5 Changes between 0.9.7 and 0.9.8 [xx XXX 2002]
7 *) Add ECDSA in new directory crypto/ecdsa/.
9 Add applications 'openssl ecdsaparam' and 'openssl ecdsa'
10 (these are variants of 'openssl dsaparam' and 'openssl dsa').
12 ECDSA support is also included in various other files across the
13 library. Most notably,
14 - 'openssl req' now has a '-newkey ecdsa:file' option;
15 - EVP_PKCS82PKEY (crypto/evp/evp_pkey.c) now can handle ECDSA;
16 - X509_PUBKEY_get (crypto/asn1/x_pubkey.c) and
17 d2i_PublicKey (crypto/asn1/d2i_pu.c) have been modified to make
18 them suitable for ECDSA where domain parameters must be
19 extracted before the specific public key.
20 [Nils Larsch <nla@trustcenter.de>]
22 *) Add reference counting for EC_GROUP objects.
23 [Nils Larsch <nla@trustcenter.de>]
25 *) Include some named elliptic curves. These can be obtained from
28 EC_GROUP_new_by_name()
29 Also add a 'nid' field to EC_GROUP objects, which can be accessed
33 [Nils Larsch <nla@trustcenter.de, Bodo Moeller]
35 Changes between 0.9.6 and 0.9.7 [xx XXX 2002]
37 OpenSSL 0.9.6a/0.9.6b/0.9.6c/0.9.6d (bugfix releases, 5 Apr 2001,
38 9 July 2001, 21 Dec 2001 and xx XXX 2002) and OpenSSL 0.9.7 were
39 developed in parallel, based on OpenSSL 0.9.6.
41 Change log entries are tagged as follows:
42 -) applies to 0.9.6a ... 0.9.6d only
43 *) applies to 0.9.6a ... 0.9.6d and 0.9.7
44 +) applies to 0.9.7 only
46 +) Add and OPENSSL_LOAD_CONF define which will cause
47 OpenSSL_add_all_algorithms() to load the openssl.cnf config file.
48 This allows older applications to transparently support certain
49 OpenSSL features: such as crypto acceleration and dynamic ENGINE loading.
50 Two new functions OPENSSL_add_all_algorithms_noconf() which will never
51 load the config file and OPENSSL_add_all_algorithms_conf() which will
52 always load it have also been added.
55 +) Config modules support in openssl utility.
57 Most commands now load modules from the config file,
58 though in a few (such as version) this isn't done
59 because it couldn't be used for anything.
61 In the case of ca and req the config file used is
62 the same as the utility itself: that is the -config
63 command line option can be used to specify an
67 +) Move default behaviour from OPENSSL_config(). If appname is NULL
68 use "openssl_conf" if filename is NULL use default openssl config file.
71 +) Add an argument to OPENSSL_config() to allow the use of an alternative
72 config section name. Add a new flag to tolerate a missing config file
73 and move code to CONF_modules_load_file().
76 *) Add information about CygWin 1.3 and on, and preserve proper
77 configuration for the versions before that.
78 [Corinna Vinschen <vinschen@redhat.com> and Richard Levitte]
80 *) Make removal from session cache (SSL_CTX_remove_session()) more robust:
81 check whether we deal with a copy of a session and do not delete from
82 the cache in this case. Problem reported by "Izhar Shoshani Levi"
83 <izhar@checkpoint.com>.
86 *) Do not store session data into the internal session cache, if it
87 is never intended to be looked up (SSL_SESS_CACHE_NO_INTERNAL_LOOKUP
88 flag is set). Proposed by Aslam <aslam@funk.com>.
91 +) Support for crypto accelerator cards from Accelerated Encryption
92 Processing, www.aep.ie. (Use engine 'aep')
93 The support was copied from 0.9.6c [engine] and adapted/corrected
94 to work with the new engine framework.
95 [AEP Inc. and Richard Levitte]
97 +) Support for SureWare crypto accelerator cards from Baltimore
98 Technologies. (Use engine 'sureware')
99 The support was copied from 0.9.6c [engine] and adapted
100 to work with the new engine framework.
103 *) Have ASN1_BIT_STRING_set_bit() really clear a bit when the requested
107 +) Have the CHIL engine fork-safe (as defined by nCipher) and actually
108 make the newer ENGINE framework commands for the CHIL engine work.
109 [Toomas Kiisk <vix@cyber.ee> and Richard Levitte]
111 +) Make it possible to produce shared libraries on ReliantUNIX.
112 [Robert Dahlem <Robert.Dahlem@ffm2.siemens.de> via Richard Levitte]
114 *) Add the configuration target linux-s390x.
115 [Neale Ferguson <Neale.Ferguson@SoftwareAG-USA.com> via Richard Levitte]
117 +) Add the configuration target debug-linux-ppro.
118 Make 'openssl rsa' use the general key loading routines
119 implemented in apps.c, and make those routines able to
120 handle the key format FORMAT_NETSCAPE and the variant
122 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
124 *) Fix a crashbug and a logic bug in hwcrhk_load_pubkey().
125 [Toomas Kiisk <vix@cyber.ee> via Richard Levitte]
127 +) Add -keyform to rsautl, and document -engine.
128 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
130 +) Change BIO_new_file (crypto/bio/bss_file.c) to use new
131 BIO_R_NO_SUCH_FILE error code rather than the generic
132 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
137 ERR_peek_last_error_line
138 ERR_peek_last_error_line_data.
142 ERR_peek_error_line_data,
143 but report on the latest error recorded rather than the first one
144 still in the error queue.
145 [Ben Laurie, Bodo Moeller]
147 +) default_algorithms option in ENGINE config module. This allows things
149 default_algorithms = ALL
150 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
153 +) Prelminary ENGINE config module.
156 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
157 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
158 variable as an indication that a ClientHello message has been
159 received. As the flag value will be lost between multiple
160 invocations of ssl3_accept when using non-blocking I/O, the
161 function may not be aware that a handshake has actually taken
162 place, thus preventing a new session from being added to the
165 To avoid this problem, we now set s->new_session to 2 instead of
166 using a local variable.
167 [Lutz Jaenicke, Bodo Moeller]
169 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
170 if the SSL_R_LENGTH_MISMATCH error is detected.
171 [Geoff Thorpe, Bodo Moeller]
173 +) New experimental application configuration code.
176 *) New 'shared_ldflag' column in Configure platform table.
179 *) Fix EVP_CIPHER_mode macro.
180 ["Dan S. Camper" <dan@bti.net>]
182 +) Change the AES code to follow the same name structure as all other
183 symmetric ciphers, and behave the same way. Move everything to
184 the directory crypto/aes, thereby obsoleting crypto/rijndael.
185 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
187 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
188 type, we must throw them away by setting rr->length to 0.
189 [D P Chang <dpc@qualys.com>]
191 -) OpenSSL 0.9.6c released [21 dec 2001]
193 +) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
194 [Ben Laurie and Theo de Raadt]
196 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
197 <Dominikus.Scherkl@biodata.com>. (The previous implementation
198 worked incorrectly for those cases where range = 10..._2 and
199 3*range is two bits longer than range.)
202 *) Only add signing time to PKCS7 structures if it is not already
206 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
207 OBJ_ld_ce should be OBJ_id_ce.
208 Also some ip-pda OIDs in crypto/objects/objects.txt were
209 incorrect (cf. RFC 3039).
210 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
212 +) Add option to output public keys in req command.
213 [Massimiliano Pala madwolf@openca.org]
215 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
216 returns early because it has nothing to do.
217 [Andy Schneider <andy.schneider@bjss.co.uk>]
219 *) [In 0.9.6c-engine and 0.9.7 release:]
220 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
221 [Andy Schneider <andy.schneider@bjss.co.uk>]
223 -) [In 0.9.6c-engine release:]
224 Add support for Cryptographic Appliance's keyserver technology.
225 (Use engine 'keyclient')
226 [Cryptographic Appliances and Geoff Thorpe]
228 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
229 is called via tools/c89.sh because arguments have to be
230 rearranged (all '-L' options must appear before the first object
232 [Richard Shapiro <rshapiro@abinitio.com>]
234 +) Use wNAFs in EC_POINTs_mul() for improved efficiency
235 (up to about 10% better than before for P-192 and P-224).
238 -) [In 0.9.6c-engine release:]
239 Add support for Broadcom crypto accelerator cards, backported
241 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
243 -) [In 0.9.6c-engine release:]
244 Add support for SureWare crypto accelerator cards from
245 Baltimore Technologies. (Use engine 'sureware')
246 [Baltimore Technologies and Mark Cox]
248 -) [In 0.9.6c-engine release:]
249 Add support for crypto accelerator cards from Accelerated
250 Encryption Processing, www.aep.ie. (Use engine 'aep')
251 [AEP Inc. and Mark Cox]
253 *) Add a configuration entry for gcc on UnixWare.
254 [Gary Benson <gbenson@redhat.com>]
256 +) New functions/macros
258 SSL_CTX_set_msg_callback(ctx, cb)
259 SSL_CTX_set_msg_callback_arg(ctx, arg)
260 SSL_set_msg_callback(ssl, cb)
261 SSL_set_msg_callback_arg(ssl, arg)
263 to request calling a callback function
265 void cb(int write_p, int version, int content_type,
266 const void *buf, size_t len, SSL *ssl, void *arg)
268 whenever a protocol message has been completely received
269 (write_p == 0) or sent (write_p == 1). Here 'version' is the
270 protocol version according to which the SSL library interprets
271 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
272 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
273 the content type as defined in the SSL 3.0/TLS 1.0 protocol
274 specification (change_cipher_spec(20), alert(21), handshake(22)).
275 'buf' and 'len' point to the actual message, 'ssl' to the
276 SSL object, and 'arg' is the application-defined value set by
277 SSL[_CTX]_set_msg_callback_arg().
279 'openssl s_client' and 'openssl s_server' have new '-msg' options
280 to enable a callback that displays all protocol messages.
283 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
284 messages are stored in a single piece (fixed-length part and
285 variable-length part combined) and fix various bugs found on the way.
288 +) Change the shared library support so shared libraries are built as
289 soon as the corresponding static library is finished, and thereby get
290 openssl and the test programs linked against the shared library.
291 This still only happens when the keyword "shard" has been given to
292 the configuration scripts.
294 NOTE: shared library support is still an experimental thing, and
295 backward binary compatibility is still not guaranteed.
296 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
298 +) Add support for Subject Information Access extension.
299 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
301 +) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
302 additional bytes when new memory had to be allocated, not just
303 when reusing an existing buffer.
306 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
307 instead. BIO_gethostbyname() does not know what timeouts are
308 appropriate, so entries would stay in cache even when they have
310 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
312 +) New command line and configuration option 'utf8' for the req command.
313 This allows field values to be specified as UTF8 strings.
316 +) Add -multi and -mr options to "openssl speed" - giving multiple parallel
317 runs for the former and machine-readable output for the latter.
320 +) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
321 of the e-mail address in the DN (i.e., it will go into a certificate
322 extension only). The new configuration file option 'email_in_dn = no'
324 [Massimiliano Pala madwolf@openca.org]
326 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
327 faced with a pathologically small ClientHello fragment that does
328 not contain client_version: Instead of aborting with an error,
329 simply choose the highest available protocol version (i.e.,
330 TLS 1.0 unless it is disabled). In practice, ClientHello
331 messages are never sent like this, but this change gives us
332 strictly correct behaviour at least for TLS.
335 +) Change all functions with names starting with des_ to be starting
336 with DES_ instead. Add wrappers that are compatible with libdes,
337 but are named _ossl_old_des_*. Finally, add macros that map the
338 des_* symbols to the corresponding _ossl_old_des_*.
340 All this is done because there are increasing clashes with libdes
341 and other DES libraries that are currently used by other projects.
342 The old libdes interface (including crypt()) is provided if
343 <openssl/des_old.h> is included. For now, this automatically
344 happens in <openssl/des.h> unless OPENSSL_DISABLE_OLD_DES_SUPPORT is
345 defined. Note that crypt() is no longer declared in <openssl/des.h>.
347 NOTE: This is a major break of an old API into a new one. Software
348 authors are encouraged to switch to the DES_ style functions. Some
349 time in the future, des_old.h and the libdes compatibility functions
350 will be completely removed.
353 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
354 never resets s->method to s->ctx->method when called from within
355 one of the SSL handshake functions.
356 [Bodo Moeller; problem pointed out by Niko Baric]
358 +) Test for certificates which contain unsupported critical extensions.
359 If such a certificate is found during a verify operation it is
360 rejected by default: this behaviour can be overridden by either
361 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
362 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
363 X509_supported_extension() has also been added which returns 1 if a
364 particular extension is supported.
367 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
368 (sent using the client's version number) if client_version is
369 smaller than the protocol version in use. Also change
370 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
371 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
372 the client will at least see that alert.
375 +) Modify the behaviour of EVP cipher functions in similar way to digests
376 to retain compatibility with existing code.
379 +) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
380 compatibility with existing code. In particular the 'ctx' parameter does
381 not have to be to be initialized before the call to EVP_DigestInit() and
382 it is tidied up after a call to EVP_DigestFinal(). New function
383 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
384 EVP_MD_CTX_copy() changed to not require the destination to be
385 initialized valid and new function EVP_MD_CTX_copy_ex() added which
386 requires the destination to be valid.
388 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
389 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
392 +) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
393 so that complete 'Handshake' protocol structures are kept in memory
394 instead of overwriting 'msg_type' and 'length' with 'body' data.
