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 *) Only add signing time to PKCS7 structures if it is not already present.
18 *) Fix crypto/objects/objects.h: "ld-ce" should be "id-ce",
19 OBJ_ld_ce should be OBJ_id_ce.
20 Also some ip-pda OIDs in crypto/objects/objects.txt were
21 incorrect (cf. RFC 3039).
22 [Matt Cooper, Frederic Giudicelli, Bodo Moeller]
24 +) Add option to output public keys in req command.
25 [Massimiliano Pala madwolf@openca.org]
27 *) Release CRYPTO_LOCK_DYNLOCK when CRYPTO_destroy_dynlockid()
28 returns early because it has nothing to do.
29 [Andy Schneider <andy.schneider@bjss.co.uk>]
31 *) [In 0.9.6c-engine and 0.9.7 release:]
32 Fix mutex callback return values in crypto/engine/hw_ncipher.c.
33 [Andy Schneider <andy.schneider@bjss.co.uk>]
35 -) [In 0.9.6c-engine release:]
36 Add support for Cryptographic Appliance's keyserver technology.
37 (Use engine 'keyclient')
38 [Cryptographic Appliances and Geoff Thorpe]
40 *) Add a configuration entry for OS/390 Unix. The C compiler 'c89'
41 is called via tools/c89.sh because arguments have to be
42 rearranged (all '-L' options must appear before the first object
44 [Richard Shapiro <rshapiro@abinitio.com>]
46 +) Use wNAFs in EC_POINTs_mul() for improved efficiency
47 (up to about 10% better than before for P-192 and P-224).
50 -) [In 0.9.6c-engine release:]
51 Add support for Broadcom crypto accelerator cards, backported
53 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
55 -) [In 0.9.6c-engine release:]
56 Add support for SureWare crypto accelerator cards from
57 Baltimore Technologies. (Use engine 'sureware')
58 [Baltimore Technologies and Mark Cox]
60 -) [In 0.9.6c-engine release:]
61 Add support for crypto accelerator cards from Accelerated
62 Encryption Processing, www.aep.ie. (Use engine 'aep')
63 [AEP Inc. and Mark Cox]
65 *) Add a configuration entry for gcc on UnixWare.
66 [Gary Benson <gbenson@redhat.com>]
68 +) New functions/macros
70 SSL_CTX_set_msg_callback(ctx, cb)
71 SSL_CTX_set_msg_callback_arg(ctx, arg)
72 SSL_set_msg_callback(ssl, cb)
73 SSL_set_msg_callback_arg(ssl, arg)
75 to request calling a callback function
77 void cb(int write_p, int version, int content_type,
78 const void *buf, size_t len, SSL *ssl, void *arg)
80 whenever a protocol message has been completely received
81 (write_p == 0) or sent (write_p == 1). Here 'version' is the
82 protocol version according to which the SSL library interprets
83 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
84 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
85 the content type as defined in the SSL 3.0/TLS 1.0 protocol
86 specification (change_cipher_spec(20), alert(21), handshake(22)).
87 'buf' and 'len' point to the actual message, 'ssl' to the
88 SSL object, and 'arg' is the application-defined value set by
89 SSL[_CTX]_set_msg_callback_arg().
91 'openssl s_client' and 'openssl s_server' have new '-msg' options
92 to enable a callback that displays all protocol messages.
95 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
96 messages are stored in a single piece (fixed-length part and
97 variable-length part combined) and fix various bugs found on the way.
100 +) Change the shared library support so shared libraries are built as
101 soon as the corresponding static library is finished, and thereby get
102 openssl and the test programs linked against the shared library.
103 This still only happens when the keyword "shard" has been given to
104 the configuration scripts.
106 NOTE: shared library support is still an experimental thing, and
107 backward binary compatibility is still not guaranteed.
108 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
110 +) Add support for Subject Information Access extension.
111 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
113 +) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
114 additional bytes when new memory had to be allocated, not just
115 when reusing an existing buffer.
118 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
119 instead. BIO_gethostbyname() does not know what timeouts are
120 appropriate, so entries would stay in cache even when they have
122 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
124 +) New command line and configuration option 'utf8' for the req command.
125 This allows field values to be specified as UTF8 strings.
128 +) Add -multi and -mr options to "openssl speed" - giving multiple parallel
129 runs for the former and machine-readable output for the latter.
132 +) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
133 of the e-mail address in the DN (i.e., it will go into a certificate
134 extension only). The new configuration file option 'email_in_dn = no'
136 [Massimiliano Pala madwolf@openca.org]
138 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
139 faced with a pathologically small ClientHello fragment that does
140 not contain client_version: Instead of aborting with an error,
141 simply choose the highest available protocol version (i.e.,
142 TLS 1.0 unless it is disabled). In practice, ClientHello
143 messages are never sent like this, but this change gives us
144 strictly correct behaviour at least for TLS.
147 +) Change all functions with names starting with des_ to be starting
148 with DES_ instead. This because there are increasing clashes with
149 libdes and other des libraries that are currently used by other
150 projects. The old libdes interface is provided, as well as crypt(),
151 if openssl/des_old.h is included. Note that crypt() is no longer
152 declared in openssl/des.h.
154 NOTE: This is a major break of an old API into a new one. Software
155 authors are encouraged to switch to the DES_ style functions. Some
156 time in the future, des_old.h and the libdes compatibility functions
157 will be completely removed.
160 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
161 never resets s->method to s->ctx->method when called from within
162 one of the SSL handshake functions.
163 [Bodo Moeller; problem pointed out by Niko Baric]
165 +) Test for certificates which contain unsupported critical extensions.
166 If such a certificate is found during a verify operation it is
167 rejected by default: this behaviour can be overridden by either
168 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
169 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
170 X509_supported_extension() has also been added which returns 1 if a
171 particular extension is supported.
174 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
175 (sent using the client's version number) if client_version is
176 smaller than the protocol version in use. Also change
177 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
178 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
179 the client will at least see that alert.
182 +) Modify the behaviour of EVP cipher functions in similar way to digests
183 to retain compatibility with existing code.
186 +) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
187 compatibility with existing code. In particular the 'ctx' parameter does
188 not have to be to be initialized before the call to EVP_DigestInit() and
189 it is tidied up after a call to EVP_DigestFinal(). New function
190 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
191 EVP_MD_CTX_copy() changed to not require the destination to be
192 initialized valid and new function EVP_MD_CTX_copy_ex() added which
193 requires the destination to be valid.
