5 Changes between 0.9.6 and 0.9.7 [xx XXX 2001]
7 OpenSSL 0.9.6a/0.9.6b/0.9.6c (bugfix releases, 5 Apr 2001, 9 July 2001
8 and 21 Dec 2001) and OpenSSL 0.9.7 were developed in parallel, based
11 Change log entries are tagged as follows:
12 -) applies to 0.9.6a/0.9.6b/0.9.6c only
13 *) applies to 0.9.6a/0.9.6b/0.9.6c and 0.9.7
14 +) applies to 0.9.7 only
16 *) Add information about CygWin32 1.3 and on, and preserve proper
17 configuration for the versions before that.
18 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
20 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
21 check whether we deal with a copy of a session and do not delete from
22 the cache in this case. Problem reported by "Izhar Shoshani Levi"
23 <izhar@checkpoint.com>.
26 *) Do not store session data into the internal session cache, if it
27 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
28 flag is set). Proposed by Aslam <aslam@funk.com>.
31 +) Support for crypto accelerator cards from Accelerated Encryption
32 Processing, www.aep.ie. (Use engine 'aep')
33 The support was copied from 0.9.6c [engine] and adapted/corrected
34 to work with the new engine framework.
35 [AEP Inc. and Richard Levitte]
37 +) Support for SureWare crypto accelerator cards from Baltimore
38 Technologies. (Use engine 'sureware')
39 The support was copied from 0.9.6c [engine] and adapted
40 to work with the new engine framework.
43 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
47 +) Have the CHIL engine fork-safe (as defined by nCipher) and actually
48 make the newer ENGINE framework commands for the CHIL engine work.
49 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
51 +) Make it possible to produce shared libraries on ReliantUNIX.
52 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
54 *) Add the configuration target linux-s390x.
55 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
57 +) Add the configuration target debug-linux-ppro.
58 Make 'openssl rsa' use the general key loading routines
59 implemented in apps.c, and make those routines able to
60 handle the key format FORMAT_NETSCAPE and the variant
62 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
64 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
65 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
67 +) Add -keyform to rsautl, and document -engine.
68 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
70 +) Change BIO_new_file (crypto/bio/bss_file.c) to use new
71 BIO_R_NO_SUCH_FILE error code rather than the generic
72 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
77 ERR_peek_last_error_line
78 ERR_peek_last_error_line_data.
82 ERR_peek_error_line_data,
83 but report on the latest error recorded rather than the first one
84 still in the error queue.
85 [Ben Laurie, Bodo Moeller]
87 +) default_algorithms option in ENGINE config module. This allows things
89 default_algorithms = ALL
90 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
93 +) Prelminary ENGINE config module.
96 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
97 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
98 variable as an indication that a ClientHello message has been
99 received. As the flag value will be lost between multiple
100 invocations of ssl3_accept when using non-blocking I/O, the
101 function may not be aware that a handshake has actually taken
102 place, thus preventing a new session from being added to the
105 To avoid this problem, we now set s->new_session to 2 instead of
106 using a local variable.
107 [Lutz Jaenicke, Bodo Moeller]
109 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
110 if the SSL_R_LENGTH_MISMATCH error is detected.
111 [Geoff Thorpe, Bodo Moeller]
113 +) New experimental application configuration code.
116 *) New 'shared_ldflag' column in Configure platform table.
119 *) Fix EVP_CIPHER_mode macro.
120 ["Dan S. Camper" <dan@bti.net>]
122 +) Change the AES code to follow the same name structure as all other
123 symmetric ciphers, and behave the same way. Move everything to
124 the directory crypto/aes, thereby obsoleting crypto/rijndael.
125 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
127 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
128 type, we must throw them away by setting rr->length to 0.
129 [D P Chang <dpc@qualys.com>]
131 -) OpenSSL 0.9.6c released [21 dec 2001]
133 +) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
134 [Ben Laurie and Theo de Raadt]
136 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
137 <Dominikus.Scherkl@biodata.com>. (The previous implementation
138 worked incorrectly for those cases where range = 10..._2 and
139 3*range is two bits longer than range.)
142 *) Only add signing time to PKCS7 structures if it is not already
146 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
147 OBJ_ld_ce should be OBJ_id_ce.
148 Also some ip-pda OIDs in crypto/objects/objects.txt were
149 incorrect (cf. RFC 3039).
150 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
152 +) Add option to output public keys in req command.
153 [Massimiliano Pala madwolf@openca.org]
155 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
156 returns early because it has nothing to do.
157 [Andy Schneider <andy.schneider@bjss.co.uk>]
159 *) [In 0.9.6c-engine and 0.9.7 release:]
160 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
161 [Andy Schneider <andy.schneider@bjss.co.uk>]
163 -) [In 0.9.6c-engine release:]
164 Add support for Cryptographic Appliance's keyserver technology.
165 (Use engine 'keyclient')
166 [Cryptographic Appliances and Geoff Thorpe]
168 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
169 is called via tools/c89.sh because arguments have to be
170 rearranged (all '-L' options must appear before the first object
172 [Richard Shapiro <rshapiro@abinitio.com>]
174 +) Use wNAFs in EC_POINTs_mul() for improved efficiency
175 (up to about 10% better than before for P-192 and P-224).
178 -) [In 0.9.6c-engine release:]
179 Add support for Broadcom crypto accelerator cards, backported
181 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
183 -) [In 0.9.6c-engine release:]
184 Add support for SureWare crypto accelerator cards from
185 Baltimore Technologies. (Use engine 'sureware')
186 [Baltimore Technologies and Mark Cox]
188 -) [In 0.9.6c-engine release:]
189 Add support for crypto accelerator cards from Accelerated
190 Encryption Processing, www.aep.ie. (Use engine 'aep')
191 [AEP Inc. and Mark Cox]
193 *) Add a configuration entry for gcc on UnixWare.
194 [Gary Benson <gbenson@redhat.com>]
196 +) New functions/macros
198 SSL_CTX_set_msg_callback(ctx, cb)
199 SSL_CTX_set_msg_callback_arg(ctx, arg)
200 SSL_set_msg_callback(ssl, cb)
201 SSL_set_msg_callback_arg(ssl, arg)
203 to request calling a callback function
205 void cb(int write_p, int version, int content_type,
206 const void *buf, size_t len, SSL *ssl, void *arg)
208 whenever a protocol message has been completely received
209 (write_p == 0) or sent (write_p == 1). Here 'version' is the
210 protocol version according to which the SSL library interprets
211 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
212 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
213 the content type as defined in the SSL 3.0/TLS 1.0 protocol
214 specification (change_cipher_spec(20), alert(21), handshake(22)).
215 'buf' and 'len' point to the actual message, 'ssl' to the
216 SSL object, and 'arg' is the application-defined value set by
217 SSL[_CTX]_set_msg_callback_arg().
219 'openssl s_client' and 'openssl s_server' have new '-msg' options
220 to enable a callback that displays all protocol messages.
223 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
224 messages are stored in a single piece (fixed-length part and
225 variable-length part combined) and fix various bugs found on the way.
228 +) Change the shared library support so shared libraries are built as
229 soon as the corresponding static library is finished, and thereby get
230 openssl and the test programs linked against the shared library.
231 This still only happens when the keyword "shard" has been given to
232 the configuration scripts.
234 NOTE: shared library support is still an experimental thing, and
235 backward binary compatibility is still not guaranteed.
236 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
238 +) Add support for Subject Information Access extension.
239 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
241 +) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
242 additional bytes when new memory had to be allocated, not just
243 when reusing an existing buffer.
246 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
247 instead. BIO_gethostbyname() does not know what timeouts are
248 appropriate, so entries would stay in cache even when they have
250 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
252 +) New command line and configuration option 'utf8' for the req command.
253 This allows field values to be specified as UTF8 strings.
256 +) Add -multi and -mr options to "openssl speed" - giving multiple parallel
257 runs for the former and machine-readable output for the latter.
260 +) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
261 of the e-mail address in the DN (i.e., it will go into a certificate
262 extension only). The new configuration file option 'email_in_dn = no'
264 [Massimiliano Pala madwolf@openca.org]
266 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
267 faced with a pathologically small ClientHello fragment that does
268 not contain client_version: Instead of aborting with an error,
269 simply choose the highest available protocol version (i.e.,
270 TLS 1.0 unless it is disabled). In practice, ClientHello
271 messages are never sent like this, but this change gives us
272 strictly correct behaviour at least for TLS.
275 +) Change all functions with names starting with des_ to be starting
276 with DES_ instead. Add wrappers that are compatible with libdes,
277 but are named _ossl_old_des_*. Finally, add macros that map the
278 des_* symbols to the corresponding _ossl_old_des_*.
280 All this is done because there are increasing clashes with libdes
281 and other DES libraries that are currently used by other projects.
282 The old libdes interface (including crypt()) is provided if
283 <openssl/des_old.h> is included. For now, this automatically
284 happens in <openssl/des.h> unless OPENSSL_DISABLE_OLD_DES_SUPPORT is
285 defined. Note that crypt() is no longer declared in <openssl/des.h>.
287 NOTE: This is a major break of an old API into a new one. Software
288 authors are encouraged to switch to the DES_ style functions. Some
289 time in the future, des_old.h and the libdes compatibility functions
290 will be completely removed.
293 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
294 never resets s->method to s->ctx->method when called from within
295 one of the SSL handshake functions.
296 [Bodo Moeller; problem pointed out by Niko Baric]
298 +) Test for certificates which contain unsupported critical extensions.
299 If such a certificate is found during a verify operation it is
300 rejected by default: this behaviour can be overridden by either
301 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
302 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
303 X509_supported_extension() has also been added which returns 1 if a
304 particular extension is supported.
307 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
308 (sent using the client's version number) if client_version is
309 smaller than the protocol version in use. Also change
310 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
311 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
312 the client will at least see that alert.
315 +) Modify the behaviour of EVP cipher functions in similar way to digests
316 to retain compatibility with existing code.
319 +) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
320 compatibility with existing code. In particular the 'ctx' parameter does
321 not have to be to be initialized before the call to EVP_DigestInit() and
322 it is tidied up after a call to EVP_DigestFinal(). New function
323 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
324 EVP_MD_CTX_copy() changed to not require the destination to be
325 initialized valid and new function EVP_MD_CTX_copy_ex() added which
326 requires the destination to be valid.
328 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
329 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
332 +) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
333 so that complete 'Handshake' protocol structures are kept in memory
334 instead of overwriting 'msg_type' and 'length' with 'body' data.
337 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
341 +) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
342 [Massimo Santin via Richard Levitte]
344 +) Major restructuring to the underlying ENGINE code. This includes
345 reduction of linker bloat, separation of pure "ENGINE" manipulation
346 (initialisation, etc) from functionality dealing with implementations
347 of specific crypto iterfaces. This change also introduces integrated
348 support for symmetric ciphers and digest implementations - so ENGINEs
349 can now accelerate these by providing EVP_CIPHER and EVP_MD
350 implementations of their own. This is detailed in crypto/engine/README
351 as it couldn't be adequately described here. However, there are a few
352 API changes worth noting - some RSA, DSA, DH, and RAND functions that
353 were changed in the original introduction of ENGINE code have now
354 reverted back - the hooking from this code to ENGINE is now a good
355 deal more passive and at run-time, operations deal directly with
356 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
357 dereferencing through an ENGINE pointer any more. Also, the ENGINE
358 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
359 they were not being used by the framework as there is no concept of a
360 BIGNUM_METHOD and they could not be generalised to the new
361 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
362 ENGINE_cpy() has been removed as it cannot be consistently defined in
366 +) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
369 +) Change mkdef.pl to sort symbols that get the same entry number,
370 and make sure the automatically generated functions ERR_load_*
371 become part of libeay.num as well.
374 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
375 client receives HelloRequest while in a handshake.
376 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
378 +) New function SSL_renegotiate_pending(). This returns true once
379 renegotiation has been requested (either SSL_renegotiate() call
380 or HelloRequest/ClientHello receveived from the peer) and becomes
381 false once a handshake has been completed.
382 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
383 sends a HelloRequest, but does not ensure that a handshake takes
384 place. SSL_renegotiate_pending() is useful for checking if the
385 client has followed the request.)
388 +) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
389 By default, clients may request session resumption even during
390 renegotiation (if session ID contexts permit); with this option,
391 session resumption is possible only in the first handshake.
394 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
395 should end in 'break', not 'goto end' which circuments various
396 cleanups done in state SSL_ST_OK. But session related stuff
397 must be disabled for SSL_ST_OK in the case that we just sent a
400 Also avoid some overhead by not calling ssl_init_wbio_buffer()
401 before just sending a HelloRequest.
402 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
404 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
405 reveal whether illegal block cipher padding was found or a MAC
406 verification error occured. (Neither SSLerr() codes nor alerts
407 are directly visible to potential attackers, but the information
408 may leak via logfiles.)
410 Similar changes are not required for the SSL 2.0 implementation
411 because the number of padding bytes is sent in clear for SSL 2.0,
412 and the extra bytes are just ignored. However ssl/s2_pkt.c
413 failed to verify that the purported number of padding bytes is in
417 +) Add some demos for certificate and certificate request creation.
420 +) Make maximum certificate chain size accepted from the peer application
421 settable (SSL*_get/set_max_cert_list()), as proposed by
422 "Douglas E. Engert" <deengert@anl.gov>.
425 +) Add support for shared libraries for Unixware-7
426 (Boyd Lynn Gerber <gerberb@zenez.com>).