397 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
401 +) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
402 [Massimo Santin via Richard Levitte]
404 +) Major restructuring to the underlying ENGINE code. This includes
405 reduction of linker bloat, separation of pure "ENGINE" manipulation
406 (initialisation, etc) from functionality dealing with implementations
407 of specific crypto iterfaces. This change also introduces integrated
408 support for symmetric ciphers and digest implementations - so ENGINEs
409 can now accelerate these by providing EVP_CIPHER and EVP_MD
410 implementations of their own. This is detailed in crypto/engine/README
411 as it couldn't be adequately described here. However, there are a few
412 API changes worth noting - some RSA, DSA, DH, and RAND functions that
413 were changed in the original introduction of ENGINE code have now
414 reverted back - the hooking from this code to ENGINE is now a good
415 deal more passive and at run-time, operations deal directly with
416 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
417 dereferencing through an ENGINE pointer any more. Also, the ENGINE
418 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
419 they were not being used by the framework as there is no concept of a
420 BIGNUM_METHOD and they could not be generalised to the new
421 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
422 ENGINE_cpy() has been removed as it cannot be consistently defined in
426 +) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
429 +) Change mkdef.pl to sort symbols that get the same entry number,
430 and make sure the automatically generated functions ERR_load_*
431 become part of libeay.num as well.
434 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
435 client receives HelloRequest while in a handshake.
436 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
438 +) New function SSL_renegotiate_pending(). This returns true once
439 renegotiation has been requested (either SSL_renegotiate() call
440 or HelloRequest/ClientHello receveived from the peer) and becomes
441 false once a handshake has been completed.
442 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
443 sends a HelloRequest, but does not ensure that a handshake takes
444 place. SSL_renegotiate_pending() is useful for checking if the
445 client has followed the request.)
448 +) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
449 By default, clients may request session resumption even during
450 renegotiation (if session ID contexts permit); with this option,
451 session resumption is possible only in the first handshake.
454 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
455 should end in 'break', not 'goto end' which circuments various
456 cleanups done in state SSL_ST_OK. But session related stuff
457 must be disabled for SSL_ST_OK in the case that we just sent a
460 Also avoid some overhead by not calling ssl_init_wbio_buffer()
461 before just sending a HelloRequest.
462 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
464 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
465 reveal whether illegal block cipher padding was found or a MAC
466 verification error occured. (Neither SSLerr() codes nor alerts
467 are directly visible to potential attackers, but the information
468 may leak via logfiles.)
470 Similar changes are not required for the SSL 2.0 implementation
471 because the number of padding bytes is sent in clear for SSL 2.0,
472 and the extra bytes are just ignored. However ssl/s2_pkt.c
473 failed to verify that the purported number of padding bytes is in
477 +) Add some demos for certificate and certificate request creation.
480 +) Make maximum certificate chain size accepted from the peer application
481 settable (SSL*_get/set_max_cert_list()), as proposed by
482 "Douglas E. Engert" <deengert@anl.gov>.
485 +) Add support for shared libraries for Unixware-7
486 (Boyd Lynn Gerber <gerberb@zenez.com>).
489 *) Add OpenUNIX-8 support including shared libraries
490 (Boyd Lynn Gerber <gerberb@zenez.com>).
493 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
494 'wristwatch attack' using huge encoding parameters (cf.
495 James H. Manger's CRYPTO 2001 paper). Note that the
496 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
497 encoding parameters and hence was not vulnerable.
500 +) Add a "destroy" handler to ENGINEs that allows structural cleanup to
501 be done prior to destruction. Use this to unload error strings from
502 ENGINEs that load their own error strings. NB: This adds two new API
503 functions to "get" and "set" this destroy handler in an ENGINE.
506 +) Alter all existing ENGINE implementations (except "openssl" and
507 "openbsd") to dynamically instantiate their own error strings. This
508 makes them more flexible to be built both as statically-linked ENGINEs
509 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
510 Also, add stub code to each that makes building them as self-contained
511 shared-libraries easier (see README.ENGINE).
514 +) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
515 implementations into applications that are completely implemented in
516 self-contained shared-libraries. The "dynamic" ENGINE exposes control
517 commands that can be used to configure what shared-library to load and
518 to control aspects of the way it is handled. Also, made an update to
519 the README.ENGINE file that brings its information up-to-date and
520 provides some information and instructions on the "dynamic" ENGINE
521 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
525 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
527 +) Make it possible to unload ranges of ERR strings with a new
528 "ERR_unload_strings" function.
531 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
532 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
533 followed by modular reduction.
534 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
536 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
537 equivalent based on BN_pseudo_rand() instead of BN_rand().
540 +) Add a copy() function to EVP_MD.
543 +) Make EVP_MD routines take a context pointer instead of just the
544 md_data void pointer.
547 +) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
548 that the digest can only process a single chunk of data
549 (typically because it is provided by a piece of
550 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
551 is only going to provide a single chunk of data, and hence the
552 framework needn't accumulate the data for oneshot drivers.
555 +) As with "ERR", make it possible to replace the underlying "ex_data"
556 functions. This change also alters the storage and management of global
557 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
558 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
559 index counters. The API functions that use this state have been changed
560 to take a "class_index" rather than pointers to the class's local STACK
561 and counter, and there is now an API function to dynamically create new
562 classes. This centralisation allows us to (a) plug a lot of the
563 thread-safety problems that existed, and (b) makes it possible to clean
564 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
565 such data would previously have always leaked in application code and
566 workarounds were in place to make the memory debugging turn a blind eye
567 to it. Application code that doesn't use this new function will still
568 leak as before, but their memory debugging output will announce it now
569 rather than letting it slide.
571 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
572 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
573 has a return value to indicate success or failure.
576 +) Make it possible to replace the underlying "ERR" functions such that the
577 global state (2 LHASH tables and 2 locks) is only used by the "default"
578 implementation. This change also adds two functions to "get" and "set"
579 the implementation prior to it being automatically set the first time
580 any other ERR function takes place. Ie. an application can call "get",
581 pass the return value to a module it has just loaded, and that module
582 can call its own "set" function using that value. This means the
583 module's "ERR" operations will use (and modify) the error state in the
584 application and not in its own statically linked copy of OpenSSL code.
587 +) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
588 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
589 the operation, and provides a more encapsulated way for external code
590 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
591 to use these functions rather than manually incrementing the counts.
593 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
596 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
597 This function was broken, as the check for a new client hello message
598 to handle SGC did not allow these large messages.
599 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
602 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
605 +) Add EVP test program.
608 +) Add symmetric cipher support to ENGINE. Expect the API to change!
611 +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
612 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
613 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
614 These allow a CRL to be built without having to access X509_CRL fields
615 directly. Modify 'ca' application to use new functions.
618 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
619 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
622 *) Rework the configuration and shared library support for Tru64 Unix.
623 The configuration part makes use of modern compiler features and
624 still retains old compiler behavior for those that run older versions
625 of the OS. The shared library support part includes a variant that
626 uses the RPATH feature, and is available through the special
627 configuration target "alpha-cc-rpath", which will never be selected
629 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
631 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
632 with the same message size as in ssl3_get_certificate_request().
633 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
634 messages might inadvertently be reject as too long.
635 [Petr Lampa <lampa@fee.vutbr.cz>]
637 +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
638 bug workarounds. Rollback attack detection is a security feature.
639 The problem will only arise on OpenSSL servers when TLSv1 is not
640 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
641 Software authors not wanting to support TLSv1 will have special reasons
642 for their choice and can explicitly enable this option.
643 [Bodo Moeller, Lutz Jaenicke]
645 +) Rationalise EVP so it can be extended: don't include a union of
646 cipher/digest structures, add init/cleanup functions. This also reduces
647 the number of header dependencies.
650 +) Make DES key schedule conform to the usual scheme, as well as
651 correcting its structure. This means that calls to DES functions
652 now have to pass a pointer to a des_key_schedule instead of a
653 plain des_key_schedule (which was actually always a pointer
657 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
660 *) Modified SSL library such that the verify_callback that has been set
661 specificly for an SSL object with SSL_set_verify() is actually being
662 used. Before the change, a verify_callback set with this function was
663 ignored and the verify_callback() set in the SSL_CTX at the time of
664 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
665 to allow the necessary settings.
668 +) Initial reduction of linker bloat: the use of some functions, such as
669 PEM causes large amounts of unused functions to be linked in due to
670 poor organisation. For example pem_all.c contains every PEM function
671 which has a knock on effect of linking in large amounts of (unused)
672 ASN1 code. Grouping together similar functions and splitting unrelated
673 functions prevents this.
676 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
677 explicitly to NULL, as at least on Solaris 8 this seems not always to be
678 done automatically (in contradiction to the requirements of the C
679 standard). This made problems when used from OpenSSH.
682 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
683 dh->length and always used
685 BN_rand_range(priv_key, dh->p).
687 BN_rand_range() is not necessary for Diffie-Hellman, and this
688 specific range makes Diffie-Hellman unnecessarily inefficient if
689 dh->length (recommended exponent length) is much smaller than the
690 length of dh->p. We could use BN_rand_range() if the order of
691 the subgroup was stored in the DH structure, but we only have
696 BN_rand(priv_key, l, ...)
698 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
704 RSA_eay_public_encrypt
705 RSA_eay_private_decrypt
706 RSA_eay_private_encrypt (signing)
707 RSA_eay_public_decrypt (signature verification)
709 (default implementations for RSA_public_encrypt,
710 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
711 always reject numbers >= n.
714 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
715 to synchronize access to 'locking_thread'. This is necessary on
716 systems where access to 'locking_thread' (an 'unsigned long'
717 variable) is not atomic.
720 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
721 *before* setting the 'crypto_lock_rand' flag. The previous code had
722 a race condition if 0 is a valid thread ID.
723 [Travis Vitek <vitek@roguewave.com>]
725 +) Cleanup of EVP macros.
728 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
732 +) Add initial OCSP responder support to ocsp application. The
733 revocation information is handled using the text based index
734 use by the ca application. The responder can either handle
735 requests generated internally, supplied in files (for example
736 via a CGI script) or using an internal minimal server.
739 +) Add configuration choices to get zlib compression for TLS.
742 +) Changes to Kerberos SSL for RFC 2712 compliance:
743 1. Implemented real KerberosWrapper, instead of just using
744 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
745 2. Implemented optional authenticator field of KerberosWrapper.
747 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
748 and authenticator structs; see crypto/krb5/.
750 Generalized Kerberos calls to support multiple Kerberos libraries.
751 [Vern Staats <staatsvr@asc.hpc.mil>,
752 Jeffrey Altman <jaltman@columbia.edu>
755 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
756 already does with RSA. testdsa.h now has 'priv_key/pub_key'
757 values for each of the key sizes rather than having just
758 parameters (and 'speed' generating keys each time).
761 *) Add support for shared libraries under Irix.
762 [Albert Chin-A-Young <china@thewrittenword.com>]
764 *) Add configuration option to build on Linux on both big-endian and
766 [Ralf Baechle <ralf@uni-koblenz.de>]
768 *) Add the possibility to create shared libraries on HP-UX.
771 -) OpenSSL 0.9.6b released [9 July 2001]
773 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
774 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
775 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
776 PRNG state recovery was possible based on the output of
777 one PRNG request appropriately sized to gain knowledge on
778 'md' followed by enough consecutive 1-byte PRNG requests
779 to traverse all of 'state'.
781 1. When updating 'md_local' (the current thread's copy of 'md')
782 during PRNG output generation, hash all of the previous
783 'md_local' value, not just the half used for PRNG output.
785 2. Make the number of bytes from 'state' included into the hash
786 independent from the number of PRNG bytes requested.
788 The first measure alone would be sufficient to avoid
789 Markku-Juhani's attack. (Actually it had never occurred
790 to me that the half of 'md_local' used for chaining was the
791 half from which PRNG output bytes were taken -- I had always
792 assumed that the secret half would be used.) The second
793 measure makes sure that additional data from 'state' is never
794 mixed into 'md_local' in small portions; this heuristically
795 further strengthens the PRNG.
798 +) Speed up EVP routines.
801 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
802 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
803 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
804 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
806 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
807 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
808 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
811 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
813 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
816 *) Fix crypto/bn/asm/mips3.s.
819 *) When only the key is given to "enc", the IV is undefined. Print out
820 an error message in this case.
823 +) Added the OS2-EMX target.
824 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
826 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
827 to support NCONF routines in extension code. New function CONF_set_nconf()
828 to allow functions which take an NCONF to also handle the old LHASH
829 structure: this means that the old CONF compatible routines can be
830 retained (in particular wrt extensions) without having to duplicate the
831 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
834 *) Handle special case when X509_NAME is empty in X509 printing routines.
837 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
838 positive and less than q.
841 +) Enhance the general user interface with mechanisms for inner control
842 and with possibilities to have yes/no kind of prompts.
845 +) Change all calls to low level digest routines in the library and
846 applications to use EVP. Add missing calls to HMAC_cleanup() and
847 don't assume HMAC_CTX can be copied using memcpy().
848 [Verdon Walker <VWalker@novell.com>, Steve Henson]
850 +) Add the possibility to control engines through control names but with
851 arbitrary arguments instead of just a string.
852 Change the key loaders to take a UI_METHOD instead of a callback
853 function pointer. NOTE: this breaks binary compatibility with earlier
854 versions of OpenSSL [engine].
855 Adapt the nCipher code for these new conditions and add a card insertion
859 +) Enhance the general user interface with mechanisms to better support
860 dialog box interfaces, application-defined prompts, the possibility
861 to use defaults (for example default passwords from somewhere else)
862 and interrupts/cancellations.
865 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
866 used: it isn't thread safe and the add_lock_callback should handle
868 [Paul Rose <Paul.Rose@bridge.com>]
870 *) Verify that incoming data obeys the block size in
871 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
874 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
875 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
879 [Ulf Möller, Bodo Möller]
881 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
882 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
883 when fixing the server behaviour for backwards-compatible 'client
884 hello' messages. (Note that the attack is impractical against
885 SSL 3.0 and TLS 1.0 anyway because length and version checking
886 means that the probability of guessing a valid ciphertext is
887 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
890 Before 0.9.5, the countermeasure (hide the error by generating a
891 random 'decryption result') did not work properly because
892 ERR_clear_error() was missing, meaning that SSL_get_error() would
893 detect the supposedly ignored error.