195 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
196 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
199 +) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
200 so that complete 'Handshake' protocol structures are kept in memory
201 instead of overwriting 'msg_type' and 'length' with 'body' data.
204 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
208 +) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
209 [Massimo Santin via Richard Levitte]
211 +) Major restructuring to the underlying ENGINE code. This includes
212 reduction of linker bloat, separation of pure "ENGINE" manipulation
213 (initialisation, etc) from functionality dealing with implementations
214 of specific crypto iterfaces. This change also introduces integrated
215 support for symmetric ciphers and digest implementations - so ENGINEs
216 can now accelerate these by providing EVP_CIPHER and EVP_MD
217 implementations of their own. This is detailed in crypto/engine/README
218 as it couldn't be adequately described here. However, there are a few
219 API changes worth noting - some RSA, DSA, DH, and RAND functions that
220 were changed in the original introduction of ENGINE code have now
221 reverted back - the hooking from this code to ENGINE is now a good
222 deal more passive and at run-time, operations deal directly with
223 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
224 dereferencing through an ENGINE pointer any more. Also, the ENGINE
225 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
226 they were not being used by the framework as there is no concept of a
227 BIGNUM_METHOD and they could not be generalised to the new
228 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
229 ENGINE_cpy() has been removed as it cannot be consistently defined in
233 +) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
236 +) Change mkdef.pl to sort symbols that get the same entry number,
237 and make sure the automatically generated functions ERR_load_*
238 become part of libeay.num as well.
241 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
242 client receives HelloRequest while in a handshake.
243 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
245 +) New function SSL_renegotiate_pending(). This returns true once
246 renegotiation has been requested (either SSL_renegotiate() call
247 or HelloRequest/ClientHello receveived from the peer) and becomes
248 false once a handshake has been completed.
249 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
250 sends a HelloRequest, but does not ensure that a handshake takes
251 place. SSL_renegotiate_pending() is useful for checking if the
252 client has followed the request.)
255 +) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
256 By default, clients may request session resumption even during
257 renegotiation (if session ID contexts permit); with this option,
258 session resumption is possible only in the first handshake.
261 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
262 should end in 'break', not 'goto end' which circuments various
263 cleanups done in state SSL_ST_OK. But session related stuff
264 must be disabled for SSL_ST_OK in the case that we just sent a
267 Also avoid some overhead by not calling ssl_init_wbio_buffer()
268 before just sending a HelloRequest.
269 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
271 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
272 reveal whether illegal block cipher padding was found or a MAC
273 verification error occured. (Neither SSLerr() codes nor alerts
274 are directly visible to potential attackers, but the information
275 may leak via logfiles.)
277 Similar changes are not required for the SSL 2.0 implementation
278 because the number of padding bytes is sent in clear for SSL 2.0,
279 and the extra bytes are just ignored. However ssl/s2_pkt.c
280 failed to verify that the purported number of padding bytes is in
284 +) Add some demos for certificate and certificate request creation.
287 +) Make maximum certificate chain size accepted from the peer application
288 settable (SSL*_get/set_max_cert_list()), as proposed by
289 "Douglas E. Engert" <deengert@anl.gov>.
292 +) Add support for shared libraries for Unixware-7
293 (Boyd Lynn Gerber <gerberb@zenez.com>).
296 *) Add OpenUNIX-8 support including shared libraries
297 (Boyd Lynn Gerber <gerberb@zenez.com>).
300 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
301 'wristwatch attack' using huge encoding parameters (cf.
302 James H. Manger's CRYPTO 2001 paper). Note that the
303 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
304 encoding parameters and hence was not vulnerable.
307 +) Add a "destroy" handler to ENGINEs that allows structural cleanup to
308 be done prior to destruction. Use this to unload error strings from
309 ENGINEs that load their own error strings. NB: This adds two new API
310 functions to "get" and "set" this destroy handler in an ENGINE.
313 +) Alter all existing ENGINE implementations (except "openssl" and
314 "openbsd") to dynamically instantiate their own error strings. This
315 makes them more flexible to be built both as statically-linked ENGINEs
316 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
317 Also, add stub code to each that makes building them as self-contained
318 shared-libraries easier (see README.ENGINE).
321 +) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
322 implementations into applications that are completely implemented in
323 self-contained shared-libraries. The "dynamic" ENGINE exposes control
324 commands that can be used to configure what shared-library to load and
325 to control aspects of the way it is handled. Also, made an update to
326 the README.ENGINE file that brings its information up-to-date and
327 provides some information and instructions on the "dynamic" ENGINE
328 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
332 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
334 *) Make it possible to unload ranges of ERR strings with a new
335 "ERR_unload_strings" function.
338 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
339 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
340 followed by modular reduction.
341 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
343 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
344 equivalent based on BN_pseudo_rand() instead of BN_rand().
347 +) Add a copy() function to EVP_MD.
350 +) Make EVP_MD routines take a context pointer instead of just the
351 md_data void pointer.
354 +) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
355 that the digest can only process a single chunk of data
356 (typically because it is provided by a piece of
357 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
358 is only going to provide a single chunk of data, and hence the
359 framework needn't accumulate the data for oneshot drivers.
362 +) As with "ERR", make it possible to replace the underlying "ex_data"
363 functions. This change also alters the storage and management of global
364 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
365 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
366 index counters. The API functions that use this state have been changed
367 to take a "class_index" rather than pointers to the class's local STACK
368 and counter, and there is now an API function to dynamically create new
369 classes. This centralisation allows us to (a) plug a lot of the
370 thread-safety problems that existed, and (b) makes it possible to clean
371 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
372 such data would previously have always leaked in application code and
373 workarounds were in place to make the memory debugging turn a blind eye
374 to it. Application code that doesn't use this new function will still
375 leak as before, but their memory debugging output will announce it now
376 rather than letting it slide.
378 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
379 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
380 has a return value to indicate success or failure.
383 +) Make it possible to replace the underlying "ERR" functions such that the
384 global state (2 LHASH tables and 2 locks) is only used by the "default"
385 implementation. This change also adds two functions to "get" and "set"
386 the implementation prior to it being automatically set the first time
387 any other ERR function takes place. Ie. an application can call "get",
388 pass the return value to a module it has just loaded, and that module
389 can call its own "set" function using that value. This means the
390 module's "ERR" operations will use (and modify) the error state in the
391 application and not in its own statically linked copy of OpenSSL code.
394 +) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
395 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
396 the operation, and provides a more encapsulated way for external code
397 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
398 to use these functions rather than manually incrementing the counts.