429 *) Add OpenUNIX-8 support including shared libraries
430 (Boyd Lynn Gerber <gerberb@zenez.com>).
433 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
434 'wristwatch attack' using huge encoding parameters (cf.
435 James H. Manger's CRYPTO 2001 paper). Note that the
436 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
437 encoding parameters and hence was not vulnerable.
440 +) Add a "destroy" handler to ENGINEs that allows structural cleanup to
441 be done prior to destruction. Use this to unload error strings from
442 ENGINEs that load their own error strings. NB: This adds two new API
443 functions to "get" and "set" this destroy handler in an ENGINE.
446 +) Alter all existing ENGINE implementations (except "openssl" and
447 "openbsd") to dynamically instantiate their own error strings. This
448 makes them more flexible to be built both as statically-linked ENGINEs
449 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
450 Also, add stub code to each that makes building them as self-contained
451 shared-libraries easier (see README.ENGINE).
454 +) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
455 implementations into applications that are completely implemented in
456 self-contained shared-libraries. The "dynamic" ENGINE exposes control
457 commands that can be used to configure what shared-library to load and
458 to control aspects of the way it is handled. Also, made an update to
459 the README.ENGINE file that brings its information up-to-date and
460 provides some information and instructions on the "dynamic" ENGINE
461 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
465 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
467 +) Make it possible to unload ranges of ERR strings with a new
468 "ERR_unload_strings" function.
471 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
472 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
473 followed by modular reduction.
474 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
476 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
477 equivalent based on BN_pseudo_rand() instead of BN_rand().
480 +) Add a copy() function to EVP_MD.
483 +) Make EVP_MD routines take a context pointer instead of just the
484 md_data void pointer.
487 +) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
488 that the digest can only process a single chunk of data
489 (typically because it is provided by a piece of
490 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
491 is only going to provide a single chunk of data, and hence the
492 framework needn't accumulate the data for oneshot drivers.
495 +) As with "ERR", make it possible to replace the underlying "ex_data"
496 functions. This change also alters the storage and management of global
497 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
498 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
499 index counters. The API functions that use this state have been changed
500 to take a "class_index" rather than pointers to the class's local STACK
501 and counter, and there is now an API function to dynamically create new
502 classes. This centralisation allows us to (a) plug a lot of the
503 thread-safety problems that existed, and (b) makes it possible to clean
504 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
505 such data would previously have always leaked in application code and
506 workarounds were in place to make the memory debugging turn a blind eye
507 to it. Application code that doesn't use this new function will still
508 leak as before, but their memory debugging output will announce it now
509 rather than letting it slide.
511 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
512 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
513 has a return value to indicate success or failure.
516 +) Make it possible to replace the underlying "ERR" functions such that the
517 global state (2 LHASH tables and 2 locks) is only used by the "default"
518 implementation. This change also adds two functions to "get" and "set"
519 the implementation prior to it being automatically set the first time
520 any other ERR function takes place. Ie. an application can call "get",
521 pass the return value to a module it has just loaded, and that module
522 can call its own "set" function using that value. This means the
523 module's "ERR" operations will use (and modify) the error state in the
524 application and not in its own statically linked copy of OpenSSL code.
527 +) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
528 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
529 the operation, and provides a more encapsulated way for external code
530 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
531 to use these functions rather than manually incrementing the counts.
533 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
536 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
537 This function was broken, as the check for a new client hello message
538 to handle SGC did not allow these large messages.
539 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
542 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
545 +) Add EVP test program.
548 +) Add symmetric cipher support to ENGINE. Expect the API to change!
551 +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
552 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
553 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
554 These allow a CRL to be built without having to access X509_CRL fields
555 directly. Modify 'ca' application to use new functions.
558 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
559 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
562 *) Rework the configuration and shared library support for Tru64 Unix.
563 The configuration part makes use of modern compiler features and
564 still retains old compiler behavior for those that run older versions
565 of the OS. The shared library support part includes a variant that
566 uses the RPATH feature, and is available through the special
567 configuration target "alpha-cc-rpath", which will never be selected
569 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
571 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
572 with the same message size as in ssl3_get_certificate_request().
573 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
574 messages might inadvertently be reject as too long.
575 [Petr Lampa <lampa@fee.vutbr.cz>]
577 +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
578 bug workarounds. Rollback attack detection is a security feature.
579 The problem will only arise on OpenSSL servers when TLSv1 is not
580 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
581 Software authors not wanting to support TLSv1 will have special reasons
582 for their choice and can explicitly enable this option.
583 [Bodo Moeller, Lutz Jaenicke]
585 +) Rationalise EVP so it can be extended: don't include a union of
586 cipher/digest structures, add init/cleanup functions. This also reduces
587 the number of header dependencies.
590 +) Make DES key schedule conform to the usual scheme, as well as
591 correcting its structure. This means that calls to DES functions
592 now have to pass a pointer to a des_key_schedule instead of a
593 plain des_key_schedule (which was actually always a pointer
597 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
600 *) Modified SSL library such that the verify_callback that has been set
601 specificly for an SSL object with SSL_set_verify() is actually being
602 used. Before the change, a verify_callback set with this function was
603 ignored and the verify_callback() set in the SSL_CTX at the time of
604 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
605 to allow the necessary settings.
608 +) Initial reduction of linker bloat: the use of some functions, such as
609 PEM causes large amounts of unused functions to be linked in due to
610 poor organisation. For example pem_all.c contains every PEM function
611 which has a knock on effect of linking in large amounts of (unused)
612 ASN1 code. Grouping together similar functions and splitting unrelated
613 functions prevents this.
616 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
617 explicitly to NULL, as at least on Solaris 8 this seems not always to be
618 done automatically (in contradiction to the requirements of the C
619 standard). This made problems when used from OpenSSH.
622 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
623 dh->length and always used
625 BN_rand_range(priv_key, dh->p).
627 BN_rand_range() is not necessary for Diffie-Hellman, and this
628 specific range makes Diffie-Hellman unnecessarily inefficient if
629 dh->length (recommended exponent length) is much smaller than the
630 length of dh->p. We could use BN_rand_range() if the order of
631 the subgroup was stored in the DH structure, but we only have
636 BN_rand(priv_key, l, ...)
638 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
644 RSA_eay_public_encrypt
645 RSA_eay_private_decrypt
646 RSA_eay_private_encrypt (signing)
647 RSA_eay_public_decrypt (signature verification)
649 (default implementations for RSA_public_encrypt,
650 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
651 always reject numbers >= n.
654 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
655 to synchronize access to 'locking_thread'. This is necessary on
656 systems where access to 'locking_thread' (an 'unsigned long'
657 variable) is not atomic.
660 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
661 *before* setting the 'crypto_lock_rand' flag. The previous code had
662 a race condition if 0 is a valid thread ID.
663 [Travis Vitek <vitek@roguewave.com>]
665 +) Cleanup of EVP macros.
668 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
672 +) Add initial OCSP responder support to ocsp application. The
673 revocation information is handled using the text based index
674 use by the ca application. The responder can either handle
675 requests generated internally, supplied in files (for example
676 via a CGI script) or using an internal minimal server.
679 +) Add configuration choices to get zlib compression for TLS.
682 +) Changes to Kerberos SSL for RFC 2712 compliance:
683 1. Implemented real KerberosWrapper, instead of just using
684 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
685 2. Implemented optional authenticator field of KerberosWrapper.
687 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
688 and authenticator structs; see crypto/krb5/.
690 Generalized Kerberos calls to support multiple Kerberos libraries.
691 [Vern Staats <staatsvr@asc.hpc.mil>,
692 Jeffrey Altman <jaltman@columbia.edu>
695 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
696 already does with RSA. testdsa.h now has 'priv_key/pub_key'
697 values for each of the key sizes rather than having just
698 parameters (and 'speed' generating keys each time).
701 *) Add support for shared libraries under Irix.
702 [Albert Chin-A-Young <china@thewrittenword.com>]
704 *) Add configuration option to build on Linux on both big-endian and
706 [Ralf Baechle <ralf@uni-koblenz.de>]
708 *) Add the possibility to create shared libraries on HP-UX.
711 -) OpenSSL 0.9.6b released [9 July 2001]
713 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
714 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
715 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
716 PRNG state recovery was possible based on the output of
717 one PRNG request appropriately sized to gain knowledge on
718 'md' followed by enough consecutive 1-byte PRNG requests
719 to traverse all of 'state'.
721 1. When updating 'md_local' (the current thread's copy of 'md')
722 during PRNG output generation, hash all of the previous
723 'md_local' value, not just the half used for PRNG output.
725 2. Make the number of bytes from 'state' included into the hash
726 independent from the number of PRNG bytes requested.
728 The first measure alone would be sufficient to avoid
729 Markku-Juhani's attack. (Actually it had never occurred
730 to me that the half of 'md_local' used for chaining was the
731 half from which PRNG output bytes were taken -- I had always
732 assumed that the secret half would be used.) The second
733 measure makes sure that additional data from 'state' is never
734 mixed into 'md_local' in small portions; this heuristically
735 further strengthens the PRNG.
738 +) Speed up EVP routines.
741 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
742 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
743 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
744 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
746 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
747 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
748 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
751 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
753 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
756 *) Fix crypto/bn/asm/mips3.s.
759 *) When only the key is given to "enc", the IV is undefined. Print out
760 an error message in this case.
763 +) Added the OS2-EMX target.
764 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
766 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
767 to support NCONF routines in extension code. New function CONF_set_nconf()
768 to allow functions which take an NCONF to also handle the old LHASH
769 structure: this means that the old CONF compatible routines can be
770 retained (in particular wrt extensions) without having to duplicate the
771 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
774 *) Handle special case when X509_NAME is empty in X509 printing routines.
777 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
778 positive and less than q.
781 +) Enhance the general user interface with mechanisms for inner control
782 and with possibilities to have yes/no kind of prompts.
785 +) Change all calls to low level digest routines in the library and
786 applications to use EVP. Add missing calls to HMAC_cleanup() and
787 don't assume HMAC_CTX can be copied using memcpy().
788 [Verdon Walker <VWalker@novell.com>, Steve Henson]
790 +) Add the possibility to control engines through control names but with
791 arbitrary arguments instead of just a string.
792 Change the key loaders to take a UI_METHOD instead of a callback
793 function pointer. NOTE: this breaks binary compatibility with earlier
794 versions of OpenSSL [engine].
795 Adapt the nCipher code for these new conditions and add a card insertion
799 +) Enhance the general user interface with mechanisms to better support
800 dialog box interfaces, application-defined prompts, the possibility
801 to use defaults (for example default passwords from somewhere else)
802 and interrupts/cancellations.
805 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
806 used: it isn't thread safe and the add_lock_callback should handle
808 [Paul Rose <Paul.Rose@bridge.com>]
810 *) Verify that incoming data obeys the block size in
811 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
814 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
815 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
819 [Ulf Möller, Bodo Möller]
821 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
822 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
823 when fixing the server behaviour for backwards-compatible 'client
824 hello' messages. (Note that the attack is impractical against
825 SSL 3.0 and TLS 1.0 anyway because length and version checking
826 means that the probability of guessing a valid ciphertext is
827 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
830 Before 0.9.5, the countermeasure (hide the error by generating a
831 random 'decryption result') did not work properly because
832 ERR_clear_error() was missing, meaning that SSL_get_error() would
833 detect the supposedly ignored error.
835 Both problems are now fixed.
838 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
839 (previously it was 1024).
842 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
843 tidy up some unnecessarily weird code in 'sk_new()').
844 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
846 +) Change the key loading routines for ENGINEs to use the same kind
847 callback (pem_password_cb) as all other routines that need this
851 *) Fix for compatibility mode trust settings: ignore trust settings
852 unless some valid trust or reject settings are present.
855 *) Fix for blowfish EVP: its a variable length cipher.
858 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
859 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
860 than this minimum value is recommended.
863 +) New random seeder for OpenVMS, using the system process statistics
864 that are easily reachable.
867 +) Windows apparently can't transparently handle global
868 variables defined in DLLs. Initialisations such as:
870 const ASN1_ITEM *it = &ASN1_INTEGER_it;
872 wont compile. This is used by the any applications that need to
873 declare their own ASN1 modules. This was fixed by adding the option
874 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
875 needed for static libraries under Win32.
878 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
879 setting of purpose and trust fields. New X509_STORE trust and
880 purpose functions and tidy up setting in other SSL functions.
883 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
884 structure. These are inherited by X509_STORE_CTX when it is
885 initialised. This allows various defaults to be set in the
886 X509_STORE structure (such as flags for CRL checking and custom
887 purpose or trust settings) for functions which only use X509_STORE_CTX
888 internally such as S/MIME.
890 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
891 trust settings if they are not set in X509_STORE. This allows X509_STORE
892 purposes and trust (in S/MIME for example) to override any set by default.
894 Add command line options for CRL checking to smime, s_client and s_server
898 +) Initial CRL based revocation checking. If the CRL checking flag(s)
899 are set then the CRL is looked up in the X509_STORE structure and
900 its validity and signature checked, then if the certificate is found
901 in the CRL the verify fails with a revoked error.
903 Various new CRL related callbacks added to X509_STORE_CTX structure.
905 Command line options added to 'verify' application to support this.
907 This needs some additional work, such as being able to handle multiple
908 CRLs with different times, extension based lookup (rather than just
909 by subject name) and ultimately more complete V2 CRL extension
913 +) Add a general user interface API (crypto/ui/). This is designed
914 to replace things like des_read_password and friends (backward
915 compatibility functions using this new API are provided).
