5 Changes between 0.9.6 and 0.9.7 [xx XXX 2001]
7 OpenSSL 0.9.6a/0.9.6b (bugfix releases, 5 Apr 2001 and 9 July 2001)
8 and OpenSSL 0.9.7 were developed in parallel, based on OpenSSL 0.9.6.
10 Change log entries are tagged as follows:
11 -) applies to 0.9.6a/0.9.6b/0.9.6c only
12 *) applies to 0.9.6a/0.9.6b/0.9.6c and 0.9.7
13 +) applies to 0.9.7 only
15 +) Add -keyform to rsautl, and document -engine.
16 [Richard Levitte, inspired by Toomas Kiisk <vix@cyber.ee>]
18 +) Change BIO_new_file (crypto/bio/bss_file.c) to use new
19 BIO_R_NO_SUCH_FILE error code rather than the generic
20 ERR_R_SYS_LIB error code if fopen() fails with ENOENT.
25 ERR_peek_last_error_line
26 ERR_peek_last_error_line_data.
30 ERR_peek_error_line_data,
31 but report on the latest error recorded rather than the first one
32 still in the error queue.
33 [Ben Laurie, Bodo Moeller]
35 +) default_algorithms option in ENGINE config module. This allows things
37 default_algorithms = ALL
38 default_algorithms = RSA, DSA, RAND, CIPHERS, DIGESTS
41 +) Prelminary ENGINE config module.
44 *) The earlier bugfix for the SSL3_ST_SW_HELLO_REQ_C case of
45 ssl3_accept (ssl/s3_srvr.c) incorrectly used a local flag
46 variable as an indication that a ClientHello message has been
47 received. As the flag value will be lost between multiple
48 invocations of ssl3_accept when using non-blocking I/O, the
49 function may not be aware that a handshake has actually taken
50 place, thus preventing a new session from being added to the
53 To avoid this problem, we now set s->new_session to 2 instead of
54 using a local variable.
55 [Lutz Jaenicke, Bodo Moeller]
57 *) Bugfix: Return -1 from ssl3_get_server_done (ssl3/s3_clnt.c)
58 if the SSL_R_LENGTH_MISMATCH error is detected.
59 [Geoff Thorpe, Bodo Moeller]
61 +) New experimental application configuration code.
64 *) New 'shared_ldflag' column in Configure platform table.
67 *) Fix EVP_CIPHER_mode macro.
68 ["Dan S. Camper" <dan@bti.net>]
70 +) Change the AES code to follow the same name structure as all other
71 symmetric ciphers, and behave the same way. Move everything to
72 the directory crypto/aes, thereby obsoleting crypto/rijndael.
73 [Stephen Sprunk <stephen@sprunk.org> and Richard Levitte]
75 *) Fix ssl3_read_bytes (ssl/s3_pkt.c): To ignore messages of unknown
76 type, we must throw them away by setting rr->length to 0.
77 [D P Chang <dpc@qualys.com>]
79 -) OpenSSL 0.9.6c released [21 dec 2001]
81 +) SECURITY: remove unsafe setjmp/signal interaction from ui_openssl.c.
82 [Ben Laurie and Theo de Raadt]
84 *) Fix BN_rand_range bug pointed out by Dominikus Scherkl
85 <Dominikus.Scherkl@biodata.com>. (The previous implementation
86 worked incorrectly for those cases where range = 10..._2 and
87 3*range is two bits longer than range.)
90 *) Only add signing time to PKCS7 structures if it is not already
94 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
95 OBJ_ld_ce should be OBJ_id_ce.
96 Also some ip-pda OIDs in crypto/objects/objects.txt were
97 incorrect (cf. RFC 3039).
98 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
100 +) Add option to output public keys in req command.
101 [Massimiliano Pala madwolf@openca.org]
103 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
104 returns early because it has nothing to do.
105 [Andy Schneider <andy.schneider@bjss.co.uk>]
107 *) [In 0.9.6c-engine and 0.9.7 release:]
108 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
109 [Andy Schneider <andy.schneider@bjss.co.uk>]
111 -) [In 0.9.6c-engine release:]
112 Add support for Cryptographic Appliance's keyserver technology.
113 (Use engine 'keyclient')
114 [Cryptographic Appliances and Geoff Thorpe]
116 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
117 is called via tools/c89.sh because arguments have to be
118 rearranged (all '-L' options must appear before the first object
120 [Richard Shapiro <rshapiro@abinitio.com>]
122 +) Use wNAFs in EC_POINTs_mul() for improved efficiency
123 (up to about 10% better than before for P-192 and P-224).
126 -) [In 0.9.6c-engine release:]
127 Add support for Broadcom crypto accelerator cards, backported
129 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
131 -) [In 0.9.6c-engine release:]
132 Add support for SureWare crypto accelerator cards from
133 Baltimore Technologies. (Use engine 'sureware')
134 [Baltimore Technologies and Mark Cox]
136 -) [In 0.9.6c-engine release:]
137 Add support for crypto accelerator cards from Accelerated
138 Encryption Processing, www.aep.ie. (Use engine 'aep')
139 [AEP Inc. and Mark Cox]
141 *) Add a configuration entry for gcc on UnixWare.
142 [Gary Benson <gbenson@redhat.com>]
144 +) New functions/macros
146 SSL_CTX_set_msg_callback(ctx, cb)
147 SSL_CTX_set_msg_callback_arg(ctx, arg)
148 SSL_set_msg_callback(ssl, cb)
149 SSL_set_msg_callback_arg(ssl, arg)
151 to request calling a callback function
153 void cb(int write_p, int version, int content_type,
154 const void *buf, size_t len, SSL *ssl, void *arg)
156 whenever a protocol message has been completely received
157 (write_p == 0) or sent (write_p == 1). Here 'version' is the
158 protocol version according to which the SSL library interprets
159 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
160 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
161 the content type as defined in the SSL 3.0/TLS 1.0 protocol
162 specification (change_cipher_spec(20), alert(21), handshake(22)).
163 'buf' and 'len' point to the actual message, 'ssl' to the
164 SSL object, and 'arg' is the application-defined value set by
165 SSL[_CTX]_set_msg_callback_arg().
167 'openssl s_client' and 'openssl s_server' have new '-msg' options
168 to enable a callback that displays all protocol messages.
171 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
172 messages are stored in a single piece (fixed-length part and
173 variable-length part combined) and fix various bugs found on the way.
176 +) Change the shared library support so shared libraries are built as
177 soon as the corresponding static library is finished, and thereby get
178 openssl and the test programs linked against the shared library.
179 This still only happens when the keyword "shard" has been given to
180 the configuration scripts.
182 NOTE: shared library support is still an experimental thing, and
183 backward binary compatibility is still not guaranteed.
184 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
186 +) Add support for Subject Information Access extension.
187 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
189 +) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
190 additional bytes when new memory had to be allocated, not just
191 when reusing an existing buffer.
194 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
195 instead. BIO_gethostbyname() does not know what timeouts are
196 appropriate, so entries would stay in cache even when they have
198 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
200 +) New command line and configuration option 'utf8' for the req command.
201 This allows field values to be specified as UTF8 strings.
204 +) Add -multi and -mr options to "openssl speed" - giving multiple parallel
205 runs for the former and machine-readable output for the latter.
208 +) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
209 of the e-mail address in the DN (i.e., it will go into a certificate
210 extension only). The new configuration file option 'email_in_dn = no'
212 [Massimiliano Pala madwolf@openca.org]
214 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
215 faced with a pathologically small ClientHello fragment that does
216 not contain client_version: Instead of aborting with an error,
217 simply choose the highest available protocol version (i.e.,
218 TLS 1.0 unless it is disabled). In practice, ClientHello
219 messages are never sent like this, but this change gives us
220 strictly correct behaviour at least for TLS.
223 +) Change all functions with names starting with des_ to be starting
224 with DES_ instead. Add wrappers that are compatible with libdes,
225 but are named _old_des_*. Finally, add macros that map the des_*
226 symbols to the corresponding _old_des_*.
228 All this is done because there are increasing clashes with libdes
229 and other DES libraries that are currently used by other projects.
230 The old libdes interface (including crypt()) is provided if
231 <openssl/des_old.h> is included. For now, this automatically
232 happens in <openssl/des.h> unless OPENSSL_DISABLE_OLD_DES_SUPPORT is
233 defined. Note that crypt() is no longer declared in <openssl/des.h>.
235 NOTE: This is a major break of an old API into a new one. Software
236 authors are encouraged to switch to the DES_ style functions. Some
237 time in the future, des_old.h and the libdes compatibility functions
238 will be completely removed.
241 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
242 never resets s->method to s->ctx->method when called from within
243 one of the SSL handshake functions.
244 [Bodo Moeller; problem pointed out by Niko Baric]
246 +) Test for certificates which contain unsupported critical extensions.
247 If such a certificate is found during a verify operation it is
248 rejected by default: this behaviour can be overridden by either
249 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
250 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
251 X509_supported_extension() has also been added which returns 1 if a
252 particular extension is supported.
255 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
256 (sent using the client's version number) if client_version is
257 smaller than the protocol version in use. Also change
258 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
259 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
260 the client will at least see that alert.
263 +) Modify the behaviour of EVP cipher functions in similar way to digests
264 to retain compatibility with existing code.
267 +) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
268 compatibility with existing code. In particular the 'ctx' parameter does
269 not have to be to be initialized before the call to EVP_DigestInit() and
270 it is tidied up after a call to EVP_DigestFinal(). New function
271 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
272 EVP_MD_CTX_copy() changed to not require the destination to be
273 initialized valid and new function EVP_MD_CTX_copy_ex() added which
274 requires the destination to be valid.
276 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
277 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
280 +) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
281 so that complete 'Handshake' protocol structures are kept in memory
282 instead of overwriting 'msg_type' and 'length' with 'body' data.
285 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
289 +) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
290 [Massimo Santin via Richard Levitte]
292 +) Major restructuring to the underlying ENGINE code. This includes
293 reduction of linker bloat, separation of pure "ENGINE" manipulation
294 (initialisation, etc) from functionality dealing with implementations
295 of specific crypto iterfaces. This change also introduces integrated
296 support for symmetric ciphers and digest implementations - so ENGINEs
297 can now accelerate these by providing EVP_CIPHER and EVP_MD
298 implementations of their own. This is detailed in crypto/engine/README
299 as it couldn't be adequately described here. However, there are a few
300 API changes worth noting - some RSA, DSA, DH, and RAND functions that
301 were changed in the original introduction of ENGINE code have now
302 reverted back - the hooking from this code to ENGINE is now a good
303 deal more passive and at run-time, operations deal directly with
304 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
305 dereferencing through an ENGINE pointer any more. Also, the ENGINE
306 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
307 they were not being used by the framework as there is no concept of a
308 BIGNUM_METHOD and they could not be generalised to the new
309 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
310 ENGINE_cpy() has been removed as it cannot be consistently defined in
314 +) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
317 +) Change mkdef.pl to sort symbols that get the same entry number,
318 and make sure the automatically generated functions ERR_load_*
319 become part of libeay.num as well.
322 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
323 client receives HelloRequest while in a handshake.
324 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
326 +) New function SSL_renegotiate_pending(). This returns true once
327 renegotiation has been requested (either SSL_renegotiate() call
328 or HelloRequest/ClientHello receveived from the peer) and becomes
329 false once a handshake has been completed.
330 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
331 sends a HelloRequest, but does not ensure that a handshake takes
332 place. SSL_renegotiate_pending() is useful for checking if the
333 client has followed the request.)
336 +) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
337 By default, clients may request session resumption even during
338 renegotiation (if session ID contexts permit); with this option,
339 session resumption is possible only in the first handshake.
342 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
343 should end in 'break', not 'goto end' which circuments various
344 cleanups done in state SSL_ST_OK. But session related stuff
345 must be disabled for SSL_ST_OK in the case that we just sent a
348 Also avoid some overhead by not calling ssl_init_wbio_buffer()
349 before just sending a HelloRequest.
350 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
352 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
353 reveal whether illegal block cipher padding was found or a MAC
354 verification error occured. (Neither SSLerr() codes nor alerts
355 are directly visible to potential attackers, but the information
356 may leak via logfiles.)
358 Similar changes are not required for the SSL 2.0 implementation
359 because the number of padding bytes is sent in clear for SSL 2.0,
360 and the extra bytes are just ignored. However ssl/s2_pkt.c
361 failed to verify that the purported number of padding bytes is in
365 +) Add some demos for certificate and certificate request creation.
368 +) Make maximum certificate chain size accepted from the peer application
369 settable (SSL*_get/set_max_cert_list()), as proposed by
370 "Douglas E. Engert" <deengert@anl.gov>.
373 +) Add support for shared libraries for Unixware-7
374 (Boyd Lynn Gerber <gerberb@zenez.com>).
377 *) Add OpenUNIX-8 support including shared libraries
378 (Boyd Lynn Gerber <gerberb@zenez.com>).
381 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
382 'wristwatch attack' using huge encoding parameters (cf.
383 James H. Manger's CRYPTO 2001 paper). Note that the
384 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
385 encoding parameters and hence was not vulnerable.
388 +) Add a "destroy" handler to ENGINEs that allows structural cleanup to
389 be done prior to destruction. Use this to unload error strings from
390 ENGINEs that load their own error strings. NB: This adds two new API
391 functions to "get" and "set" this destroy handler in an ENGINE.
394 +) Alter all existing ENGINE implementations (except "openssl" and
395 "openbsd") to dynamically instantiate their own error strings. This
396 makes them more flexible to be built both as statically-linked ENGINEs
397 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
398 Also, add stub code to each that makes building them as self-contained
399 shared-libraries easier (see README.ENGINE).