895 Both problems are now fixed.
898 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
899 (previously it was 1024).
902 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
903 tidy up some unnecessarily weird code in 'sk_new()').
904 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
906 +) Change the key loading routines for ENGINEs to use the same kind
907 callback (pem_password_cb) as all other routines that need this
911 *) Fix for compatibility mode trust settings: ignore trust settings
912 unless some valid trust or reject settings are present.
915 *) Fix for blowfish EVP: its a variable length cipher.
918 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
919 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
920 than this minimum value is recommended.
923 +) New random seeder for OpenVMS, using the system process statistics
924 that are easily reachable.
927 +) Windows apparently can't transparently handle global
928 variables defined in DLLs. Initialisations such as:
930 const ASN1_ITEM *it = &ASN1_INTEGER_it;
932 wont compile. This is used by the any applications that need to
933 declare their own ASN1 modules. This was fixed by adding the option
934 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
935 needed for static libraries under Win32.
938 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
939 setting of purpose and trust fields. New X509_STORE trust and
940 purpose functions and tidy up setting in other SSL functions.
943 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
944 structure. These are inherited by X509_STORE_CTX when it is
945 initialised. This allows various defaults to be set in the
946 X509_STORE structure (such as flags for CRL checking and custom
947 purpose or trust settings) for functions which only use X509_STORE_CTX
948 internally such as S/MIME.
950 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
951 trust settings if they are not set in X509_STORE. This allows X509_STORE
952 purposes and trust (in S/MIME for example) to override any set by default.
954 Add command line options for CRL checking to smime, s_client and s_server
958 +) Initial CRL based revocation checking. If the CRL checking flag(s)
959 are set then the CRL is looked up in the X509_STORE structure and
960 its validity and signature checked, then if the certificate is found
961 in the CRL the verify fails with a revoked error.
963 Various new CRL related callbacks added to X509_STORE_CTX structure.
965 Command line options added to 'verify' application to support this.
967 This needs some additional work, such as being able to handle multiple
968 CRLs with different times, extension based lookup (rather than just
969 by subject name) and ultimately more complete V2 CRL extension
973 +) Add a general user interface API (crypto/ui/). This is designed
974 to replace things like des_read_password and friends (backward
975 compatibility functions using this new API are provided).
976 The purpose is to remove prompting functions from the DES code
977 section as well as provide for prompting through dialog boxes in
978 a window system and the like.
981 *) Fix various bugs related to DSA S/MIME verification. Handle missing
982 parameters in DSA public key structures and return an error in the
983 DSA routines if parameters are absent.
986 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
987 in the current directory if neither $RANDFILE nor $HOME was set.
988 RAND_file_name() in 0.9.6a returned NULL in this case. This has
989 caused some confusion to Windows users who haven't defined $HOME.
990 Thus RAND_file_name() is changed again: e_os.h can define a
991 DEFAULT_HOME, which will be used if $HOME is not set.
992 For Windows, we use "C:"; on other platforms, we still require
993 environment variables.
995 +) Add "ex_data" support to ENGINE so implementations can add state at a
996 per-structure level rather than having to store it globally.
999 +) Make it possible for ENGINE structures to be copied when retrieved by
1000 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
1001 This causes the "original" ENGINE structure to act like a template,
1002 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
1003 operational state can be localised to each ENGINE structure, despite the
1004 fact they all share the same "methods". New ENGINE structures returned in
1005 this case have no functional references and the return value is the single
1006 structural reference. This matches the single structural reference returned
1007 by ENGINE_by_id() normally, when it is incremented on the pre-existing
1011 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
1012 needs to match any other type at all we need to manually clear the
1016 +) Changes to the "openssl engine" utility to include;
1017 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
1018 about an ENGINE's available control commands.
1019 - executing control commands from command line arguments using the
1020 '-pre' and '-post' switches. '-post' is only used if '-t' is
1021 specified and the ENGINE is successfully initialised. The syntax for
1022 the individual commands are colon-separated, for example;
1023 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
1026 +) New dynamic control command support for ENGINEs. ENGINEs can now
1027 declare their own commands (numbers), names (strings), descriptions,
1028 and input types for run-time discovery by calling applications. A
1029 subset of these commands are implicitly classed as "executable"
1030 depending on their input type, and only these can be invoked through
1031 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
1032 can be based on user input, config files, etc). The distinction is
1033 that "executable" commands cannot return anything other than a boolean
1034 result and can only support numeric or string input, whereas some
1035 discoverable commands may only be for direct use through
1036 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
1037 pointers, or other custom uses. The "executable" commands are to
1038 support parameterisations of ENGINE behaviour that can be
1039 unambiguously defined by ENGINEs and used consistently across any
1040 OpenSSL-based application. Commands have been added to all the
1041 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
1042 control over shared-library paths without source code alterations.
1045 +) Changed all ENGINE implementations to dynamically allocate their
1046 ENGINEs rather than declaring them statically. Apart from this being
1047 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
1048 this also allows the implementations to compile without using the
1049 internal engine_int.h header.
1052 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
1053 'const' value. Any code that should be able to modify a RAND_METHOD
1054 should already have non-const pointers to it (ie. they should only
1055 modify their own ones).
1058 +) Made a variety of little tweaks to the ENGINE code.
1059 - "atalla" and "ubsec" string definitions were moved from header files
1060 to C code. "nuron" string definitions were placed in variables
1061 rather than hard-coded - allowing parameterisation of these values
1062 later on via ctrl() commands.
1063 - Removed unused "#if 0"'d code.
1064 - Fixed engine list iteration code so it uses ENGINE_free() to release
1065 structural references.
1066 - Constified the RAND_METHOD element of ENGINE structures.
1067 - Constified various get/set functions as appropriate and added
1068 missing functions (including a catch-all ENGINE_cpy that duplicates
1069 all ENGINE values onto a new ENGINE except reference counts/state).
1070 - Removed NULL parameter checks in get/set functions. Setting a method
1071 or function to NULL is a way of cancelling out a previously set
1072 value. Passing a NULL ENGINE parameter is just plain stupid anyway
1073 and doesn't justify the extra error symbols and code.
1074 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
1075 flags from engine_int.h to engine.h.
1076 - Changed prototypes for ENGINE handler functions (init(), finish(),
1077 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
1080 *) Move 'if (!initialized) RAND_poll()' into regions protected by
1081 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
1082 having multiple threads call RAND_poll() concurrently.
1085 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
1086 combination of a flag and a thread ID variable.
1087 Otherwise while one thread is in ssleay_rand_bytes (which sets the
1088 flag), *other* threads can enter ssleay_add_bytes without obeying
1089 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
1090 that they do not hold after the first thread unsets add_do_not_lock).
1093 +) Implement binary inversion algorithm for BN_mod_inverse in addition
1094 to the algorithm using long division. The binary algorithm can be
1095 used only if the modulus is odd. On 32-bit systems, it is faster
1096 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
1097 roughly 5-15% for 256-bit moduli), so we use it only for moduli
1098 up to 450 bits. In 64-bit environments, the binary algorithm
1099 appears to be advantageous for much longer moduli; here we use it
1100 for moduli up to 2048 bits.
1103 *) Change bctest again: '-x' expressions are not available in all
1107 -) OpenSSL 0.9.6a released [5 Apr 2001]
1109 *) Fix a couple of memory leaks in PKCS7_dataDecode()
1110 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
1112 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
1113 the default extension for executables, if any. Also, make the perl
1114 scripts that use symlink() to test if it really exists and use "cp"
1115 if it doesn't. All this made OpenSSL compilable and installable in
1119 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1120 could not support the combine flag in choice fields.
1123 -) Fix for asn1_GetSequence() for indefinite length constructed data.
1124 If SEQUENCE is length is indefinite just set c->slen to the total
1125 amount of data available.
1126 [Steve Henson, reported by shige@FreeBSD.org]
1127 [This change does not apply to 0.9.7.]
1129 *) Change bctest to avoid here-documents inside command substitution
1130 (workaround for FreeBSD /bin/sh bug).
1131 For compatibility with Ultrix, avoid shell functions (introduced
1132 in the bctest version that searches along $PATH).
1135 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
1136 with des_encrypt() defined on some operating systems, like Solaris
1140 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
1141 On the Importance of Eliminating Errors in Cryptographic
1142 Computations, J. Cryptology 14 (2001) 2, 101-119,
1143 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
1146 *) MIPS assembler BIGNUM division bug fix.
1149 *) Disabled incorrect Alpha assembler code.
1152 -) Fix PKCS#7 decode routines so they correctly update the length
1153 after reading an EOC for the EXPLICIT tag.
1155 [This change does not apply to 0.9.7.]
1157 *) Fix bug in PKCS#12 key generation routines. This was triggered
1158 if a 3DES key was generated with a 0 initial byte. Include
1159 PKCS12_BROKEN_KEYGEN compilation option to retain the old
1160 (but broken) behaviour.
1163 *) Enhance bctest to search for a working bc along $PATH and print
1165 [Tim Rice <tim@multitalents.net> via Richard Levitte]
1167 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
1168 extensions from a certificate request to the certificate.
1171 +) Allow multiple 'certopt' and 'nameopt' options to be separated
1172 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1173 file: this allows the display of the certificate about to be
1174 signed to be customised, to allow certain fields to be included
1175 or excluded and extension details. The old system didn't display
1176 multicharacter strings properly, omitted fields not in the policy
1177 and couldn't display additional details such as extensions.
1180 *) Fix memory leaks in err.c: free err_data string if necessary;
1181 don't write to the wrong index in ERR_set_error_data.
1184 +) Function EC_POINTs_mul for multiple scalar multiplication
1185 of an arbitrary number of elliptic curve points
1186 \sum scalars[i]*points[i],
1187 optionally including the generator defined for the EC_GROUP:
1188 scalar*generator + \sum scalars[i]*points[i].
1190 EC_POINT_mul is a simple wrapper function for the typical case
1191 that the point list has just one item (besides the optional
1195 +) First EC_METHODs for curves over GF(p):
1197 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1198 operations and provides various method functions that can also
1199 operate with faster implementations of modular arithmetic.
1201 EC_GFp_mont_method() reuses most functions that are part of
1202 EC_GFp_simple_method, but uses Montgomery arithmetic.
1204 [Bodo Moeller; point addition and point doubling
1205 implementation directly derived from source code provided by
1206 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1208 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1209 crypto/ec/ec_lib.c):
1211 Curves are EC_GROUP objects (with an optional group generator)
1212 based on EC_METHODs that are built into the library.
1214 Points are EC_POINT objects based on EC_GROUP objects.
1216 Most of the framework would be able to handle curves over arbitrary
1217 finite fields, but as there are no obvious types for fields other
1218 than GF(p), some functions are limited to that for now.
1221 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1222 that the file contains a complete HTTP response.
1225 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1226 change the def and num file printf format specifier from "%-40sXXX"
1227 to "%-39s XXX". The latter will always guarantee a space after the
1228 field while the former will cause them to run together if the field
1229 is 40 of more characters long.
1232 +) Constify the cipher and digest 'method' functions and structures
1233 and modify related functions to take constant EVP_MD and EVP_CIPHER
1237 *) Implement ssl23_peek (analogous to ssl23_read), which previously
1241 *) Replace rdtsc with _emit statements for VC++ version 5.
1242 [Jeremy Cooper <jeremy@baymoo.org>]
1244 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1245 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1248 +) Modify EVP_Digest*() routines so they now return values. Although the
1249 internal software routines can never fail additional hardware versions
1253 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1255 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1256 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1259 ERR_R_NESTED_ASN1_ERROR
1261 ERR_R_MISSING_ASN1_EOS
1262 were 4 .. 9, conflicting with
1263 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1265 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1266 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1268 Add new error code 'ERR_R_INTERNAL_ERROR'.
1271 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1275 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1276 sets the subject name for a new request or supersedes the
1277 subject name in a given request. Formats that can be parsed are
1278 'CN=Some Name, OU=myOU, C=IT'
1280 'CN=Some Name/OU=myOU/C=IT'.
1282 Add options '-batch' and '-verbose' to 'openssl req'.
1283 [Massimiliano Pala <madwolf@hackmasters.net>]
1285 *) Make it possible to reuse SSLv2 sessions.
1288 +) Introduce the possibility to access global variables through
1289 functions on platform were that's the best way to handle exporting
1290 global variables in shared libraries. To enable this functionality,
1291 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1292 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1293 is normally done by Configure or something similar).
1295 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1296 in the source file (foo.c) like this:
1298 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1299 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1301 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1302 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1304 OPENSSL_DECLARE_GLOBAL(int,foo);
1305 #define foo OPENSSL_GLOBAL_REF(foo)
1306 OPENSSL_DECLARE_GLOBAL(double,bar);
1307 #define bar OPENSSL_GLOBAL_REF(bar)
1309 The #defines are very important, and therefore so is including the
1310 header file everywhere where the defined globals are used.
1312 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1313 of ASN.1 items, but that structure is a bit different.
1315 The largest change is in util/mkdef.pl which has been enhanced with
1316 better and easier to understand logic to choose which symbols should
1317 go into the Windows .def files as well as a number of fixes and code
1318 cleanup (among others, algorithm keywords are now sorted
1319 lexicographically to avoid constant rewrites).
1322 *) In copy_email() check for >= 0 as a return value for
1323 X509_NAME_get_index_by_NID() since 0 is a valid index.
1324 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
1326 +) In BN_div() keep a copy of the sign of 'num' before writing the
1327 result to 'rm' because if rm==num the value will be overwritten
1328 and produce the wrong result if 'num' is negative: this caused
1329 problems with BN_mod() and BN_nnmod().