916 The purpose is to remove prompting functions from the DES code
917 section as well as provide for prompting through dialog boxes in
918 a window system and the like.
921 *) Fix various bugs related to DSA S/MIME verification. Handle missing
922 parameters in DSA public key structures and return an error in the
923 DSA routines if parameters are absent.
926 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
927 in the current directory if neither $RANDFILE nor $HOME was set.
928 RAND_file_name() in 0.9.6a returned NULL in this case. This has
929 caused some confusion to Windows users who haven't defined $HOME.
930 Thus RAND_file_name() is changed again: e_os.h can define a
931 DEFAULT_HOME, which will be used if $HOME is not set.
932 For Windows, we use "C:"; on other platforms, we still require
933 environment variables.
935 +) Add "ex_data" support to ENGINE so implementations can add state at a
936 per-structure level rather than having to store it globally.
939 +) Make it possible for ENGINE structures to be copied when retrieved by
940 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
941 This causes the "original" ENGINE structure to act like a template,
942 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
943 operational state can be localised to each ENGINE structure, despite the
944 fact they all share the same "methods". New ENGINE structures returned in
945 this case have no functional references and the return value is the single
946 structural reference. This matches the single structural reference returned
947 by ENGINE_by_id() normally, when it is incremented on the pre-existing
951 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
952 needs to match any other type at all we need to manually clear the
956 +) Changes to the "openssl engine" utility to include;
957 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
958 about an ENGINE's available control commands.
959 - executing control commands from command line arguments using the
960 '-pre' and '-post' switches. '-post' is only used if '-t' is
961 specified and the ENGINE is successfully initialised. The syntax for
962 the individual commands are colon-separated, for example;
963 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
966 +) New dynamic control command support for ENGINEs. ENGINEs can now
967 declare their own commands (numbers), names (strings), descriptions,
968 and input types for run-time discovery by calling applications. A
969 subset of these commands are implicitly classed as "executable"
970 depending on their input type, and only these can be invoked through
971 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
972 can be based on user input, config files, etc). The distinction is
973 that "executable" commands cannot return anything other than a boolean
974 result and can only support numeric or string input, whereas some
975 discoverable commands may only be for direct use through
976 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
977 pointers, or other custom uses. The "executable" commands are to
978 support parameterisations of ENGINE behaviour that can be
979 unambiguously defined by ENGINEs and used consistently across any
980 OpenSSL-based application. Commands have been added to all the
981 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
982 control over shared-library paths without source code alterations.
985 +) Changed all ENGINE implementations to dynamically allocate their
986 ENGINEs rather than declaring them statically. Apart from this being
987 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
988 this also allows the implementations to compile without using the
989 internal engine_int.h header.
992 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
993 'const' value. Any code that should be able to modify a RAND_METHOD
994 should already have non-const pointers to it (ie. they should only
995 modify their own ones).
998 +) Made a variety of little tweaks to the ENGINE code.
999 - "atalla" and "ubsec" string definitions were moved from header files
1000 to C code. "nuron" string definitions were placed in variables
1001 rather than hard-coded - allowing parameterisation of these values
1002 later on via ctrl() commands.
1003 - Removed unused "#if 0"'d code.
1004 - Fixed engine list iteration code so it uses ENGINE_free() to release
1005 structural references.
1006 - Constified the RAND_METHOD element of ENGINE structures.
1007 - Constified various get/set functions as appropriate and added
1008 missing functions (including a catch-all ENGINE_cpy that duplicates
1009 all ENGINE values onto a new ENGINE except reference counts/state).
1010 - Removed NULL parameter checks in get/set functions. Setting a method
1011 or function to NULL is a way of cancelling out a previously set
1012 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1013 and doesn't justify the extra error symbols and code.
1014 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1015 flags from engine_int.h to engine.h.
1016 - Changed prototypes for ENGINE handler functions (init(), finish(),
1017 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1020 *) Move 'if (!initialized) RAND_poll()' into regions protected by
1021 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
1022 having multiple threads call RAND_poll() concurrently.
1025 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
1026 combination of a flag and a thread ID variable.
1027 Otherwise while one thread is in ssleay_rand_bytes (which sets the
1028 flag), *other* threads can enter ssleay_add_bytes without obeying
1029 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
1030 that they do not hold after the first thread unsets add_do_not_lock).
1033 +) Implement binary inversion algorithm for BN_mod_inverse in addition
1034 to the algorithm using long division. The binary algorithm can be
1035 used only if the modulus is odd. On 32-bit systems, it is faster
1036 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1037 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1038 up to 450 bits. In 64-bit environments, the binary algorithm
1039 appears to be advantageous for much longer moduli; here we use it
1040 for moduli up to 2048 bits.
1043 *) Change bctest again: '-x' expressions are not available in all
1047 -) OpenSSL 0.9.6a released [5 Apr 2001]
1049 *) Fix a couple of memory leaks in PKCS7_dataDecode()
1050 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
1052 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
1053 the default extension for executables, if any. Also, make the perl
1054 scripts that use symlink() to test if it really exists and use "cp"
1055 if it doesn't. All this made OpenSSL compilable and installable in
1059 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1060 could not support the combine flag in choice fields.
1063 -) Fix for asn1_GetSequence() for indefinite length constructed data.
1064 If SEQUENCE is length is indefinite just set c->slen to the total
1065 amount of data available.
1066 [Steve Henson, reported by shige@FreeBSD.org]
1067 [This change does not apply to 0.9.7.]
1069 *) Change bctest to avoid here-documents inside command substitution
1070 (workaround for FreeBSD /bin/sh bug).
1071 For compatibility with Ultrix, avoid shell functions (introduced
1072 in the bctest version that searches along $PATH).
1075 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
1076 with des_encrypt() defined on some operating systems, like Solaris
1080 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
1081 On the Importance of Eliminating Errors in Cryptographic
1082 Computations, J. Cryptology 14 (2001) 2, 101-119,
1083 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
1086 *) MIPS assembler BIGNUM division bug fix.
1089 *) Disabled incorrect Alpha assembler code.
1092 -) Fix PKCS#7 decode routines so they correctly update the length
1093 after reading an EOC for the EXPLICIT tag.
1095 [This change does not apply to 0.9.7.]
1097 *) Fix bug in PKCS#12 key generation routines. This was triggered
1098 if a 3DES key was generated with a 0 initial byte. Include
1099 PKCS12_BROKEN_KEYGEN compilation option to retain the old
1100 (but broken) behaviour.
1103 *) Enhance bctest to search for a working bc along $PATH and print
1105 [Tim Rice <tim@multitalents.net> via Richard Levitte]
1107 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
1108 extensions from a certificate request to the certificate.
1111 +) Allow multiple 'certopt' and 'nameopt' options to be separated
1112 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1113 file: this allows the display of the certificate about to be
1114 signed to be customised, to allow certain fields to be included
1115 or excluded and extension details. The old system didn't display
1116 multicharacter strings properly, omitted fields not in the policy
1117 and couldn't display additional details such as extensions.
1120 *) Fix memory leaks in err.c: free err_data string if necessary;
1121 don't write to the wrong index in ERR_set_error_data.
1124 +) Function EC_POINTs_mul for multiple scalar multiplication
1125 of an arbitrary number of elliptic curve points
1126 \sum scalars[i]*points[i],
1127 optionally including the generator defined for the EC_GROUP:
1128 scalar*generator + \sum scalars[i]*points[i].
1130 EC_POINT_mul is a simple wrapper function for the typical case
1131 that the point list has just one item (besides the optional
1135 +) First EC_METHODs for curves over GF(p):
1137 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1138 operations and provides various method functions that can also
1139 operate with faster implementations of modular arithmetic.
1141 EC_GFp_mont_method() reuses most functions that are part of
1142 EC_GFp_simple_method, but uses Montgomery arithmetic.
1144 [Bodo Moeller; point addition and point doubling
1145 implementation directly derived from source code provided by
1146 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1148 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1149 crypto/ec/ec_lib.c):
1151 Curves are EC_GROUP objects (with an optional group generator)
1152 based on EC_METHODs that are built into the library.
1154 Points are EC_POINT objects based on EC_GROUP objects.
1156 Most of the framework would be able to handle curves over arbitrary
1157 finite fields, but as there are no obvious types for fields other
1158 than GF(p), some functions are limited to that for now.
1161 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1162 that the file contains a complete HTTP response.
1165 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1166 change the def and num file printf format specifier from "%-40sXXX"
1167 to "%-39s XXX". The latter will always guarantee a space after the
1168 field while the former will cause them to run together if the field
1169 is 40 of more characters long.
1172 +) Constify the cipher and digest 'method' functions and structures
1173 and modify related functions to take constant EVP_MD and EVP_CIPHER
1177 *) Implement ssl23_peek (analogous to ssl23_read), which previously
1181 *) Replace rdtsc with _emit statements for VC++ version 5.
1182 [Jeremy Cooper <jeremy@baymoo.org>]
1184 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1185 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1188 +) Modify EVP_Digest*() routines so they now return values. Although the
1189 internal software routines can never fail additional hardware versions
1193 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1195 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1196 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1199 ERR_R_NESTED_ASN1_ERROR
1201 ERR_R_MISSING_ASN1_EOS
1202 were 4 .. 9, conflicting with
1203 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1205 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1206 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1208 Add new error code 'ERR_R_INTERNAL_ERROR'.
1211 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1215 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1216 sets the subject name for a new request or supersedes the
1217 subject name in a given request. Formats that can be parsed are
1218 'CN=Some Name, OU=myOU, C=IT'
1220 'CN=Some Name/OU=myOU/C=IT'.
1222 Add options '-batch' and '-verbose' to 'openssl req'.
1223 [Massimiliano Pala <madwolf@hackmasters.net>]
1225 *) Make it possible to reuse SSLv2 sessions.
1228 +) Introduce the possibility to access global variables through
1229 functions on platform were that's the best way to handle exporting
1230 global variables in shared libraries. To enable this functionality,
1231 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1232 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1233 is normally done by Configure or something similar).
1235 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1236 in the source file (foo.c) like this:
1238 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1239 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1241 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1242 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1244 OPENSSL_DECLARE_GLOBAL(int,foo);
1245 #define foo OPENSSL_GLOBAL_REF(foo)
1246 OPENSSL_DECLARE_GLOBAL(double,bar);
1247 #define bar OPENSSL_GLOBAL_REF(bar)
1249 The #defines are very important, and therefore so is including the
1250 header file everywhere where the defined globals are used.
1252 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1253 of ASN.1 items, but that structure is a bit different.
1255 The largest change is in util/mkdef.pl which has been enhanced with
1256 better and easier to understand logic to choose which symbols should
1257 go into the Windows .def files as well as a number of fixes and code
1258 cleanup (among others, algorithm keywords are now sorted
1259 lexicographically to avoid constant rewrites).
1262 *) In copy_email() check for >= 0 as a return value for
1263 X509_NAME_get_index_by_NID() since 0 is a valid index.
1264 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
1266 +) In BN_div() keep a copy of the sign of 'num' before writing the
1267 result to 'rm' because if rm==num the value will be overwritten
1268 and produce the wrong result if 'num' is negative: this caused
1269 problems with BN_mod() and BN_nnmod().
1272 +) Function OCSP_request_verify(). This checks the signature on an
1273 OCSP request and verifies the signer certificate. The signer
1274 certificate is just checked for a generic purpose and OCSP request
1278 +) Add OCSP_check_validity() function to check the validity of OCSP
1279 responses. OCSP responses are prepared in real time and may only
1280 be a few seconds old. Simply checking that the current time lies
1281 between thisUpdate and nextUpdate max reject otherwise valid responses
1282 caused by either OCSP responder or client clock inaccuracy. Instead
1283 we allow thisUpdate and nextUpdate to fall within a certain period of
1284 the current time. The age of the response can also optionally be
1285 checked. Two new options -validity_period and -status_age added to
1289 +) If signature or public key algorithm is unrecognized print out its
1290 OID rather that just UNKNOWN.
1293 *) Avoid coredump with unsupported or invalid public keys by checking if
1294 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
1295 PKCS7_verify() fails with non detached data.
1298 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1299 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1300 ID to be generated from the issuer certificate alone which can then be
1301 passed to OCSP_id_issuer_cmp().
1304 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1305 ASN1 modules to export functions returning ASN1_ITEM pointers
1306 instead of the ASN1_ITEM structures themselves. This adds several
1307 new macros which allow the underlying ASN1 function/structure to
1308 be accessed transparently. As a result code should not use ASN1_ITEM
1309 references directly (such as &X509_it) but instead use the relevant
1310 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1311 use of the new ASN1 code on platforms where exporting structures
1312 is problematical (for example in shared libraries) but exporting
1313 functions returning pointers to structures is not.
1316 +) Add support for overriding the generation of SSL/TLS session IDs.
1317 These callbacks can be registered either in an SSL_CTX or per SSL.
1318 The purpose of this is to allow applications to control, if they wish,
1319 the arbitrary values chosen for use as session IDs, particularly as it
1320 can be useful for session caching in multiple-server environments. A
1321 command-line switch for testing this (and any client code that wishes
1322 to use such a feature) has been added to "s_server".
1323 [Geoff Thorpe, Lutz Jaenicke]
1325 +) Modify mkdef.pl to recognise and parse preprocessor conditionals
1326 of the form '#if defined(...) || defined(...) || ...' and
1327 '#if !defined(...) && !defined(...) && ...'. This also avoids
1328 the growing number of special cases it was previously handling.
1331 *) Don't use getenv in library functions when run as setuid/setgid.