402 +) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
403 implementations into applications that are completely implemented in
404 self-contained shared-libraries. The "dynamic" ENGINE exposes control
405 commands that can be used to configure what shared-library to load and
406 to control aspects of the way it is handled. Also, made an update to
407 the README.ENGINE file that brings its information up-to-date and
408 provides some information and instructions on the "dynamic" ENGINE
409 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
413 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
415 +) Make it possible to unload ranges of ERR strings with a new
416 "ERR_unload_strings" function.
419 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
420 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
421 followed by modular reduction.
422 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
424 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
425 equivalent based on BN_pseudo_rand() instead of BN_rand().
428 +) Add a copy() function to EVP_MD.
431 +) Make EVP_MD routines take a context pointer instead of just the
432 md_data void pointer.
435 +) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
436 that the digest can only process a single chunk of data
437 (typically because it is provided by a piece of
438 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
439 is only going to provide a single chunk of data, and hence the
440 framework needn't accumulate the data for oneshot drivers.
443 +) As with "ERR", make it possible to replace the underlying "ex_data"
444 functions. This change also alters the storage and management of global
445 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
446 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
447 index counters. The API functions that use this state have been changed
448 to take a "class_index" rather than pointers to the class's local STACK
449 and counter, and there is now an API function to dynamically create new
450 classes. This centralisation allows us to (a) plug a lot of the
451 thread-safety problems that existed, and (b) makes it possible to clean
452 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
453 such data would previously have always leaked in application code and
454 workarounds were in place to make the memory debugging turn a blind eye
455 to it. Application code that doesn't use this new function will still
456 leak as before, but their memory debugging output will announce it now
457 rather than letting it slide.
459 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
460 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
461 has a return value to indicate success or failure.
464 +) Make it possible to replace the underlying "ERR" functions such that the
465 global state (2 LHASH tables and 2 locks) is only used by the "default"
466 implementation. This change also adds two functions to "get" and "set"
467 the implementation prior to it being automatically set the first time
468 any other ERR function takes place. Ie. an application can call "get",
469 pass the return value to a module it has just loaded, and that module
470 can call its own "set" function using that value. This means the
471 module's "ERR" operations will use (and modify) the error state in the
472 application and not in its own statically linked copy of OpenSSL code.
475 +) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
476 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
477 the operation, and provides a more encapsulated way for external code
478 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
479 to use these functions rather than manually incrementing the counts.
481 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
484 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
485 This function was broken, as the check for a new client hello message
486 to handle SGC did not allow these large messages.
487 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
490 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
493 +) Add EVP test program.
496 +) Add symmetric cipher support to ENGINE. Expect the API to change!
499 +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
500 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
501 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
502 These allow a CRL to be built without having to access X509_CRL fields
503 directly. Modify 'ca' application to use new functions.
506 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
507 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
510 *) Rework the configuration and shared library support for Tru64 Unix.
511 The configuration part makes use of modern compiler features and
512 still retains old compiler behavior for those that run older versions
513 of the OS. The shared library support part includes a variant that
514 uses the RPATH feature, and is available through the special
515 configuration target "alpha-cc-rpath", which will never be selected
517 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
519 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
520 with the same message size as in ssl3_get_certificate_request().
521 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
522 messages might inadvertently be reject as too long.
523 [Petr Lampa <lampa@fee.vutbr.cz>]
525 +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
526 bug workarounds. Rollback attack detection is a security feature.
527 The problem will only arise on OpenSSL servers when TLSv1 is not
528 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
529 Software authors not wanting to support TLSv1 will have special reasons
530 for their choice and can explicitly enable this option.
531 [Bodo Moeller, Lutz Jaenicke]
533 +) Rationalise EVP so it can be extended: don't include a union of
534 cipher/digest structures, add init/cleanup functions. This also reduces
535 the number of header dependencies.
538 +) Make DES key schedule conform to the usual scheme, as well as
539 correcting its structure. This means that calls to DES functions
540 now have to pass a pointer to a des_key_schedule instead of a
541 plain des_key_schedule (which was actually always a pointer
545 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
548 *) Modified SSL library such that the verify_callback that has been set
549 specificly for an SSL object with SSL_set_verify() is actually being
550 used. Before the change, a verify_callback set with this function was
551 ignored and the verify_callback() set in the SSL_CTX at the time of
552 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
553 to allow the necessary settings.
556 +) Initial reduction of linker bloat: the use of some functions, such as
557 PEM causes large amounts of unused functions to be linked in due to
558 poor organisation. For example pem_all.c contains every PEM function
559 which has a knock on effect of linking in large amounts of (unused)
560 ASN1 code. Grouping together similar functions and splitting unrelated
561 functions prevents this.
564 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
565 explicitly to NULL, as at least on Solaris 8 this seems not always to be
566 done automatically (in contradiction to the requirements of the C
567 standard). This made problems when used from OpenSSH.
570 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
571 dh->length and always used
573 BN_rand_range(priv_key, dh->p).
575 BN_rand_range() is not necessary for Diffie-Hellman, and this
576 specific range makes Diffie-Hellman unnecessarily inefficient if
577 dh->length (recommended exponent length) is much smaller than the
578 length of dh->p. We could use BN_rand_range() if the order of
579 the subgroup was stored in the DH structure, but we only have
584 BN_rand(priv_key, l, ...)
586 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
592 RSA_eay_public_encrypt
593 RSA_eay_private_decrypt
594 RSA_eay_private_encrypt (signing)
595 RSA_eay_public_decrypt (signature verification)
597 (default implementations for RSA_public_encrypt,
598 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
599 always reject numbers >= n.
602 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
603 to synchronize access to 'locking_thread'. This is necessary on
604 systems where access to 'locking_thread' (an 'unsigned long'
605 variable) is not atomic.
608 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
609 *before* setting the 'crypto_lock_rand' flag. The previous code had
610 a race condition if 0 is a valid thread ID.
611 [Travis Vitek <vitek@roguewave.com>]
613 +) Cleanup of EVP macros.
616 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
620 +) Add initial OCSP responder support to ocsp application. The
621 revocation information is handled using the text based index
622 use by the ca application. The responder can either handle
623 requests generated internally, supplied in files (for example
624 via a CGI script) or using an internal minimal server.
627 +) Add configuration choices to get zlib compression for TLS.
630 +) Changes to Kerberos SSL for RFC 2712 compliance:
631 1. Implemented real KerberosWrapper, instead of just using
632 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
633 2. Implemented optional authenticator field of KerberosWrapper.
635 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
636 and authenticator structs; see crypto/krb5/.
638 Generalized Kerberos calls to support multiple Kerberos libraries.
639 [Vern Staats <staatsvr@asc.hpc.mil>,
640 Jeffrey Altman <jaltman@columbia.edu>
643 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
644 already does with RSA. testdsa.h now has 'priv_key/pub_key'
645 values for each of the key sizes rather than having just
646 parameters (and 'speed' generating keys each time).
649 *) Add support for shared libraries under Irix.
650 [Albert Chin-A-Young <china@thewrittenword.com>]
652 *) Add configuration option to build on Linux on both big-endian and
654 [Ralf Baechle <ralf@uni-koblenz.de>]
656 *) Add the possibility to create shared libraries on HP-UX.
659 -) OpenSSL 0.9.6b released [9 July 2001]
661 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
662 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
663 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
664 PRNG state recovery was possible based on the output of
665 one PRNG request appropriately sized to gain knowledge on
666 'md' followed by enough consecutive 1-byte PRNG requests
667 to traverse all of 'state'.
669 1. When updating 'md_local' (the current thread's copy of 'md')
670 during PRNG output generation, hash all of the previous
671 'md_local' value, not just the half used for PRNG output.
673 2. Make the number of bytes from 'state' included into the hash
674 independent from the number of PRNG bytes requested.
676 The first measure alone would be sufficient to avoid
677 Markku-Juhani's attack. (Actually it had never occurred
678 to me that the half of 'md_local' used for chaining was the
679 half from which PRNG output bytes were taken -- I had always
680 assumed that the secret half would be used.) The second
681 measure makes sure that additional data from 'state' is never
682 mixed into 'md_local' in small portions; this heuristically
683 further strengthens the PRNG.
686 +) Speed up EVP routines.
689 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
690 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
691 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
692 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
694 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
695 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
696 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
699 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
701 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
704 *) Fix crypto/bn/asm/mips3.s.
707 *) When only the key is given to "enc", the IV is undefined. Print out
708 an error message in this case.
711 +) Added the OS2-EMX target.
712 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
714 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
715 to support NCONF routines in extension code. New function CONF_set_nconf()
716 to allow functions which take an NCONF to also handle the old LHASH
717 structure: this means that the old CONF compatible routines can be
718 retained (in particular wrt extensions) without having to duplicate the
719 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
722 *) Handle special case when X509_NAME is empty in X509 printing routines.
725 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
726 positive and less than q.
729 +) Enhance the general user interface with mechanisms for inner control
730 and with possibilities to have yes/no kind of prompts.
733 +) Change all calls to low level digest routines in the library and
734 applications to use EVP. Add missing calls to HMAC_cleanup() and
735 don't assume HMAC_CTX can be copied using memcpy().
736 [Verdon Walker <VWalker@novell.com>, Steve Henson]
738 +) Add the possibility to control engines through control names but with
739 arbitrary arguments instead of just a string.
740 Change the key loaders to take a UI_METHOD instead of a callback
741 function pointer. NOTE: this breaks binary compatibility with earlier
742 versions of OpenSSL [engine].
743 Adapt the nCipher code for these new conditions and add a card insertion
747 +) Enhance the general user interface with mechanisms to better support
748 dialog box interfaces, application-defined prompts, the possibility
749 to use defaults (for example default passwords from somewhere else)
750 and interrupts/cancellations.
753 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
754 used: it isn't thread safe and the add_lock_callback should handle
756 [Paul Rose <Paul.Rose@bridge.com>]
758 *) Verify that incoming data obeys the block size in
759 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
762 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
763 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
767 [Ulf Möller, Bodo Möller]
769 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
770 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
771 when fixing the server behaviour for backwards-compatible 'client
772 hello' messages. (Note that the attack is impractical against
773 SSL 3.0 and TLS 1.0 anyway because length and version checking
774 means that the probability of guessing a valid ciphertext is
775 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
778 Before 0.9.5, the countermeasure (hide the error by generating a
779 random 'decryption result') did not work properly because
780 ERR_clear_error() was missing, meaning that SSL_get_error() would
781 detect the supposedly ignored error.
783 Both problems are now fixed.
786 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
787 (previously it was 1024).
790 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
791 tidy up some unnecessarily weird code in 'sk_new()').
792 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
794 +) Change the key loading routines for ENGINEs to use the same kind
795 callback (pem_password_cb) as all other routines that need this
799 *) Fix for compatibility mode trust settings: ignore trust settings
800 unless some valid trust or reject settings are present.
803 *) Fix for blowfish EVP: its a variable length cipher.
806 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
807 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
808 than this minimum value is recommended.
811 +) New random seeder for OpenVMS, using the system process statistics
812 that are easily reachable.
815 +) Windows apparently can't transparently handle global
816 variables defined in DLLs. Initialisations such as:
818 const ASN1_ITEM *it = &ASN1_INTEGER_it;
820 wont compile. This is used by the any applications that need to
821 declare their own ASN1 modules. This was fixed by adding the option
822 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
823 needed for static libraries under Win32.
826 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
827 setting of purpose and trust fields. New X509_STORE trust and
828 purpose functions and tidy up setting in other SSL functions.
831 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
832 structure. These are inherited by X509_STORE_CTX when it is
833 initialised. This allows various defaults to be set in the
834 X509_STORE structure (such as flags for CRL checking and custom
835 purpose or trust settings) for functions which only use X509_STORE_CTX
836 internally such as S/MIME.
838 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
839 trust settings if they are not set in X509_STORE. This allows X509_STORE
840 purposes and trust (in S/MIME for example) to override any set by default.
842 Add command line options for CRL checking to smime, s_client and s_server
846 +) Initial CRL based revocation checking. If the CRL checking flag(s)
847 are set then the CRL is looked up in the X509_STORE structure and
848 its validity and signature checked, then if the certificate is found
849 in the CRL the verify fails with a revoked error.
851 Various new CRL related callbacks added to X509_STORE_CTX structure.
853 Command line options added to 'verify' application to support this.
855 This needs some additional work, such as being able to handle multiple
856 CRLs with different times, extension based lookup (rather than just
857 by subject name) and ultimately more complete V2 CRL extension
861 +) Add a general user interface API (crypto/ui/). This is designed
862 to replace things like des_read_password and friends (backward
863 compatibility functions using this new API are provided).
864 The purpose is to remove prompting functions from the DES code
865 section as well as provide for prompting through dialog boxes in
866 a window system and the like.
869 *) Fix various bugs related to DSA S/MIME verification. Handle missing
870 parameters in DSA public key structures and return an error in the
871 DSA routines if parameters are absent.
874 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
875 in the current directory if neither $RANDFILE nor $HOME was set.
876 RAND_file_name() in 0.9.6a returned NULL in this case. This has
877 caused some confusion to Windows users who haven't defined $HOME.
878 Thus RAND_file_name() is changed again: e_os.h can define a
879 DEFAULT_HOME, which will be used if $HOME is not set.
880 For Windows, we use "C:"; on other platforms, we still require
881 environment variables.
883 +) Add "ex_data" support to ENGINE so implementations can add state at a
884 per-structure level rather than having to store it globally.