1332 +) Function OCSP_request_verify(). This checks the signature on an
1333 OCSP request and verifies the signer certificate. The signer
1334 certificate is just checked for a generic purpose and OCSP request
1338 +) Add OCSP_check_validity() function to check the validity of OCSP
1339 responses. OCSP responses are prepared in real time and may only
1340 be a few seconds old. Simply checking that the current time lies
1341 between thisUpdate and nextUpdate max reject otherwise valid responses
1342 caused by either OCSP responder or client clock inaccuracy. Instead
1343 we allow thisUpdate and nextUpdate to fall within a certain period of
1344 the current time. The age of the response can also optionally be
1345 checked. Two new options -validity_period and -status_age added to
1349 +) If signature or public key algorithm is unrecognized print out its
1350 OID rather that just UNKNOWN.
1353 *) Avoid coredump with unsupported or invalid public keys by checking if
1354 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
1355 PKCS7_verify() fails with non detached data.
1358 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1359 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1360 ID to be generated from the issuer certificate alone which can then be
1361 passed to OCSP_id_issuer_cmp().
1364 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1365 ASN1 modules to export functions returning ASN1_ITEM pointers
1366 instead of the ASN1_ITEM structures themselves. This adds several
1367 new macros which allow the underlying ASN1 function/structure to
1368 be accessed transparently. As a result code should not use ASN1_ITEM
1369 references directly (such as &X509_it) but instead use the relevant
1370 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1371 use of the new ASN1 code on platforms where exporting structures
1372 is problematical (for example in shared libraries) but exporting
1373 functions returning pointers to structures is not.
1376 +) Add support for overriding the generation of SSL/TLS session IDs.
1377 These callbacks can be registered either in an SSL_CTX or per SSL.
1378 The purpose of this is to allow applications to control, if they wish,
1379 the arbitrary values chosen for use as session IDs, particularly as it
1380 can be useful for session caching in multiple-server environments. A
1381 command-line switch for testing this (and any client code that wishes
1382 to use such a feature) has been added to "s_server".
1383 [Geoff Thorpe, Lutz Jaenicke]
1385 +) Modify mkdef.pl to recognise and parse preprocessor conditionals
1386 of the form '#if defined(...) || defined(...) || ...' and
1387 '#if !defined(...) && !defined(...) && ...'. This also avoids
1388 the growing number of special cases it was previously handling.
1391 *) Don't use getenv in library functions when run as setuid/setgid.
1392 New function OPENSSL_issetugid().
1395 +) Make all configuration macros available for application by making
1396 sure they are available in opensslconf.h, by giving them names starting
1397 with "OPENSSL_" to avoid conflicts with other packages and by making
1398 sure e_os2.h will cover all platform-specific cases together with
1400 Additionally, it is now possible to define configuration/platform-
1401 specific names (called "system identities"). In the C code, these
1402 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1403 macro with the name beginning with "OPENSSL_SYS_", which is determined
1404 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1408 +) New option -set_serial to 'req' and 'x509' this allows the serial
1409 number to use to be specified on the command line. Previously self
1410 signed certificates were hard coded with serial number 0 and the
1411 CA options of 'x509' had to use a serial number in a file which was
1415 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
1416 due to incorrect handling of multi-threading:
1418 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
1420 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
1422 3. Count how many times MemCheck_off() has been called so that
1423 nested use can be treated correctly. This also avoids
1424 inband-signalling in the previous code (which relied on the
1425 assumption that thread ID 0 is impossible).
1428 +) New options to 'ca' utility to support V2 CRL entry extensions.
1429 Currently CRL reason, invalidity date and hold instruction are
1430 supported. Add new CRL extensions to V3 code and some new objects.
1433 *) Add "-rand" option also to s_client and s_server.
1436 +) New function EVP_CIPHER_CTX_set_padding() this is used to
1437 disable standard block padding (aka PKCS#5 padding) in the EVP
1438 API, which was previously mandatory. This means that the data is
1439 not padded in any way and so the total length much be a multiple
1440 of the block size, otherwise an error occurs.
1443 +) Initial (incomplete) OCSP SSL support.
1446 *) Fix CPU detection on Irix 6.x.
1447 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
1448 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1450 +) New function OCSP_parse_url(). This splits up a URL into its host,
1451 port and path components: primarily to parse OCSP URLs. New -url
1452 option to ocsp utility.
1455 +) New nonce behavior. The return value of OCSP_check_nonce() now
1456 reflects the various checks performed. Applications can decide
1457 whether to tolerate certain situations such as an absent nonce
1458 in a response when one was present in a request: the ocsp application
1459 just prints out a warning. New function OCSP_add1_basic_nonce()
1460 this is to allow responders to include a nonce in a response even if
1461 the request is nonce-less.
1464 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
1467 [This change does not apply to 0.9.7.]
1469 *) Use the cached encoding of an X509_NAME structure rather than
1470 copying it. This is apparently the reason for the libsafe "errors"
1471 but the code is actually correct.
1474 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1475 skipped when using openssl x509 multiple times on a single input file,
1476 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1479 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1480 set string type: to handle setting ASN1_TIME structures. Fix ca
1481 utility to correctly initialize revocation date of CRLs.
1484 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1485 the clients preferred ciphersuites and rather use its own preferences.
1486 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1487 Internet Explorer by ensuring unchanged hash method during stepup.
1488 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1491 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1492 to aes and add a new 'exist' option to print out symbols that don't
1496 +) Additional options to ocsp utility to allow flags to be set and
1497 additional certificates supplied.
1500 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
1501 OCSP client a number of certificate to only verify the response
1505 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
1506 Bleichenbacher's DSA attack.
1507 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
1508 to be set and top=0 forces the highest bit to be set; top=-1 is new
1509 and leaves the highest bit random.
1510 [Ulf Moeller, Bodo Moeller]
1512 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1513 handle the new API. Currently only ECB, CBC modes supported. Add new
1514 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
1515 for TLS" draft-ietf-tls-ciphersuite-03.txt.
1516 [Ben Laurie, Steve Henson]
1518 *) In the NCONF_...-based implementations for CONF_... queries
1519 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
1520 a temporary CONF structure with the data component set to NULL
1521 (which gives segmentation faults in lh_retrieve).
1522 Instead, use NULL for the CONF pointer in CONF_get_string and
1523 CONF_get_number (which may use environment variables) and directly
1524 return NULL from CONF_get_section.
1527 *) Fix potential buffer overrun for EBCDIC.
1530 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
1531 request to response.
1534 +) Functions for OCSP responders. OCSP_request_onereq_count(),
1535 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1536 extract information from a certificate request. OCSP_response_create()
1537 creates a response and optionally adds a basic response structure.
1538 OCSP_basic_add1_status() adds a complete single response to a basic
1539 response and returns the OCSP_SINGLERESP structure just added (to allow
1540 extensions to be included for example). OCSP_basic_add1_cert() adds a
1541 certificate to a basic response and OCSP_basic_sign() signs a basic
1542 response with various flags. New helper functions ASN1_TIME_check()
1543 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1544 (converts ASN1_TIME to GeneralizedTime).
1547 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1548 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1549 structure from a certificate. X509_pubkey_digest() digests the public_key
1550 contents: this is used in various key identifiers.
1553 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
1554 keyUsage if basicConstraints absent for a CA.
1557 *) Make SMIME_write_PKCS7() write mail header values with a format that
1558 is more generally accepted (no spaces before the semicolon), since
1559 some programs can't parse those values properly otherwise. Also make
1560 sure BIO's that break lines after each write do not create invalid
1564 +) Make sk_sort() tolerate a NULL argument.
1565 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1567 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1568 passed by the function are trusted implicitly. If any of them signed the
1569 response then it is assumed to be valid and is not verified.
1572 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
1573 macros previously used would not encode an empty SEQUENCE OF
1574 and break the signature.
1576 [This change does not apply to 0.9.7.]
1578 *) Zero the premaster secret after deriving the master secret in
1582 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1583 to data. This was previously part of the PKCS7 ASN1 code. This
1584 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1585 [Steve Henson, reported by Kenneth R. Robinette
1586 <support@securenetterm.com>]
1588 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1589 routines: without these tracing memory leaks is very painful.
1590 Fix leaks in PKCS12 and PKCS7 routines.
1593 *) Add some EVP_add_digest_alias registrations (as found in
1594 OpenSSL_add_all_digests()) to SSL_library_init()
1595 aka OpenSSL_add_ssl_algorithms(). This provides improved
1596 compatibility with peers using X.509 certificates
1597 with unconventional AlgorithmIdentifier OIDs.
1600 *) Fix for Irix with NO_ASM.
1601 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1603 *) ./config script fixes.
1604 [Ulf Moeller, Richard Levitte]
1606 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1607 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1608 effectively meant GeneralizedTime would never be used. Now it
1609 is initialised to -1 but X509_time_adj() now has to check the value
1610 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1611 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1612 [Steve Henson, reported by Kenneth R. Robinette
1613 <support@securenetterm.com>]
1615 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1616 result in a zero length in the ASN1_INTEGER structure which was
1617 not consistent with the structure when d2i_ASN1_INTEGER() was used
1618 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1619 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1620 where it did not print out a minus for negative ASN1_INTEGER.
1623 *) Fix 'openssl passwd -1'.
1626 +) Add summary printout to ocsp utility. The various functions which
1627 convert status values to strings have been renamed to:
1628 OCSP_response_status_str(), OCSP_cert_status_str() and
1629 OCSP_crl_reason_str() and are no longer static. New options
1630 to verify nonce values and to disable verification. OCSP response
1631 printout format cleaned up.
1634 +) Add additional OCSP certificate checks. These are those specified
1635 in RFC2560. This consists of two separate checks: the CA of the
1636 certificate being checked must either be the OCSP signer certificate
1637 or the issuer of the OCSP signer certificate. In the latter case the
1638 OCSP signer certificate must contain the OCSP signing extended key
1639 usage. This check is performed by attempting to match the OCSP
1640 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1641 in the OCSP_CERTID structures of the response.
1644 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1645 and related routines. This uses the standard OpenSSL certificate
1646 verify routines to perform initial checks (just CA validity) and
1647 to obtain the certificate chain. Then additional checks will be
1648 performed on the chain. Currently the root CA is checked to see
1649 if it is explicitly trusted for OCSP signing. This is used to set
1650 a root CA as a global signing root: that is any certificate that
1651 chains to that CA is an acceptable OCSP signing certificate.
1654 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1655 extensions from a separate configuration file.
1656 As when reading extensions from the main configuration file,
1657 the '-extensions ...' option may be used for specifying the
1659 [Massimiliano Pala <madwolf@comune.modena.it>]
1661 *) Change PKCS12_key_gen_asc() so it can cope with non null
1662 terminated strings whose length is passed in the passlen
1663 parameter, for example from PEM callbacks. This was done
1664 by adding an extra length parameter to asc2uni().
1665 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1667 +) New OCSP utility. Allows OCSP requests to be generated or
1668 read. The request can be sent to a responder and the output
1669 parsed, outputed or printed in text form. Not complete yet:
1670 still needs to check the OCSP response validity.
1673 +) New subcommands for 'openssl ca':
1674 'openssl ca -status <serial>' prints the status of the cert with
1675 the given serial number (according to the index file).
1676 'openssl ca -updatedb' updates the expiry status of certificates
1678 [Massimiliano Pala <madwolf@comune.modena.it>]
1680 +) New '-newreq-nodes' command option to CA.pl. This is like
1681 '-newreq', but calls 'openssl req' with the '-nodes' option
1682 so that the resulting key is not encrypted.
1683 [Damien Miller <djm@mindrot.org>]
1685 +) New configuration for the GNU Hurd.
1686 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1688 +) Initial code to implement OCSP basic response verify. This
1689 is currently incomplete. Currently just finds the signer's
1690 certificate and verifies the signature on the response.
1693 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1694 value of OPENSSLDIR. This is available via the new '-d' option
1695 to 'openssl version', and is also included in 'openssl version -a'.
1698 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1699 call failed, free the DSA structure.
1702 +) Allowing defining memory allocation callbacks that will be given
1703 file name and line number information in additional arguments
1704 (a const char* and an int). The basic functionality remains, as
1705 well as the original possibility to just replace malloc(),
1706 realloc() and free() by functions that do not know about these
1707 additional arguments. To register and find out the current
1708 settings for extended allocation functions, the following
1709 functions are provided:
1711 CRYPTO_set_mem_ex_functions
1712 CRYPTO_set_locked_mem_ex_functions
1713 CRYPTO_get_mem_ex_functions
1714 CRYPTO_get_locked_mem_ex_functions
1716 These work the same way as CRYPTO_set_mem_functions and friends.
1717 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1718 extended allocation function is enabled.
1719 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1720 a conventional allocation function is enabled.
1721 [Richard Levitte, Bodo Moeller]
1723 *) Fix to uni2asc() to cope with zero length Unicode strings.
1724 These are present in some PKCS#12 files.
1727 +) Finish off removing the remaining LHASH function pointer casts.
1728 There should no longer be any prototype-casting required when using
1729 the LHASH abstraction, and any casts that remain are "bugs". See
1730 the callback types and macros at the head of lhash.h for details
1731 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1734 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1735 If /dev/[u]random devices are not available or do not return enough
1736 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1738 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1739 /etc/entropy will be queried once each in this sequence, quering stops
1740 when enough entropy was collected without querying more sockets.
1743 +) Change the Unix RAND_poll() variant to be able to poll several
1744 random devices, as specified by DEVRANDOM, until a sufficient amount
1745 of data has been collected. We spend at most 10 ms on each file
1746 (select timeout) and read in non-blocking mode. DEVRANDOM now
1747 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1748 (previously it was just the string "/dev/urandom"), so on typical
1749 platforms the 10 ms delay will never occur.