1332 New function OPENSSL_issetugid().
1335 +) Make all configuration macros available for application by making
1336 sure they are available in opensslconf.h, by giving them names starting
1337 with "OPENSSL_" to avoid conflicts with other packages and by making
1338 sure e_os2.h will cover all platform-specific cases together with
1340 Additionally, it is now possible to define configuration/platform-
1341 specific names (called "system identities"). In the C code, these
1342 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1343 macro with the name beginning with "OPENSSL_SYS_", which is determined
1344 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1348 +) New option -set_serial to 'req' and 'x509' this allows the serial
1349 number to use to be specified on the command line. Previously self
1350 signed certificates were hard coded with serial number 0 and the
1351 CA options of 'x509' had to use a serial number in a file which was
1355 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
1356 due to incorrect handling of multi-threading:
1358 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
1360 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
1362 3. Count how many times MemCheck_off() has been called so that
1363 nested use can be treated correctly. This also avoids
1364 inband-signalling in the previous code (which relied on the
1365 assumption that thread ID 0 is impossible).
1368 +) New options to 'ca' utility to support V2 CRL entry extensions.
1369 Currently CRL reason, invalidity date and hold instruction are
1370 supported. Add new CRL extensions to V3 code and some new objects.
1373 *) Add "-rand" option also to s_client and s_server.
1376 +) New function EVP_CIPHER_CTX_set_padding() this is used to
1377 disable standard block padding (aka PKCS#5 padding) in the EVP
1378 API, which was previously mandatory. This means that the data is
1379 not padded in any way and so the total length much be a multiple
1380 of the block size, otherwise an error occurs.
1383 +) Initial (incomplete) OCSP SSL support.
1386 *) Fix CPU detection on Irix 6.x.
1387 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
1388 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1390 +) New function OCSP_parse_url(). This splits up a URL into its host,
1391 port and path components: primarily to parse OCSP URLs. New -url
1392 option to ocsp utility.
1395 +) New nonce behavior. The return value of OCSP_check_nonce() now
1396 reflects the various checks performed. Applications can decide
1397 whether to tolerate certain situations such as an absent nonce
1398 in a response when one was present in a request: the ocsp application
1399 just prints out a warning. New function OCSP_add1_basic_nonce()
1400 this is to allow responders to include a nonce in a response even if
1401 the request is nonce-less.
1404 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
1407 [This change does not apply to 0.9.7.]
1409 *) Use the cached encoding of an X509_NAME structure rather than
1410 copying it. This is apparently the reason for the libsafe "errors"
1411 but the code is actually correct.
1414 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1415 skipped when using openssl x509 multiple times on a single input file,
1416 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1419 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1420 set string type: to handle setting ASN1_TIME structures. Fix ca
1421 utility to correctly initialize revocation date of CRLs.
1424 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1425 the clients preferred ciphersuites and rather use its own preferences.
1426 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1427 Internet Explorer by ensuring unchanged hash method during stepup.
1428 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1431 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1432 to aes and add a new 'exist' option to print out symbols that don't
1436 +) Additional options to ocsp utility to allow flags to be set and
1437 additional certificates supplied.
1440 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
1441 OCSP client a number of certificate to only verify the response
1445 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
1446 Bleichenbacher's DSA attack.
1447 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
1448 to be set and top=0 forces the highest bit to be set; top=-1 is new
1449 and leaves the highest bit random.
1450 [Ulf Moeller, Bodo Moeller]
1452 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1453 handle the new API. Currently only ECB, CBC modes supported. Add new
1454 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
1455 for TLS" draft-ietf-tls-ciphersuite-03.txt.
1456 [Ben Laurie, Steve Henson]
1458 *) In the NCONF_...-based implementations for CONF_... queries
1459 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
1460 a temporary CONF structure with the data component set to NULL
1461 (which gives segmentation faults in lh_retrieve).
1462 Instead, use NULL for the CONF pointer in CONF_get_string and
1463 CONF_get_number (which may use environment variables) and directly
1464 return NULL from CONF_get_section.
1467 *) Fix potential buffer overrun for EBCDIC.
1470 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
1471 request to response.
1474 +) Functions for OCSP responders. OCSP_request_onereq_count(),
1475 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1476 extract information from a certificate request. OCSP_response_create()
1477 creates a response and optionally adds a basic response structure.
1478 OCSP_basic_add1_status() adds a complete single response to a basic
1479 response and returns the OCSP_SINGLERESP structure just added (to allow
1480 extensions to be included for example). OCSP_basic_add1_cert() adds a
1481 certificate to a basic response and OCSP_basic_sign() signs a basic
1482 response with various flags. New helper functions ASN1_TIME_check()
1483 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1484 (converts ASN1_TIME to GeneralizedTime).
1487 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1488 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1489 structure from a certificate. X509_pubkey_digest() digests the public_key
1490 contents: this is used in various key identifiers.
1493 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
1494 keyUsage if basicConstraints absent for a CA.
1497 *) Make SMIME_write_PKCS7() write mail header values with a format that
1498 is more generally accepted (no spaces before the semicolon), since
1499 some programs can't parse those values properly otherwise. Also make
1500 sure BIO's that break lines after each write do not create invalid
1504 +) Make sk_sort() tolerate a NULL argument.
1505 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1507 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1508 passed by the function are trusted implicitly. If any of them signed the
1509 response then it is assumed to be valid and is not verified.
1512 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
1513 macros previously used would not encode an empty SEQUENCE OF
1514 and break the signature.
1516 [This change does not apply to 0.9.7.]
1518 *) Zero the premaster secret after deriving the master secret in
1522 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1523 to data. This was previously part of the PKCS7 ASN1 code. This
1524 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1525 [Steve Henson, reported by Kenneth R. Robinette
1526 <support@securenetterm.com>]
1528 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1529 routines: without these tracing memory leaks is very painful.
1530 Fix leaks in PKCS12 and PKCS7 routines.
1533 *) Add some EVP_add_digest_alias registrations (as found in
1534 OpenSSL_add_all_digests()) to SSL_library_init()
1535 aka OpenSSL_add_ssl_algorithms(). This provides improved
1536 compatibility with peers using X.509 certificates
1537 with unconventional AlgorithmIdentifier OIDs.
1540 *) Fix for Irix with NO_ASM.
1541 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1543 *) ./config script fixes.
1544 [Ulf Moeller, Richard Levitte]
1546 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1547 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1548 effectively meant GeneralizedTime would never be used. Now it
1549 is initialised to -1 but X509_time_adj() now has to check the value
1550 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1551 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1552 [Steve Henson, reported by Kenneth R. Robinette
1553 <support@securenetterm.com>]
1555 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1556 result in a zero length in the ASN1_INTEGER structure which was
1557 not consistent with the structure when d2i_ASN1_INTEGER() was used
1558 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1559 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1560 where it did not print out a minus for negative ASN1_INTEGER.
1563 *) Fix 'openssl passwd -1'.
1566 +) Add summary printout to ocsp utility. The various functions which
1567 convert status values to strings have been renamed to:
1568 OCSP_response_status_str(), OCSP_cert_status_str() and
1569 OCSP_crl_reason_str() and are no longer static. New options
1570 to verify nonce values and to disable verification. OCSP response
1571 printout format cleaned up.
1574 +) Add additional OCSP certificate checks. These are those specified
1575 in RFC2560. This consists of two separate checks: the CA of the
1576 certificate being checked must either be the OCSP signer certificate
1577 or the issuer of the OCSP signer certificate. In the latter case the
1578 OCSP signer certificate must contain the OCSP signing extended key
1579 usage. This check is performed by attempting to match the OCSP
1580 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1581 in the OCSP_CERTID structures of the response.
1584 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1585 and related routines. This uses the standard OpenSSL certificate
1586 verify routines to perform initial checks (just CA validity) and
1587 to obtain the certificate chain. Then additional checks will be
1588 performed on the chain. Currently the root CA is checked to see
1589 if it is explicitly trusted for OCSP signing. This is used to set
1590 a root CA as a global signing root: that is any certificate that
1591 chains to that CA is an acceptable OCSP signing certificate.
1594 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1595 extensions from a separate configuration file.
1596 As when reading extensions from the main configuration file,
1597 the '-extensions ...' option may be used for specifying the
1599 [Massimiliano Pala <madwolf@comune.modena.it>]
1601 *) Change PKCS12_key_gen_asc() so it can cope with non null
1602 terminated strings whose length is passed in the passlen
1603 parameter, for example from PEM callbacks. This was done
1604 by adding an extra length parameter to asc2uni().
1605 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1607 +) New OCSP utility. Allows OCSP requests to be generated or
1608 read. The request can be sent to a responder and the output
1609 parsed, outputed or printed in text form. Not complete yet:
1610 still needs to check the OCSP response validity.
1613 +) New subcommands for 'openssl ca':
1614 'openssl ca -status <serial>' prints the status of the cert with
1615 the given serial number (according to the index file).
1616 'openssl ca -updatedb' updates the expiry status of certificates
1618 [Massimiliano Pala <madwolf@comune.modena.it>]
1620 +) New '-newreq-nodes' command option to CA.pl. This is like
1621 '-newreq', but calls 'openssl req' with the '-nodes' option
1622 so that the resulting key is not encrypted.
1623 [Damien Miller <djm@mindrot.org>]
1625 +) New configuration for the GNU Hurd.
1626 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1628 +) Initial code to implement OCSP basic response verify. This
1629 is currently incomplete. Currently just finds the signer's
1630 certificate and verifies the signature on the response.
1633 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1634 value of OPENSSLDIR. This is available via the new '-d' option
1635 to 'openssl version', and is also included in 'openssl version -a'.
1638 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1639 call failed, free the DSA structure.
1642 +) Allowing defining memory allocation callbacks that will be given
1643 file name and line number information in additional arguments
1644 (a const char* and an int). The basic functionality remains, as
1645 well as the original possibility to just replace malloc(),
1646 realloc() and free() by functions that do not know about these
1647 additional arguments. To register and find out the current
1648 settings for extended allocation functions, the following
1649 functions are provided:
1651 CRYPTO_set_mem_ex_functions
1652 CRYPTO_set_locked_mem_ex_functions
1653 CRYPTO_get_mem_ex_functions
1654 CRYPTO_get_locked_mem_ex_functions
1656 These work the same way as CRYPTO_set_mem_functions and friends.
1657 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1658 extended allocation function is enabled.
1659 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1660 a conventional allocation function is enabled.
1661 [Richard Levitte, Bodo Moeller]
1663 *) Fix to uni2asc() to cope with zero length Unicode strings.
1664 These are present in some PKCS#12 files.
1667 +) Finish off removing the remaining LHASH function pointer casts.
1668 There should no longer be any prototype-casting required when using
1669 the LHASH abstraction, and any casts that remain are "bugs". See
1670 the callback types and macros at the head of lhash.h for details
1671 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1674 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1675 If /dev/[u]random devices are not available or do not return enough
1676 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1678 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1679 /etc/entropy will be queried once each in this sequence, quering stops
1680 when enough entropy was collected without querying more sockets.
1683 +) Change the Unix RAND_poll() variant to be able to poll several
1684 random devices, as specified by DEVRANDOM, until a sufficient amount
1685 of data has been collected. We spend at most 10 ms on each file
1686 (select timeout) and read in non-blocking mode. DEVRANDOM now
1687 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1688 (previously it was just the string "/dev/urandom"), so on typical
1689 platforms the 10 ms delay will never occur.
1690 Also separate out the Unix variant to its own file, rand_unix.c.
1691 For VMS, there's a currently-empty rand_vms.c.
1694 +) Move OCSP client related routines to ocsp_cl.c. These
1695 provide utility functions which an application needing
1696 to issue a request to an OCSP responder and analyse the
1697 response will typically need: as opposed to those which an
1698 OCSP responder itself would need which will be added later.
1700 OCSP_request_sign() signs an OCSP request with an API similar
1701 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1702 response. OCSP_response_get1_basic() extracts basic response
1703 from response. OCSP_resp_find_status(): finds and extracts status
1704 information from an OCSP_CERTID structure (which will be created
1705 when the request structure is built). These are built from lower
1706 level functions which work on OCSP_SINGLERESP structures but
1707 wont normally be used unless the application wishes to examine
1708 extensions in the OCSP response for example.
1710 Replace nonce routines with a pair of functions.
1711 OCSP_request_add1_nonce() adds a nonce value and optionally
1712 generates a random value. OCSP_check_nonce() checks the
1713 validity of the nonce in an OCSP response.
1716 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1717 This doesn't copy the supplied OCSP_CERTID and avoids the
1718 need to free up the newly created id. Change return type
1719 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1720 This can then be used to add extensions to the request.
1721 Deleted OCSP_request_new(), since most of its functionality
1722 is now in OCSP_REQUEST_new() (and the case insensitive name
1723 clash) apart from the ability to set the request name which
1724 will be added elsewhere.
1727 +) Update OCSP API. Remove obsolete extensions argument from
1728 various functions. Extensions are now handled using the new
1729 OCSP extension code. New simple OCSP HTTP function which
1730 can be used to send requests and parse the response.
1733 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1734 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1735 uses the special reorder version of SET OF to sort the attributes
1736 and reorder them to match the encoded order. This resolves a long
1737 standing problem: a verify on a PKCS7 structure just after signing
1738 it used to fail because the attribute order did not match the
1739 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1740 it uses the received order. This is necessary to tolerate some broken
1741 software that does not order SET OF. This is handled by encoding
1742 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1743 to produce the required SET OF.
1746 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1747 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1748 files to get correct declarations of the ASN.1 item variables.