400 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
403 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
404 This function was broken, as the check for a new client hello message
405 to handle SGC did not allow these large messages.
406 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
409 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
412 +) Add EVP test program.
415 +) Add symmetric cipher support to ENGINE. Expect the API to change!
418 +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
419 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
420 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
421 These allow a CRL to be built without having to access X509_CRL fields
422 directly. Modify 'ca' application to use new functions.
425 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
426 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
429 *) Rework the configuration and shared library support for Tru64 Unix.
430 The configuration part makes use of modern compiler features and
431 still retains old compiler behavior for those that run older versions
432 of the OS. The shared library support part includes a variant that
433 uses the RPATH feature, and is available through the special
434 configuration target "alpha-cc-rpath", which will never be selected
436 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
438 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
439 with the same message size as in ssl3_get_certificate_request().
440 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
441 messages might inadvertently be reject as too long.
442 [Petr Lampa <lampa@fee.vutbr.cz>]
444 +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
445 bug workarounds. Rollback attack detection is a security feature.
446 The problem will only arise on OpenSSL servers when TLSv1 is not
447 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
448 Software authors not wanting to support TLSv1 will have special reasons
449 for their choice and can explicitly enable this option.
450 [Bodo Moeller, Lutz Jaenicke]
452 +) Rationalise EVP so it can be extended: don't include a union of
453 cipher/digest structures, add init/cleanup functions. This also reduces
454 the number of header dependencies.
457 +) Make DES key schedule conform to the usual scheme, as well as
458 correcting its structure. This means that calls to DES functions
459 now have to pass a pointer to a des_key_schedule instead of a
460 plain des_key_schedule (which was actually always a pointer
464 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
467 *) Modified SSL library such that the verify_callback that has been set
468 specificly for an SSL object with SSL_set_verify() is actually being
469 used. Before the change, a verify_callback set with this function was
470 ignored and the verify_callback() set in the SSL_CTX at the time of
471 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
472 to allow the necessary settings.
475 +) Initial reduction of linker bloat: the use of some functions, such as
476 PEM causes large amounts of unused functions to be linked in due to
477 poor organisation. For example pem_all.c contains every PEM function
478 which has a knock on effect of linking in large amounts of (unused)
479 ASN1 code. Grouping together similar functions and splitting unrelated
480 functions prevents this.
483 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
484 explicitly to NULL, as at least on Solaris 8 this seems not always to be
485 done automatically (in contradiction to the requirements of the C
486 standard). This made problems when used from OpenSSH.
489 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
490 dh->length and always used
492 BN_rand_range(priv_key, dh->p).
494 BN_rand_range() is not necessary for Diffie-Hellman, and this
495 specific range makes Diffie-Hellman unnecessarily inefficient if
496 dh->length (recommended exponent length) is much smaller than the
497 length of dh->p. We could use BN_rand_range() if the order of
498 the subgroup was stored in the DH structure, but we only have
503 BN_rand(priv_key, l, ...)
505 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
511 RSA_eay_public_encrypt
512 RSA_eay_private_decrypt
513 RSA_eay_private_encrypt (signing)
514 RSA_eay_public_decrypt (signature verification)
516 (default implementations for RSA_public_encrypt,
517 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
518 always reject numbers >= n.
521 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
522 to synchronize access to 'locking_thread'. This is necessary on
523 systems where access to 'locking_thread' (an 'unsigned long'
524 variable) is not atomic.
527 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
528 *before* setting the 'crypto_lock_rand' flag. The previous code had
529 a race condition if 0 is a valid thread ID.
530 [Travis Vitek <vitek@roguewave.com>]
532 +) Cleanup of EVP macros.
535 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
539 +) Add initial OCSP responder support to ocsp application. The
540 revocation information is handled using the text based index
541 use by the ca application. The responder can either handle
542 requests generated internally, supplied in files (for example
543 via a CGI script) or using an internal minimal server.
546 +) Add configuration choices to get zlib compression for TLS.
549 +) Changes to Kerberos SSL for RFC 2712 compliance:
550 1. Implemented real KerberosWrapper, instead of just using
551 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
552 2. Implemented optional authenticator field of KerberosWrapper.
554 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
555 and authenticator structs; see crypto/krb5/.
557 Generalized Kerberos calls to support multiple Kerberos libraries.
558 [Vern Staats <staatsvr@asc.hpc.mil>,
559 Jeffrey Altman <jaltman@columbia.edu>
562 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
563 already does with RSA. testdsa.h now has 'priv_key/pub_key'
564 values for each of the key sizes rather than having just
565 parameters (and 'speed' generating keys each time).
568 -) OpenSSL 0.9.6b released [9 July 2001]
570 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
571 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
572 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
573 PRNG state recovery was possible based on the output of
574 one PRNG request appropriately sized to gain knowledge on
575 'md' followed by enough consecutive 1-byte PRNG requests
576 to traverse all of 'state'.
578 1. When updating 'md_local' (the current thread's copy of 'md')
579 during PRNG output generation, hash all of the previous
580 'md_local' value, not just the half used for PRNG output.
582 2. Make the number of bytes from 'state' included into the hash
583 independent from the number of PRNG bytes requested.
585 The first measure alone would be sufficient to avoid
586 Markku-Juhani's attack. (Actually it had never occurred
587 to me that the half of 'md_local' used for chaining was the
588 half from which PRNG output bytes were taken -- I had always
589 assumed that the secret half would be used.) The second
590 measure makes sure that additional data from 'state' is never
591 mixed into 'md_local' in small portions; this heuristically
592 further strengthens the PRNG.
595 +) Speed up EVP routines.
598 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
599 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
600 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
601 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
603 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
604 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
605 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
608 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
610 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
613 *) Fix crypto/bn/asm/mips3.s.
616 *) When only the key is given to "enc", the IV is undefined. Print out
617 an error message in this case.
620 +) Added the OS2-EMX target.
621 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
623 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
624 to support NCONF routines in extension code. New function CONF_set_nconf()
625 to allow functions which take an NCONF to also handle the old LHASH
626 structure: this means that the old CONF compatible routines can be
627 retained (in particular wrt extensions) without having to duplicate the
628 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
631 *) Handle special case when X509_NAME is empty in X509 printing routines.
634 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
635 positive and less than q.
638 +) Enhance the general user interface with mechanisms for inner control
639 and with possibilities to have yes/no kind of prompts.
642 +) Change all calls to low level digest routines in the library and
643 applications to use EVP. Add missing calls to HMAC_cleanup() and
644 don't assume HMAC_CTX can be copied using memcpy().