1750 Also separate out the Unix variant to its own file, rand_unix.c.
1751 For VMS, there's a currently-empty rand_vms.c.
1754 +) Move OCSP client related routines to ocsp_cl.c. These
1755 provide utility functions which an application needing
1756 to issue a request to an OCSP responder and analyse the
1757 response will typically need: as opposed to those which an
1758 OCSP responder itself would need which will be added later.
1760 OCSP_request_sign() signs an OCSP request with an API similar
1761 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1762 response. OCSP_response_get1_basic() extracts basic response
1763 from response. OCSP_resp_find_status(): finds and extracts status
1764 information from an OCSP_CERTID structure (which will be created
1765 when the request structure is built). These are built from lower
1766 level functions which work on OCSP_SINGLERESP structures but
1767 wont normally be used unless the application wishes to examine
1768 extensions in the OCSP response for example.
1770 Replace nonce routines with a pair of functions.
1771 OCSP_request_add1_nonce() adds a nonce value and optionally
1772 generates a random value. OCSP_check_nonce() checks the
1773 validity of the nonce in an OCSP response.
1776 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1777 This doesn't copy the supplied OCSP_CERTID and avoids the
1778 need to free up the newly created id. Change return type
1779 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1780 This can then be used to add extensions to the request.
1781 Deleted OCSP_request_new(), since most of its functionality
1782 is now in OCSP_REQUEST_new() (and the case insensitive name
1783 clash) apart from the ability to set the request name which
1784 will be added elsewhere.
1787 +) Update OCSP API. Remove obsolete extensions argument from
1788 various functions. Extensions are now handled using the new
1789 OCSP extension code. New simple OCSP HTTP function which
1790 can be used to send requests and parse the response.
1793 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1794 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1795 uses the special reorder version of SET OF to sort the attributes
1796 and reorder them to match the encoded order. This resolves a long
1797 standing problem: a verify on a PKCS7 structure just after signing
1798 it used to fail because the attribute order did not match the
1799 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1800 it uses the received order. This is necessary to tolerate some broken
1801 software that does not order SET OF. This is handled by encoding
1802 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1803 to produce the required SET OF.
1806 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1807 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1808 files to get correct declarations of the ASN.1 item variables.
1811 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1812 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1813 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1814 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1815 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1816 ASN1_ITEM and no wrapper functions.
1819 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1820 replace the old function pointer based I/O routines. Change most of
1821 the *_d2i_bio() and *_d2i_fp() functions to use these.
1824 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1825 lines, recognice more "algorithms" that can be deselected, and make
1826 it complain about algorithm deselection that isn't recognised.
1829 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1830 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1831 to use new functions. Add NO_ASN1_OLD which can be set to remove
1832 some old style ASN1 functions: this can be used to determine if old
1833 code will still work when these eventually go away.
1836 +) New extension functions for OCSP structures, these follow the
1837 same conventions as certificates and CRLs.
1840 +) New function X509V3_add1_i2d(). This automatically encodes and
1841 adds an extension. Its behaviour can be customised with various
1842 flags to append, replace or delete. Various wrappers added for
1843 certifcates and CRLs.
1846 +) Fix to avoid calling the underlying ASN1 print routine when
1847 an extension cannot be parsed. Correct a typo in the
1848 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1851 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1852 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1853 when writing a 32767 byte record.
1854 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1856 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1857 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1859 (RSA objects have a reference count access to which is protected
1860 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1861 so they are meant to be shared between threads.)
1862 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1863 "Reddie, Steven" <Steven.Reddie@ca.com>]
1865 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1866 entries for variables.
1869 *) Fix a deadlock in CRYPTO_mem_leaks().
1872 +) Add functionality to apps/openssl.c for detecting locking
1873 problems: As the program is single-threaded, all we have
1874 to do is register a locking callback using an array for
1875 storing which locks are currently held by the program.
1878 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1879 SSL_get_ex_data_X509_STORE_idx(), which is used in
1880 ssl_verify_cert_chain() and thus can be called at any time
1881 during TLS/SSL handshakes so that thread-safety is essential.
1882 Unfortunately, the ex_data design is not at all suited
1883 for multi-threaded use, so it probably should be abolished.
1886 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1887 [Broadcom, tweaked and integrated by Geoff Thorpe]
1889 +) Move common extension printing code to new function
1890 X509V3_print_extensions(). Reorganise OCSP print routines and
1891 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1894 +) New function X509_signature_print() to remove duplication in some
1898 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1899 set (this was treated exactly the same as SET OF previously). This
1900 is used to reorder the STACK representing the structure to match the
1901 encoding. This will be used to get round a problem where a PKCS7
1902 structure which was signed could not be verified because the STACK
1903 order did not reflect the encoded order.
1906 +) Reimplement the OCSP ASN1 module using the new code.
1909 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1910 for its ASN1 operations. The old style function pointers still exist
1911 for now but they will eventually go away.
1914 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1915 completely replaces the old ASN1 functionality with a table driven
1916 encoder and decoder which interprets an ASN1_ITEM structure describing
1917 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1918 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1919 has also been converted to the new form.
1922 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1923 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1924 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1925 for negative moduli.
1928 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1929 of not touching the result's sign bit.
1932 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1936 +) Changed the LHASH code to use prototypes for callbacks, and created
1937 macros to declare and implement thin (optionally static) functions
1938 that provide type-safety and avoid function pointer casting for the
1939 type-specific callbacks.
1942 *) Use better test patterns in bntest.
1945 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1947 [Veers Staats <staatsvr@asc.hpc.mil>,
1948 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1950 *) rand_win.c fix for Borland C.
1953 *) BN_rshift bugfix for n == 0.
1956 +) Reformat the FAQ so the different questions and answers can be divided
1957 in sections depending on the subject.
1960 +) Have the zlib compression code load ZLIB.DLL dynamically under
1964 +) New function BN_mod_sqrt for computing square roots modulo a prime
1965 (using the probabilistic Tonelli-Shanks algorithm unless
1966 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1967 be handled deterministically).
1968 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1970 *) Add a 'bctest' script that checks for some known 'bc' bugs
1971 so that 'make test' does not abort just because 'bc' is broken.
1974 *) Store verify_result within SSL_SESSION also for client side to
1975 avoid potential security hole. (Re-used sessions on the client side
1976 always resulted in verify_result==X509_V_OK, not using the original
1977 result of the server certificate verification.)
1980 +) Make BN_mod_inverse faster by explicitly handling small quotients
1981 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1982 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1985 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1986 SSL3_RT_APPLICATION_DATA, return 0.
1987 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1991 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1992 releases, have been re-implemented by renaming the previous
1993 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1994 and ssl3_read_internal, respectively, and adding 'peek' parameters
1995 to them. The new ssl[23]_{read,peek} functions are calls to
1996 ssl[23]_read_internal with the 'peek' flag set appropriately.
1997 A 'peek' parameter has also been added to ssl3_read_bytes, which
1998 does the actual work for ssl3_read_internal.
2001 +) New function BN_kronecker.
2004 +) Fix BN_gcd so that it works on negative inputs; the result is
2005 positive unless both parameters are zero.
2006 Previously something reasonably close to an infinite loop was
2007 possible because numbers could be growing instead of shrinking
2008 in the implementation of Euclid's algorithm.
2011 +) Fix BN_is_word() and BN_is_one() macros to take into account the
2012 sign of the number in question.
2014 Fix BN_is_word(a,w) to work correctly for w == 0.
2016 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
2017 because its test if the absolute value of 'a' equals 'w'.
2018 Note that BN_abs_is_word does *not* handle w == 0 reliably;
2019 it exists mostly for use in the implementations of BN_is_zero(),
2020 BN_is_one(), and BN_is_word().
2023 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
2024 the method-specific "init()" handler. Also clean up ex_data after
2025 calling the method-specific "finish()" handler. Previously, this was
2026 happening the other way round.
2029 +) New function BN_swap.
2032 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
2033 the exponentiation functions are more likely to produce reasonable
2034 results on negative inputs.
2037 +) Change BN_mod_mul so that the result is always non-negative.
2038 Previously, it could be negative if one of the factors was negative;
2039 I don't think anyone really wanted that behaviour.
2042 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
2043 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
2044 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
2045 and add new functions:
2054 BN_mod_lshift1_quick
2058 These functions always generate non-negative results.
2060 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
2061 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
2063 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
2064 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
2065 be reduced modulo m.
2066 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
2068 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
2069 was actually never needed) and in BN_mul(). The removal in BN_mul()
2070 required a small change in bn_mul_part_recursive() and the addition
2071 of the functions bn_cmp_part_words(), bn_sub_part_words() and
2072 bn_add_part_words(), which do the same thing as bn_cmp_words(),
2073 bn_sub_words() and bn_add_words() except they take arrays with
2077 +) In 'openssl passwd', verify passwords read from the terminal
2078 unless the '-salt' option is used (which usually means that
2079 verification would just waste user's time since the resulting
2080 hash is going to be compared with some given password hash)
2081 or the new '-noverify' option is used.
2083 This is an incompatible change, but it does not affect
2084 non-interactive use of 'openssl passwd' (passwords on the command
2085 line, '-stdin' option, '-in ...' option) and thus should not
2089 +) Remove all references to RSAref, since there's no more need for it.
2092 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
2093 The previous value, 12, was not always sufficient for BN_mod_exp().
2096 +) Make DSO load along a path given through an environment variable
2097 (SHLIB_PATH) with shl_load().
2100 +) Constify the ENGINE code as a result of BIGNUM constification.
2101 Also constify the RSA code and most things related to it. In a
2102 few places, most notable in the depth of the ASN.1 code, ugly
2103 casts back to non-const were required (to be solved at a later
2107 +) Make it so the openssl application has all engines loaded by default.
2110 +) Constify the BIGNUM routines a little more.
2113 *) Make sure that shared libraries get the internal name engine with
2114 the full version number and not just 0. This should mark the
2115 shared libraries as not backward compatible. Of course, this should
2116 be changed again when we can guarantee backward binary compatibility.
2119 +) Add the following functions:
2121 ENGINE_load_cswift()
2123 ENGINE_load_atalla()
2125 ENGINE_load_builtin_engines()
2127 That way, an application can itself choose if external engines that
2128 are built-in in OpenSSL shall ever be used or not. The benefit is
2129 that applications won't have to be linked with libdl or other dso
2130 libraries unless it's really needed.
2132 Changed 'openssl engine' to load all engines on demand.
2133 Changed the engine header files to avoid the duplication of some
2134 declarations (they differed!).
2137 +) 'openssl engine' can now list capabilities.
2140 +) Better error reporting in 'openssl engine'.
2143 +) Never call load_dh_param(NULL) in s_server.
2146 +) Add engine application. It can currently list engines by name and
2147 identity, and test if they are actually available.
2150 +) Improve RPM specification file by forcing symbolic linking and making
2151 sure the installed documentation is also owned by root.root.
2152 [Damien Miller <djm@mindrot.org>]
2154 +) Give the OpenSSL applications more possibilities to make use of
2155 keys (public as well as private) handled by engines.
2158 +) Add OCSP code that comes from CertCo.
2161 +) Add VMS support for the Rijndael code.
2164 +) Added untested support for Nuron crypto accelerator.
2167 +) Add support for external cryptographic devices. This code was
2168 previously distributed separately as the "engine" branch.
2169 [Geoff Thorpe, Richard Levitte]
2171 +) Rework the filename-translation in the DSO code. It is now possible to
2172 have far greater control over how a "name" is turned into a filename
2173 depending on the operating environment and any oddities about the
2174 different shared library filenames on each system.
2177 +) Support threads on FreeBSD-elf in Configure.
2180 +) Fix for SHA1 assembly problem with MASM: it produces
2181 warnings about corrupt line number information when assembling
2182 with debugging information. This is caused by the overlapping
2184 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2187 NCONF_get_number() has no error checking at all. As a replacement,
2188 NCONF_get_number_e() is defined (_e for "error checking") and is
2189 promoted strongly. The old NCONF_get_number is kept around for
2190 binary backward compatibility.
2191 Make it possible for methods to load from something other than a BIO,
2192 by providing a function pointer that is given a name instead of a BIO.
2193 For example, this could be used to load configuration data from an
2197 *) Fix typo in get_cert_by_subject() in by_dir.c
2198 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
2200 *) Rework the system to generate shared libraries:
2202 - Make note of the expected extension for the shared libraries and
2203 if there is a need for symbolic links from for example libcrypto.so.0
2204 to libcrypto.so.0.9.7. There is extended info in Configure for
2207 - Make as few rebuilds of the shared libraries as possible.
2209 - Still avoid linking the OpenSSL programs with the shared libraries.
2211 - When installing, install the shared libraries separately from the
2215 +) Fix for non blocking accept BIOs. Added new I/O special reason
2216 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2217 with non blocking I/O was not possible because no retry code was
2218 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2222 +) Added the beginnings of Rijndael support.
2225 +) Fix for bug in DirectoryString mask setting. Add support for
2226 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2227 to allow certificate printing to more controllable, additional
2228 'certopt' option to 'x509' to allow new printing options to be
2232 +) Clean old EAY MD5 hack from e_os.h.
2235 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
2237 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
2238 and not in SSL_clear because the latter is also used by the
2239 accept/connect functions; previously, the settings made by
2240 SSL_set_read_ahead would be lost during the handshake.
2241 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
2243 *) Correct util/mkdef.pl to be selective about disabled algorithms.
2244 Previously, it would create entries for disableed algorithms no
2248 *) Added several new manual pages for SSL_* function.
2251 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
2253 *) In ssl23_get_client_hello, generate an error message when faced
2254 with an initial SSL 3.0/TLS record that is too small to contain the
2255 first two bytes of the ClientHello message, i.e. client_version.