887 +) Make it possible for ENGINE structures to be copied when retrieved by
888 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
889 This causes the "original" ENGINE structure to act like a template,
890 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
891 operational state can be localised to each ENGINE structure, despite the
892 fact they all share the same "methods". New ENGINE structures returned in
893 this case have no functional references and the return value is the single
894 structural reference. This matches the single structural reference returned
895 by ENGINE_by_id() normally, when it is incremented on the pre-existing
899 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
900 needs to match any other type at all we need to manually clear the
904 +) Changes to the "openssl engine" utility to include;
905 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
906 about an ENGINE's available control commands.
907 - executing control commands from command line arguments using the
908 '-pre' and '-post' switches. '-post' is only used if '-t' is
909 specified and the ENGINE is successfully initialised. The syntax for
910 the individual commands are colon-separated, for example;
911 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
914 +) New dynamic control command support for ENGINEs. ENGINEs can now
915 declare their own commands (numbers), names (strings), descriptions,
916 and input types for run-time discovery by calling applications. A
917 subset of these commands are implicitly classed as "executable"
918 depending on their input type, and only these can be invoked through
919 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
920 can be based on user input, config files, etc). The distinction is
921 that "executable" commands cannot return anything other than a boolean
922 result and can only support numeric or string input, whereas some
923 discoverable commands may only be for direct use through
924 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
925 pointers, or other custom uses. The "executable" commands are to
926 support parameterisations of ENGINE behaviour that can be
927 unambiguously defined by ENGINEs and used consistently across any
928 OpenSSL-based application. Commands have been added to all the
929 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
930 control over shared-library paths without source code alterations.
933 +) Changed all ENGINE implementations to dynamically allocate their
934 ENGINEs rather than declaring them statically. Apart from this being
935 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
936 this also allows the implementations to compile without using the
937 internal engine_int.h header.
940 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
941 'const' value. Any code that should be able to modify a RAND_METHOD
942 should already have non-const pointers to it (ie. they should only
943 modify their own ones).
946 +) Made a variety of little tweaks to the ENGINE code.
947 - "atalla" and "ubsec" string definitions were moved from header files
948 to C code. "nuron" string definitions were placed in variables
949 rather than hard-coded - allowing parameterisation of these values
950 later on via ctrl() commands.
951 - Removed unused "#if 0"'d code.
952 - Fixed engine list iteration code so it uses ENGINE_free() to release
953 structural references.
954 - Constified the RAND_METHOD element of ENGINE structures.
955 - Constified various get/set functions as appropriate and added
956 missing functions (including a catch-all ENGINE_cpy that duplicates
957 all ENGINE values onto a new ENGINE except reference counts/state).
958 - Removed NULL parameter checks in get/set functions. Setting a method
959 or function to NULL is a way of cancelling out a previously set
960 value. Passing a NULL ENGINE parameter is just plain stupid anyway
961 and doesn't justify the extra error symbols and code.
962 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
963 flags from engine_int.h to engine.h.
964 - Changed prototypes for ENGINE handler functions (init(), finish(),
965 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
968 *) Move 'if (!initialized) RAND_poll()' into regions protected by
969 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
970 having multiple threads call RAND_poll() concurrently.
973 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
974 combination of a flag and a thread ID variable.
975 Otherwise while one thread is in ssleay_rand_bytes (which sets the
976 flag), *other* threads can enter ssleay_add_bytes without obeying
977 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
978 that they do not hold after the first thread unsets add_do_not_lock).
981 +) Implement binary inversion algorithm for BN_mod_inverse in addition
982 to the algorithm using long division. The binary algorithm can be
983 used only if the modulus is odd. On 32-bit systems, it is faster
984 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
985 roughly 5-15% for 256-bit moduli), so we use it only for moduli
986 up to 450 bits. In 64-bit environments, the binary algorithm
987 appears to be advantageous for much longer moduli; here we use it
988 for moduli up to 2048 bits.
991 *) Change bctest again: '-x' expressions are not available in all
995 -) OpenSSL 0.9.6a released [5 Apr 2001]
997 *) Fix a couple of memory leaks in PKCS7_dataDecode()
998 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
1000 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
1001 the default extension for executables, if any. Also, make the perl
1002 scripts that use symlink() to test if it really exists and use "cp"
1003 if it doesn't. All this made OpenSSL compilable and installable in
1007 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
1008 could not support the combine flag in choice fields.
1011 -) Fix for asn1_GetSequence() for indefinite length constructed data.
1012 If SEQUENCE is length is indefinite just set c->slen to the total
1013 amount of data available.
1014 [Steve Henson, reported by shige@FreeBSD.org]
1015 [This change does not apply to 0.9.7.]
1017 *) Change bctest to avoid here-documents inside command substitution
1018 (workaround for FreeBSD /bin/sh bug).
1019 For compatibility with Ultrix, avoid shell functions (introduced
1020 in the bctest version that searches along $PATH).
1023 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
1024 with des_encrypt() defined on some operating systems, like Solaris
1028 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
1029 On the Importance of Eliminating Errors in Cryptographic
1030 Computations, J. Cryptology 14 (2001) 2, 101-119,
1031 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
1034 *) MIPS assembler BIGNUM division bug fix.
1037 *) Disabled incorrect Alpha assembler code.
1040 -) Fix PKCS#7 decode routines so they correctly update the length
1041 after reading an EOC for the EXPLICIT tag.
1043 [This change does not apply to 0.9.7.]
1045 *) Fix bug in PKCS#12 key generation routines. This was triggered
1046 if a 3DES key was generated with a 0 initial byte. Include
1047 PKCS12_BROKEN_KEYGEN compilation option to retain the old
1048 (but broken) behaviour.
1051 *) Enhance bctest to search for a working bc along $PATH and print
1053 [Tim Rice <tim@multitalents.net> via Richard Levitte]
1055 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
1056 extensions from a certificate request to the certificate.
1059 +) Allow multiple 'certopt' and 'nameopt' options to be separated
1060 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
1061 file: this allows the display of the certificate about to be
1062 signed to be customised, to allow certain fields to be included
1063 or excluded and extension details. The old system didn't display
1064 multicharacter strings properly, omitted fields not in the policy
1065 and couldn't display additional details such as extensions.
1068 *) Fix memory leaks in err.c: free err_data string if necessary;
1069 don't write to the wrong index in ERR_set_error_data.
1072 +) Function EC_POINTs_mul for multiple scalar multiplication
1073 of an arbitrary number of elliptic curve points
1074 \sum scalars[i]*points[i],
1075 optionally including the generator defined for the EC_GROUP:
1076 scalar*generator + \sum scalars[i]*points[i].
1078 EC_POINT_mul is a simple wrapper function for the typical case
1079 that the point list has just one item (besides the optional
1083 +) First EC_METHODs for curves over GF(p):
1085 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
1086 operations and provides various method functions that can also
1087 operate with faster implementations of modular arithmetic.
1089 EC_GFp_mont_method() reuses most functions that are part of
1090 EC_GFp_simple_method, but uses Montgomery arithmetic.
1092 [Bodo Moeller; point addition and point doubling
1093 implementation directly derived from source code provided by
1094 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1096 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1097 crypto/ec/ec_lib.c):
1099 Curves are EC_GROUP objects (with an optional group generator)
1100 based on EC_METHODs that are built into the library.
1102 Points are EC_POINT objects based on EC_GROUP objects.
1104 Most of the framework would be able to handle curves over arbitrary
1105 finite fields, but as there are no obvious types for fields other
1106 than GF(p), some functions are limited to that for now.
1109 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1110 that the file contains a complete HTTP response.
1113 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1114 change the def and num file printf format specifier from "%-40sXXX"
1115 to "%-39s XXX". The latter will always guarantee a space after the
1116 field while the former will cause them to run together if the field
1117 is 40 of more characters long.
1120 +) Constify the cipher and digest 'method' functions and structures
1121 and modify related functions to take constant EVP_MD and EVP_CIPHER
1125 *) Implement ssl23_peek (analogous to ssl23_read), which previously
1129 *) Replace rdtsc with _emit statements for VC++ version 5.
1130 [Jeremy Cooper <jeremy@baymoo.org>]
1132 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1133 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1136 +) Modify EVP_Digest*() routines so they now return values. Although the
1137 internal software routines can never fail additional hardware versions
1141 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1143 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1144 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1147 ERR_R_NESTED_ASN1_ERROR
1149 ERR_R_MISSING_ASN1_EOS
1150 were 4 .. 9, conflicting with
1151 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1153 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1154 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1156 Add new error code 'ERR_R_INTERNAL_ERROR'.
1159 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1163 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1164 sets the subject name for a new request or supersedes the
1165 subject name in a given request. Formats that can be parsed are
1166 'CN=Some Name, OU=myOU, C=IT'
1168 'CN=Some Name/OU=myOU/C=IT'.
1170 Add options '-batch' and '-verbose' to 'openssl req'.
1171 [Massimiliano Pala <madwolf@hackmasters.net>]
1173 *) Make it possible to reuse SSLv2 sessions.
1176 +) Introduce the possibility to access global variables through
1177 functions on platform were that's the best way to handle exporting
1178 global variables in shared libraries. To enable this functionality,
1179 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1180 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1181 is normally done by Configure or something similar).
1183 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1184 in the source file (foo.c) like this:
1186 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1187 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1189 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1190 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1192 OPENSSL_DECLARE_GLOBAL(int,foo);
1193 #define foo OPENSSL_GLOBAL_REF(foo)
1194 OPENSSL_DECLARE_GLOBAL(double,bar);
1195 #define bar OPENSSL_GLOBAL_REF(bar)
1197 The #defines are very important, and therefore so is including the
1198 header file everywhere where the defined globals are used.
1200 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1201 of ASN.1 items, but that structure is a bit different.
1203 The largest change is in util/mkdef.pl which has been enhanced with
1204 better and easier to understand logic to choose which symbols should
1205 go into the Windows .def files as well as a number of fixes and code
1206 cleanup (among others, algorithm keywords are now sorted
1207 lexicographically to avoid constant rewrites).
1210 *) In copy_email() check for >= 0 as a return value for
1211 X509_NAME_get_index_by_NID() since 0 is a valid index.
1212 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
1214 +) In BN_div() keep a copy of the sign of 'num' before writing the
1215 result to 'rm' because if rm==num the value will be overwritten
1216 and produce the wrong result if 'num' is negative: this caused
1217 problems with BN_mod() and BN_nnmod().
1220 +) Function OCSP_request_verify(). This checks the signature on an
1221 OCSP request and verifies the signer certificate. The signer
1222 certificate is just checked for a generic purpose and OCSP request
1226 +) Add OCSP_check_validity() function to check the validity of OCSP
1227 responses. OCSP responses are prepared in real time and may only
1228 be a few seconds old. Simply checking that the current time lies
1229 between thisUpdate and nextUpdate max reject otherwise valid responses
1230 caused by either OCSP responder or client clock inaccuracy. Instead
1231 we allow thisUpdate and nextUpdate to fall within a certain period of
1232 the current time. The age of the response can also optionally be
1233 checked. Two new options -validity_period and -status_age added to
1237 +) If signature or public key algorithm is unrecognized print out its
1238 OID rather that just UNKNOWN.
1241 *) Avoid coredump with unsupported or invalid public keys by checking if
1242 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
1243 PKCS7_verify() fails with non detached data.
1246 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1247 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1248 ID to be generated from the issuer certificate alone which can then be
1249 passed to OCSP_id_issuer_cmp().
1252 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1253 ASN1 modules to export functions returning ASN1_ITEM pointers
1254 instead of the ASN1_ITEM structures themselves. This adds several
1255 new macros which allow the underlying ASN1 function/structure to
1256 be accessed transparently. As a result code should not use ASN1_ITEM
1257 references directly (such as &X509_it) but instead use the relevant
1258 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1259 use of the new ASN1 code on platforms where exporting structures
1260 is problematical (for example in shared libraries) but exporting
1261 functions returning pointers to structures is not.
1264 +) Add support for overriding the generation of SSL/TLS session IDs.
1265 These callbacks can be registered either in an SSL_CTX or per SSL.
1266 The purpose of this is to allow applications to control, if they wish,
1267 the arbitrary values chosen for use as session IDs, particularly as it
1268 can be useful for session caching in multiple-server environments. A
1269 command-line switch for testing this (and any client code that wishes
1270 to use such a feature) has been added to "s_server".
1271 [Geoff Thorpe, Lutz Jaenicke]
1273 +) Modify mkdef.pl to recognise and parse preprocessor conditionals
1274 of the form '#if defined(...) || defined(...) || ...' and
1275 '#if !defined(...) && !defined(...) && ...'. This also avoids
1276 the growing number of special cases it was previously handling.
1279 *) Don't use getenv in library functions when run as setuid/setgid.
1280 New function OPENSSL_issetugid().
1283 +) Make all configuration macros available for application by making
1284 sure they are available in opensslconf.h, by giving them names starting
1285 with "OPENSSL_" to avoid conflicts with other packages and by making
1286 sure e_os2.h will cover all platform-specific cases together with
1288 Additionally, it is now possible to define configuration/platform-
1289 specific names (called "system identities"). In the C code, these
1290 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1291 macro with the name beginning with "OPENSSL_SYS_", which is determined
1292 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1296 +) New option -set_serial to 'req' and 'x509' this allows the serial
1297 number to use to be specified on the command line. Previously self
1298 signed certificates were hard coded with serial number 0 and the
1299 CA options of 'x509' had to use a serial number in a file which was
1303 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
1304 due to incorrect handling of multi-threading:
1306 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
1308 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
1310 3. Count how many times MemCheck_off() has been called so that
1311 nested use can be treated correctly. This also avoids
1312 inband-signalling in the previous code (which relied on the
1313 assumption that thread ID 0 is impossible).
1316 +) New options to 'ca' utility to support V2 CRL entry extensions.
1317 Currently CRL reason, invalidity date and hold instruction are
1318 supported. Add new CRL extensions to V3 code and some new objects.