645 [Verdon Walker <VWalker@novell.com>, Steve Henson]
647 +) Add the possibility to control engines through control names but with
648 arbitrary arguments instead of just a string.
649 Change the key loaders to take a UI_METHOD instead of a callback
650 function pointer. NOTE: this breaks binary compatibility with earlier
651 versions of OpenSSL [engine].
652 Adapt the nCipher code for these new conditions and add a card insertion
656 +) Enhance the general user interface with mechanisms to better support
657 dialog box interfaces, application-defined prompts, the possibility
658 to use defaults (for example default passwords from somewhere else)
659 and interrupts/cancellations.
662 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
663 used: it isn't thread safe and the add_lock_callback should handle
665 [Paul Rose <Paul.Rose@bridge.com>]
667 *) Verify that incoming data obeys the block size in
668 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
671 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
672 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
676 [Ulf Möller, Bodo Möller]
678 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
679 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
680 when fixing the server behaviour for backwards-compatible 'client
681 hello' messages. (Note that the attack is impractical against
682 SSL 3.0 and TLS 1.0 anyway because length and version checking
683 means that the probability of guessing a valid ciphertext is
684 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
687 Before 0.9.5, the countermeasure (hide the error by generating a
688 random 'decryption result') did not work properly because
689 ERR_clear_error() was missing, meaning that SSL_get_error() would
690 detect the supposedly ignored error.
692 Both problems are now fixed.
695 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
696 (previously it was 1024).
699 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
700 tidy up some unnecessarily weird code in 'sk_new()').
701 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
703 +) Change the key loading routines for ENGINEs to use the same kind
704 callback (pem_password_cb) as all other routines that need this
708 *) Fix for compatibility mode trust settings: ignore trust settings
709 unless some valid trust or reject settings are present.
712 *) Fix for blowfish EVP: its a variable length cipher.
715 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
716 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
717 than this minimum value is recommended.
720 +) New random seeder for OpenVMS, using the system process statistics
721 that are easily reachable.
724 +) Windows apparently can't transparently handle global
725 variables defined in DLLs. Initialisations such as:
727 const ASN1_ITEM *it = &ASN1_INTEGER_it;
729 wont compile. This is used by the any applications that need to
730 declare their own ASN1 modules. This was fixed by adding the option
731 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
732 needed for static libraries under Win32.
735 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
736 setting of purpose and trust fields. New X509_STORE trust and
737 purpose functions and tidy up setting in other SSL functions.
740 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
741 structure. These are inherited by X509_STORE_CTX when it is
742 initialised. This allows various defaults to be set in the
743 X509_STORE structure (such as flags for CRL checking and custom
744 purpose or trust settings) for functions which only use X509_STORE_CTX
745 internally such as S/MIME.
747 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
748 trust settings if they are not set in X509_STORE. This allows X509_STORE
749 purposes and trust (in S/MIME for example) to override any set by default.
751 Add command line options for CRL checking to smime, s_client and s_server
755 +) Initial CRL based revocation checking. If the CRL checking flag(s)
756 are set then the CRL is looked up in the X509_STORE structure and
757 its validity and signature checked, then if the certificate is found
758 in the CRL the verify fails with a revoked error.
760 Various new CRL related callbacks added to X509_STORE_CTX structure.
762 Command line options added to 'verify' application to support this.
764 This needs some additional work, such as being able to handle multiple
765 CRLs with different times, extension based lookup (rather than just
766 by subject name) and ultimately more complete V2 CRL extension
770 +) Add a general user interface API (crypto/ui/). This is designed
771 to replace things like des_read_password and friends (backward
772 compatibility functions using this new API are provided).
773 The purpose is to remove prompting functions from the DES code
774 section as well as provide for prompting through dialog boxes in
775 a window system and the like.
778 *) Fix various bugs related to DSA S/MIME verification. Handle missing
779 parameters in DSA public key structures and return an error in the
780 DSA routines if parameters are absent.
783 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
784 in the current directory if neither $RANDFILE nor $HOME was set.
785 RAND_file_name() in 0.9.6a returned NULL in this case. This has
786 caused some confusion to Windows users who haven't defined $HOME.
787 Thus RAND_file_name() is changed again: e_os.h can define a
788 DEFAULT_HOME, which will be used if $HOME is not set.
789 For Windows, we use "C:"; on other platforms, we still require
790 environment variables.
792 +) Add "ex_data" support to ENGINE so implementations can add state at a
793 per-structure level rather than having to store it globally.
796 +) Make it possible for ENGINE structures to be copied when retrieved by
797 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
798 This causes the "original" ENGINE structure to act like a template,
799 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
800 operational state can be localised to each ENGINE structure, despite the
801 fact they all share the same "methods". New ENGINE structures returned in
802 this case have no functional references and the return value is the single
803 structural reference. This matches the single structural reference returned
804 by ENGINE_by_id() normally, when it is incremented on the pre-existing
808 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
809 needs to match any other type at all we need to manually clear the
813 +) Changes to the "openssl engine" utility to include;
814 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
815 about an ENGINE's available control commands.
816 - executing control commands from command line arguments using the
817 '-pre' and '-post' switches. '-post' is only used if '-t' is
818 specified and the ENGINE is successfully initialised. The syntax for
819 the individual commands are colon-separated, for example;
820 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
823 +) New dynamic control command support for ENGINEs. ENGINEs can now
824 declare their own commands (numbers), names (strings), descriptions,
825 and input types for run-time discovery by calling applications. A
826 subset of these commands are implicitly classed as "executable"
827 depending on their input type, and only these can be invoked through
828 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
829 can be based on user input, config files, etc). The distinction is
830 that "executable" commands cannot return anything other than a boolean
831 result and can only support numeric or string input, whereas some
832 discoverable commands may only be for direct use through
833 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
834 pointers, or other custom uses. The "executable" commands are to
835 support parameterisations of ENGINE behaviour that can be
836 unambiguously defined by ENGINEs and used consistently across any
837 OpenSSL-based application. Commands have been added to all the
838 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
839 control over shared-library paths without source code alterations.
842 +) Changed all ENGINE implementations to dynamically allocate their
843 ENGINEs rather than declaring them statically. Apart from this being
844 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
845 this also allows the implementations to compile without using the
846 internal engine_int.h header.
849 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
850 'const' value. Any code that should be able to modify a RAND_METHOD
851 should already have non-const pointers to it (ie. they should only
852 modify their own ones).