2256 (Note that this is a pathologic case that probably has never happened
2257 in real life.) The previous approach was to use the version number
2258 from the record header as a substitute; but our protocol choice
2259 should not depend on that one because it is not authenticated
2260 by the Finished messages.
2263 *) More robust randomness gathering functions for Windows.
2264 [Jeffrey Altman <jaltman@columbia.edu>]
2266 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
2267 not set then we don't setup the error code for issuer check errors
2268 to avoid possibly overwriting other errors which the callback does
2269 handle. If an application does set the flag then we assume it knows
2270 what it is doing and can handle the new informational codes
2274 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
2275 a general "ANY" type, as such it should be able to decode anything
2276 including tagged types. However it didn't check the class so it would
2277 wrongly interpret tagged types in the same way as their universal
2278 counterpart and unknown types were just rejected. Changed so that the
2279 tagged and unknown types are handled in the same way as a SEQUENCE:
2280 that is the encoding is stored intact. There is also a new type
2281 "V_ASN1_OTHER" which is used when the class is not universal, in this
2282 case we have no idea what the actual type is so we just lump them all
2286 *) On VMS, stdout may very well lead to a file that is written to
2287 in a record-oriented fashion. That means that every write() will
2288 write a separate record, which will be read separately by the
2289 programs trying to read from it. This can be very confusing.
2291 The solution is to put a BIO filter in the way that will buffer
2292 text until a linefeed is reached, and then write everything a
2293 line at a time, so every record written will be an actual line,
2294 not chunks of lines and not (usually doesn't happen, but I've
2295 seen it once) several lines in one record. BIO_f_linebuffer() is
2298 Currently, it's a VMS-only method, because that's where it has
2299 been tested well enough.
2302 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
2303 it can return incorrect results.
2304 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
2305 but it was in 0.9.6-beta[12].)
2308 *) Disable the check for content being present when verifying detached
2309 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
2310 include zero length content when signing messages.
2313 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
2314 BIO_ctrl (for BIO pairs).
2317 *) Add DSO method for VMS.
2320 *) Bug fix: Montgomery multiplication could produce results with the
2324 *) Add RPM specification openssl.spec and modify it to build three
2325 packages. The default package contains applications, application
2326 documentation and run-time libraries. The devel package contains
2327 include files, static libraries and function documentation. The
2328 doc package contains the contents of the doc directory. The original
2329 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
2332 *) Add a large number of documentation files for many SSL routines.
2333 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2335 *) Add a configuration entry for Sony News 4.
2336 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
2338 *) Don't set the two most significant bits to one when generating a
2339 random number < q in the DSA library.
2342 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
2343 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
2344 the underlying transport is blocking) if a handshake took place.
2345 (The default behaviour is needed by applications such as s_client
2346 and s_server that use select() to determine when to use SSL_read;
2347 but for applications that know in advance when to expect data, it
2348 just makes things more complicated.)
2351 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
2355 *) Add a few more EBCDIC conditionals that make `req' and `x509'
2356 work better on such systems.
2357 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
2359 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
2360 Update PKCS12_parse() so it copies the friendlyName and the
2361 keyid to the certificates aux info.
2364 *) Fix bug in PKCS7_verify() which caused an infinite loop
2365 if there was more than one signature.
2366 [Sven Uszpelkat <su@celocom.de>]
2368 *) Major change in util/mkdef.pl to include extra information
2369 about each symbol, as well as presentig variables as well
2370 as functions. This change means that there's n more need
2371 to rebuild the .num files when some algorithms are excluded.
2374 *) Allow the verify time to be set by an application,
2375 rather than always using the current time.
2378 *) Phase 2 verify code reorganisation. The certificate
2379 verify code now looks up an issuer certificate by a
2380 number of criteria: subject name, authority key id
2381 and key usage. It also verifies self signed certificates
2382 by the same criteria. The main comparison function is
2383 X509_check_issued() which performs these checks.
2385 Lot of changes were necessary in order to support this
2386 without completely rewriting the lookup code.
2388 Authority and subject key identifier are now cached.
2390 The LHASH 'certs' is X509_STORE has now been replaced
2391 by a STACK_OF(X509_OBJECT). This is mainly because an
2392 LHASH can't store or retrieve multiple objects with
2393 the same hash value.
2395 As a result various functions (which were all internal
2396 use only) have changed to handle the new X509_STORE
2397 structure. This will break anything that messed round
2398 with X509_STORE internally.
2400 The functions X509_STORE_add_cert() now checks for an
2401 exact match, rather than just subject name.
2403 The X509_STORE API doesn't directly support the retrieval
2404 of multiple certificates matching a given criteria, however
2405 this can be worked round by performing a lookup first
2406 (which will fill the cache with candidate certificates)
2407 and then examining the cache for matches. This is probably
2408 the best we can do without throwing out X509_LOOKUP
2409 entirely (maybe later...).
2411 The X509_VERIFY_CTX structure has been enhanced considerably.
2413 All certificate lookup operations now go via a get_issuer()
2414 callback. Although this currently uses an X509_STORE it
2415 can be replaced by custom lookups. This is a simple way
2416 to bypass the X509_STORE hackery necessary to make this
2417 work and makes it possible to use more efficient techniques
2418 in future. A very simple version which uses a simple
2419 STACK for its trusted certificate store is also provided
2420 using X509_STORE_CTX_trusted_stack().
2422 The verify_cb() and verify() callbacks now have equivalents
2423 in the X509_STORE_CTX structure.
2425 X509_STORE_CTX also has a 'flags' field which can be used
2426 to customise the verify behaviour.
2429 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
2430 excludes S/MIME capabilities.
2433 *) When a certificate request is read in keep a copy of the
2434 original encoding of the signed data and use it when outputing
2435 again. Signatures then use the original encoding rather than
2436 a decoded, encoded version which may cause problems if the
2437 request is improperly encoded.
2440 *) For consistency with other BIO_puts implementations, call
2441 buffer_write(b, ...) directly in buffer_puts instead of calling
2444 In BIO_puts, increment b->num_write as in BIO_write.
2445 [Peter.Sylvester@EdelWeb.fr]
2447 *) Fix BN_mul_word for the case where the word is 0. (We have to use
2448 BN_zero, we may not return a BIGNUM with an array consisting of
2452 *) Avoid calling abort() from within the library when problems are
2453 detected, except if preprocessor symbols have been defined
2454 (such as REF_CHECK, BN_DEBUG etc.).
2457 *) New openssl application 'rsautl'. This utility can be
2458 used for low level RSA operations. DER public key
2459 BIO/fp routines also added.
2462 *) New Configure entry and patches for compiling on QNX 4.
2463 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
2465 *) A demo state-machine implementation was sponsored by
2466 Nuron (http://www.nuron.com/) and is now available in
2467 demos/state_machine.
2470 *) New options added to the 'dgst' utility for signature
2471 generation and verification.
2474 *) Unrecognized PKCS#7 content types are now handled via a
2475 catch all ASN1_TYPE structure. This allows unsupported
2476 types to be stored as a "blob" and an application can
2477 encode and decode it manually.
2480 *) Fix various signed/unsigned issues to make a_strex.c
2482 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
2484 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
2485 length if passed a buffer. ASN1_INTEGER_to_BN failed
2486 if passed a NULL BN and its argument was negative.
2487 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
2489 *) Modification to PKCS#7 encoding routines to output definite
2490 length encoding. Since currently the whole structures are in
2491 memory there's not real point in using indefinite length
2492 constructed encoding. However if OpenSSL is compiled with
2493 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
2496 *) Added BIO_vprintf() and BIO_vsnprintf().
2499 *) Added more prefixes to parse for in the the strings written
2500 through a logging bio, to cover all the levels that are available
2501 through syslog. The prefixes are now:
2503 PANIC, EMERG, EMR => LOG_EMERG
2504 ALERT, ALR => LOG_ALERT
2505 CRIT, CRI => LOG_CRIT
2506 ERROR, ERR => LOG_ERR
2507 WARNING, WARN, WAR => LOG_WARNING
2508 NOTICE, NOTE, NOT => LOG_NOTICE
2509 INFO, INF => LOG_INFO
2510 DEBUG, DBG => LOG_DEBUG
2512 and as before, if none of those prefixes are present at the
2513 beginning of the string, LOG_ERR is chosen.
2515 On Win32, the LOG_* levels are mapped according to this:
2517 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
2518 LOG_WARNING => EVENTLOG_WARNING_TYPE
2519 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
2523 *) Made it possible to reconfigure with just the configuration
2524 argument "reconf" or "reconfigure". The command line arguments
2525 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
2526 and are retrieved from there when reconfiguring.
2530 [Assar Westerlund <assar@sics.se>, Richard Levitte]
2532 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
2535 *) The obj_dat.pl script was messing up the sorting of object
2536 names. The reason was that it compared the quoted version
2537 of strings as a result "OCSP" > "OCSP Signing" because
2538 " > SPACE. Changed script to store unquoted versions of
2539 names and add quotes on output. It was also omitting some
2540 names from the lookup table if they were given a default
2541 value (that is if SN is missing it is given the same
2542 value as LN and vice versa), these are now added on the
2543 grounds that if an object has a name we should be able to
2544 look it up. Finally added warning output when duplicate
2545 short or long names are found.
2548 *) Changes needed for Tandem NSK.
2549 [Scott Uroff <scott@xypro.com>]
2551 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
2552 RSA_padding_check_SSLv23(), special padding was never detected
2553 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
2554 version rollback attacks was not effective.
2556 In s23_clnt.c, don't use special rollback-attack detection padding
2557 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
2558 client; similarly, in s23_srvr.c, don't do the rollback check if
2559 SSL 2.0 is the only protocol enabled in the server.
2562 *) Make it possible to get hexdumps of unprintable data with 'openssl
2563 asn1parse'. By implication, the functions ASN1_parse_dump() and
2564 BIO_dump_indent() are added.
2567 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
2568 these print out strings and name structures based on various
2569 flags including RFC2253 support and proper handling of
2570 multibyte characters. Added options to the 'x509' utility
2571 to allow the various flags to be set.
2574 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
2575 Also change the functions X509_cmp_current_time() and
2576 X509_gmtime_adj() work with an ASN1_TIME structure,
2577 this will enable certificates using GeneralizedTime in validity
2578 dates to be checked.
2581 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
2582 negative public key encodings) on by default,
2583 NO_NEG_PUBKEY_BUG can be set to disable it.
2586 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
2587 content octets. An i2c_ASN1_OBJECT is unnecessary because
2588 the encoding can be trivially obtained from the structure.
2591 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
2592 not read locks (CRYPTO_r_[un]lock).
2595 *) A first attempt at creating official support for shared
2596 libraries through configuration. I've kept it so the
2597 default is static libraries only, and the OpenSSL programs
2598 are always statically linked for now, but there are
2599 preparations for dynamic linking in place.
2600 This has been tested on Linux and Tru64.
2603 *) Randomness polling function for Win9x, as described in:
2604 Peter Gutmann, Software Generation of Practically Strong
2608 *) Fix so PRNG is seeded in req if using an already existing
2612 *) New options to smime application. -inform and -outform
2613 allow alternative formats for the S/MIME message including
2614 PEM and DER. The -content option allows the content to be
2615 specified separately. This should allow things like Netscape
2616 form signing output easier to verify.
2619 *) Fix the ASN1 encoding of tags using the 'long form'.
2622 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
2623 STRING types. These convert content octets to and from the
2624 underlying type. The actual tag and length octets are
2625 already assumed to have been read in and checked. These
2626 are needed because all other string types have virtually
2627 identical handling apart from the tag. By having versions
2628 of the ASN1 functions that just operate on content octets
2629 IMPLICIT tagging can be handled properly. It also allows
2630 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
2631 and ASN1_INTEGER are identical apart from the tag.
2634 *) Change the handling of OID objects as follows:
2636 - New object identifiers are inserted in objects.txt, following
2637 the syntax given in objects.README.
2638 - objects.pl is used to process obj_mac.num and create a new
2640 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2643 This is currently kind of a hack, and the perl code in objects.pl
2644 isn't very elegant, but it works as I intended. The simplest way
2645 to check that it worked correctly is to look in obj_dat.h and
2646 check the array nid_objs and make sure the objects haven't moved
2647 around (this is important!). Additions are OK, as well as
2648 consistent name changes.
2651 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2654 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2655 The given file adds to whatever has already been seeded into the
2656 random pool through the RANDFILE configuration file option or
2657 environment variable, or the default random state file.
2660 *) mkstack.pl now sorts each macro group into lexical order.
2661 Previously the output order depended on the order the files
2662 appeared in the directory, resulting in needless rewriting
2666 *) Patches to make OpenSSL compile under Win32 again. Mostly
2667 work arounds for the VC++ problem that it treats func() as
2668 func(void). Also stripped out the parts of mkdef.pl that
2669 added extra typesafe functions: these no longer exist.
2672 *) Reorganisation of the stack code. The macros are now all
2673 collected in safestack.h . Each macro is defined in terms of
2674 a "stack macro" of the form SKM_<name>(type, a, b). The
2675 DEBUG_SAFESTACK is now handled in terms of function casts,
2676 this has the advantage of retaining type safety without the
2677 use of additional functions. If DEBUG_SAFESTACK is not defined
2678 then the non typesafe macros are used instead. Also modified the
2679 mkstack.pl script to handle the new form. Needs testing to see
2680 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2681 the default if no major problems. Similar behaviour for ASN1_SET_OF
2682 and PKCS12_STACK_OF.