1321 *) Add "-rand" option also to s_client and s_server.
1324 +) New function EVP_CIPHER_CTX_set_padding() this is used to
1325 disable standard block padding (aka PKCS#5 padding) in the EVP
1326 API, which was previously mandatory. This means that the data is
1327 not padded in any way and so the total length much be a multiple
1328 of the block size, otherwise an error occurs.
1331 +) Initial (incomplete) OCSP SSL support.
1334 *) Fix CPU detection on Irix 6.x.
1335 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
1336 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1338 +) New function OCSP_parse_url(). This splits up a URL into its host,
1339 port and path components: primarily to parse OCSP URLs. New -url
1340 option to ocsp utility.
1343 +) New nonce behavior. The return value of OCSP_check_nonce() now
1344 reflects the various checks performed. Applications can decide
1345 whether to tolerate certain situations such as an absent nonce
1346 in a response when one was present in a request: the ocsp application
1347 just prints out a warning. New function OCSP_add1_basic_nonce()
1348 this is to allow responders to include a nonce in a response even if
1349 the request is nonce-less.
1352 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
1355 [This change does not apply to 0.9.7.]
1357 *) Use the cached encoding of an X509_NAME structure rather than
1358 copying it. This is apparently the reason for the libsafe "errors"
1359 but the code is actually correct.
1362 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1363 skipped when using openssl x509 multiple times on a single input file,
1364 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1367 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1368 set string type: to handle setting ASN1_TIME structures. Fix ca
1369 utility to correctly initialize revocation date of CRLs.
1372 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1373 the clients preferred ciphersuites and rather use its own preferences.
1374 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1375 Internet Explorer by ensuring unchanged hash method during stepup.
1376 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1379 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1380 to aes and add a new 'exist' option to print out symbols that don't
1384 +) Additional options to ocsp utility to allow flags to be set and
1385 additional certificates supplied.
1388 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
1389 OCSP client a number of certificate to only verify the response
1393 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
1394 Bleichenbacher's DSA attack.
1395 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
1396 to be set and top=0 forces the highest bit to be set; top=-1 is new
1397 and leaves the highest bit random.
1398 [Ulf Moeller, Bodo Moeller]
1400 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1401 handle the new API. Currently only ECB, CBC modes supported. Add new
1402 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
1403 for TLS" draft-ietf-tls-ciphersuite-03.txt.
1404 [Ben Laurie, Steve Henson]
1406 *) In the NCONF_...-based implementations for CONF_... queries
1407 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
1408 a temporary CONF structure with the data component set to NULL
1409 (which gives segmentation faults in lh_retrieve).
1410 Instead, use NULL for the CONF pointer in CONF_get_string and
1411 CONF_get_number (which may use environment variables) and directly
1412 return NULL from CONF_get_section.
1415 *) Fix potential buffer overrun for EBCDIC.
1418 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
1419 request to response.
1422 +) Functions for OCSP responders. OCSP_request_onereq_count(),
1423 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1424 extract information from a certificate request. OCSP_response_create()
1425 creates a response and optionally adds a basic response structure.
1426 OCSP_basic_add1_status() adds a complete single response to a basic
1427 response and returns the OCSP_SINGLERESP structure just added (to allow
1428 extensions to be included for example). OCSP_basic_add1_cert() adds a
1429 certificate to a basic response and OCSP_basic_sign() signs a basic
1430 response with various flags. New helper functions ASN1_TIME_check()
1431 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1432 (converts ASN1_TIME to GeneralizedTime).
1435 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1436 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1437 structure from a certificate. X509_pubkey_digest() digests the public_key
1438 contents: this is used in various key identifiers.
1441 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
1442 keyUsage if basicConstraints absent for a CA.
1445 *) Make SMIME_write_PKCS7() write mail header values with a format that
1446 is more generally accepted (no spaces before the semicolon), since
1447 some programs can't parse those values properly otherwise. Also make
1448 sure BIO's that break lines after each write do not create invalid
1452 +) Make sk_sort() tolerate a NULL argument.
1453 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1455 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1456 passed by the function are trusted implicitly. If any of them signed the
1457 response then it is assumed to be valid and is not verified.
1460 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
1461 macros previously used would not encode an empty SEQUENCE OF
1462 and break the signature.
1464 [This change does not apply to 0.9.7.]
1466 *) Zero the premaster secret after deriving the master secret in
1470 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1471 to data. This was previously part of the PKCS7 ASN1 code. This
1472 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1473 [Steve Henson, reported by Kenneth R. Robinette
1474 <support@securenetterm.com>]
1476 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1477 routines: without these tracing memory leaks is very painful.
1478 Fix leaks in PKCS12 and PKCS7 routines.
1481 *) Add some EVP_add_digest_alias registrations (as found in
1482 OpenSSL_add_all_digests()) to SSL_library_init()
1483 aka OpenSSL_add_ssl_algorithms(). This provides improved
1484 compatibility with peers using X.509 certificates
1485 with unconventional AlgorithmIdentifier OIDs.
1488 *) Fix for Irix with NO_ASM.
1489 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1491 *) ./config script fixes.
1492 [Ulf Moeller, Richard Levitte]
1494 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1495 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1496 effectively meant GeneralizedTime would never be used. Now it
1497 is initialised to -1 but X509_time_adj() now has to check the value
1498 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1499 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1500 [Steve Henson, reported by Kenneth R. Robinette
1501 <support@securenetterm.com>]
1503 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1504 result in a zero length in the ASN1_INTEGER structure which was
1505 not consistent with the structure when d2i_ASN1_INTEGER() was used
1506 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1507 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1508 where it did not print out a minus for negative ASN1_INTEGER.
1511 *) Fix 'openssl passwd -1'.
1514 +) Add summary printout to ocsp utility. The various functions which
1515 convert status values to strings have been renamed to:
1516 OCSP_response_status_str(), OCSP_cert_status_str() and
1517 OCSP_crl_reason_str() and are no longer static. New options
1518 to verify nonce values and to disable verification. OCSP response
1519 printout format cleaned up.
1522 +) Add additional OCSP certificate checks. These are those specified
1523 in RFC2560. This consists of two separate checks: the CA of the
1524 certificate being checked must either be the OCSP signer certificate
1525 or the issuer of the OCSP signer certificate. In the latter case the
1526 OCSP signer certificate must contain the OCSP signing extended key
1527 usage. This check is performed by attempting to match the OCSP
1528 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1529 in the OCSP_CERTID structures of the response.
1532 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1533 and related routines. This uses the standard OpenSSL certificate
1534 verify routines to perform initial checks (just CA validity) and
1535 to obtain the certificate chain. Then additional checks will be
1536 performed on the chain. Currently the root CA is checked to see
1537 if it is explicitly trusted for OCSP signing. This is used to set
1538 a root CA as a global signing root: that is any certificate that
1539 chains to that CA is an acceptable OCSP signing certificate.
1542 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1543 extensions from a separate configuration file.
1544 As when reading extensions from the main configuration file,
1545 the '-extensions ...' option may be used for specifying the
1547 [Massimiliano Pala <madwolf@comune.modena.it>]
1549 *) Change PKCS12_key_gen_asc() so it can cope with non null
1550 terminated strings whose length is passed in the passlen
1551 parameter, for example from PEM callbacks. This was done
1552 by adding an extra length parameter to asc2uni().
1553 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1555 +) New OCSP utility. Allows OCSP requests to be generated or
1556 read. The request can be sent to a responder and the output
1557 parsed, outputed or printed in text form. Not complete yet:
1558 still needs to check the OCSP response validity.
1561 +) New subcommands for 'openssl ca':
1562 'openssl ca -status <serial>' prints the status of the cert with
1563 the given serial number (according to the index file).
1564 'openssl ca -updatedb' updates the expiry status of certificates
1566 [Massimiliano Pala <madwolf@comune.modena.it>]
1568 +) New '-newreq-nodes' command option to CA.pl. This is like
1569 '-newreq', but calls 'openssl req' with the '-nodes' option
1570 so that the resulting key is not encrypted.
1571 [Damien Miller <djm@mindrot.org>]
1573 +) New configuration for the GNU Hurd.
1574 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1576 +) Initial code to implement OCSP basic response verify. This
1577 is currently incomplete. Currently just finds the signer's
1578 certificate and verifies the signature on the response.
1581 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1582 value of OPENSSLDIR. This is available via the new '-d' option
1583 to 'openssl version', and is also included in 'openssl version -a'.
1586 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1587 call failed, free the DSA structure.
1590 +) Allowing defining memory allocation callbacks that will be given
1591 file name and line number information in additional arguments
1592 (a const char* and an int). The basic functionality remains, as
1593 well as the original possibility to just replace malloc(),
1594 realloc() and free() by functions that do not know about these
1595 additional arguments. To register and find out the current
1596 settings for extended allocation functions, the following
1597 functions are provided:
1599 CRYPTO_set_mem_ex_functions
1600 CRYPTO_set_locked_mem_ex_functions
1601 CRYPTO_get_mem_ex_functions
1602 CRYPTO_get_locked_mem_ex_functions
1604 These work the same way as CRYPTO_set_mem_functions and friends.
1605 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1606 extended allocation function is enabled.
1607 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1608 a conventional allocation function is enabled.
1609 [Richard Levitte, Bodo Moeller]
1611 *) Fix to uni2asc() to cope with zero length Unicode strings.
1612 These are present in some PKCS#12 files.
1615 +) Finish off removing the remaining LHASH function pointer casts.
1616 There should no longer be any prototype-casting required when using
1617 the LHASH abstraction, and any casts that remain are "bugs". See
1618 the callback types and macros at the head of lhash.h for details
1619 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1622 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1623 If /dev/[u]random devices are not available or do not return enough
1624 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1626 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1627 /etc/entropy will be queried once each in this sequence, quering stops
1628 when enough entropy was collected without querying more sockets.
1631 +) Change the Unix RAND_poll() variant to be able to poll several
1632 random devices, as specified by DEVRANDOM, until a sufficient amount
1633 of data has been collected. We spend at most 10 ms on each file
1634 (select timeout) and read in non-blocking mode. DEVRANDOM now
1635 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1636 (previously it was just the string "/dev/urandom"), so on typical
1637 platforms the 10 ms delay will never occur.
1638 Also separate out the Unix variant to its own file, rand_unix.c.
1639 For VMS, there's a currently-empty rand_vms.c.
1642 +) Move OCSP client related routines to ocsp_cl.c. These
1643 provide utility functions which an application needing
1644 to issue a request to an OCSP responder and analyse the
1645 response will typically need: as opposed to those which an
1646 OCSP responder itself would need which will be added later.
1648 OCSP_request_sign() signs an OCSP request with an API similar
1649 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1650 response. OCSP_response_get1_basic() extracts basic response
1651 from response. OCSP_resp_find_status(): finds and extracts status
1652 information from an OCSP_CERTID structure (which will be created
1653 when the request structure is built). These are built from lower
1654 level functions which work on OCSP_SINGLERESP structures but
1655 wont normally be used unless the application wishes to examine
1656 extensions in the OCSP response for example.
1658 Replace nonce routines with a pair of functions.
1659 OCSP_request_add1_nonce() adds a nonce value and optionally
1660 generates a random value. OCSP_check_nonce() checks the
1661 validity of the nonce in an OCSP response.
1664 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1665 This doesn't copy the supplied OCSP_CERTID and avoids the
1666 need to free up the newly created id. Change return type
1667 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1668 This can then be used to add extensions to the request.
1669 Deleted OCSP_request_new(), since most of its functionality
1670 is now in OCSP_REQUEST_new() (and the case insensitive name
1671 clash) apart from the ability to set the request name which
1672 will be added elsewhere.
1675 +) Update OCSP API. Remove obsolete extensions argument from
1676 various functions. Extensions are now handled using the new
1677 OCSP extension code. New simple OCSP HTTP function which
1678 can be used to send requests and parse the response.
1681 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1682 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1683 uses the special reorder version of SET OF to sort the attributes
1684 and reorder them to match the encoded order. This resolves a long
1685 standing problem: a verify on a PKCS7 structure just after signing
1686 it used to fail because the attribute order did not match the
1687 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1688 it uses the received order. This is necessary to tolerate some broken
1689 software that does not order SET OF. This is handled by encoding
1690 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1691 to produce the required SET OF.
1694 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1695 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1696 files to get correct declarations of the ASN.1 item variables.
1699 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1700 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1701 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1702 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1703 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1704 ASN1_ITEM and no wrapper functions.
1707 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1708 replace the old function pointer based I/O routines. Change most of
1709 the *_d2i_bio() and *_d2i_fp() functions to use these.
1712 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1713 lines, recognice more "algorithms" that can be deselected, and make
1714 it complain about algorithm deselection that isn't recognised.
1717 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1718 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1719 to use new functions. Add NO_ASN1_OLD which can be set to remove
1720 some old style ASN1 functions: this can be used to determine if old
1721 code will still work when these eventually go away.
1724 +) New extension functions for OCSP structures, these follow the
1725 same conventions as certificates and CRLs.
1728 +) New function X509V3_add1_i2d(). This automatically encodes and
1729 adds an extension. Its behaviour can be customised with various
1730 flags to append, replace or delete. Various wrappers added for
1731 certifcates and CRLs.
1734 +) Fix to avoid calling the underlying ASN1 print routine when
1735 an extension cannot be parsed. Correct a typo in the
1736 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1739 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1740 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1741 when writing a 32767 byte record.
1742 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1744 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1745 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1747 (RSA objects have a reference count access to which is protected
1748 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1749 so they are meant to be shared between threads.)
1750 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1751 "Reddie, Steven" <Steven.Reddie@ca.com>]
1753 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1754 entries for variables.