855 +) Made a variety of little tweaks to the ENGINE code.
856 - "atalla" and "ubsec" string definitions were moved from header files
857 to C code. "nuron" string definitions were placed in variables
858 rather than hard-coded - allowing parameterisation of these values
859 later on via ctrl() commands.
860 - Removed unused "#if 0"'d code.
861 - Fixed engine list iteration code so it uses ENGINE_free() to release
862 structural references.
863 - Constified the RAND_METHOD element of ENGINE structures.
864 - Constified various get/set functions as appropriate and added
865 missing functions (including a catch-all ENGINE_cpy that duplicates
866 all ENGINE values onto a new ENGINE except reference counts/state).
867 - Removed NULL parameter checks in get/set functions. Setting a method
868 or function to NULL is a way of cancelling out a previously set
869 value. Passing a NULL ENGINE parameter is just plain stupid anyway
870 and doesn't justify the extra error symbols and code.
871 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
872 flags from engine_int.h to engine.h.
873 - Changed prototypes for ENGINE handler functions (init(), finish(),
874 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
877 *) Move 'if (!initialized) RAND_poll()' into regions protected by
878 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
879 having multiple threads call RAND_poll() concurrently.
882 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
883 combination of a flag and a thread ID variable.
884 Otherwise while one thread is in ssleay_rand_bytes (which sets the
885 flag), *other* threads can enter ssleay_add_bytes without obeying
886 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
887 that they do not hold after the first thread unsets add_do_not_lock).
890 +) Implement binary inversion algorithm for BN_mod_inverse in addition
891 to the algorithm using long division. The binary algorithm can be
892 used only if the modulus is odd. On 32-bit systems, it is faster
893 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
894 roughly 5-15% for 256-bit moduli), so we use it only for moduli
895 up to 450 bits. In 64-bit environments, the binary algorithm
896 appears to be advantageous for much longer moduli; here we use it
897 for moduli up to 2048 bits.
900 *) Change bctest again: '-x' expressions are not available in all
904 -) OpenSSL 0.9.6a released [5 Apr 2001]
906 *) Fix a couple of memory leaks in PKCS7_dataDecode()
907 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
909 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
910 the default extension for executables, if any. Also, make the perl
911 scripts that use symlink() to test if it really exists and use "cp"
912 if it doesn't. All this made OpenSSL compilable and installable in
916 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
917 could not support the combine flag in choice fields.
920 -) Fix for asn1_GetSequence() for indefinite length constructed data.
921 If SEQUENCE is length is indefinite just set c->slen to the total
922 amount of data available.
923 [Steve Henson, reported by shige@FreeBSD.org]
924 [This change does not apply to 0.9.7.]
926 *) Change bctest to avoid here-documents inside command substitution
927 (workaround for FreeBSD /bin/sh bug).
928 For compatibility with Ultrix, avoid shell functions (introduced
929 in the bctest version that searches along $PATH).
932 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
933 with des_encrypt() defined on some operating systems, like Solaris
937 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
938 On the Importance of Eliminating Errors in Cryptographic
939 Computations, J. Cryptology 14 (2001) 2, 101-119,
940 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
943 *) MIPS assembler BIGNUM division bug fix.
946 *) Disabled incorrect Alpha assembler code.
949 -) Fix PKCS#7 decode routines so they correctly update the length
950 after reading an EOC for the EXPLICIT tag.
952 [This change does not apply to 0.9.7.]
954 *) Fix bug in PKCS#12 key generation routines. This was triggered
955 if a 3DES key was generated with a 0 initial byte. Include
956 PKCS12_BROKEN_KEYGEN compilation option to retain the old
957 (but broken) behaviour.
960 *) Enhance bctest to search for a working bc along $PATH and print
962 [Tim Rice <tim@multitalents.net> via Richard Levitte]
964 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
965 extensions from a certificate request to the certificate.
968 +) Allow multiple 'certopt' and 'nameopt' options to be separated
969 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
970 file: this allows the display of the certificate about to be
971 signed to be customised, to allow certain fields to be included
972 or excluded and extension details. The old system didn't display
973 multicharacter strings properly, omitted fields not in the policy
974 and couldn't display additional details such as extensions.
977 *) Fix memory leaks in err.c: free err_data string if necessary;
978 don't write to the wrong index in ERR_set_error_data.
981 +) Function EC_POINTs_mul for multiple scalar multiplication
982 of an arbitrary number of elliptic curve points
983 \sum scalars[i]*points[i],
984 optionally including the generator defined for the EC_GROUP:
985 scalar*generator + \sum scalars[i]*points[i].
987 EC_POINT_mul is a simple wrapper function for the typical case
988 that the point list has just one item (besides the optional
992 +) First EC_METHODs for curves over GF(p):
994 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
995 operations and provides various method functions that can also
996 operate with faster implementations of modular arithmetic.
998 EC_GFp_mont_method() reuses most functions that are part of
999 EC_GFp_simple_method, but uses Montgomery arithmetic.
1001 [Bodo Moeller; point addition and point doubling
1002 implementation directly derived from source code provided by
1003 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
1005 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
1006 crypto/ec/ec_lib.c):
1008 Curves are EC_GROUP objects (with an optional group generator)
1009 based on EC_METHODs that are built into the library.
1011 Points are EC_POINT objects based on EC_GROUP objects.
1013 Most of the framework would be able to handle curves over arbitrary
1014 finite fields, but as there are no obvious types for fields other
1015 than GF(p), some functions are limited to that for now.
1018 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
1019 that the file contains a complete HTTP response.
1022 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
1023 change the def and num file printf format specifier from "%-40sXXX"
1024 to "%-39s XXX". The latter will always guarantee a space after the
1025 field while the former will cause them to run together if the field
1026 is 40 of more characters long.
1029 +) Constify the cipher and digest 'method' functions and structures
1030 and modify related functions to take constant EVP_MD and EVP_CIPHER
1034 *) Implement ssl23_peek (analogous to ssl23_read), which previously
1038 *) Replace rdtsc with _emit statements for VC++ version 5.
1039 [Jeremy Cooper <jeremy@baymoo.org>]
1041 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1042 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1045 +) Modify EVP_Digest*() routines so they now return values. Although the
1046 internal software routines can never fail additional hardware versions
1050 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1052 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1053 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1056 ERR_R_NESTED_ASN1_ERROR
1058 ERR_R_MISSING_ASN1_EOS
1059 were 4 .. 9, conflicting with
1060 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1062 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1063 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1065 Add new error code 'ERR_R_INTERNAL_ERROR'.