2685 *) When some versions of IIS use the 'NET' form of private key the
2686 key derivation algorithm is different. Normally MD5(password) is
2687 used as a 128 bit RC4 key. In the modified case
2688 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2689 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2690 as the old Netscape_RSA functions except they have an additional
2691 'sgckey' parameter which uses the modified algorithm. Also added
2692 an -sgckey command line option to the rsa utility. Thanks to
2693 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2694 algorithm to openssl-dev.
2697 *) The evp_local.h macros were using 'c.##kname' which resulted in
2698 invalid expansion on some systems (SCO 5.0.5 for example).
2699 Corrected to 'c.kname'.
2700 [Phillip Porch <root@theporch.com>]
2702 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2703 a STACK of email addresses from a certificate or request, these look
2704 in the subject name and the subject alternative name extensions and
2705 omit any duplicate addresses.
2708 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2709 This makes DSA verification about 2 % faster.
2712 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2713 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2714 plus overhead for 1024 bit moduli).
2715 This makes exponentiations about 0.5 % faster for 1024 bit
2716 exponents (as measured by "openssl speed rsa2048").
2719 *) Rename memory handling macros to avoid conflicts with other
2721 Malloc => OPENSSL_malloc
2722 Malloc_locked => OPENSSL_malloc_locked
2723 Realloc => OPENSSL_realloc
2724 Free => OPENSSL_free
2727 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2728 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2731 *) CygWin32 support.
2732 [John Jarvie <jjarvie@newsguy.com>]
2734 *) The type-safe stack code has been rejigged. It is now only compiled
2735 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2736 by default all type-specific stack functions are "#define"d back to
2737 standard stack functions. This results in more streamlined output
2738 but retains the type-safety checking possibilities of the original
2742 *) The STACK code has been cleaned up, and certain type declarations
2743 that didn't make a lot of sense have been brought in line. This has
2744 also involved a cleanup of sorts in safestack.h to more correctly
2745 map type-safe stack functions onto their plain stack counterparts.
2746 This work has also resulted in a variety of "const"ifications of
2747 lots of the code, especially "_cmp" operations which should normally
2748 be prototyped with "const" parameters anyway.
2751 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2752 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2753 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2754 where all of 'md' is used each time the PRNG is used, but 'state'
2755 is used only indexed by a cyclic counter. As entropy may not be
2756 well distributed from the beginning, 'md' is important as a
2757 chaining variable. However, the output function chains only half
2758 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2759 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2760 in all of 'state' being rewritten, with the new values depending
2761 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2764 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2765 the handshake is continued after ssl_verify_cert_chain();
2766 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2767 can lead to 'unexplainable' connection aborts later.
2768 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2770 *) Major EVP API cipher revision.
2771 Add hooks for extra EVP features. This allows various cipher
2772 parameters to be set in the EVP interface. Support added for variable
2773 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2774 setting of RC2 and RC5 parameters.
2776 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2779 Remove lots of duplicated code from the EVP library. For example *every*
2780 cipher init() function handles the 'iv' in the same way according to the
2781 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2782 for CFB and OFB modes they zero ctx->num.
2784 New functionality allows removal of S/MIME code RC2 hack.
2786 Most of the routines have the same form and so can be declared in terms
2789 By shifting this to the top level EVP_CipherInit() it can be removed from
2790 all individual ciphers. If the cipher wants to handle IVs or keys
2791 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2794 Change lots of functions like EVP_EncryptUpdate() to now return a
2795 value: although software versions of the algorithms cannot fail
2796 any installed hardware versions can.
2799 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2800 this option is set, tolerate broken clients that send the negotiated
2801 protocol version number instead of the requested protocol version
2805 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2806 i.e. non-zero for export ciphersuites, zero otherwise.
2807 Previous versions had this flag inverted, inconsistent with
2808 rsa_tmp_cb (..._TMP_RSA_CB).
2809 [Bodo Moeller; problem reported by Amit Chopra]
2811 *) Add missing DSA library text string. Work around for some IIS
2812 key files with invalid SEQUENCE encoding.
2815 *) Add a document (doc/standards.txt) that list all kinds of standards
2816 and so on that are implemented in OpenSSL.
2819 *) Enhance c_rehash script. Old version would mishandle certificates
2820 with the same subject name hash and wouldn't handle CRLs at all.
2821 Added -fingerprint option to crl utility, to support new c_rehash
2825 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2828 *) Fix for SSL server purpose checking. Server checking was
2829 rejecting certificates which had extended key usage present
2830 but no ssl client purpose.
2831 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2833 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2834 is a little unclear about how a blank password is handled.
2835 Since the password in encoded as a BMPString with terminating
2836 double NULL a zero length password would end up as just the
2837 double NULL. However no password at all is different and is
2838 handled differently in the PKCS#12 key generation code. NS
2839 treats a blank password as zero length. MSIE treats it as no
2840 password on export: but it will try both on import. We now do
2841 the same: PKCS12_parse() tries zero length and no password if
2842 the password is set to "" or NULL (NULL is now a valid password:
2843 it wasn't before) as does the pkcs12 application.
2846 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2847 perror when PEM_read_bio_X509_REQ fails, the error message must
2848 be obtained from the error queue.
2851 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2852 it in ERR_remove_state if appropriate, and change ERR_get_state
2853 accordingly to avoid race conditions (this is necessary because
2854 thread_hash is no longer constant once set).
2857 *) Bugfix for linux-elf makefile.one.
2860 *) RSA_get_default_method() will now cause a default
2861 RSA_METHOD to be chosen if one doesn't exist already.
2862 Previously this was only set during a call to RSA_new()
2863 or RSA_new_method(NULL) meaning it was possible for
2864 RSA_get_default_method() to return NULL.
2867 *) Added native name translation to the existing DSO code
2868 that will convert (if the flag to do so is set) filenames
2869 that are sufficiently small and have no path information
2870 into a canonical native form. Eg. "blah" converted to
2871 "libblah.so" or "blah.dll" etc.
2874 *) New function ERR_error_string_n(e, buf, len) which is like
2875 ERR_error_string(e, buf), but writes at most 'len' bytes
2876 including the 0 terminator. For ERR_error_string_n, 'buf'
2878 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2880 *) CONF library reworked to become more general. A new CONF
2881 configuration file reader "class" is implemented as well as a
2882 new functions (NCONF_*, for "New CONF") to handle it. The now
2883 old CONF_* functions are still there, but are reimplemented to
2884 work in terms of the new functions. Also, a set of functions
2885 to handle the internal storage of the configuration data is
2886 provided to make it easier to write new configuration file
2887 reader "classes" (I can definitely see something reading a
2888 configuration file in XML format, for example), called _CONF_*,
2889 or "the configuration storage API"...
2891 The new configuration file reading functions are:
2893 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2894 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2896 NCONF_default, NCONF_WIN32
2898 NCONF_dump_fp, NCONF_dump_bio
2900 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2901 NCONF_new creates a new CONF object. This works in the same way
2902 as other interfaces in OpenSSL, like the BIO interface.
2903 NCONF_dump_* dump the internal storage of the configuration file,
2904 which is useful for debugging. All other functions take the same
2905 arguments as the old CONF_* functions wth the exception of the
2906 first that must be a `CONF *' instead of a `LHASH *'.
2908 To make it easer to use the new classes with the old CONF_* functions,
2909 the function CONF_set_default_method is provided.
2912 *) Add '-tls1' option to 'openssl ciphers', which was already
2913 mentioned in the documentation but had not been implemented.
2914 (This option is not yet really useful because even the additional
2915 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2918 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2919 OpenSSL-based applications) load shared libraries and bind to
2920 them in a portable way.
2921 [Geoff Thorpe, with contributions from Richard Levitte]
2923 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2925 *) Make sure _lrotl and _lrotr are only used with MSVC.
2927 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2928 (the default implementation of RAND_status).
2930 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2931 to '-clrext' (= clear extensions), as intended and documented.
2932 [Bodo Moeller; inconsistency pointed out by Michael Attili
2935 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2936 was larger than the MD block size.
2937 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2939 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2940 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2941 using the passed key: if the passed key was a private key the result
2942 of X509_print(), for example, would be to print out all the private key
2946 *) des_quad_cksum() byte order bug fix.
2947 [Ulf Möller, using the problem description in krb4-0.9.7, where
2948 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2950 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2952 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2954 *) For easily testing in shell scripts whether some command
2955 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2956 returns with exit code 0 iff no command of the given name is available.
2957 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2958 the output goes to stdout and nothing is printed to stderr.
2959 Additional arguments are always ignored.
2961 Since for each cipher there is a command of the same name,
2962 the 'no-cipher' compilation switches can be tested this way.
2964 ('openssl no-XXX' is not able to detect pseudo-commands such
2965 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2968 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2971 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2972 is set; it will be thrown away anyway because each handshake creates
2974 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2975 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2976 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2977 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2980 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2981 'Q' and 'R' lose their special meanings (quit/renegotiate).
2982 This is part of what -quiet does; unlike -quiet, -ign_eof
2983 does not suppress any output.
2986 *) Add compatibility options to the purpose and trust code. The
2987 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2988 accepts a certificate or CA, this was the previous behaviour,
2989 with all the associated security issues.
2991 X509_TRUST_COMPAT is the old trust behaviour: only and
2992 automatically trust self signed roots in certificate store. A
2993 new trust setting X509_TRUST_DEFAULT is used to specify that
2994 a purpose has no associated trust setting and it should instead
2995 use the value in the default purpose.
2998 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2999 and fix a memory leak.
3002 *) In util/mkerr.pl (which implements 'make errors'), preserve
3003 reason strings from the previous version of the .c file, as
3004 the default to have only downcase letters (and digits) in
3005 automatically generated reasons codes is not always appropriate.
3008 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
3009 using strerror. Previously, ERR_reason_error_string() returned
3010 library names as reason strings for SYSerr; but SYSerr is a special
3011 case where small numbers are errno values, not library numbers.
3014 *) Add '-dsaparam' option to 'openssl dhparam' application. This
3015 converts DSA parameters into DH parameters. (When creating parameters,
3016 DSA_generate_parameters is used.)
3019 *) Include 'length' (recommended exponent length) in C code generated
3020 by 'openssl dhparam -C'.
3023 *) The second argument to set_label in perlasm was already being used
3024 so couldn't be used as a "file scope" flag. Moved to third argument
3028 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
3029 instead of RAND_bytes for encryption IVs and salts.
3032 *) Include RAND_status() into RAND_METHOD instead of implementing
3033 it only for md_rand.c Otherwise replacing the PRNG by calling
3034 RAND_set_rand_method would be impossible.
3037 *) Don't let DSA_generate_key() enter an infinite loop if the random
3038 number generation fails.
3041 *) New 'rand' application for creating pseudo-random output.
3044 *) Added configuration support for Linux/IA64
3045 [Rolf Haberrecker <rolf@suse.de>]
3047 *) Assembler module support for Mingw32.
3050 *) Shared library support for HPUX (in shlib/).
3051 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
3053 *) Shared library support for Solaris gcc.
3054 [Lutz Behnke <behnke@trustcenter.de>]
3056 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
3058 *) PKCS7_encrypt() was adding text MIME headers twice because they
3059 were added manually and by SMIME_crlf_copy().
3062 *) In bntest.c don't call BN_rand with zero bits argument.
3063 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
3065 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
3066 case was implemented. This caused BN_div_recp() to fail occasionally.
3069 *) Add an optional second argument to the set_label() in the perl
3070 assembly language builder. If this argument exists and is set
3071 to 1 it signals that the assembler should use a symbol whose
3072 scope is the entire file, not just the current function. This
3073 is needed with MASM which uses the format label:: for this scope.
3074 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
3076 *) Change the ASN1 types so they are typedefs by default. Before
3077 almost all types were #define'd to ASN1_STRING which was causing
3078 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
3082 *) Change names of new functions to the new get1/get0 naming
3083 convention: After 'get1', the caller owns a reference count
3084 and has to call ..._free; 'get0' returns a pointer to some
3085 data structure without incrementing reference counters.
3086 (Some of the existing 'get' functions increment a reference
3087 counter, some don't.)
3088 Similarly, 'set1' and 'add1' functions increase reference
3089 counters or duplicate objects.
3092 *) Allow for the possibility of temp RSA key generation failure:
3093 the code used to assume it always worked and crashed on failure.
3096 *) Fix potential buffer overrun problem in BIO_printf().
3097 [Ulf Möller, using public domain code by Patrick Powell; problem
3098 pointed out by David Sacerdote <das33@cornell.edu>]
3100 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
3101 RAND_egd() and RAND_status(). In the command line application,
3102 the EGD socket can be specified like a seed file using RANDFILE
3106 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
3107 Some CAs (e.g. Verisign) distribute certificates in this form.
3110 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
3111 list to exclude them. This means that no special compilation option
3112 is needed to use anonymous DH: it just needs to be included in the
3116 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
3117 EVP_MD_type. The old functionality is available in a new macro called
3118 EVP_MD_md(). Change code that uses it and update docs.
3121 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
3122 where the 'void *' argument is replaced by a function pointer argument.
3123 Previously 'void *' was abused to point to functions, which works on
3124 many platforms, but is not correct. As these functions are usually
3125 called by macros defined in OpenSSL header files, most source code
3126 should work without changes.
3129 *) <openssl/opensslconf.h> (which is created by Configure) now contains
3130 sections with information on -D... compiler switches used for
3131 compiling the library so that applications can see them. To enable
3132 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
3133 must be defined. E.g.,
3134 #define OPENSSL_ALGORITHM_DEFINES
3135 #include <openssl/opensslconf.h>
3136 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
3137 [Richard Levitte, Ulf and Bodo Möller]
3139 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
3143 *) Change the 'other' type in certificate aux info to a STACK_OF
3144 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
3145 the required ASN1 format: arbitrary types determined by an OID.
3148 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
3149 argument to 'req'. This is not because the function is newer or
3150 better than others it just uses the work 'NEW' in the certificate
3151 request header lines. Some software needs this.