1751 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1752 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1753 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1754 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1755 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1756 ASN1_ITEM and no wrapper functions.
1759 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1760 replace the old function pointer based I/O routines. Change most of
1761 the *_d2i_bio() and *_d2i_fp() functions to use these.
1764 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1765 lines, recognice more "algorithms" that can be deselected, and make
1766 it complain about algorithm deselection that isn't recognised.
1769 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1770 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1771 to use new functions. Add NO_ASN1_OLD which can be set to remove
1772 some old style ASN1 functions: this can be used to determine if old
1773 code will still work when these eventually go away.
1776 +) New extension functions for OCSP structures, these follow the
1777 same conventions as certificates and CRLs.
1780 +) New function X509V3_add1_i2d(). This automatically encodes and
1781 adds an extension. Its behaviour can be customised with various
1782 flags to append, replace or delete. Various wrappers added for
1783 certifcates and CRLs.
1786 +) Fix to avoid calling the underlying ASN1 print routine when
1787 an extension cannot be parsed. Correct a typo in the
1788 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1791 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1792 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1793 when writing a 32767 byte record.
1794 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1796 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1797 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1799 (RSA objects have a reference count access to which is protected
1800 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1801 so they are meant to be shared between threads.)
1802 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1803 "Reddie, Steven" <Steven.Reddie@ca.com>]
1805 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1806 entries for variables.
1809 *) Fix a deadlock in CRYPTO_mem_leaks().
1812 +) Add functionality to apps/openssl.c for detecting locking
1813 problems: As the program is single-threaded, all we have
1814 to do is register a locking callback using an array for
1815 storing which locks are currently held by the program.
1818 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1819 SSL_get_ex_data_X509_STORE_idx(), which is used in
1820 ssl_verify_cert_chain() and thus can be called at any time
1821 during TLS/SSL handshakes so that thread-safety is essential.
1822 Unfortunately, the ex_data design is not at all suited
1823 for multi-threaded use, so it probably should be abolished.
1826 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1827 [Broadcom, tweaked and integrated by Geoff Thorpe]
1829 +) Move common extension printing code to new function
1830 X509V3_print_extensions(). Reorganise OCSP print routines and
1831 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1834 +) New function X509_signature_print() to remove duplication in some
1838 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1839 set (this was treated exactly the same as SET OF previously). This
1840 is used to reorder the STACK representing the structure to match the
1841 encoding. This will be used to get round a problem where a PKCS7
1842 structure which was signed could not be verified because the STACK
1843 order did not reflect the encoded order.
1846 +) Reimplement the OCSP ASN1 module using the new code.
1849 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1850 for its ASN1 operations. The old style function pointers still exist
1851 for now but they will eventually go away.
1854 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1855 completely replaces the old ASN1 functionality with a table driven
1856 encoder and decoder which interprets an ASN1_ITEM structure describing
1857 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1858 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1859 has also been converted to the new form.
1862 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1863 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1864 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1865 for negative moduli.
1868 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1869 of not touching the result's sign bit.
1872 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1876 +) Changed the LHASH code to use prototypes for callbacks, and created
1877 macros to declare and implement thin (optionally static) functions
1878 that provide type-safety and avoid function pointer casting for the
1879 type-specific callbacks.
1882 *) Use better test patterns in bntest.
1885 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1887 [Veers Staats <staatsvr@asc.hpc.mil>,
1888 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1890 *) rand_win.c fix for Borland C.
1893 *) BN_rshift bugfix for n == 0.
1896 +) Reformat the FAQ so the different questions and answers can be divided
1897 in sections depending on the subject.
1900 +) Have the zlib compression code load ZLIB.DLL dynamically under
1904 +) New function BN_mod_sqrt for computing square roots modulo a prime
1905 (using the probabilistic Tonelli-Shanks algorithm unless
1906 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1907 be handled deterministically).
1908 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1910 *) Add a 'bctest' script that checks for some known 'bc' bugs
1911 so that 'make test' does not abort just because 'bc' is broken.
1914 *) Store verify_result within SSL_SESSION also for client side to
1915 avoid potential security hole. (Re-used sessions on the client side
1916 always resulted in verify_result==X509_V_OK, not using the original
1917 result of the server certificate verification.)
1920 +) Make BN_mod_inverse faster by explicitly handling small quotients
1921 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1922 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1925 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1926 SSL3_RT_APPLICATION_DATA, return 0.
1927 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1931 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1932 releases, have been re-implemented by renaming the previous
1933 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1934 and ssl3_read_internal, respectively, and adding 'peek' parameters
1935 to them. The new ssl[23]_{read,peek} functions are calls to
1936 ssl[23]_read_internal with the 'peek' flag set appropriately.
1937 A 'peek' parameter has also been added to ssl3_read_bytes, which
1938 does the actual work for ssl3_read_internal.
1941 +) New function BN_kronecker.
1944 +) Fix BN_gcd so that it works on negative inputs; the result is
1945 positive unless both parameters are zero.
1946 Previously something reasonably close to an infinite loop was
1947 possible because numbers could be growing instead of shrinking
1948 in the implementation of Euclid's algorithm.
1951 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1952 sign of the number in question.
1954 Fix BN_is_word(a,w) to work correctly for w == 0.
1956 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1957 because its test if the absolute value of 'a' equals 'w'.
1958 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1959 it exists mostly for use in the implementations of BN_is_zero(),
1960 BN_is_one(), and BN_is_word().
1963 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1964 the method-specific "init()" handler. Also clean up ex_data after
1965 calling the method-specific "finish()" handler. Previously, this was
1966 happening the other way round.
1969 +) New function BN_swap.
1972 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1973 the exponentiation functions are more likely to produce reasonable
1974 results on negative inputs.
1977 +) Change BN_mod_mul so that the result is always non-negative.
1978 Previously, it could be negative if one of the factors was negative;
1979 I don't think anyone really wanted that behaviour.
1982 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1983 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1984 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1985 and add new functions:
1994 BN_mod_lshift1_quick
1998 These functions always generate non-negative results.
2000 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2001 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2003 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2004 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2005 be reduced modulo m.
2006 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2008 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2009 was actually never needed) and in BN_mul(). The removal in BN_mul()
2010 required a small change in bn_mul_part_recursive() and the addition
2011 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2012 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2013 bn_sub_words() and bn_add_words() except they take arrays with
2017 +) In 'openssl passwd', verify passwords read from the terminal
2018 unless the '-salt' option is used (which usually means that
2019 verification would just waste user's time since the resulting
2020 hash is going to be compared with some given password hash)
2021 or the new '-noverify' option is used.
2023 This is an incompatible change, but it does not affect
2024 non-interactive use of 'openssl passwd' (passwords on the command
2025 line, '-stdin' option, '-in ...' option) and thus should not
2029 +) Remove all references to RSAref, since there's no more need for it.
2032 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
2033 The previous value, 12, was not always sufficient for BN_mod_exp().
2036 +) Make DSO load along a path given through an environment variable
2037 (SHLIB_PATH) with shl_load().
2040 +) Constify the ENGINE code as a result of BIGNUM constification.
2041 Also constify the RSA code and most things related to it. In a
2042 few places, most notable in the depth of the ASN.1 code, ugly
2043 casts back to non-const were required (to be solved at a later
2047 +) Make it so the openssl application has all engines loaded by default.
2050 +) Constify the BIGNUM routines a little more.
2053 *) Make sure that shared libraries get the internal name engine with
2054 the full version number and not just 0. This should mark the
2055 shared libraries as not backward compatible. Of course, this should
2056 be changed again when we can guarantee backward binary compatibility.
2059 +) Add the following functions:
2061 ENGINE_load_cswift()
2063 ENGINE_load_atalla()
2065 ENGINE_load_builtin_engines()
2067 That way, an application can itself choose if external engines that
2068 are built-in in OpenSSL shall ever be used or not. The benefit is
2069 that applications won't have to be linked with libdl or other dso
2070 libraries unless it's really needed.
2072 Changed 'openssl engine' to load all engines on demand.
2073 Changed the engine header files to avoid the duplication of some
2074 declarations (they differed!).
2077 +) 'openssl engine' can now list capabilities.
2080 +) Better error reporting in 'openssl engine'.
2083 +) Never call load_dh_param(NULL) in s_server.
2086 +) Add engine application. It can currently list engines by name and
2087 identity, and test if they are actually available.
2090 +) Improve RPM specification file by forcing symbolic linking and making
2091 sure the installed documentation is also owned by root.root.
2092 [Damien Miller <djm@mindrot.org>]
2094 +) Give the OpenSSL applications more possibilities to make use of
2095 keys (public as well as private) handled by engines.
2098 +) Add OCSP code that comes from CertCo.
2101 +) Add VMS support for the Rijndael code.
2104 +) Added untested support for Nuron crypto accelerator.
2107 +) Add support for external cryptographic devices. This code was
2108 previously distributed separately as the "engine" branch.
2109 [Geoff Thorpe, Richard Levitte]
2111 +) Rework the filename-translation in the DSO code. It is now possible to
2112 have far greater control over how a "name" is turned into a filename
2113 depending on the operating environment and any oddities about the
2114 different shared library filenames on each system.
2117 +) Support threads on FreeBSD-elf in Configure.
2120 +) Fix for SHA1 assembly problem with MASM: it produces
2121 warnings about corrupt line number information when assembling
2122 with debugging information. This is caused by the overlapping
2124 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2127 NCONF_get_number() has no error checking at all. As a replacement,
2128 NCONF_get_number_e() is defined (_e for "error checking") and is
2129 promoted strongly. The old NCONF_get_number is kept around for
2130 binary backward compatibility.
2131 Make it possible for methods to load from something other than a BIO,
2132 by providing a function pointer that is given a name instead of a BIO.
2133 For example, this could be used to load configuration data from an
2137 *) Fix typo in get_cert_by_subject() in by_dir.c
2138 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
2140 *) Rework the system to generate shared libraries:
2142 - Make note of the expected extension for the shared libraries and
2143 if there is a need for symbolic links from for example libcrypto.so.0
2144 to libcrypto.so.0.9.7. There is extended info in Configure for
2147 - Make as few rebuilds of the shared libraries as possible.
2149 - Still avoid linking the OpenSSL programs with the shared libraries.
2151 - When installing, install the shared libraries separately from the
2155 +) Fix for non blocking accept BIOs. Added new I/O special reason
2156 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2157 with non blocking I/O was not possible because no retry code was
2158 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2162 +) Added the beginnings of Rijndael support.
2165 +) Fix for bug in DirectoryString mask setting. Add support for
2166 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2167 to allow certificate printing to more controllable, additional
2168 'certopt' option to 'x509' to allow new printing options to be
2172 +) Clean old EAY MD5 hack from e_os.h.
2175 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
2177 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
2178 and not in SSL_clear because the latter is also used by the
2179 accept/connect functions; previously, the settings made by
2180 SSL_set_read_ahead would be lost during the handshake.
2181 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
2183 *) Correct util/mkdef.pl to be selective about disabled algorithms.
2184 Previously, it would create entries for disableed algorithms no
2188 *) Added several new manual pages for SSL_* function.
2191 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
2193 *) In ssl23_get_client_hello, generate an error message when faced
2194 with an initial SSL 3.0/TLS record that is too small to contain the
2195 first two bytes of the ClientHello message, i.e. client_version.
2196 (Note that this is a pathologic case that probably has never happened
2197 in real life.) The previous approach was to use the version number
2198 from the record header as a substitute; but our protocol choice
2199 should not depend on that one because it is not authenticated
2200 by the Finished messages.
2203 *) More robust randomness gathering functions for Windows.
2204 [Jeffrey Altman <jaltman@columbia.edu>]
2206 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
2207 not set then we don't setup the error code for issuer check errors
2208 to avoid possibly overwriting other errors which the callback does
2209 handle. If an application does set the flag then we assume it knows
2210 what it is doing and can handle the new informational codes
2214 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
2215 a general "ANY" type, as such it should be able to decode anything
2216 including tagged types. However it didn't check the class so it would
2217 wrongly interpret tagged types in the same way as their universal
2218 counterpart and unknown types were just rejected. Changed so that the
2219 tagged and unknown types are handled in the same way as a SEQUENCE:
2220 that is the encoding is stored intact. There is also a new type
2221 "V_ASN1_OTHER" which is used when the class is not universal, in this
2222 case we have no idea what the actual type is so we just lump them all
2226 *) On VMS, stdout may very well lead to a file that is written to
2227 in a record-oriented fashion. That means that every write() will
2228 write a separate record, which will be read separately by the
2229 programs trying to read from it. This can be very confusing.
2231 The solution is to put a BIO filter in the way that will buffer
2232 text until a linefeed is reached, and then write everything a
2233 line at a time, so every record written will be an actual line,
2234 not chunks of lines and not (usually doesn't happen, but I've
2235 seen it once) several lines in one record. BIO_f_linebuffer() is
2238 Currently, it's a VMS-only method, because that's where it has
2239 been tested well enough.
2242 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
2243 it can return incorrect results.
2244 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
2245 but it was in 0.9.6-beta[12].)
2248 *) Disable the check for content being present when verifying detached
2249 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
2250 include zero length content when signing messages.
2253 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
2254 BIO_ctrl (for BIO pairs).
2257 *) Add DSO method for VMS.
2260 *) Bug fix: Montgomery multiplication could produce results with the
2264 *) Add RPM specification openssl.spec and modify it to build three
2265 packages. The default package contains applications, application
2266 documentation and run-time libraries. The devel package contains
2267 include files, static libraries and function documentation. The
2268 doc package contains the contents of the doc directory. The original
2269 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
2272 *) Add a large number of documentation files for many SSL routines.