1068 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1072 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1073 sets the subject name for a new request or supersedes the
1074 subject name in a given request. Formats that can be parsed are
1075 'CN=Some Name, OU=myOU, C=IT'
1077 'CN=Some Name/OU=myOU/C=IT'.
1079 Add options '-batch' and '-verbose' to 'openssl req'.
1080 [Massimiliano Pala <madwolf@hackmasters.net>]
1082 *) Make it possible to reuse SSLv2 sessions.
1085 +) Introduce the possibility to access global variables through
1086 functions on platform were that's the best way to handle exporting
1087 global variables in shared libraries. To enable this functionality,
1088 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1089 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1090 is normally done by Configure or something similar).
1092 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1093 in the source file (foo.c) like this:
1095 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1096 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1098 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1099 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1101 OPENSSL_DECLARE_GLOBAL(int,foo);
1102 #define foo OPENSSL_GLOBAL_REF(foo)
1103 OPENSSL_DECLARE_GLOBAL(double,bar);
1104 #define bar OPENSSL_GLOBAL_REF(bar)
1106 The #defines are very important, and therefore so is including the
1107 header file everywhere where the defined globals are used.
1109 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1110 of ASN.1 items, but that structure is a bit different.
1112 The largest change is in util/mkdef.pl which has been enhanced with
1113 better and easier to understand logic to choose which symbols should
1114 go into the Windows .def files as well as a number of fixes and code
1115 cleanup (among others, algorithm keywords are now sorted
1116 lexicographically to avoid constant rewrites).
1119 *) In copy_email() check for >= 0 as a return value for
1120 X509_NAME_get_index_by_NID() since 0 is a valid index.
1121 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
1123 +) In BN_div() keep a copy of the sign of 'num' before writing the
1124 result to 'rm' because if rm==num the value will be overwritten
1125 and produce the wrong result if 'num' is negative: this caused
1126 problems with BN_mod() and BN_nnmod().
1129 +) Function OCSP_request_verify(). This checks the signature on an
1130 OCSP request and verifies the signer certificate. The signer
1131 certificate is just checked for a generic purpose and OCSP request
1135 +) Add OCSP_check_validity() function to check the validity of OCSP
1136 responses. OCSP responses are prepared in real time and may only
1137 be a few seconds old. Simply checking that the current time lies
1138 between thisUpdate and nextUpdate max reject otherwise valid responses
1139 caused by either OCSP responder or client clock inaccuracy. Instead
1140 we allow thisUpdate and nextUpdate to fall within a certain period of
1141 the current time. The age of the response can also optionally be
1142 checked. Two new options -validity_period and -status_age added to
1146 +) If signature or public key algorithm is unrecognized print out its
1147 OID rather that just UNKNOWN.
1150 *) Avoid coredump with unsupported or invalid public keys by checking if
1151 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
1152 PKCS7_verify() fails with non detached data.
1155 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1156 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1157 ID to be generated from the issuer certificate alone which can then be
1158 passed to OCSP_id_issuer_cmp().
1161 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1162 ASN1 modules to export functions returning ASN1_ITEM pointers
1163 instead of the ASN1_ITEM structures themselves. This adds several
1164 new macros which allow the underlying ASN1 function/structure to
1165 be accessed transparently. As a result code should not use ASN1_ITEM
1166 references directly (such as &X509_it) but instead use the relevant
1167 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1168 use of the new ASN1 code on platforms where exporting structures
1169 is problematical (for example in shared libraries) but exporting
1170 functions returning pointers to structures is not.
1173 +) Add support for overriding the generation of SSL/TLS session IDs.
1174 These callbacks can be registered either in an SSL_CTX or per SSL.
1175 The purpose of this is to allow applications to control, if they wish,
1176 the arbitrary values chosen for use as session IDs, particularly as it
1177 can be useful for session caching in multiple-server environments. A
1178 command-line switch for testing this (and any client code that wishes
1179 to use such a feature) has been added to "s_server".
1180 [Geoff Thorpe, Lutz Jaenicke]
1182 +) Modify mkdef.pl to recognise and parse preprocessor conditionals
1183 of the form '#if defined(...) || defined(...) || ...' and
1184 '#if !defined(...) && !defined(...) && ...'. This also avoids
1185 the growing number of special cases it was previously handling.
1188 *) Don't use getenv in library functions when run as setuid/setgid.
1189 New function OPENSSL_issetugid().
1192 +) Make all configuration macros available for application by making
1193 sure they are available in opensslconf.h, by giving them names starting
1194 with "OPENSSL_" to avoid conflicts with other packages and by making
1195 sure e_os2.h will cover all platform-specific cases together with
1197 Additionally, it is now possible to define configuration/platform-
1198 specific names (called "system identities"). In the C code, these
1199 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1200 macro with the name beginning with "OPENSSL_SYS_", which is determined
1201 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1205 +) New option -set_serial to 'req' and 'x509' this allows the serial
1206 number to use to be specified on the command line. Previously self
1207 signed certificates were hard coded with serial number 0 and the
1208 CA options of 'x509' had to use a serial number in a file which was
1212 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
1213 due to incorrect handling of multi-threading:
1215 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
1217 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
1219 3. Count how many times MemCheck_off() has been called so that
1220 nested use can be treated correctly. This also avoids
1221 inband-signalling in the previous code (which relied on the
1222 assumption that thread ID 0 is impossible).
1225 +) New options to 'ca' utility to support V2 CRL entry extensions.
1226 Currently CRL reason, invalidity date and hold instruction are
1227 supported. Add new CRL extensions to V3 code and some new objects.
1230 *) Add "-rand" option also to s_client and s_server.
1233 +) New function EVP_CIPHER_CTX_set_padding() this is used to
1234 disable standard block padding (aka PKCS#5 padding) in the EVP
1235 API, which was previously mandatory. This means that the data is
1236 not padded in any way and so the total length much be a multiple
1237 of the block size, otherwise an error occurs.
1240 +) Initial (incomplete) OCSP SSL support.
1243 *) Fix CPU detection on Irix 6.x.
1244 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
1245 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1247 +) New function OCSP_parse_url(). This splits up a URL into its host,
1248 port and path components: primarily to parse OCSP URLs. New -url
1249 option to ocsp utility.
1252 +) New nonce behavior. The return value of OCSP_check_nonce() now
1253 reflects the various checks performed. Applications can decide
1254 whether to tolerate certain situations such as an absent nonce
1255 in a response when one was present in a request: the ocsp application
1256 just prints out a warning. New function OCSP_add1_basic_nonce()
1257 this is to allow responders to include a nonce in a response even if
1258 the request is nonce-less.