1757 *) Fix a deadlock in CRYPTO_mem_leaks().
1760 +) Add functionality to apps/openssl.c for detecting locking
1761 problems: As the program is single-threaded, all we have
1762 to do is register a locking callback using an array for
1763 storing which locks are currently held by the program.
1766 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1767 SSL_get_ex_data_X509_STORE_idx(), which is used in
1768 ssl_verify_cert_chain() and thus can be called at any time
1769 during TLS/SSL handshakes so that thread-safety is essential.
1770 Unfortunately, the ex_data design is not at all suited
1771 for multi-threaded use, so it probably should be abolished.
1774 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1775 [Broadcom, tweaked and integrated by Geoff Thorpe]
1777 +) Move common extension printing code to new function
1778 X509V3_print_extensions(). Reorganise OCSP print routines and
1779 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1782 +) New function X509_signature_print() to remove duplication in some
1786 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1787 set (this was treated exactly the same as SET OF previously). This
1788 is used to reorder the STACK representing the structure to match the
1789 encoding. This will be used to get round a problem where a PKCS7
1790 structure which was signed could not be verified because the STACK
1791 order did not reflect the encoded order.
1794 +) Reimplement the OCSP ASN1 module using the new code.
1797 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1798 for its ASN1 operations. The old style function pointers still exist
1799 for now but they will eventually go away.
1802 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1803 completely replaces the old ASN1 functionality with a table driven
1804 encoder and decoder which interprets an ASN1_ITEM structure describing
1805 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1806 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1807 has also been converted to the new form.
1810 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1811 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1812 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1813 for negative moduli.
1816 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1817 of not touching the result's sign bit.
1820 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1824 +) Changed the LHASH code to use prototypes for callbacks, and created
1825 macros to declare and implement thin (optionally static) functions
1826 that provide type-safety and avoid function pointer casting for the
1827 type-specific callbacks.
1830 *) Use better test patterns in bntest.
1833 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1835 [Veers Staats <staatsvr@asc.hpc.mil>,
1836 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1838 *) rand_win.c fix for Borland C.
1841 *) BN_rshift bugfix for n == 0.
1844 +) Reformat the FAQ so the different questions and answers can be divided
1845 in sections depending on the subject.
1848 +) Have the zlib compression code load ZLIB.DLL dynamically under
1852 +) New function BN_mod_sqrt for computing square roots modulo a prime
1853 (using the probabilistic Tonelli-Shanks algorithm unless
1854 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1855 be handled deterministically).
1856 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1858 *) Add a 'bctest' script that checks for some known 'bc' bugs
1859 so that 'make test' does not abort just because 'bc' is broken.
1862 *) Store verify_result within SSL_SESSION also for client side to
1863 avoid potential security hole. (Re-used sessions on the client side
1864 always resulted in verify_result==X509_V_OK, not using the original
1865 result of the server certificate verification.)
1868 +) Make BN_mod_inverse faster by explicitly handling small quotients
1869 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1870 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1873 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1874 SSL3_RT_APPLICATION_DATA, return 0.
1875 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1879 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1880 releases, have been re-implemented by renaming the previous
1881 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1882 and ssl3_read_internal, respectively, and adding 'peek' parameters
1883 to them. The new ssl[23]_{read,peek} functions are calls to
1884 ssl[23]_read_internal with the 'peek' flag set appropriately.
1885 A 'peek' parameter has also been added to ssl3_read_bytes, which
1886 does the actual work for ssl3_read_internal.
1889 +) New function BN_kronecker.
1892 +) Fix BN_gcd so that it works on negative inputs; the result is
1893 positive unless both parameters are zero.
1894 Previously something reasonably close to an infinite loop was
1895 possible because numbers could be growing instead of shrinking
1896 in the implementation of Euclid's algorithm.
1899 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1900 sign of the number in question.
1902 Fix BN_is_word(a,w) to work correctly for w == 0.
1904 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1905 because its test if the absolute value of 'a' equals 'w'.
1906 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1907 it exists mostly for use in the implementations of BN_is_zero(),
1908 BN_is_one(), and BN_is_word().
1911 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1912 the method-specific "init()" handler. Also clean up ex_data after
1913 calling the method-specific "finish()" handler. Previously, this was
1914 happening the other way round.
1917 +) New function BN_swap.
1920 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1921 the exponentiation functions are more likely to produce reasonable
1922 results on negative inputs.
1925 +) Change BN_mod_mul so that the result is always non-negative.
1926 Previously, it could be negative if one of the factors was negative;
1927 I don't think anyone really wanted that behaviour.
1930 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1931 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1932 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1933 and add new functions:
1942 BN_mod_lshift1_quick
1946 These functions always generate non-negative results.
1948 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1949 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1951 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1952 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1953 be reduced modulo m.
1954 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1956 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1957 was actually never needed) and in BN_mul(). The removal in BN_mul()
1958 required a small change in bn_mul_part_recursive() and the addition
1959 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1960 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1961 bn_sub_words() and bn_add_words() except they take arrays with
1965 +) In 'openssl passwd', verify passwords read from the terminal
1966 unless the '-salt' option is used (which usually means that
1967 verification would just waste user's time since the resulting
1968 hash is going to be compared with some given password hash)
1969 or the new '-noverify' option is used.
1971 This is an incompatible change, but it does not affect
1972 non-interactive use of 'openssl passwd' (passwords on the command
1973 line, '-stdin' option, '-in ...' option) and thus should not
1977 +) Remove all references to RSAref, since there's no more need for it.
1980 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
1981 The previous value, 12, was not always sufficient for BN_mod_exp().
1984 +) Make DSO load along a path given through an environment variable
1985 (SHLIB_PATH) with shl_load().
1988 +) Constify the ENGINE code as a result of BIGNUM constification.
1989 Also constify the RSA code and most things related to it. In a
1990 few places, most notable in the depth of the ASN.1 code, ugly
1991 casts back to non-const were required (to be solved at a later
1995 +) Make it so the openssl application has all engines loaded by default.
1998 +) Constify the BIGNUM routines a little more.
2001 *) Make sure that shared libraries get the internal name engine with
2002 the full version number and not just 0. This should mark the
2003 shared libraries as not backward compatible. Of course, this should
2004 be changed again when we can guarantee backward binary compatibility.
2007 +) Add the following functions:
2009 ENGINE_load_cswift()
2011 ENGINE_load_atalla()
2013 ENGINE_load_builtin_engines()
2015 That way, an application can itself choose if external engines that
2016 are built-in in OpenSSL shall ever be used or not. The benefit is
2017 that applications won't have to be linked with libdl or other dso
2018 libraries unless it's really needed.
2020 Changed 'openssl engine' to load all engines on demand.
2021 Changed the engine header files to avoid the duplication of some
2022 declarations (they differed!).
2025 +) 'openssl engine' can now list capabilities.
2028 +) Better error reporting in 'openssl engine'.
2031 +) Never call load_dh_param(NULL) in s_server.
2034 +) Add engine application. It can currently list engines by name and
2035 identity, and test if they are actually available.
2038 +) Improve RPM specification file by forcing symbolic linking and making
2039 sure the installed documentation is also owned by root.root.
2040 [Damien Miller <djm@mindrot.org>]
2042 +) Give the OpenSSL applications more possibilities to make use of
2043 keys (public as well as private) handled by engines.
2046 +) Add OCSP code that comes from CertCo.
2049 +) Add VMS support for the Rijndael code.
2052 +) Added untested support for Nuron crypto accelerator.
2055 +) Add support for external cryptographic devices. This code was
2056 previously distributed separately as the "engine" branch.
2057 [Geoff Thorpe, Richard Levitte]
2059 +) Rework the filename-translation in the DSO code. It is now possible to
2060 have far greater control over how a "name" is turned into a filename
2061 depending on the operating environment and any oddities about the
2062 different shared library filenames on each system.
2065 +) Support threads on FreeBSD-elf in Configure.
2068 +) Fix for SHA1 assembly problem with MASM: it produces
2069 warnings about corrupt line number information when assembling
2070 with debugging information. This is caused by the overlapping
2072 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
2075 NCONF_get_number() has no error checking at all. As a replacement,
2076 NCONF_get_number_e() is defined (_e for "error checking") and is
2077 promoted strongly. The old NCONF_get_number is kept around for
2078 binary backward compatibility.
2079 Make it possible for methods to load from something other than a BIO,
2080 by providing a function pointer that is given a name instead of a BIO.
2081 For example, this could be used to load configuration data from an
2085 *) Fix typo in get_cert_by_subject() in by_dir.c
2086 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
2088 *) Rework the system to generate shared libraries:
2090 - Make note of the expected extension for the shared libraries and
2091 if there is a need for symbolic links from for example libcrypto.so.0
2092 to libcrypto.so.0.9.7. There is extended info in Configure for
2095 - Make as few rebuilds of the shared libraries as possible.
2097 - Still avoid linking the OpenSSL programs with the shared libraries.
2099 - When installing, install the shared libraries separately from the
2103 +) Fix for non blocking accept BIOs. Added new I/O special reason
2104 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2105 with non blocking I/O was not possible because no retry code was
2106 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2110 +) Added the beginnings of Rijndael support.
2113 +) Fix for bug in DirectoryString mask setting. Add support for
2114 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2115 to allow certificate printing to more controllable, additional
2116 'certopt' option to 'x509' to allow new printing options to be
2120 +) Clean old EAY MD5 hack from e_os.h.
2123 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
2125 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
2126 and not in SSL_clear because the latter is also used by the
2127 accept/connect functions; previously, the settings made by
2128 SSL_set_read_ahead would be lost during the handshake.
2129 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
2131 *) Correct util/mkdef.pl to be selective about disabled algorithms.
2132 Previously, it would create entries for disableed algorithms no
2136 *) Added several new manual pages for SSL_* function.
2139 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
2141 *) In ssl23_get_client_hello, generate an error message when faced
2142 with an initial SSL 3.0/TLS record that is too small to contain the
2143 first two bytes of the ClientHello message, i.e. client_version.
2144 (Note that this is a pathologic case that probably has never happened
2145 in real life.) The previous approach was to use the version number
2146 from the record header as a substitute; but our protocol choice
2147 should not depend on that one because it is not authenticated
2148 by the Finished messages.
2151 *) More robust randomness gathering functions for Windows.
2152 [Jeffrey Altman <jaltman@columbia.edu>]
2154 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
2155 not set then we don't setup the error code for issuer check errors
2156 to avoid possibly overwriting other errors which the callback does
2157 handle. If an application does set the flag then we assume it knows
2158 what it is doing and can handle the new informational codes
2162 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
2163 a general "ANY" type, as such it should be able to decode anything
2164 including tagged types. However it didn't check the class so it would
2165 wrongly interpret tagged types in the same way as their universal
2166 counterpart and unknown types were just rejected. Changed so that the
2167 tagged and unknown types are handled in the same way as a SEQUENCE:
2168 that is the encoding is stored intact. There is also a new type
2169 "V_ASN1_OTHER" which is used when the class is not universal, in this
2170 case we have no idea what the actual type is so we just lump them all
2174 *) On VMS, stdout may very well lead to a file that is written to
2175 in a record-oriented fashion. That means that every write() will
2176 write a separate record, which will be read separately by the
2177 programs trying to read from it. This can be very confusing.
2179 The solution is to put a BIO filter in the way that will buffer
2180 text until a linefeed is reached, and then write everything a
2181 line at a time, so every record written will be an actual line,
2182 not chunks of lines and not (usually doesn't happen, but I've
2183 seen it once) several lines in one record. BIO_f_linebuffer() is
2186 Currently, it's a VMS-only method, because that's where it has
2187 been tested well enough.
2190 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
2191 it can return incorrect results.
2192 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
2193 but it was in 0.9.6-beta[12].)
2196 *) Disable the check for content being present when verifying detached
2197 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
2198 include zero length content when signing messages.
2201 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
2202 BIO_ctrl (for BIO pairs).
2205 *) Add DSO method for VMS.
2208 *) Bug fix: Montgomery multiplication could produce results with the
2212 *) Add RPM specification openssl.spec and modify it to build three
2213 packages. The default package contains applications, application
2214 documentation and run-time libraries. The devel package contains
2215 include files, static libraries and function documentation. The
2216 doc package contains the contents of the doc directory. The original
2217 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
2220 *) Add a large number of documentation files for many SSL routines.
2221 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2223 *) Add a configuration entry for Sony News 4.
2224 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
2226 *) Don't set the two most significant bits to one when generating a
2227 random number < q in the DSA library.
2230 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
2231 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
2232 the underlying transport is blocking) if a handshake took place.
2233 (The default behaviour is needed by applications such as s_client
2234 and s_server that use select() to determine when to use SSL_read;
2235 but for applications that know in advance when to expect data, it
2236 just makes things more complicated.)
2239 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
2243 *) Add a few more EBCDIC conditionals that make `req' and `x509'
2244 work better on such systems.
2245 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
2247 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
2248 Update PKCS12_parse() so it copies the friendlyName and the
2249 keyid to the certificates aux info.
2252 *) Fix bug in PKCS7_verify() which caused an infinite loop
2253 if there was more than one signature.
2254 [Sven Uszpelkat <su@celocom.de>]
2256 *) Major change in util/mkdef.pl to include extra information
2257 about each symbol, as well as presentig variables as well
2258 as functions. This change means that there's n more need
2259 to rebuild the .num files when some algorithms are excluded.
2262 *) Allow the verify time to be set by an application,
2263 rather than always using the current time.
2266 *) Phase 2 verify code reorganisation. The certificate
2267 verify code now looks up an issuer certificate by a
2268 number of criteria: subject name, authority key id
2269 and key usage. It also verifies self signed certificates
2270 by the same criteria. The main comparison function is
2271 X509_check_issued() which performs these checks.
2273 Lot of changes were necessary in order to support this
2274 without completely rewriting the lookup code.