2273 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2275 *) Add a configuration entry for Sony News 4.
2276 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
2278 *) Don't set the two most significant bits to one when generating a
2279 random number < q in the DSA library.
2282 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
2283 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
2284 the underlying transport is blocking) if a handshake took place.
2285 (The default behaviour is needed by applications such as s_client
2286 and s_server that use select() to determine when to use SSL_read;
2287 but for applications that know in advance when to expect data, it
2288 just makes things more complicated.)
2291 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
2295 *) Add a few more EBCDIC conditionals that make `req' and `x509'
2296 work better on such systems.
2297 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
2299 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
2300 Update PKCS12_parse() so it copies the friendlyName and the
2301 keyid to the certificates aux info.
2304 *) Fix bug in PKCS7_verify() which caused an infinite loop
2305 if there was more than one signature.
2306 [Sven Uszpelkat <su@celocom.de>]
2308 *) Major change in util/mkdef.pl to include extra information
2309 about each symbol, as well as presentig variables as well
2310 as functions. This change means that there's n more need
2311 to rebuild the .num files when some algorithms are excluded.
2314 *) Allow the verify time to be set by an application,
2315 rather than always using the current time.
2318 *) Phase 2 verify code reorganisation. The certificate
2319 verify code now looks up an issuer certificate by a
2320 number of criteria: subject name, authority key id
2321 and key usage. It also verifies self signed certificates
2322 by the same criteria. The main comparison function is
2323 X509_check_issued() which performs these checks.
2325 Lot of changes were necessary in order to support this
2326 without completely rewriting the lookup code.
2328 Authority and subject key identifier are now cached.
2330 The LHASH 'certs' is X509_STORE has now been replaced
2331 by a STACK_OF(X509_OBJECT). This is mainly because an
2332 LHASH can't store or retrieve multiple objects with
2333 the same hash value.
2335 As a result various functions (which were all internal
2336 use only) have changed to handle the new X509_STORE
2337 structure. This will break anything that messed round
2338 with X509_STORE internally.
2340 The functions X509_STORE_add_cert() now checks for an
2341 exact match, rather than just subject name.
2343 The X509_STORE API doesn't directly support the retrieval
2344 of multiple certificates matching a given criteria, however
2345 this can be worked round by performing a lookup first
2346 (which will fill the cache with candidate certificates)
2347 and then examining the cache for matches. This is probably
2348 the best we can do without throwing out X509_LOOKUP
2349 entirely (maybe later...).
2351 The X509_VERIFY_CTX structure has been enhanced considerably.
2353 All certificate lookup operations now go via a get_issuer()
2354 callback. Although this currently uses an X509_STORE it
2355 can be replaced by custom lookups. This is a simple way
2356 to bypass the X509_STORE hackery necessary to make this
2357 work and makes it possible to use more efficient techniques
2358 in future. A very simple version which uses a simple
2359 STACK for its trusted certificate store is also provided
2360 using X509_STORE_CTX_trusted_stack().
2362 The verify_cb() and verify() callbacks now have equivalents
2363 in the X509_STORE_CTX structure.
2365 X509_STORE_CTX also has a 'flags' field which can be used
2366 to customise the verify behaviour.
2369 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
2370 excludes S/MIME capabilities.
2373 *) When a certificate request is read in keep a copy of the
2374 original encoding of the signed data and use it when outputing
2375 again. Signatures then use the original encoding rather than
2376 a decoded, encoded version which may cause problems if the
2377 request is improperly encoded.
2380 *) For consistency with other BIO_puts implementations, call
2381 buffer_write(b, ...) directly in buffer_puts instead of calling
2384 In BIO_puts, increment b->num_write as in BIO_write.
2385 [Peter.Sylvester@EdelWeb.fr]
2387 *) Fix BN_mul_word for the case where the word is 0. (We have to use
2388 BN_zero, we may not return a BIGNUM with an array consisting of
2392 *) Avoid calling abort() from within the library when problems are
2393 detected, except if preprocessor symbols have been defined
2394 (such as REF_CHECK, BN_DEBUG etc.).
2397 *) New openssl application 'rsautl'. This utility can be
2398 used for low level RSA operations. DER public key
2399 BIO/fp routines also added.
2402 *) New Configure entry and patches for compiling on QNX 4.
2403 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
2405 *) A demo state-machine implementation was sponsored by
2406 Nuron (http://www.nuron.com/) and is now available in
2407 demos/state_machine.
2410 *) New options added to the 'dgst' utility for signature
2411 generation and verification.
2414 *) Unrecognized PKCS#7 content types are now handled via a
2415 catch all ASN1_TYPE structure. This allows unsupported
2416 types to be stored as a "blob" and an application can
2417 encode and decode it manually.
2420 *) Fix various signed/unsigned issues to make a_strex.c
2422 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
2424 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
2425 length if passed a buffer. ASN1_INTEGER_to_BN failed
2426 if passed a NULL BN and its argument was negative.
2427 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
2429 *) Modification to PKCS#7 encoding routines to output definite
2430 length encoding. Since currently the whole structures are in
2431 memory there's not real point in using indefinite length
2432 constructed encoding. However if OpenSSL is compiled with
2433 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
2436 *) Added BIO_vprintf() and BIO_vsnprintf().
2439 *) Added more prefixes to parse for in the the strings written
2440 through a logging bio, to cover all the levels that are available
2441 through syslog. The prefixes are now:
2443 PANIC, EMERG, EMR => LOG_EMERG
2444 ALERT, ALR => LOG_ALERT
2445 CRIT, CRI => LOG_CRIT
2446 ERROR, ERR => LOG_ERR
2447 WARNING, WARN, WAR => LOG_WARNING
2448 NOTICE, NOTE, NOT => LOG_NOTICE
2449 INFO, INF => LOG_INFO
2450 DEBUG, DBG => LOG_DEBUG
2452 and as before, if none of those prefixes are present at the
2453 beginning of the string, LOG_ERR is chosen.
2455 On Win32, the LOG_* levels are mapped according to this:
2457 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
2458 LOG_WARNING => EVENTLOG_WARNING_TYPE
2459 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
2463 *) Made it possible to reconfigure with just the configuration
2464 argument "reconf" or "reconfigure". The command line arguments
2465 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
2466 and are retrieved from there when reconfiguring.
2470 [Assar Westerlund <assar@sics.se>, Richard Levitte]
2472 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
2475 *) The obj_dat.pl script was messing up the sorting of object
2476 names. The reason was that it compared the quoted version
2477 of strings as a result "OCSP" > "OCSP Signing" because
2478 " > SPACE. Changed script to store unquoted versions of
2479 names and add quotes on output. It was also omitting some
2480 names from the lookup table if they were given a default
2481 value (that is if SN is missing it is given the same
2482 value as LN and vice versa), these are now added on the
2483 grounds that if an object has a name we should be able to
2484 look it up. Finally added warning output when duplicate
2485 short or long names are found.
2488 *) Changes needed for Tandem NSK.
2489 [Scott Uroff <scott@xypro.com>]
2491 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
2492 RSA_padding_check_SSLv23(), special padding was never detected
2493 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
2494 version rollback attacks was not effective.
2496 In s23_clnt.c, don't use special rollback-attack detection padding
2497 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
2498 client; similarly, in s23_srvr.c, don't do the rollback check if
2499 SSL 2.0 is the only protocol enabled in the server.
2502 *) Make it possible to get hexdumps of unprintable data with 'openssl
2503 asn1parse'. By implication, the functions ASN1_parse_dump() and
2504 BIO_dump_indent() are added.
2507 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
2508 these print out strings and name structures based on various
2509 flags including RFC2253 support and proper handling of
2510 multibyte characters. Added options to the 'x509' utility
2511 to allow the various flags to be set.
2514 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
2515 Also change the functions X509_cmp_current_time() and
2516 X509_gmtime_adj() work with an ASN1_TIME structure,
2517 this will enable certificates using GeneralizedTime in validity
2518 dates to be checked.
2521 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
2522 negative public key encodings) on by default,
2523 NO_NEG_PUBKEY_BUG can be set to disable it.
2526 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
2527 content octets. An i2c_ASN1_OBJECT is unnecessary because
2528 the encoding can be trivially obtained from the structure.
2531 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
2532 not read locks (CRYPTO_r_[un]lock).
2535 *) A first attempt at creating official support for shared
2536 libraries through configuration. I've kept it so the
2537 default is static libraries only, and the OpenSSL programs
2538 are always statically linked for now, but there are
2539 preparations for dynamic linking in place.
2540 This has been tested on Linux and Tru64.
2543 *) Randomness polling function for Win9x, as described in:
2544 Peter Gutmann, Software Generation of Practically Strong
2548 *) Fix so PRNG is seeded in req if using an already existing
2552 *) New options to smime application. -inform and -outform
2553 allow alternative formats for the S/MIME message including
2554 PEM and DER. The -content option allows the content to be
2555 specified separately. This should allow things like Netscape
2556 form signing output easier to verify.
2559 *) Fix the ASN1 encoding of tags using the 'long form'.
2562 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
2563 STRING types. These convert content octets to and from the
2564 underlying type. The actual tag and length octets are
2565 already assumed to have been read in and checked. These
2566 are needed because all other string types have virtually
2567 identical handling apart from the tag. By having versions
2568 of the ASN1 functions that just operate on content octets
2569 IMPLICIT tagging can be handled properly. It also allows
2570 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
2571 and ASN1_INTEGER are identical apart from the tag.
2574 *) Change the handling of OID objects as follows:
2576 - New object identifiers are inserted in objects.txt, following
2577 the syntax given in objects.README.
2578 - objects.pl is used to process obj_mac.num and create a new
2580 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2583 This is currently kind of a hack, and the perl code in objects.pl
2584 isn't very elegant, but it works as I intended. The simplest way
2585 to check that it worked correctly is to look in obj_dat.h and
2586 check the array nid_objs and make sure the objects haven't moved
2587 around (this is important!). Additions are OK, as well as
2588 consistent name changes.
2591 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2594 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2595 The given file adds to whatever has already been seeded into the
2596 random pool through the RANDFILE configuration file option or
2597 environment variable, or the default random state file.
2600 *) mkstack.pl now sorts each macro group into lexical order.
2601 Previously the output order depended on the order the files
2602 appeared in the directory, resulting in needless rewriting
2606 *) Patches to make OpenSSL compile under Win32 again. Mostly
2607 work arounds for the VC++ problem that it treats func() as
2608 func(void). Also stripped out the parts of mkdef.pl that
2609 added extra typesafe functions: these no longer exist.
2612 *) Reorganisation of the stack code. The macros are now all
2613 collected in safestack.h . Each macro is defined in terms of
2614 a "stack macro" of the form SKM_<name>(type, a, b). The
2615 DEBUG_SAFESTACK is now handled in terms of function casts,
2616 this has the advantage of retaining type safety without the
2617 use of additional functions. If DEBUG_SAFESTACK is not defined
2618 then the non typesafe macros are used instead. Also modified the
2619 mkstack.pl script to handle the new form. Needs testing to see
2620 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2621 the default if no major problems. Similar behaviour for ASN1_SET_OF
2622 and PKCS12_STACK_OF.
2625 *) When some versions of IIS use the 'NET' form of private key the
2626 key derivation algorithm is different. Normally MD5(password) is
2627 used as a 128 bit RC4 key. In the modified case
2628 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2629 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2630 as the old Netscape_RSA functions except they have an additional
2631 'sgckey' parameter which uses the modified algorithm. Also added
2632 an -sgckey command line option to the rsa utility. Thanks to
2633 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2634 algorithm to openssl-dev.
2637 *) The evp_local.h macros were using 'c.##kname' which resulted in
2638 invalid expansion on some systems (SCO 5.0.5 for example).
2639 Corrected to 'c.kname'.
2640 [Phillip Porch <root@theporch.com>]
2642 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2643 a STACK of email addresses from a certificate or request, these look
2644 in the subject name and the subject alternative name extensions and
2645 omit any duplicate addresses.
2648 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2649 This makes DSA verification about 2 % faster.
2652 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2653 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2654 plus overhead for 1024 bit moduli).
2655 This makes exponentiations about 0.5 % faster for 1024 bit
2656 exponents (as measured by "openssl speed rsa2048").
2659 *) Rename memory handling macros to avoid conflicts with other
2661 Malloc => OPENSSL_malloc
2662 Malloc_locked => OPENSSL_malloc_locked
2663 Realloc => OPENSSL_realloc
2664 Free => OPENSSL_free
2667 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2668 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2671 *) CygWin32 support.
2672 [John Jarvie <jjarvie@newsguy.com>]
2674 *) The type-safe stack code has been rejigged. It is now only compiled
2675 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2676 by default all type-specific stack functions are "#define"d back to
2677 standard stack functions. This results in more streamlined output
2678 but retains the type-safety checking possibilities of the original
2682 *) The STACK code has been cleaned up, and certain type declarations
2683 that didn't make a lot of sense have been brought in line. This has
2684 also involved a cleanup of sorts in safestack.h to more correctly
2685 map type-safe stack functions onto their plain stack counterparts.
2686 This work has also resulted in a variety of "const"ifications of
2687 lots of the code, especially "_cmp" operations which should normally
2688 be prototyped with "const" parameters anyway.