1261 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
1264 [This change does not apply to 0.9.7.]
1266 *) Use the cached encoding of an X509_NAME structure rather than
1267 copying it. This is apparently the reason for the libsafe "errors"
1268 but the code is actually correct.
1271 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1272 skipped when using openssl x509 multiple times on a single input file,
1273 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1276 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1277 set string type: to handle setting ASN1_TIME structures. Fix ca
1278 utility to correctly initialize revocation date of CRLs.
1281 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1282 the clients preferred ciphersuites and rather use its own preferences.
1283 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1284 Internet Explorer by ensuring unchanged hash method during stepup.
1285 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1288 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1289 to aes and add a new 'exist' option to print out symbols that don't
1293 +) Additional options to ocsp utility to allow flags to be set and
1294 additional certificates supplied.
1297 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
1298 OCSP client a number of certificate to only verify the response
1302 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
1303 Bleichenbacher's DSA attack.
1304 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
1305 to be set and top=0 forces the highest bit to be set; top=-1 is new
1306 and leaves the highest bit random.
1307 [Ulf Moeller, Bodo Moeller]
1309 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1310 handle the new API. Currently only ECB, CBC modes supported. Add new
1311 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
1312 for TLS" draft-ietf-tls-ciphersuite-03.txt.
1313 [Ben Laurie, Steve Henson]
1315 *) In the NCONF_...-based implementations for CONF_... queries
1316 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
1317 a temporary CONF structure with the data component set to NULL
1318 (which gives segmentation faults in lh_retrieve).
1319 Instead, use NULL for the CONF pointer in CONF_get_string and
1320 CONF_get_number (which may use environment variables) and directly
1321 return NULL from CONF_get_section.
1324 *) Fix potential buffer overrun for EBCDIC.
1327 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
1328 request to response.
1331 +) Functions for OCSP responders. OCSP_request_onereq_count(),
1332 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1333 extract information from a certificate request. OCSP_response_create()
1334 creates a response and optionally adds a basic response structure.
1335 OCSP_basic_add1_status() adds a complete single response to a basic
1336 response and returns the OCSP_SINGLERESP structure just added (to allow
1337 extensions to be included for example). OCSP_basic_add1_cert() adds a
1338 certificate to a basic response and OCSP_basic_sign() signs a basic
1339 response with various flags. New helper functions ASN1_TIME_check()
1340 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1341 (converts ASN1_TIME to GeneralizedTime).
1344 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1345 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1346 structure from a certificate. X509_pubkey_digest() digests the public_key
1347 contents: this is used in various key identifiers.
1350 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
1351 keyUsage if basicConstraints absent for a CA.
1354 *) Make SMIME_write_PKCS7() write mail header values with a format that
1355 is more generally accepted (no spaces before the semicolon), since
1356 some programs can't parse those values properly otherwise. Also make
1357 sure BIO's that break lines after each write do not create invalid
1361 +) Make sk_sort() tolerate a NULL argument.
1362 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1364 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1365 passed by the function are trusted implicitly. If any of them signed the
1366 response then it is assumed to be valid and is not verified.
1369 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
1370 macros previously used would not encode an empty SEQUENCE OF
1371 and break the signature.
1373 [This change does not apply to 0.9.7.]
1375 *) Zero the premaster secret after deriving the master secret in
1379 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1380 to data. This was previously part of the PKCS7 ASN1 code. This
1381 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1382 [Steve Henson, reported by Kenneth R. Robinette
1383 <support@securenetterm.com>]
1385 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1386 routines: without these tracing memory leaks is very painful.
1387 Fix leaks in PKCS12 and PKCS7 routines.
1390 *) Add some EVP_add_digest_alias registrations (as found in
1391 OpenSSL_add_all_digests()) to SSL_library_init()
1392 aka OpenSSL_add_ssl_algorithms(). This provides improved
1393 compatibility with peers using X.509 certificates
1394 with unconventional AlgorithmIdentifier OIDs.
1397 *) Fix for Irix with NO_ASM.
1398 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1400 *) ./config script fixes.
1401 [Ulf Moeller, Richard Levitte]
1403 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1404 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1405 effectively meant GeneralizedTime would never be used. Now it
1406 is initialised to -1 but X509_time_adj() now has to check the value
1407 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1408 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1409 [Steve Henson, reported by Kenneth R. Robinette
1410 <support@securenetterm.com>]
1412 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1413 result in a zero length in the ASN1_INTEGER structure which was
1414 not consistent with the structure when d2i_ASN1_INTEGER() was used
1415 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1416 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1417 where it did not print out a minus for negative ASN1_INTEGER.
1420 *) Fix 'openssl passwd -1'.
1423 +) Add summary printout to ocsp utility. The various functions which
1424 convert status values to strings have been renamed to:
1425 OCSP_response_status_str(), OCSP_cert_status_str() and
1426 OCSP_crl_reason_str() and are no longer static. New options
1427 to verify nonce values and to disable verification. OCSP response
1428 printout format cleaned up.
1431 +) Add additional OCSP certificate checks. These are those specified
1432 in RFC2560. This consists of two separate checks: the CA of the
1433 certificate being checked must either be the OCSP signer certificate
1434 or the issuer of the OCSP signer certificate. In the latter case the
1435 OCSP signer certificate must contain the OCSP signing extended key
1436 usage. This check is performed by attempting to match the OCSP
1437 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1438 in the OCSP_CERTID structures of the response.
1441 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1442 and related routines. This uses the standard OpenSSL certificate
1443 verify routines to perform initial checks (just CA validity) and
1444 to obtain the certificate chain. Then additional checks will be
1445 performed on the chain. Currently the root CA is checked to see
1446 if it is explicitly trusted for OCSP signing. This is used to set
1447 a root CA as a global signing root: that is any certificate that
1448 chains to that CA is an acceptable OCSP signing certificate.
1451 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1452 extensions from a separate configuration file.
1453 As when reading extensions from the main configuration file,
1454 the '-extensions ...' option may be used for specifying the
1456 [Massimiliano Pala <madwolf@comune.modena.it>]
1458 *) Change PKCS12_key_gen_asc() so it can cope with non null
1459 terminated strings whose length is passed in the passlen
1460 parameter, for example from PEM callbacks. This was done
1461 by adding an extra length parameter to asc2uni().