2276 Authority and subject key identifier are now cached.
2278 The LHASH 'certs' is X509_STORE has now been replaced
2279 by a STACK_OF(X509_OBJECT). This is mainly because an
2280 LHASH can't store or retrieve multiple objects with
2281 the same hash value.
2283 As a result various functions (which were all internal
2284 use only) have changed to handle the new X509_STORE
2285 structure. This will break anything that messed round
2286 with X509_STORE internally.
2288 The functions X509_STORE_add_cert() now checks for an
2289 exact match, rather than just subject name.
2291 The X509_STORE API doesn't directly support the retrieval
2292 of multiple certificates matching a given criteria, however
2293 this can be worked round by performing a lookup first
2294 (which will fill the cache with candidate certificates)
2295 and then examining the cache for matches. This is probably
2296 the best we can do without throwing out X509_LOOKUP
2297 entirely (maybe later...).
2299 The X509_VERIFY_CTX structure has been enhanced considerably.
2301 All certificate lookup operations now go via a get_issuer()
2302 callback. Although this currently uses an X509_STORE it
2303 can be replaced by custom lookups. This is a simple way
2304 to bypass the X509_STORE hackery necessary to make this
2305 work and makes it possible to use more efficient techniques
2306 in future. A very simple version which uses a simple
2307 STACK for its trusted certificate store is also provided
2308 using X509_STORE_CTX_trusted_stack().
2310 The verify_cb() and verify() callbacks now have equivalents
2311 in the X509_STORE_CTX structure.
2313 X509_STORE_CTX also has a 'flags' field which can be used
2314 to customise the verify behaviour.
2317 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
2318 excludes S/MIME capabilities.
2321 *) When a certificate request is read in keep a copy of the
2322 original encoding of the signed data and use it when outputing
2323 again. Signatures then use the original encoding rather than
2324 a decoded, encoded version which may cause problems if the
2325 request is improperly encoded.
2328 *) For consistency with other BIO_puts implementations, call
2329 buffer_write(b, ...) directly in buffer_puts instead of calling
2332 In BIO_puts, increment b->num_write as in BIO_write.
2333 [Peter.Sylvester@EdelWeb.fr]
2335 *) Fix BN_mul_word for the case where the word is 0. (We have to use
2336 BN_zero, we may not return a BIGNUM with an array consisting of
2340 *) Avoid calling abort() from within the library when problems are
2341 detected, except if preprocessor symbols have been defined
2342 (such as REF_CHECK, BN_DEBUG etc.).
2345 *) New openssl application 'rsautl'. This utility can be
2346 used for low level RSA operations. DER public key
2347 BIO/fp routines also added.
2350 *) New Configure entry and patches for compiling on QNX 4.
2351 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
2353 *) A demo state-machine implementation was sponsored by
2354 Nuron (http://www.nuron.com/) and is now available in
2355 demos/state_machine.
2358 *) New options added to the 'dgst' utility for signature
2359 generation and verification.
2362 *) Unrecognized PKCS#7 content types are now handled via a
2363 catch all ASN1_TYPE structure. This allows unsupported
2364 types to be stored as a "blob" and an application can
2365 encode and decode it manually.
2368 *) Fix various signed/unsigned issues to make a_strex.c
2370 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
2372 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
2373 length if passed a buffer. ASN1_INTEGER_to_BN failed
2374 if passed a NULL BN and its argument was negative.
2375 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
2377 *) Modification to PKCS#7 encoding routines to output definite
2378 length encoding. Since currently the whole structures are in
2379 memory there's not real point in using indefinite length
2380 constructed encoding. However if OpenSSL is compiled with
2381 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
2384 *) Added BIO_vprintf() and BIO_vsnprintf().
2387 *) Added more prefixes to parse for in the the strings written
2388 through a logging bio, to cover all the levels that are available
2389 through syslog. The prefixes are now:
2391 PANIC, EMERG, EMR => LOG_EMERG
2392 ALERT, ALR => LOG_ALERT
2393 CRIT, CRI => LOG_CRIT
2394 ERROR, ERR => LOG_ERR
2395 WARNING, WARN, WAR => LOG_WARNING
2396 NOTICE, NOTE, NOT => LOG_NOTICE
2397 INFO, INF => LOG_INFO
2398 DEBUG, DBG => LOG_DEBUG
2400 and as before, if none of those prefixes are present at the
2401 beginning of the string, LOG_ERR is chosen.
2403 On Win32, the LOG_* levels are mapped according to this:
2405 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
2406 LOG_WARNING => EVENTLOG_WARNING_TYPE
2407 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
2411 *) Made it possible to reconfigure with just the configuration
2412 argument "reconf" or "reconfigure". The command line arguments
2413 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
2414 and are retrieved from there when reconfiguring.
2418 [Assar Westerlund <assar@sics.se>, Richard Levitte]
2420 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
2423 *) The obj_dat.pl script was messing up the sorting of object
2424 names. The reason was that it compared the quoted version
2425 of strings as a result "OCSP" > "OCSP Signing" because
2426 " > SPACE. Changed script to store unquoted versions of
2427 names and add quotes on output. It was also omitting some
2428 names from the lookup table if they were given a default
2429 value (that is if SN is missing it is given the same
2430 value as LN and vice versa), these are now added on the
2431 grounds that if an object has a name we should be able to
2432 look it up. Finally added warning output when duplicate
2433 short or long names are found.
2436 *) Changes needed for Tandem NSK.
2437 [Scott Uroff <scott@xypro.com>]
2439 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
2440 RSA_padding_check_SSLv23(), special padding was never detected
2441 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
2442 version rollback attacks was not effective.
2444 In s23_clnt.c, don't use special rollback-attack detection padding
2445 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
2446 client; similarly, in s23_srvr.c, don't do the rollback check if
2447 SSL 2.0 is the only protocol enabled in the server.
2450 *) Make it possible to get hexdumps of unprintable data with 'openssl
2451 asn1parse'. By implication, the functions ASN1_parse_dump() and
2452 BIO_dump_indent() are added.
2455 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
2456 these print out strings and name structures based on various
2457 flags including RFC2253 support and proper handling of
2458 multibyte characters. Added options to the 'x509' utility
2459 to allow the various flags to be set.
2462 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
2463 Also change the functions X509_cmp_current_time() and
2464 X509_gmtime_adj() work with an ASN1_TIME structure,
2465 this will enable certificates using GeneralizedTime in validity
2466 dates to be checked.
2469 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
2470 negative public key encodings) on by default,
2471 NO_NEG_PUBKEY_BUG can be set to disable it.
2474 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
2475 content octets. An i2c_ASN1_OBJECT is unnecessary because
2476 the encoding can be trivially obtained from the structure.
2479 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
2480 not read locks (CRYPTO_r_[un]lock).
2483 *) A first attempt at creating official support for shared
2484 libraries through configuration. I've kept it so the
2485 default is static libraries only, and the OpenSSL programs
2486 are always statically linked for now, but there are
2487 preparations for dynamic linking in place.
2488 This has been tested on Linux and Tru64.
2491 *) Randomness polling function for Win9x, as described in:
2492 Peter Gutmann, Software Generation of Practically Strong
2496 *) Fix so PRNG is seeded in req if using an already existing
2500 *) New options to smime application. -inform and -outform
2501 allow alternative formats for the S/MIME message including
2502 PEM and DER. The -content option allows the content to be
2503 specified separately. This should allow things like Netscape
2504 form signing output easier to verify.
2507 *) Fix the ASN1 encoding of tags using the 'long form'.
2510 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
2511 STRING types. These convert content octets to and from the
2512 underlying type. The actual tag and length octets are
2513 already assumed to have been read in and checked. These
2514 are needed because all other string types have virtually
2515 identical handling apart from the tag. By having versions
2516 of the ASN1 functions that just operate on content octets
2517 IMPLICIT tagging can be handled properly. It also allows
2518 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
2519 and ASN1_INTEGER are identical apart from the tag.
2522 *) Change the handling of OID objects as follows:
2524 - New object identifiers are inserted in objects.txt, following
2525 the syntax given in objects.README.
2526 - objects.pl is used to process obj_mac.num and create a new
2528 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2531 This is currently kind of a hack, and the perl code in objects.pl
2532 isn't very elegant, but it works as I intended. The simplest way
2533 to check that it worked correctly is to look in obj_dat.h and
2534 check the array nid_objs and make sure the objects haven't moved
2535 around (this is important!). Additions are OK, as well as
2536 consistent name changes.
2539 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2542 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2543 The given file adds to whatever has already been seeded into the
2544 random pool through the RANDFILE configuration file option or
2545 environment variable, or the default random state file.
2548 *) mkstack.pl now sorts each macro group into lexical order.
2549 Previously the output order depended on the order the files
2550 appeared in the directory, resulting in needless rewriting
2554 *) Patches to make OpenSSL compile under Win32 again. Mostly
2555 work arounds for the VC++ problem that it treats func() as
2556 func(void). Also stripped out the parts of mkdef.pl that
2557 added extra typesafe functions: these no longer exist.
2560 *) Reorganisation of the stack code. The macros are now all
2561 collected in safestack.h . Each macro is defined in terms of
2562 a "stack macro" of the form SKM_<name>(type, a, b). The
2563 DEBUG_SAFESTACK is now handled in terms of function casts,
2564 this has the advantage of retaining type safety without the
2565 use of additional functions. If DEBUG_SAFESTACK is not defined
2566 then the non typesafe macros are used instead. Also modified the
2567 mkstack.pl script to handle the new form. Needs testing to see
2568 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2569 the default if no major problems. Similar behaviour for ASN1_SET_OF
2570 and PKCS12_STACK_OF.
2573 *) When some versions of IIS use the 'NET' form of private key the
2574 key derivation algorithm is different. Normally MD5(password) is
2575 used as a 128 bit RC4 key. In the modified case
2576 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2577 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2578 as the old Netscape_RSA functions except they have an additional
2579 'sgckey' parameter which uses the modified algorithm. Also added
2580 an -sgckey command line option to the rsa utility. Thanks to
2581 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2582 algorithm to openssl-dev.
2585 *) The evp_local.h macros were using 'c.##kname' which resulted in
2586 invalid expansion on some systems (SCO 5.0.5 for example).
2587 Corrected to 'c.kname'.
2588 [Phillip Porch <root@theporch.com>]
2590 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2591 a STACK of email addresses from a certificate or request, these look
2592 in the subject name and the subject alternative name extensions and
2593 omit any duplicate addresses.
2596 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2597 This makes DSA verification about 2 % faster.
2600 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2601 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2602 plus overhead for 1024 bit moduli).
2603 This makes exponentiations about 0.5 % faster for 1024 bit
2604 exponents (as measured by "openssl speed rsa2048").
2607 *) Rename memory handling macros to avoid conflicts with other
2609 Malloc => OPENSSL_malloc
2610 Malloc_locked => OPENSSL_malloc_locked
2611 Realloc => OPENSSL_realloc
2612 Free => OPENSSL_free
2615 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2616 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2619 *) CygWin32 support.
2620 [John Jarvie <jjarvie@newsguy.com>]
2622 *) The type-safe stack code has been rejigged. It is now only compiled
2623 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2624 by default all type-specific stack functions are "#define"d back to
2625 standard stack functions. This results in more streamlined output
2626 but retains the type-safety checking possibilities of the original
2630 *) The STACK code has been cleaned up, and certain type declarations
2631 that didn't make a lot of sense have been brought in line. This has
2632 also involved a cleanup of sorts in safestack.h to more correctly
2633 map type-safe stack functions onto their plain stack counterparts.
2634 This work has also resulted in a variety of "const"ifications of
2635 lots of the code, especially "_cmp" operations which should normally
2636 be prototyped with "const" parameters anyway.
2639 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2640 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2641 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2642 where all of 'md' is used each time the PRNG is used, but 'state'
2643 is used only indexed by a cyclic counter. As entropy may not be
2644 well distributed from the beginning, 'md' is important as a
2645 chaining variable. However, the output function chains only half
2646 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2647 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2648 in all of 'state' being rewritten, with the new values depending
2649 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2652 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2653 the handshake is continued after ssl_verify_cert_chain();
2654 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2655 can lead to 'unexplainable' connection aborts later.
2656 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2658 *) Major EVP API cipher revision.
2659 Add hooks for extra EVP features. This allows various cipher
2660 parameters to be set in the EVP interface. Support added for variable
2661 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2662 setting of RC2 and RC5 parameters.
2664 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2667 Remove lots of duplicated code from the EVP library. For example *every*
2668 cipher init() function handles the 'iv' in the same way according to the
2669 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2670 for CFB and OFB modes they zero ctx->num.
2672 New functionality allows removal of S/MIME code RC2 hack.
2674 Most of the routines have the same form and so can be declared in terms
2677 By shifting this to the top level EVP_CipherInit() it can be removed from
2678 all individual ciphers. If the cipher wants to handle IVs or keys
2679 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2682 Change lots of functions like EVP_EncryptUpdate() to now return a
2683 value: although software versions of the algorithms cannot fail
2684 any installed hardware versions can.
2687 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2688 this option is set, tolerate broken clients that send the negotiated
2689 protocol version number instead of the requested protocol version
2693 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2694 i.e. non-zero for export ciphersuites, zero otherwise.
2695 Previous versions had this flag inverted, inconsistent with
2696 rsa_tmp_cb (..._TMP_RSA_CB).
2697 [Bodo Moeller; problem reported by Amit Chopra]
2699 *) Add missing DSA library text string. Work around for some IIS
2700 key files with invalid SEQUENCE encoding.