3154 *) Reorganise password command line arguments: now passwords can be
3155 obtained from various sources. Delete the PEM_cb function and make
3156 it the default behaviour: i.e. if the callback is NULL and the
3157 usrdata argument is not NULL interpret it as a null terminated pass
3158 phrase. If usrdata and the callback are NULL then the pass phrase
3159 is prompted for as usual.
3162 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
3163 the support is automatically enabled. The resulting binaries will
3164 autodetect the card and use it if present.
3165 [Ben Laurie and Compaq Inc.]
3167 *) Work around for Netscape hang bug. This sends certificate request
3168 and server done in one record. Since this is perfectly legal in the
3169 SSL/TLS protocol it isn't a "bug" option and is on by default. See
3170 the bugs/SSLv3 entry for more info.
3173 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
3176 *) Add -rand argument to smime and pkcs12 applications and read/write
3180 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
3183 *) Add command line password options to the remaining applications.
3186 *) Bug fix for BN_div_recp() for numerators with an even number of
3190 *) More tests in bntest.c, and changed test_bn output.
3193 *) ./config recognizes MacOS X now.
3196 *) Bug fix for BN_div() when the first words of num and divsor are
3197 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
3200 *) Add support for various broken PKCS#8 formats, and command line
3201 options to produce them.
3204 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
3205 get temporary BIGNUMs from a BN_CTX.
3208 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
3212 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
3213 include a #define from the old name to the new. The original intent
3214 was that statically linked binaries could for example just call
3215 SSLeay_add_all_ciphers() to just add ciphers to the table and not
3216 link with digests. This never worked becayse SSLeay_add_all_digests()
3217 and SSLeay_add_all_ciphers() were in the same source file so calling
3218 one would link with the other. They are now in separate source files.
3221 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
3224 *) Use a less unusual form of the Miller-Rabin primality test (it used
3225 a binary algorithm for exponentiation integrated into the Miller-Rabin
3226 loop, our standard modexp algorithms are faster).
3229 *) Support for the EBCDIC character set completed.
3230 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
3232 *) Source code cleanups: use const where appropriate, eliminate casts,
3233 use void * instead of char * in lhash.
3236 *) Bugfix: ssl3_send_server_key_exchange was not restartable
3237 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
3238 this the server could overwrite ephemeral keys that the client
3242 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
3243 using 50 iterations of the Rabin-Miller test.
3245 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
3246 iterations of the Rabin-Miller test as required by the appendix
3247 to FIPS PUB 186[-1]) instead of DSA_is_prime.
3248 As BN_is_prime_fasttest includes trial division, DSA parameter
3249 generation becomes much faster.
3251 This implies a change for the callback functions in DSA_is_prime
3252 and DSA_generate_parameters: The callback function is called once
3253 for each positive witness in the Rabin-Miller test, not just
3254 occasionally in the inner loop; and the parameters to the
3255 callback function now provide an iteration count for the outer
3256 loop rather than for the current invocation of the inner loop.
3257 DSA_generate_parameters additionally can call the callback
3258 function with an 'iteration count' of -1, meaning that a
3259 candidate has passed the trial division test (when q is generated
3260 from an application-provided seed, trial division is skipped).
3263 *) New function BN_is_prime_fasttest that optionally does trial
3264 division before starting the Rabin-Miller test and has
3265 an additional BN_CTX * argument (whereas BN_is_prime always
3266 has to allocate at least one BN_CTX).
3267 'callback(1, -1, cb_arg)' is called when a number has passed the
3268 trial division stage.
3271 *) Fix for bug in CRL encoding. The validity dates weren't being handled
3275 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
3278 *) New function BN_pseudo_rand().
3281 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
3282 bignum version of BN_from_montgomery() with the working code from
3283 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
3287 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
3288 made it impossible to use the same SSL_SESSION data structure in
3289 SSL2 clients in multiple threads.
3292 *) The return value of RAND_load_file() no longer counts bytes obtained
3293 by stat(). RAND_load_file(..., -1) is new and uses the complete file
3294 to seed the PRNG (previously an explicit byte count was required).
3295 [Ulf Möller, Bodo Möller]
3297 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
3298 used (char *) instead of (void *) and had casts all over the place.
3301 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
3304 *) Retain source code compatibility for BN_prime_checks macro:
3305 BN_is_prime(..., BN_prime_checks, ...) now uses
3306 BN_prime_checks_for_size to determine the appropriate number of
3307 Rabin-Miller iterations.
3310 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
3311 DH_CHECK_P_NOT_SAFE_PRIME.
3312 (Check if this is true? OpenPGP calls them "strong".)
3315 *) Merge the functionality of "dh" and "gendh" programs into a new program
3316 "dhparam". The old programs are retained for now but will handle DH keys
3317 (instead of parameters) in future.
3320 *) Make the ciphers, s_server and s_client programs check the return values
3321 when a new cipher list is set.
3324 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
3325 ciphers. Before when the 56bit ciphers were enabled the sorting was
3328 The syntax for the cipher sorting has been extended to support sorting by
3329 cipher-strength (using the strength_bits hard coded in the tables).
3330 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
3332 Fix a bug in the cipher-command parser: when supplying a cipher command
3333 string with an "undefined" symbol (neither command nor alphanumeric
3334 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
3335 an error is flagged.
3337 Due to the strength-sorting extension, the code of the
3338 ssl_create_cipher_list() function was completely rearranged. I hope that
3339 the readability was also increased :-)
3340 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3342 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
3343 for the first serial number and places 2 in the serial number file. This
3344 avoids problems when the root CA is created with serial number zero and
3345 the first user certificate has the same issuer name and serial number
3349 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
3350 the new code. Add documentation for this stuff.
3353 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
3354 X509_*() to X509at_*() on the grounds that they don't handle X509
3355 structures and behave in an analagous way to the X509v3 functions:
3356 they shouldn't be called directly but wrapper functions should be used
3359 So we also now have some wrapper functions that call the X509at functions
3360 when passed certificate requests. (TO DO: similar things can be done with
3361 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
3362 things. Some of these need some d2i or i2d and print functionality
3363 because they handle more complex structures.)
3366 *) Add missing #ifndefs that caused missing symbols when building libssl
3367 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
3368 NO_RSA in ssl/s2*.c.
3369 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
3371 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
3372 has a return value which indicates the quality of the random data
3373 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
3374 error queue. New function RAND_pseudo_bytes() generates output that is
3375 guaranteed to be unique but not unpredictable. RAND_add is like
3376 RAND_seed, but takes an extra argument for an entropy estimate
3377 (RAND_seed always assumes full entropy).
3380 *) Do more iterations of Rabin-Miller probable prime test (specifically,
3381 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
3382 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
3383 in crypto/bn/bn_prime.c for the complete table). This guarantees a
3384 false-positive rate of at most 2^-80 for random input.
3387 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
3390 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
3391 in the 0.9.5 release), this returns the chain
3392 from an X509_CTX structure with a dup of the stack and all
3393 the X509 reference counts upped: so the stack will exist
3394 after X509_CTX_cleanup() has been called. Modify pkcs12.c
3397 Also make SSL_SESSION_print() print out the verify return
3401 *) Add manpage for the pkcs12 command. Also change the default
3402 behaviour so MAC iteration counts are used unless the new
3403 -nomaciter option is used. This improves file security and
3404 only older versions of MSIE (4.0 for example) need it.
3407 *) Honor the no-xxx Configure options when creating .DEF files.
3410 *) Add PKCS#10 attributes to field table: challengePassword,
3411 unstructuredName and unstructuredAddress. These are taken from
3412 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
3413 international characters are used.
3415 More changes to X509_ATTRIBUTE code: allow the setting of types
3416 based on strings. Remove the 'loc' parameter when adding
3417 attributes because these will be a SET OF encoding which is sorted
3421 *) Initial changes to the 'req' utility to allow request generation
3422 automation. This will allow an application to just generate a template
3423 file containing all the field values and have req construct the
3426 Initial support for X509_ATTRIBUTE handling. Stacks of these are
3427 used all over the place including certificate requests and PKCS#7
3428 structures. They are currently handled manually where necessary with
3429 some primitive wrappers for PKCS#7. The new functions behave in a
3430 manner analogous to the X509 extension functions: they allow
3431 attributes to be looked up by NID and added.
3433 Later something similar to the X509V3 code would be desirable to
3434 automatically handle the encoding, decoding and printing of the
3435 more complex types. The string types like challengePassword can
3436 be handled by the string table functions.
3438 Also modified the multi byte string table handling. Now there is
3439 a 'global mask' which masks out certain types. The table itself
3440 can use the flag STABLE_NO_MASK to ignore the mask setting: this
3441 is useful when for example there is only one permissible type
3442 (as in countryName) and using the mask might result in no valid
3446 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
3447 SSL_get_peer_finished to allow applications to obtain the latest
3448 Finished messages sent to the peer or expected from the peer,
3449 respectively. (SSL_get_peer_finished is usually the Finished message
3450 actually received from the peer, otherwise the protocol will be aborted.)
3452 As the Finished message are message digests of the complete handshake
3453 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
3454 be used for external authentication procedures when the authentication
3455 provided by SSL/TLS is not desired or is not enough.
3458 *) Enhanced support for Alpha Linux is added. Now ./config checks if
3459 the host supports BWX extension and if Compaq C is present on the
3460 $PATH. Just exploiting of the BWX extension results in 20-30%
3461 performance kick for some algorithms, e.g. DES and RC4 to mention
3462 a couple. Compaq C in turn generates ~20% faster code for MD5 and
3466 *) Add support for MS "fast SGC". This is arguably a violation of the
3467 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
3468 weak crypto and after checking the certificate is SGC a second one
3469 with strong crypto. MS SGC stops the first handshake after receiving
3470 the server certificate message and sends a second client hello. Since
3471 a server will typically do all the time consuming operations before
3472 expecting any further messages from the client (server key exchange
3473 is the most expensive) there is little difference between the two.
3475 To get OpenSSL to support MS SGC we have to permit a second client
3476 hello message after we have sent server done. In addition we have to
3477 reset the MAC if we do get this second client hello.
3480 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
3481 if a DER encoded private key is RSA or DSA traditional format. Changed
3482 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
3483 format DER encoded private key. Newer code should use PKCS#8 format which
3484 has the key type encoded in the ASN1 structure. Added DER private key
3485 support to pkcs8 application.
3488 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
3489 ciphersuites has been selected (as required by the SSL 3/TLS 1
3490 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
3491 is set, we interpret this as a request to violate the specification
3492 (the worst that can happen is a handshake failure, and 'correct'
3493 behaviour would result in a handshake failure anyway).
3496 *) In SSL_CTX_add_session, take into account that there might be multiple
3497 SSL_SESSION structures with the same session ID (e.g. when two threads
3498 concurrently obtain them from an external cache).
3499 The internal cache can handle only one SSL_SESSION with a given ID,
3500 so if there's a conflict, we now throw out the old one to achieve
3504 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
3505 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
3506 some routines that use cipher OIDs: some ciphers do not have OIDs
3507 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
3511 *) Simplify the trust setting structure and code. Now we just have
3512 two sequences of OIDs for trusted and rejected settings. These will
3513 typically have values the same as the extended key usage extension
3514 and any application specific purposes.
3516 The trust checking code now has a default behaviour: it will just
3517 check for an object with the same NID as the passed id. Functions can
3518 be provided to override either the default behaviour or the behaviour
3519 for a given id. SSL client, server and email already have functions
3520 in place for compatibility: they check the NID and also return "trusted"
3521 if the certificate is self signed.
3524 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
3525 traditional format into an EVP_PKEY structure.
3528 *) Add a password callback function PEM_cb() which either prompts for
3529 a password if usr_data is NULL or otherwise assumes it is a null
3530 terminated password. Allow passwords to be passed on command line
3531 environment or config files in a few more utilities.
3534 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
3535 keys. Add some short names for PKCS#8 PBE algorithms and allow them
3536 to be specified on the command line for the pkcs8 and pkcs12 utilities.
3537 Update documentation.
3540 *) Support for ASN1 "NULL" type. This could be handled before by using
3541 ASN1_TYPE but there wasn't any function that would try to read a NULL
3542 and produce an error if it couldn't. For compatibility we also have
3543 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
3544 don't allocate anything because they don't need to.
3547 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
3549 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
3551 *) Rebuild of the memory allocation routines used by OpenSSL code and
3552 possibly others as well. The purpose is to make an interface that
3553 provide hooks so anyone can build a separate set of allocation and
3554 deallocation routines to be used by OpenSSL, for example memory
3555 pool implementations, or something else, which was previously hard
3556 since Malloc(), Realloc() and Free() were defined as macros having
3557 the values malloc, realloc and free, respectively (except for Win32
3558 compilations). The same is provided for memory debugging code.
3559 OpenSSL already comes with functionality to find memory leaks, but
3560 this gives people a chance to debug other memory problems.
3562 With these changes, a new set of functions and macros have appeared:
3564 CRYPTO_set_mem_debug_functions() [F]
3565 CRYPTO_get_mem_debug_functions() [F]
3566 CRYPTO_dbg_set_options() [F]
3567 CRYPTO_dbg_get_options() [F]
3568 CRYPTO_malloc_debug_init() [M]
3570 The memory debug functions are NULL by default, unless the library
3571 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
3572 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
3573 gives the standard debugging functions that come with OpenSSL) or
3574 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
3575 provided by the library user) must be used. When the standard
3576 debugging functions are used, CRYPTO_dbg_set_options can be used to
3577 request additional information:
3578 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
3579 the CRYPTO_MDEBUG_xxx macro when compiling the library.
3581 Also, things like CRYPTO_set_mem_functions will always give the
3582 expected result (the new set of functions is used for allocation
projects / openssl.git / blob