6610 crypto/asn1/x crypto/bio/cd crypto/bio/fg crypto/bio/grep crypto/bio/vi
6611 crypto/bn/asm/......add.c crypto/bn/asm/a.out crypto/dsa/f crypto/md5/f
6612 crypto/pem/gmon.out crypto/perlasm/f crypto/pkcs7/build crypto/rsa/f
6613 crypto/sha/asm/f crypto/threads/f ms/zzz ssl/f ssl/f.mak test/f
6614 util/f.mak util/pl/f util/pl/f.mak crypto/bf/bf_locl.old apps/f
6615 [Ralf S. Engelschall]
6617 *) Added various platform portability fixes.
6620 *) The Genesis of the OpenSSL rpject:
6621 We start with the latest (unreleased) SSLeay version 0.9.1b which Eric A.
6622 Young and Tim J. Hudson created while they were working for C2Net until
6624 [The OpenSSL Project]
6627 Changes between 0.9.0b and 0.9.1b [not released]
6629 *) Updated a few CA certificates under certs/
6632 *) Changed some BIGNUM api stuff.
6635 *) Various platform ports: OpenBSD, Ultrix, IRIX 64bit, NetBSD,
6636 DGUX x86, Linux Alpha, etc.
6639 *) New COMP library [crypto/comp/] for SSL Record Layer Compression:
6640 RLE (dummy implemented) and ZLIB (really implemented when ZLIB is
6644 *) Add -strparse option to asn1pars program which parses nested
6646 [Dr Stephen Henson <shenson@bigfoot.com>]
6648 *) Added "oid_file" to ssleay.cnf for "ca" and "req" programs.
6651 *) DSA fix for "ca" program.
6654 *) Added "-genkey" option to "dsaparam" program.
6657 *) Added RIPE MD160 (rmd160) message digest.
6660 *) Added -a (all) option to "ssleay version" command.
6663 *) Added PLATFORM define which is the id given to Configure.
6666 *) Added MemCheck_XXXX functions to crypto/mem.c for memory checking.
6669 *) Extended the ASN.1 parser routines.
6672 *) Extended BIO routines to support REUSEADDR, seek, tell, etc.
6675 *) Added a BN_CTX to the BN library.
6678 *) Fixed the weak key values in DES library
6681 *) Changed API in EVP library for cipher aliases.
6684 *) Added support for RC2/64bit cipher.
6687 *) Converted the lhash library to the crypto/mem.c functions.
6690 *) Added more recognized ASN.1 object ids.
6693 *) Added more RSA padding checks for SSL/TLS.
6696 *) Added BIO proxy/filter functionality.
6699 *) Added extra_certs to SSL_CTX which can be used
6700 send extra CA certificates to the client in the CA cert chain sending
6701 process. It can be configured with SSL_CTX_add_extra_chain_cert().
6704 *) Now Fortezza is denied in the authentication phase because
6705 this is key exchange mechanism is not supported by SSLeay at all.
6708 *) Additional PKCS1 checks.
6711 *) Support the string "TLSv1" for all TLS v1 ciphers.
6714 *) Added function SSL_get_ex_data_X509_STORE_CTX_idx() which gives the
6715 ex_data index of the SSL context in the X509_STORE_CTX ex_data.
6718 *) Fixed a few memory leaks.
6721 *) Fixed various code and comment typos.
6724 *) A minor bug in ssl/s3_clnt.c where there would always be 4 0
6725 bytes sent in the client random.
6726 [Edward Bishop <ebishop@spyglass.com>]