2703 *) Add a document (doc/standards.txt) that list all kinds of standards
2704 and so on that are implemented in OpenSSL.
2707 *) Enhance c_rehash script. Old version would mishandle certificates
2708 with the same subject name hash and wouldn't handle CRLs at all.
2709 Added -fingerprint option to crl utility, to support new c_rehash
2713 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2716 *) Fix for SSL server purpose checking. Server checking was
2717 rejecting certificates which had extended key usage present
2718 but no ssl client purpose.
2719 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2721 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2722 is a little unclear about how a blank password is handled.
2723 Since the password in encoded as a BMPString with terminating
2724 double NULL a zero length password would end up as just the
2725 double NULL. However no password at all is different and is
2726 handled differently in the PKCS#12 key generation code. NS
2727 treats a blank password as zero length. MSIE treats it as no
2728 password on export: but it will try both on import. We now do
2729 the same: PKCS12_parse() tries zero length and no password if
2730 the password is set to "" or NULL (NULL is now a valid password:
2731 it wasn't before) as does the pkcs12 application.
2734 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2735 perror when PEM_read_bio_X509_REQ fails, the error message must
2736 be obtained from the error queue.
2739 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2740 it in ERR_remove_state if appropriate, and change ERR_get_state
2741 accordingly to avoid race conditions (this is necessary because
2742 thread_hash is no longer constant once set).
2745 *) Bugfix for linux-elf makefile.one.
2748 *) RSA_get_default_method() will now cause a default
2749 RSA_METHOD to be chosen if one doesn't exist already.
2750 Previously this was only set during a call to RSA_new()
2751 or RSA_new_method(NULL) meaning it was possible for
2752 RSA_get_default_method() to return NULL.
2755 *) Added native name translation to the existing DSO code
2756 that will convert (if the flag to do so is set) filenames
2757 that are sufficiently small and have no path information
2758 into a canonical native form. Eg. "blah" converted to
2759 "libblah.so" or "blah.dll" etc.
2762 *) New function ERR_error_string_n(e, buf, len) which is like
2763 ERR_error_string(e, buf), but writes at most 'len' bytes
2764 including the 0 terminator. For ERR_error_string_n, 'buf'
2766 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2768 *) CONF library reworked to become more general. A new CONF
2769 configuration file reader "class" is implemented as well as a
2770 new functions (NCONF_*, for "New CONF") to handle it. The now
2771 old CONF_* functions are still there, but are reimplemented to
2772 work in terms of the new functions. Also, a set of functions
2773 to handle the internal storage of the configuration data is
2774 provided to make it easier to write new configuration file
2775 reader "classes" (I can definitely see something reading a
2776 configuration file in XML format, for example), called _CONF_*,
2777 or "the configuration storage API"...
2779 The new configuration file reading functions are:
2781 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2782 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2784 NCONF_default, NCONF_WIN32
2786 NCONF_dump_fp, NCONF_dump_bio
2788 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2789 NCONF_new creates a new CONF object. This works in the same way
2790 as other interfaces in OpenSSL, like the BIO interface.
2791 NCONF_dump_* dump the internal storage of the configuration file,
2792 which is useful for debugging. All other functions take the same
2793 arguments as the old CONF_* functions wth the exception of the
2794 first that must be a `CONF *' instead of a `LHASH *'.
2796 To make it easer to use the new classes with the old CONF_* functions,
2797 the function CONF_set_default_method is provided.
2800 *) Add '-tls1' option to 'openssl ciphers', which was already
2801 mentioned in the documentation but had not been implemented.
2802 (This option is not yet really useful because even the additional
2803 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2806 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2807 OpenSSL-based applications) load shared libraries and bind to
2808 them in a portable way.
2809 [Geoff Thorpe, with contributions from Richard Levitte]
2811 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2813 *) Make sure _lrotl and _lrotr are only used with MSVC.
2815 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2816 (the default implementation of RAND_status).
2818 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2819 to '-clrext' (= clear extensions), as intended and documented.
2820 [Bodo Moeller; inconsistency pointed out by Michael Attili
2823 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2824 was larger than the MD block size.
2825 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2827 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2828 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2829 using the passed key: if the passed key was a private key the result
2830 of X509_print(), for example, would be to print out all the private key
2834 *) des_quad_cksum() byte order bug fix.
2835 [Ulf Möller, using the problem description in krb4-0.9.7, where
2836 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2838 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2840 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2842 *) For easily testing in shell scripts whether some command
2843 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2844 returns with exit code 0 iff no command of the given name is available.
2845 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2846 the output goes to stdout and nothing is printed to stderr.
2847 Additional arguments are always ignored.
2849 Since for each cipher there is a command of the same name,
2850 the 'no-cipher' compilation switches can be tested this way.
2852 ('openssl no-XXX' is not able to detect pseudo-commands such
2853 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2856 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2859 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2860 is set; it will be thrown away anyway because each handshake creates
2862 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2863 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2864 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2865 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2868 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2869 'Q' and 'R' lose their special meanings (quit/renegotiate).
2870 This is part of what -quiet does; unlike -quiet, -ign_eof
2871 does not suppress any output.
2874 *) Add compatibility options to the purpose and trust code. The
2875 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2876 accepts a certificate or CA, this was the previous behaviour,
2877 with all the associated security issues.
2879 X509_TRUST_COMPAT is the old trust behaviour: only and
2880 automatically trust self signed roots in certificate store. A
2881 new trust setting X509_TRUST_DEFAULT is used to specify that
2882 a purpose has no associated trust setting and it should instead
2883 use the value in the default purpose.
2886 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2887 and fix a memory leak.
2890 *) In util/mkerr.pl (which implements 'make errors'), preserve
2891 reason strings from the previous version of the .c file, as
2892 the default to have only downcase letters (and digits) in
2893 automatically generated reasons codes is not always appropriate.
2896 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2897 using strerror. Previously, ERR_reason_error_string() returned
2898 library names as reason strings for SYSerr; but SYSerr is a special
2899 case where small numbers are errno values, not library numbers.
2902 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2903 converts DSA parameters into DH parameters. (When creating parameters,
2904 DSA_generate_parameters is used.)
2907 *) Include 'length' (recommended exponent length) in C code generated
2908 by 'openssl dhparam -C'.
2911 *) The second argument to set_label in perlasm was already being used
2912 so couldn't be used as a "file scope" flag. Moved to third argument
2916 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2917 instead of RAND_bytes for encryption IVs and salts.
2920 *) Include RAND_status() into RAND_METHOD instead of implementing
2921 it only for md_rand.c Otherwise replacing the PRNG by calling
2922 RAND_set_rand_method would be impossible.
2925 *) Don't let DSA_generate_key() enter an infinite loop if the random
2926 number generation fails.
2929 *) New 'rand' application for creating pseudo-random output.
2932 *) Added configuration support for Linux/IA64
2933 [Rolf Haberrecker <rolf@suse.de>]
2935 *) Assembler module support for Mingw32.
2938 *) Shared library support for HPUX (in shlib/).
2939 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2941 *) Shared library support for Solaris gcc.
2942 [Lutz Behnke <behnke@trustcenter.de>]
2944 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2946 *) PKCS7_encrypt() was adding text MIME headers twice because they
2947 were added manually and by SMIME_crlf_copy().
2950 *) In bntest.c don't call BN_rand with zero bits argument.
2951 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
2953 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
2954 case was implemented. This caused BN_div_recp() to fail occasionally.
2957 *) Add an optional second argument to the set_label() in the perl
2958 assembly language builder. If this argument exists and is set
2959 to 1 it signals that the assembler should use a symbol whose
2960 scope is the entire file, not just the current function. This
2961 is needed with MASM which uses the format label:: for this scope.
2962 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
2964 *) Change the ASN1 types so they are typedefs by default. Before
2965 almost all types were #define'd to ASN1_STRING which was causing
2966 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
2970 *) Change names of new functions to the new get1/get0 naming
2971 convention: After 'get1', the caller owns a reference count
2972 and has to call ..._free; 'get0' returns a pointer to some
2973 data structure without incrementing reference counters.
2974 (Some of the existing 'get' functions increment a reference
2975 counter, some don't.)
2976 Similarly, 'set1' and 'add1' functions increase reference
2977 counters or duplicate objects.
2980 *) Allow for the possibility of temp RSA key generation failure:
2981 the code used to assume it always worked and crashed on failure.
2984 *) Fix potential buffer overrun problem in BIO_printf().
2985 [Ulf Möller, using public domain code by Patrick Powell; problem
2986 pointed out by David Sacerdote <das33@cornell.edu>]
2988 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
2989 RAND_egd() and RAND_status(). In the command line application,
2990 the EGD socket can be specified like a seed file using RANDFILE
2994 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
2995 Some CAs (e.g. Verisign) distribute certificates in this form.
2998 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
2999 list to exclude them. This means that no special compilation option
3000 is needed to use anonymous DH: it just needs to be included in the
3004 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
3005 EVP_MD_type. The old functionality is available in a new macro called
3006 EVP_MD_md(). Change code that uses it and update docs.
3009 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
3010 where the 'void *' argument is replaced by a function pointer argument.
3011 Previously 'void *' was abused to point to functions, which works on
3012 many platforms, but is not correct. As these functions are usually
3013 called by macros defined in OpenSSL header files, most source code
3014 should work without changes.
3017 *) <openssl/opensslconf.h> (which is created by Configure) now contains
3018 sections with information on -D... compiler switches used for
3019 compiling the library so that applications can see them. To enable
3020 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
3021 must be defined. E.g.,
3022 #define OPENSSL_ALGORITHM_DEFINES
3023 #include <openssl/opensslconf.h>
3024 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
3025 [Richard Levitte, Ulf and Bodo Möller]
3027 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
3031 *) Change the 'other' type in certificate aux info to a STACK_OF
3032 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
3033 the required ASN1 format: arbitrary types determined by an OID.
3036 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
3037 argument to 'req'. This is not because the function is newer or
3038 better than others it just uses the work 'NEW' in the certificate
3039 request header lines. Some software needs this.
3042 *) Reorganise password command line arguments: now passwords can be
3043 obtained from various sources. Delete the PEM_cb function and make
3044 it the default behaviour: i.e. if the callback is NULL and the
3045 usrdata argument is not NULL interpret it as a null terminated pass
3046 phrase. If usrdata and the callback are NULL then the pass phrase
3047 is prompted for as usual.
3050 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
3051 the support is automatically enabled. The resulting binaries will
3052 autodetect the card and use it if present.
3053 [Ben Laurie and Compaq Inc.]
3055 *) Work around for Netscape hang bug. This sends certificate request
3056 and server done in one record. Since this is perfectly legal in the
3057 SSL/TLS protocol it isn't a "bug" option and is on by default. See
3058 the bugs/SSLv3 entry for more info.
3061 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
3064 *) Add -rand argument to smime and pkcs12 applications and read/write
3068 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
3071 *) Add command line password options to the remaining applications.
3074 *) Bug fix for BN_div_recp() for numerators with an even number of
3078 *) More tests in bntest.c, and changed test_bn output.
3081 *) ./config recognizes MacOS X now.
3084 *) Bug fix for BN_div() when the first words of num and divsor are
3085 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
3088 *) Add support for various broken PKCS#8 formats, and command line
3089 options to produce them.
3092 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
3093 get temporary BIGNUMs from a BN_CTX.
3096 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
3100 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
3101 include a #define from the old name to the new. The original intent
3102 was that statically linked binaries could for example just call
3103 SSLeay_add_all_ciphers() to just add ciphers to the table and not
3104 link with digests. This never worked becayse SSLeay_add_all_digests()
3105 and SSLeay_add_all_ciphers() were in the same source file so calling
3106 one would link with the other. They are now in separate source files.
3109 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
3112 *) Use a less unusual form of the Miller-Rabin primality test (it used
3113 a binary algorithm for exponentiation integrated into the Miller-Rabin
3114 loop, our standard modexp algorithms are faster).
3117 *) Support for the EBCDIC character set completed.
3118 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
3120 *) Source code cleanups: use const where appropriate, eliminate casts,
3121 use void * instead of char * in lhash.
3124 *) Bugfix: ssl3_send_server_key_exchange was not restartable
3125 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
3126 this the server could overwrite ephemeral keys that the client
3130 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
3131 using 50 iterations of the Rabin-Miller test.
3133 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
3134 iterations of the Rabin-Miller test as required by the appendix
3135 to FIPS PUB 186[-1]) instead of DSA_is_prime.
3136 As BN_is_prime_fasttest includes trial division, DSA parameter
3137 generation becomes much faster.
3139 This implies a change for the callback functions in DSA_is_prime
3140 and DSA_generate_parameters: The callback function is called once
3141 for each positive witness in the Rabin-Miller test, not just
3142 occasionally in the inner loop; and the parameters to the
3143 callback function now provide an iteration count for the outer
3144 loop rather than for the current invocation of the inner loop.
3145 DSA_generate_parameters additionally can call the callback
3146 function with an 'iteration count' of -1, meaning that a
3147 candidate has passed the trial division test (when q is generated
3148 from an application-provided seed, trial division is skipped).
3151 *) New function BN_is_prime_fasttest that optionally does trial
3152 division before starting the Rabin-Miller test and has
3153 an additional BN_CTX * argument (whereas BN_is_prime always
3154 has to allocate at least one BN_CTX).
3155 'callback(1, -1, cb_arg)' is called when a number has passed the
3156 trial division stage.
3159 *) Fix for bug in CRL encoding. The validity dates weren't being handled
3163 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
3166 *) New function BN_pseudo_rand().