2691 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2692 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2693 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2694 where all of 'md' is used each time the PRNG is used, but 'state'
2695 is used only indexed by a cyclic counter. As entropy may not be
2696 well distributed from the beginning, 'md' is important as a
2697 chaining variable. However, the output function chains only half
2698 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2699 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2700 in all of 'state' being rewritten, with the new values depending
2701 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2704 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2705 the handshake is continued after ssl_verify_cert_chain();
2706 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2707 can lead to 'unexplainable' connection aborts later.
2708 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2710 *) Major EVP API cipher revision.
2711 Add hooks for extra EVP features. This allows various cipher
2712 parameters to be set in the EVP interface. Support added for variable
2713 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2714 setting of RC2 and RC5 parameters.
2716 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2719 Remove lots of duplicated code from the EVP library. For example *every*
2720 cipher init() function handles the 'iv' in the same way according to the
2721 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2722 for CFB and OFB modes they zero ctx->num.
2724 New functionality allows removal of S/MIME code RC2 hack.
2726 Most of the routines have the same form and so can be declared in terms
2729 By shifting this to the top level EVP_CipherInit() it can be removed from
2730 all individual ciphers. If the cipher wants to handle IVs or keys
2731 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2734 Change lots of functions like EVP_EncryptUpdate() to now return a
2735 value: although software versions of the algorithms cannot fail
2736 any installed hardware versions can.
2739 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2740 this option is set, tolerate broken clients that send the negotiated
2741 protocol version number instead of the requested protocol version
2745 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2746 i.e. non-zero for export ciphersuites, zero otherwise.
2747 Previous versions had this flag inverted, inconsistent with
2748 rsa_tmp_cb (..._TMP_RSA_CB).
2749 [Bodo Moeller; problem reported by Amit Chopra]
2751 *) Add missing DSA library text string. Work around for some IIS
2752 key files with invalid SEQUENCE encoding.
2755 *) Add a document (doc/standards.txt) that list all kinds of standards
2756 and so on that are implemented in OpenSSL.
2759 *) Enhance c_rehash script. Old version would mishandle certificates
2760 with the same subject name hash and wouldn't handle CRLs at all.
2761 Added -fingerprint option to crl utility, to support new c_rehash
2765 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2768 *) Fix for SSL server purpose checking. Server checking was
2769 rejecting certificates which had extended key usage present
2770 but no ssl client purpose.
2771 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2773 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2774 is a little unclear about how a blank password is handled.
2775 Since the password in encoded as a BMPString with terminating
2776 double NULL a zero length password would end up as just the
2777 double NULL. However no password at all is different and is
2778 handled differently in the PKCS#12 key generation code. NS
2779 treats a blank password as zero length. MSIE treats it as no
2780 password on export: but it will try both on import. We now do
2781 the same: PKCS12_parse() tries zero length and no password if
2782 the password is set to "" or NULL (NULL is now a valid password:
2783 it wasn't before) as does the pkcs12 application.
2786 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2787 perror when PEM_read_bio_X509_REQ fails, the error message must
2788 be obtained from the error queue.
2791 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2792 it in ERR_remove_state if appropriate, and change ERR_get_state
2793 accordingly to avoid race conditions (this is necessary because
2794 thread_hash is no longer constant once set).
2797 *) Bugfix for linux-elf makefile.one.
2800 *) RSA_get_default_method() will now cause a default
2801 RSA_METHOD to be chosen if one doesn't exist already.
2802 Previously this was only set during a call to RSA_new()
2803 or RSA_new_method(NULL) meaning it was possible for
2804 RSA_get_default_method() to return NULL.
2807 *) Added native name translation to the existing DSO code
2808 that will convert (if the flag to do so is set) filenames
2809 that are sufficiently small and have no path information
2810 into a canonical native form. Eg. "blah" converted to
2811 "libblah.so" or "blah.dll" etc.
2814 *) New function ERR_error_string_n(e, buf, len) which is like
2815 ERR_error_string(e, buf), but writes at most 'len' bytes
2816 including the 0 terminator. For ERR_error_string_n, 'buf'
2818 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2820 *) CONF library reworked to become more general. A new CONF
2821 configuration file reader "class" is implemented as well as a
2822 new functions (NCONF_*, for "New CONF") to handle it. The now
2823 old CONF_* functions are still there, but are reimplemented to
2824 work in terms of the new functions. Also, a set of functions
2825 to handle the internal storage of the configuration data is
2826 provided to make it easier to write new configuration file
2827 reader "classes" (I can definitely see something reading a
2828 configuration file in XML format, for example), called _CONF_*,
2829 or "the configuration storage API"...
2831 The new configuration file reading functions are:
2833 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2834 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2836 NCONF_default, NCONF_WIN32
2838 NCONF_dump_fp, NCONF_dump_bio
2840 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2841 NCONF_new creates a new CONF object. This works in the same way
2842 as other interfaces in OpenSSL, like the BIO interface.
2843 NCONF_dump_* dump the internal storage of the configuration file,
2844 which is useful for debugging. All other functions take the same
2845 arguments as the old CONF_* functions wth the exception of the
2846 first that must be a `CONF *' instead of a `LHASH *'.
2848 To make it easer to use the new classes with the old CONF_* functions,
2849 the function CONF_set_default_method is provided.
2852 *) Add '-tls1' option to 'openssl ciphers', which was already
2853 mentioned in the documentation but had not been implemented.
2854 (This option is not yet really useful because even the additional
2855 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2858 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2859 OpenSSL-based applications) load shared libraries and bind to
2860 them in a portable way.
2861 [Geoff Thorpe, with contributions from Richard Levitte]
2863 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2865 *) Make sure _lrotl and _lrotr are only used with MSVC.
2867 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2868 (the default implementation of RAND_status).
2870 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2871 to '-clrext' (= clear extensions), as intended and documented.
2872 [Bodo Moeller; inconsistency pointed out by Michael Attili
2875 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2876 was larger than the MD block size.
2877 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2879 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2880 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2881 using the passed key: if the passed key was a private key the result
2882 of X509_print(), for example, would be to print out all the private key
2886 *) des_quad_cksum() byte order bug fix.
2887 [Ulf Möller, using the problem description in krb4-0.9.7, where
2888 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2890 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2892 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2894 *) For easily testing in shell scripts whether some command
2895 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2896 returns with exit code 0 iff no command of the given name is available.
2897 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2898 the output goes to stdout and nothing is printed to stderr.
2899 Additional arguments are always ignored.
2901 Since for each cipher there is a command of the same name,
2902 the 'no-cipher' compilation switches can be tested this way.
2904 ('openssl no-XXX' is not able to detect pseudo-commands such
2905 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2908 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2911 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2912 is set; it will be thrown away anyway because each handshake creates
2914 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2915 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2916 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2917 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2920 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2921 'Q' and 'R' lose their special meanings (quit/renegotiate).
2922 This is part of what -quiet does; unlike -quiet, -ign_eof
2923 does not suppress any output.
2926 *) Add compatibility options to the purpose and trust code. The
2927 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2928 accepts a certificate or CA, this was the previous behaviour,
2929 with all the associated security issues.
2931 X509_TRUST_COMPAT is the old trust behaviour: only and
2932 automatically trust self signed roots in certificate store. A
2933 new trust setting X509_TRUST_DEFAULT is used to specify that
2934 a purpose has no associated trust setting and it should instead
2935 use the value in the default purpose.
2938 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2939 and fix a memory leak.
2942 *) In util/mkerr.pl (which implements 'make errors'), preserve
2943 reason strings from the previous version of the .c file, as
2944 the default to have only downcase letters (and digits) in
2945 automatically generated reasons codes is not always appropriate.
2948 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2949 using strerror. Previously, ERR_reason_error_string() returned
2950 library names as reason strings for SYSerr; but SYSerr is a special
2951 case where small numbers are errno values, not library numbers.
2954 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2955 converts DSA parameters into DH parameters. (When creating parameters,
2956 DSA_generate_parameters is used.)
2959 *) Include 'length' (recommended exponent length) in C code generated
2960 by 'openssl dhparam -C'.
2963 *) The second argument to set_label in perlasm was already being used
2964 so couldn't be used as a "file scope" flag. Moved to third argument
2968 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2969 instead of RAND_bytes for encryption IVs and salts.
2972 *) Include RAND_status() into RAND_METHOD instead of implementing
2973 it only for md_rand.c Otherwise replacing the PRNG by calling
2974 RAND_set_rand_method would be impossible.
2977 *) Don't let DSA_generate_key() enter an infinite loop if the random
2978 number generation fails.
2981 *) New 'rand' application for creating pseudo-random output.
2984 *) Added configuration support for Linux/IA64
2985 [Rolf Haberrecker <rolf@suse.de>]
2987 *) Assembler module support for Mingw32.
2990 *) Shared library support for HPUX (in shlib/).
2991 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2993 *) Shared library support for Solaris gcc.
2994 [Lutz Behnke <behnke@trustcenter.de>]
2996 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2998 *) PKCS7_encrypt() was adding text MIME headers twice because they
2999 were added manually and by SMIME_crlf_copy().
3002 *) In bntest.c don't call BN_rand with zero bits argument.
3003 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
3005 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
3006 case was implemented. This caused BN_div_recp() to fail occasionally.
3009 *) Add an optional second argument to the set_label() in the perl
3010 assembly language builder. If this argument exists and is set
3011 to 1 it signals that the assembler should use a symbol whose
3012 scope is the entire file, not just the current function. This
3013 is needed with MASM which uses the format label:: for this scope.
3014 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
3016 *) Change the ASN1 types so they are typedefs by default. Before
3017 almost all types were #define'd to ASN1_STRING which was causing
3018 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
3022 *) Change names of new functions to the new get1/get0 naming
3023 convention: After 'get1', the caller owns a reference count
3024 and has to call ..._free; 'get0' returns a pointer to some
3025 data structure without incrementing reference counters.
3026 (Some of the existing 'get' functions increment a reference
3027 counter, some don't.)
3028 Similarly, 'set1' and 'add1' functions increase reference
3029 counters or duplicate objects.
3032 *) Allow for the possibility of temp RSA key generation failure:
3033 the code used to assume it always worked and crashed on failure.
3036 *) Fix potential buffer overrun problem in BIO_printf().
3037 [Ulf Möller, using public domain code by Patrick Powell; problem
3038 pointed out by David Sacerdote <das33@cornell.edu>]
3040 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
3041 RAND_egd() and RAND_status(). In the command line application,
3042 the EGD socket can be specified like a seed file using RANDFILE
3046 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
3047 Some CAs (e.g. Verisign) distribute certificates in this form.
3050 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
3051 list to exclude them. This means that no special compilation option
3052 is needed to use anonymous DH: it just needs to be included in the
3056 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
3057 EVP_MD_type. The old functionality is available in a new macro called
3058 EVP_MD_md(). Change code that uses it and update docs.
3061 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
3062 where the 'void *' argument is replaced by a function pointer argument.
3063 Previously 'void *' was abused to point to functions, which works on
3064 many platforms, but is not correct. As these functions are usually
3065 called by macros defined in OpenSSL header files, most source code
3066 should work without changes.
3069 *) <openssl/opensslconf.h> (which is created by Configure) now contains
3070 sections with information on -D... compiler switches used for
3071 compiling the library so that applications can see them. To enable
3072 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
3073 must be defined. E.g.,
3074 #define OPENSSL_ALGORITHM_DEFINES
3075 #include <openssl/opensslconf.h>
3076 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
3077 [Richard Levitte, Ulf and Bodo Möller]
3079 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
3083 *) Change the 'other' type in certificate aux info to a STACK_OF
3084 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
3085 the required ASN1 format: arbitrary types determined by an OID.
3088 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
3089 argument to 'req'. This is not because the function is newer or
3090 better than others it just uses the work 'NEW' in the certificate
3091 request header lines. Some software needs this.
3094 *) Reorganise password command line arguments: now passwords can be
3095 obtained from various sources. Delete the PEM_cb function and make
3096 it the default behaviour: i.e. if the callback is NULL and the
3097 usrdata argument is not NULL interpret it as a null terminated pass
3098 phrase. If usrdata and the callback are NULL then the pass phrase
3099 is prompted for as usual.
3102 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
3103 the support is automatically enabled. The resulting binaries will
3104 autodetect the card and use it if present.
3105 [Ben Laurie and Compaq Inc.]
3107 *) Work around for Netscape hang bug. This sends certificate request
3108 and server done in one record. Since this is perfectly legal in the
3109 SSL/TLS protocol it isn't a "bug" option and is on by default. See
3110 the bugs/SSLv3 entry for more info.
3113 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
3116 *) Add -rand argument to smime and pkcs12 applications and read/write
3120 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
3123 *) Add command line password options to the remaining applications.
3126 *) Bug fix for BN_div_recp() for numerators with an even number of
3130 *) More tests in bntest.c, and changed test_bn output.
3133 *) ./config recognizes MacOS X now.
3136 *) Bug fix for BN_div() when the first words of num and divsor are
3137 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
3140 *) Add support for various broken PKCS#8 formats, and command line
3141 options to produce them.
3144 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
3145 get temporary BIGNUMs from a BN_CTX.
3148 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
3152 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
3153 include a #define from the old name to the new. The original intent
3154 was that statically linked binaries could for example just call
3155 SSLeay_add_all_ciphers() to just add ciphers to the table and not
3156 link with digests. This never worked becayse SSLeay_add_all_digests()
3157 and SSLeay_add_all_ciphers() were in the same source file so calling
3158 one would link with the other. They are now in separate source files.
3161 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
3164 *) Use a less unusual form of the Miller-Rabin primality test (it used
3165 a binary algorithm for exponentiation integrated into the Miller-Rabin
3166 loop, our standard modexp algorithms are faster).