3169 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
3170 bignum version of BN_from_montgomery() with the working code from
3171 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
3175 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
3176 made it impossible to use the same SSL_SESSION data structure in
3177 SSL2 clients in multiple threads.
3180 *) The return value of RAND_load_file() no longer counts bytes obtained
3181 by stat(). RAND_load_file(..., -1) is new and uses the complete file
3182 to seed the PRNG (previously an explicit byte count was required).
3183 [Ulf Möller, Bodo Möller]
3185 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
3186 used (char *) instead of (void *) and had casts all over the place.
3189 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
3192 *) Retain source code compatibility for BN_prime_checks macro:
3193 BN_is_prime(..., BN_prime_checks, ...) now uses
3194 BN_prime_checks_for_size to determine the appropriate number of
3195 Rabin-Miller iterations.
3198 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
3199 DH_CHECK_P_NOT_SAFE_PRIME.
3200 (Check if this is true? OpenPGP calls them "strong".)
3203 *) Merge the functionality of "dh" and "gendh" programs into a new program
3204 "dhparam". The old programs are retained for now but will handle DH keys
3205 (instead of parameters) in future.
3208 *) Make the ciphers, s_server and s_client programs check the return values
3209 when a new cipher list is set.
3212 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
3213 ciphers. Before when the 56bit ciphers were enabled the sorting was
3216 The syntax for the cipher sorting has been extended to support sorting by
3217 cipher-strength (using the strength_bits hard coded in the tables).
3218 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
3220 Fix a bug in the cipher-command parser: when supplying a cipher command
3221 string with an "undefined" symbol (neither command nor alphanumeric
3222 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
3223 an error is flagged.
3225 Due to the strength-sorting extension, the code of the
3226 ssl_create_cipher_list() function was completely rearranged. I hope that
3227 the readability was also increased :-)
3228 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3230 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
3231 for the first serial number and places 2 in the serial number file. This
3232 avoids problems when the root CA is created with serial number zero and
3233 the first user certificate has the same issuer name and serial number
3237 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
3238 the new code. Add documentation for this stuff.
3241 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
3242 X509_*() to X509at_*() on the grounds that they don't handle X509
3243 structures and behave in an analagous way to the X509v3 functions:
3244 they shouldn't be called directly but wrapper functions should be used
3247 So we also now have some wrapper functions that call the X509at functions
3248 when passed certificate requests. (TO DO: similar things can be done with
3249 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
3250 things. Some of these need some d2i or i2d and print functionality
3251 because they handle more complex structures.)
3254 *) Add missing #ifndefs that caused missing symbols when building libssl
3255 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
3256 NO_RSA in ssl/s2*.c.
3257 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
3259 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
3260 has a return value which indicates the quality of the random data
3261 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
3262 error queue. New function RAND_pseudo_bytes() generates output that is
3263 guaranteed to be unique but not unpredictable. RAND_add is like
3264 RAND_seed, but takes an extra argument for an entropy estimate
3265 (RAND_seed always assumes full entropy).
3268 *) Do more iterations of Rabin-Miller probable prime test (specifically,
3269 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
3270 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
3271 in crypto/bn/bn_prime.c for the complete table). This guarantees a
3272 false-positive rate of at most 2^-80 for random input.
3275 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
3278 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
3279 in the 0.9.5 release), this returns the chain
3280 from an X509_CTX structure with a dup of the stack and all
3281 the X509 reference counts upped: so the stack will exist
3282 after X509_CTX_cleanup() has been called. Modify pkcs12.c
3285 Also make SSL_SESSION_print() print out the verify return
3289 *) Add manpage for the pkcs12 command. Also change the default
3290 behaviour so MAC iteration counts are used unless the new
3291 -nomaciter option is used. This improves file security and
3292 only older versions of MSIE (4.0 for example) need it.
3295 *) Honor the no-xxx Configure options when creating .DEF files.
3298 *) Add PKCS#10 attributes to field table: challengePassword,
3299 unstructuredName and unstructuredAddress. These are taken from
3300 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
3301 international characters are used.
3303 More changes to X509_ATTRIBUTE code: allow the setting of types
3304 based on strings. Remove the 'loc' parameter when adding
3305 attributes because these will be a SET OF encoding which is sorted
3309 *) Initial changes to the 'req' utility to allow request generation
3310 automation. This will allow an application to just generate a template
3311 file containing all the field values and have req construct the
3314 Initial support for X509_ATTRIBUTE handling. Stacks of these are
3315 used all over the place including certificate requests and PKCS#7
3316 structures. They are currently handled manually where necessary with
3317 some primitive wrappers for PKCS#7. The new functions behave in a
3318 manner analogous to the X509 extension functions: they allow
3319 attributes to be looked up by NID and added.
3321 Later something similar to the X509V3 code would be desirable to
3322 automatically handle the encoding, decoding and printing of the
3323 more complex types. The string types like challengePassword can
3324 be handled by the string table functions.
3326 Also modified the multi byte string table handling. Now there is
3327 a 'global mask' which masks out certain types. The table itself
3328 can use the flag STABLE_NO_MASK to ignore the mask setting: this
3329 is useful when for example there is only one permissible type
3330 (as in countryName) and using the mask might result in no valid
3334 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
3335 SSL_get_peer_finished to allow applications to obtain the latest
3336 Finished messages sent to the peer or expected from the peer,
3337 respectively. (SSL_get_peer_finished is usually the Finished message
3338 actually received from the peer, otherwise the protocol will be aborted.)
3340 As the Finished message are message digests of the complete handshake
3341 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
3342 be used for external authentication procedures when the authentication
3343 provided by SSL/TLS is not desired or is not enough.
3346 *) Enhanced support for Alpha Linux is added. Now ./config checks if
3347 the host supports BWX extension and if Compaq C is present on the
3348 $PATH. Just exploiting of the BWX extension results in 20-30%
3349 performance kick for some algorithms, e.g. DES and RC4 to mention
3350 a couple. Compaq C in turn generates ~20% faster code for MD5 and
3354 *) Add support for MS "fast SGC". This is arguably a violation of the
3355 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
3356 weak crypto and after checking the certificate is SGC a second one
3357 with strong crypto. MS SGC stops the first handshake after receiving
3358 the server certificate message and sends a second client hello. Since
3359 a server will typically do all the time consuming operations before
3360 expecting any further messages from the client (server key exchange
3361 is the most expensive) there is little difference between the two.
3363 To get OpenSSL to support MS SGC we have to permit a second client
3364 hello message after we have sent server done. In addition we have to
3365 reset the MAC if we do get this second client hello.
3368 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
3369 if a DER encoded private key is RSA or DSA traditional format. Changed
3370 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
3371 format DER encoded private key. Newer code should use PKCS#8 format which
3372 has the key type encoded in the ASN1 structure. Added DER private key
3373 support to pkcs8 application.
3376 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
3377 ciphersuites has been selected (as required by the SSL 3/TLS 1
3378 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
3379 is set, we interpret this as a request to violate the specification
3380 (the worst that can happen is a handshake failure, and 'correct'
3381 behaviour would result in a handshake failure anyway).
3384 *) In SSL_CTX_add_session, take into account that there might be multiple
3385 SSL_SESSION structures with the same session ID (e.g. when two threads
3386 concurrently obtain them from an external cache).
3387 The internal cache can handle only one SSL_SESSION with a given ID,
3388 so if there's a conflict, we now throw out the old one to achieve
3392 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
3393 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
3394 some routines that use cipher OIDs: some ciphers do not have OIDs
3395 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
3399 *) Simplify the trust setting structure and code. Now we just have
3400 two sequences of OIDs for trusted and rejected settings. These will
3401 typically have values the same as the extended key usage extension
3402 and any application specific purposes.
3404 The trust checking code now has a default behaviour: it will just
3405 check for an object with the same NID as the passed id. Functions can
3406 be provided to override either the default behaviour or the behaviour
3407 for a given id. SSL client, server and email already have functions
3408 in place for compatibility: they check the NID and also return "trusted"
3409 if the certificate is self signed.
3412 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
3413 traditional format into an EVP_PKEY structure.
3416 *) Add a password callback function PEM_cb() which either prompts for
3417 a password if usr_data is NULL or otherwise assumes it is a null
3418 terminated password. Allow passwords to be passed on command line
3419 environment or config files in a few more utilities.
3422 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
3423 keys. Add some short names for PKCS#8 PBE algorithms and allow them
3424 to be specified on the command line for the pkcs8 and pkcs12 utilities.
3425 Update documentation.
3428 *) Support for ASN1 "NULL" type. This could be handled before by using
3429 ASN1_TYPE but there wasn't any function that would try to read a NULL
3430 and produce an error if it couldn't. For compatibility we also have
3431 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
3432 don't allocate anything because they don't need to.
3435 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
3437 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
3439 *) Rebuild of the memory allocation routines used by OpenSSL code and
3440 possibly others as well. The purpose is to make an interface that
3441 provide hooks so anyone can build a separate set of allocation and
3442 deallocation routines to be used by OpenSSL, for example memory
3443 pool implementations, or something else, which was previously hard
3444 since Malloc(), Realloc() and Free() were defined as macros having
3445 the values malloc, realloc and free, respectively (except for Win32
3446 compilations). The same is provided for memory debugging code.
3447 OpenSSL already comes with functionality to find memory leaks, but
3448 this gives people a chance to debug other memory problems.
3450 With these changes, a new set of functions and macros have appeared:
3452 CRYPTO_set_mem_debug_functions() [F]
3453 CRYPTO_get_mem_debug_functions() [F]
3454 CRYPTO_dbg_set_options() [F]
3455 CRYPTO_dbg_get_options() [F]
3456 CRYPTO_malloc_debug_init() [M]
3458 The memory debug functions are NULL by default, unless the library
3459 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
3460 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
3461 gives the standard debugging functions that come with OpenSSL) or
3462 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
3463 provided by the library user) must be used. When the standard
3464 debugging functions are used, CRYPTO_dbg_set_options can be used to
3465 request additional information:
3466 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
3467 the CRYPTO_MDEBUG_xxx macro when compiling the library.
3469 Also, things like CRYPTO_set_mem_functions will always give the
3470 expected result (the new set of functions is used for allocation
3471 and deallocation) at all times, regardless of platform and compiler
3474 To finish it up, some functions that were never use in any other
3475 way than through macros have a new API and new semantic:
3478 CRYPTO_dbg_realloc()
3481 All macros of value have retained their old syntax.
3482 [Richard Levitte and Bodo Moeller]
3484 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
3485 ordering of SMIMECapabilities wasn't in "strength order" and there
3486 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
3490 *) Some ASN1 types with illegal zero length encoding (INTEGER,
3491 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
3492 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
3494 *) Merge in my S/MIME library for OpenSSL. This provides a simple
3495 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
3496 functionality to handle multipart/signed properly) and a utility
3497 called 'smime' to call all this stuff. This is based on code I
3498 originally wrote for Celo who have kindly allowed it to be
3499 included in OpenSSL.
3502 *) Add variants des_set_key_checked and des_set_key_unchecked of
3503 des_set_key (aka des_key_sched). Global variable des_check_key
3504 decides which of these is called by des_set_key; this way
3505 des_check_key behaves as it always did, but applications and
3506 the library itself, which was buggy for des_check_key == 1,
3507 have a cleaner way to pick the version they need.
3510 *) New function PKCS12_newpass() which changes the password of a
3514 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
3515 dynamic mix. In both cases the ids can be used as an index into the
3516 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
3517 functions so they accept a list of the field values and the
3518 application doesn't need to directly manipulate the X509_TRUST
3522 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
3526 *) Modify the way the V3 extension code looks up extensions. This now
3527 works in a similar way to the object code: we have some "standard"
3528 extensions in a static table which is searched with OBJ_bsearch()
3529 and the application can add dynamic ones if needed. The file
3530 crypto/x509v3/ext_dat.h now has the info: this file needs to be
3531 updated whenever a new extension is added to the core code and kept
3532 in ext_nid order. There is a simple program 'tabtest.c' which checks
3533 this. New extensions are not added too often so this file can readily
3534 be maintained manually.
3536 There are two big advantages in doing things this way. The extensions
3537 can be looked up immediately and no longer need to be "added" using
3538 X509V3_add_standard_extensions(): this function now does nothing.
3539 [Side note: I get *lots* of email saying the extension code doesn't
3540 work because people forget to call this function]
3541 Also no dynamic allocation is done unless new extensions are added:
3542 so if we don't add custom extensions there is no need to call
3543 X509V3_EXT_cleanup().
3546 *) Modify enc utility's salting as follows: make salting the default. Add a
3547 magic header, so unsalted files fail gracefully instead of just decrypting
3548 to garbage. This is because not salting is a big security hole, so people
3549 should be discouraged from doing it.
3552 *) Fixes and enhancements to the 'x509' utility. It allowed a message
3553 digest to be passed on the command line but it only used this
3554 parameter when signing a certificate. Modified so all relevant
3555 operations are affected by the digest parameter including the
3556 -fingerprint and -x509toreq options. Also -x509toreq choked if a
3557 DSA key was used because it didn't fix the digest.
3560 *) Initial certificate chain verify code. Currently tests the untrusted
3561 certificates for consistency with the verify purpose (which is set
3562 when the X509_STORE_CTX structure is set up) and checks the pathlength.
3564 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
3565 this is because it will reject chains with invalid extensions whereas
3566 every previous version of OpenSSL and SSLeay made no checks at all.
3568 Trust code: checks the root CA for the relevant trust settings. Trust
3569 settings have an initial value consistent with the verify purpose: e.g.
3570 if the verify purpose is for SSL client use it expects the CA to be
3571 trusted for SSL client use. However the default value can be changed to
3572 permit custom trust settings: one example of this would be to only trust
3573 certificates from a specific "secure" set of CAs.
3575 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
3576 which should be used for version portability: especially since the
3577 verify structure is likely to change more often now.
projects / openssl.git / blob