3169 *) Support for the EBCDIC character set completed.
3170 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
3172 *) Source code cleanups: use const where appropriate, eliminate casts,
3173 use void * instead of char * in lhash.
3176 *) Bugfix: ssl3_send_server_key_exchange was not restartable
3177 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
3178 this the server could overwrite ephemeral keys that the client
3182 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
3183 using 50 iterations of the Rabin-Miller test.
3185 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
3186 iterations of the Rabin-Miller test as required by the appendix
3187 to FIPS PUB 186[-1]) instead of DSA_is_prime.
3188 As BN_is_prime_fasttest includes trial division, DSA parameter
3189 generation becomes much faster.
3191 This implies a change for the callback functions in DSA_is_prime
3192 and DSA_generate_parameters: The callback function is called once
3193 for each positive witness in the Rabin-Miller test, not just
3194 occasionally in the inner loop; and the parameters to the
3195 callback function now provide an iteration count for the outer
3196 loop rather than for the current invocation of the inner loop.
3197 DSA_generate_parameters additionally can call the callback
3198 function with an 'iteration count' of -1, meaning that a
3199 candidate has passed the trial division test (when q is generated
3200 from an application-provided seed, trial division is skipped).
3203 *) New function BN_is_prime_fasttest that optionally does trial
3204 division before starting the Rabin-Miller test and has
3205 an additional BN_CTX * argument (whereas BN_is_prime always
3206 has to allocate at least one BN_CTX).
3207 'callback(1, -1, cb_arg)' is called when a number has passed the
3208 trial division stage.
3211 *) Fix for bug in CRL encoding. The validity dates weren't being handled
3215 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
3218 *) New function BN_pseudo_rand().
3221 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
3222 bignum version of BN_from_montgomery() with the working code from
3223 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
3227 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
3228 made it impossible to use the same SSL_SESSION data structure in
3229 SSL2 clients in multiple threads.
3232 *) The return value of RAND_load_file() no longer counts bytes obtained
3233 by stat(). RAND_load_file(..., -1) is new and uses the complete file
3234 to seed the PRNG (previously an explicit byte count was required).
3235 [Ulf Möller, Bodo Möller]
3237 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
3238 used (char *) instead of (void *) and had casts all over the place.
3241 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
3244 *) Retain source code compatibility for BN_prime_checks macro:
3245 BN_is_prime(..., BN_prime_checks, ...) now uses
3246 BN_prime_checks_for_size to determine the appropriate number of
3247 Rabin-Miller iterations.
3250 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
3251 DH_CHECK_P_NOT_SAFE_PRIME.
3252 (Check if this is true? OpenPGP calls them "strong".)
3255 *) Merge the functionality of "dh" and "gendh" programs into a new program
3256 "dhparam". The old programs are retained for now but will handle DH keys
3257 (instead of parameters) in future.
3260 *) Make the ciphers, s_server and s_client programs check the return values
3261 when a new cipher list is set.
3264 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
3265 ciphers. Before when the 56bit ciphers were enabled the sorting was
3268 The syntax for the cipher sorting has been extended to support sorting by
3269 cipher-strength (using the strength_bits hard coded in the tables).
3270 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
3272 Fix a bug in the cipher-command parser: when supplying a cipher command
3273 string with an "undefined" symbol (neither command nor alphanumeric
3274 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
3275 an error is flagged.
3277 Due to the strength-sorting extension, the code of the
3278 ssl_create_cipher_list() function was completely rearranged. I hope that
3279 the readability was also increased :-)
3280 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3282 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
3283 for the first serial number and places 2 in the serial number file. This
3284 avoids problems when the root CA is created with serial number zero and
3285 the first user certificate has the same issuer name and serial number
3289 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
3290 the new code. Add documentation for this stuff.
3293 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
3294 X509_*() to X509at_*() on the grounds that they don't handle X509
3295 structures and behave in an analagous way to the X509v3 functions:
3296 they shouldn't be called directly but wrapper functions should be used
3299 So we also now have some wrapper functions that call the X509at functions
3300 when passed certificate requests. (TO DO: similar things can be done with
3301 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
3302 things. Some of these need some d2i or i2d and print functionality
3303 because they handle more complex structures.)
3306 *) Add missing #ifndefs that caused missing symbols when building libssl
3307 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
3308 NO_RSA in ssl/s2*.c.
3309 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
3311 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
3312 has a return value which indicates the quality of the random data
3313 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
3314 error queue. New function RAND_pseudo_bytes() generates output that is
3315 guaranteed to be unique but not unpredictable. RAND_add is like
3316 RAND_seed, but takes an extra argument for an entropy estimate
3317 (RAND_seed always assumes full entropy).
3320 *) Do more iterations of Rabin-Miller probable prime test (specifically,
3321 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
3322 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
3323 in crypto/bn/bn_prime.c for the complete table). This guarantees a
3324 false-positive rate of at most 2^-80 for random input.
3327 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
3330 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
3331 in the 0.9.5 release), this returns the chain
3332 from an X509_CTX structure with a dup of the stack and all
3333 the X509 reference counts upped: so the stack will exist
3334 after X509_CTX_cleanup() has been called. Modify pkcs12.c
3337 Also make SSL_SESSION_print() print out the verify return
3341 *) Add manpage for the pkcs12 command. Also change the default
3342 behaviour so MAC iteration counts are used unless the new
3343 -nomaciter option is used. This improves file security and
3344 only older versions of MSIE (4.0 for example) need it.
3347 *) Honor the no-xxx Configure options when creating .DEF files.
3350 *) Add PKCS#10 attributes to field table: challengePassword,
3351 unstructuredName and unstructuredAddress. These are taken from
3352 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
3353 international characters are used.
3355 More changes to X509_ATTRIBUTE code: allow the setting of types
3356 based on strings. Remove the 'loc' parameter when adding
3357 attributes because these will be a SET OF encoding which is sorted
3361 *) Initial changes to the 'req' utility to allow request generation
3362 automation. This will allow an application to just generate a template
3363 file containing all the field values and have req construct the
3366 Initial support for X509_ATTRIBUTE handling. Stacks of these are
3367 used all over the place including certificate requests and PKCS#7
3368 structures. They are currently handled manually where necessary with
3369 some primitive wrappers for PKCS#7. The new functions behave in a
3370 manner analogous to the X509 extension functions: they allow
3371 attributes to be looked up by NID and added.
3373 Later something similar to the X509V3 code would be desirable to
3374 automatically handle the encoding, decoding and printing of the
3375 more complex types. The string types like challengePassword can
3376 be handled by the string table functions.
3378 Also modified the multi byte string table handling. Now there is
3379 a 'global mask' which masks out certain types. The table itself
3380 can use the flag STABLE_NO_MASK to ignore the mask setting: this
3381 is useful when for example there is only one permissible type
3382 (as in countryName) and using the mask might result in no valid
3386 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
3387 SSL_get_peer_finished to allow applications to obtain the latest
3388 Finished messages sent to the peer or expected from the peer,
3389 respectively. (SSL_get_peer_finished is usually the Finished message
3390 actually received from the peer, otherwise the protocol will be aborted.)
3392 As the Finished message are message digests of the complete handshake
3393 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
3394 be used for external authentication procedures when the authentication
3395 provided by SSL/TLS is not desired or is not enough.
3398 *) Enhanced support for Alpha Linux is added. Now ./config checks if
3399 the host supports BWX extension and if Compaq C is present on the
3400 $PATH. Just exploiting of the BWX extension results in 20-30%
3401 performance kick for some algorithms, e.g. DES and RC4 to mention
3402 a couple. Compaq C in turn generates ~20% faster code for MD5 and
3406 *) Add support for MS "fast SGC". This is arguably a violation of the
3407 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
3408 weak crypto and after checking the certificate is SGC a second one
3409 with strong crypto. MS SGC stops the first handshake after receiving
3410 the server certificate message and sends a second client hello. Since
3411 a server will typically do all the time consuming operations before
3412 expecting any further messages from the client (server key exchange
3413 is the most expensive) there is little difference between the two.
3415 To get OpenSSL to support MS SGC we have to permit a second client
3416 hello message after we have sent server done. In addition we have to
3417 reset the MAC if we do get this second client hello.
3420 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
3421 if a DER encoded private key is RSA or DSA traditional format. Changed
3422 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
3423 format DER encoded private key. Newer code should use PKCS#8 format which
3424 has the key type encoded in the ASN1 structure. Added DER private key
3425 support to pkcs8 application.
3428 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
3429 ciphersuites has been selected (as required by the SSL 3/TLS 1
3430 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
3431 is set, we interpret this as a request to violate the specification
3432 (the worst that can happen is a handshake failure, and 'correct'
3433 behaviour would result in a handshake failure anyway).
3436 *) In SSL_CTX_add_session, take into account that there might be multiple
3437 SSL_SESSION structures with the same session ID (e.g. when two threads
3438 concurrently obtain them from an external cache).
3439 The internal cache can handle only one SSL_SESSION with a given ID,
3440 so if there's a conflict, we now throw out the old one to achieve
3444 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
3445 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
3446 some routines that use cipher OIDs: some ciphers do not have OIDs
3447 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
3451 *) Simplify the trust setting structure and code. Now we just have
3452 two sequences of OIDs for trusted and rejected settings. These will
3453 typically have values the same as the extended key usage extension
3454 and any application specific purposes.
3456 The trust checking code now has a default behaviour: it will just
3457 check for an object with the same NID as the passed id. Functions can
3458 be provided to override either the default behaviour or the behaviour
3459 for a given id. SSL client, server and email already have functions
3460 in place for compatibility: they check the NID and also return "trusted"
3461 if the certificate is self signed.
3464 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
3465 traditional format into an EVP_PKEY structure.
3468 *) Add a password callback function PEM_cb() which either prompts for
3469 a password if usr_data is NULL or otherwise assumes it is a null
3470 terminated password. Allow passwords to be passed on command line
3471 environment or config files in a few more utilities.
3474 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
3475 keys. Add some short names for PKCS#8 PBE algorithms and allow them
3476 to be specified on the command line for the pkcs8 and pkcs12 utilities.
3477 Update documentation.
3480 *) Support for ASN1 "NULL" type. This could be handled before by using
3481 ASN1_TYPE but there wasn't any function that would try to read a NULL
3482 and produce an error if it couldn't. For compatibility we also have
3483 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
3484 don't allocate anything because they don't need to.
3487 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
3489 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
3491 *) Rebuild of the memory allocation routines used by OpenSSL code and
3492 possibly others as well. The purpose is to make an interface that
3493 provide hooks so anyone can build a separate set of allocation and
3494 deallocation routines to be used by OpenSSL, for example memory
3495 pool implementations, or something else, which was previously hard
3496 since Malloc(), Realloc() and Free() were defined as macros having
3497 the values malloc, realloc and free, respectively (except for Win32
3498 compilations). The same is provided for memory debugging code.
3499 OpenSSL already comes with functionality to find memory leaks, but
3500 this gives people a chance to debug other memory problems.
3502 With these changes, a new set of functions and macros have appeared:
3504 CRYPTO_set_mem_debug_functions() [F]
3505 CRYPTO_get_mem_debug_functions() [F]
3506 CRYPTO_dbg_set_options() [F]
3507 CRYPTO_dbg_get_options() [F]
3508 CRYPTO_malloc_debug_init() [M]
3510 The memory debug functions are NULL by default, unless the library
3511 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
3512 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
3513 gives the standard debugging functions that come with OpenSSL) or
3514 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
3515 provided by the library user) must be used. When the standard
3516 debugging functions are used, CRYPTO_dbg_set_options can be used to
3517 request additional information:
3518 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
3519 the CRYPTO_MDEBUG_xxx macro when compiling the library.
3521 Also, things like CRYPTO_set_mem_functions will always give the
3522 expected result (the new set of functions is used for allocation
3523 and deallocation) at all times, regardless of platform and compiler
3526 To finish it up, some functions that were never use in any other
3527 way than through macros have a new API and new semantic:
3530 CRYPTO_dbg_realloc()
3533 All macros of value have retained their old syntax.
3534 [Richard Levitte and Bodo Moeller]
3536 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
3537 ordering of SMIMECapabilities wasn't in "strength order" and there
3538 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
3542 *) Some ASN1 types with illegal zero length encoding (INTEGER,
3543 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
3544 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
3546 *) Merge in my S/MIME library for OpenSSL. This provides a simple
3547 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
3548 functionality to handle multipart/signed properly) and a utility
3549 called 'smime' to call all this stuff. This is based on code I
3550 originally wrote for Celo who have kindly allowed it to be
3551 included in OpenSSL.
3554 *) Add variants des_set_key_checked and des_set_key_unchecked of
3555 des_set_key (aka des_key_sched). Global variable des_check_key
3556 decides which of these is called by des_set_key; this way
3557 des_check_key behaves as it always did, but applications and
3558 the library itself, which was buggy for des_check_key == 1,
3559 have a cleaner way to pick the version they need.
3562 *) New function PKCS12_newpass() which changes the password of a
3566 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
3567 dynamic mix. In both cases the ids can be used as an index into the
3568 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
3569 functions so they accept a list of the field values and the
3570 application doesn't need to directly manipulate the X509_TRUST
3574 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
3578 *) Modify the way the V3 extension code looks up extensions. This now
3579 works in a similar way to the object code: we have some "standard"
3580 extensions in a static table which is searched with OBJ_bsearch()
3581 and the application can add dynamic ones if needed. The file
3582 crypto/x509v3/ext_dat.h now has the info: this file&nbs
projects / openssl.git / blob