1462 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1464 +) New OCSP utility. Allows OCSP requests to be generated or
1465 read. The request can be sent to a responder and the output
1466 parsed, outputed or printed in text form. Not complete yet:
1467 still needs to check the OCSP response validity.
1470 +) New subcommands for 'openssl ca':
1471 'openssl ca -status <serial>' prints the status of the cert with
1472 the given serial number (according to the index file).
1473 'openssl ca -updatedb' updates the expiry status of certificates
1475 [Massimiliano Pala <madwolf@comune.modena.it>]
1477 +) New '-newreq-nodes' command option to CA.pl. This is like
1478 '-newreq', but calls 'openssl req' with the '-nodes' option
1479 so that the resulting key is not encrypted.
1480 [Damien Miller <djm@mindrot.org>]
1482 +) New configuration for the GNU Hurd.
1483 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1485 +) Initial code to implement OCSP basic response verify. This
1486 is currently incomplete. Currently just finds the signer's
1487 certificate and verifies the signature on the response.
1490 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1491 value of OPENSSLDIR. This is available via the new '-d' option
1492 to 'openssl version', and is also included in 'openssl version -a'.
1495 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1496 call failed, free the DSA structure.
1499 +) Allowing defining memory allocation callbacks that will be given
1500 file name and line number information in additional arguments
1501 (a const char* and an int). The basic functionality remains, as
1502 well as the original possibility to just replace malloc(),
1503 realloc() and free() by functions that do not know about these
1504 additional arguments. To register and find out the current
1505 settings for extended allocation functions, the following
1506 functions are provided:
1508 CRYPTO_set_mem_ex_functions
1509 CRYPTO_set_locked_mem_ex_functions
1510 CRYPTO_get_mem_ex_functions
1511 CRYPTO_get_locked_mem_ex_functions
1513 These work the same way as CRYPTO_set_mem_functions and friends.
1514 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1515 extended allocation function is enabled.
1516 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1517 a conventional allocation function is enabled.
1518 [Richard Levitte, Bodo Moeller]
1520 *) Fix to uni2asc() to cope with zero length Unicode strings.
1521 These are present in some PKCS#12 files.
1524 +) Finish off removing the remaining LHASH function pointer casts.
1525 There should no longer be any prototype-casting required when using
1526 the LHASH abstraction, and any casts that remain are "bugs". See
1527 the callback types and macros at the head of lhash.h for details
1528 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1531 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1532 If /dev/[u]random devices are not available or do not return enough
1533 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1535 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1536 /etc/entropy will be queried once each in this sequence, quering stops
1537 when enough entropy was collected without querying more sockets.
1540 +) Change the Unix RAND_poll() variant to be able to poll several
1541 random devices, as specified by DEVRANDOM, until a sufficient amount
1542 of data has been collected. We spend at most 10 ms on each file
1543 (select timeout) and read in non-blocking mode. DEVRANDOM now
1544 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1545 (previously it was just the string "/dev/urandom"), so on typical
1546 platforms the 10 ms delay will never occur.
1547 Also separate out the Unix variant to its own file, rand_unix.c.
1548 For VMS, there's a currently-empty rand_vms.c.
1551 +) Move OCSP client related routines to ocsp_cl.c. These
1552 provide utility functions which an application needing
1553 to issue a request to an OCSP responder and analyse the
1554 response will typically need: as opposed to those which an
1555 OCSP responder itself would need which will be added later.
1557 OCSP_request_sign() signs an OCSP request with an API similar
1558 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1559 response. OCSP_response_get1_basic() extracts basic response
1560 from response. OCSP_resp_find_status(): finds and extracts status
1561 information from an OCSP_CERTID structure (which will be created
1562 when the request structure is built). These are built from lower
1563 level functions which work on OCSP_SINGLERESP structures but
1564 wont normally be used unless the application wishes to examine
1565 extensions in the OCSP response for example.
1567 Replace nonce routines with a pair of functions.
1568 OCSP_request_add1_nonce() adds a nonce value and optionally
1569 generates a random value. OCSP_check_nonce() checks the
1570 validity of the nonce in an OCSP response.
1573 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1574 This doesn't copy the supplied OCSP_CERTID and avoids the
1575 need to free up the newly created id. Change return type
1576 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1577 This can then be used to add extensions to the request.
1578 Deleted OCSP_request_new(), since most of its functionality
1579 is now in OCSP_REQUEST_new() (and the case insensitive name
1580 clash) apart from the ability to set the request name which
1581 will be added elsewhere.
1584 +) Update OCSP API. Remove obsolete extensions argument from
1585 various functions. Extensions are now handled using the new
1586 OCSP extension code. New simple OCSP HTTP function which
1587 can be used to send requests and parse the response.
1590 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1591 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1592 uses the special reorder version of SET OF to sort the attributes
1593 and reorder them to match the encoded order. This resolves a long
1594 standing problem: a verify on a PKCS7 structure just after signing
1595 it used to fail because the attribute order did not match the
1596 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1597 it uses the received order. This is necessary to tolerate some broken
1598 software that does not order SET OF. This is handled by encoding
1599 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1600 to produce the required SET OF.
1603 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1604 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1605 files to get correct declarations of the ASN.1 item variables.
1608 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1609 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1610 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1611 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1612 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1613 ASN1_ITEM and no wrapper functions.
1616 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1617 replace the old function pointer based I/O routines. Change most of
1618 the *_d2i_bio() and *_d2i_fp() functions to use these.
1621 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1622 lines, recognice more "algorithms" that can be deselected, and make
1623 it complain about algorithm deselection that isn't recognised.
1626 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1627 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1628 to use new functions. Add NO_ASN1_OLD which can be set to remove
1629 some old style ASN1 functions: this can be used to determine if old
1630 code will still work when these eventually go away.
1633 +) New extension functions for OCSP structures, these follow the
1634 same conventions as certificates and CRLs.
1637 +) New function X509V3_add1_i2d(). This automatically encodes and
1638 adds an extension. Its behaviour can be customised with various
1639 flags to append, replace or delete. Various wrappers added for
1640 certifcates and CRLs.
1643 +) Fix to avoid calling the underlying ASN1 print routine when
1644 an extension cannot be parsed. Correct a typo in the
1645 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1648 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1649 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1650 when writing a 32767 byte record.
1651 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1653 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1654 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1656 (RSA objects have a reference count access to which is protected
1657 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1658 so they are meant to be shared between threads.)
1659 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1660 "Reddie, Steven" <Steven.Reddie@ca.com>]
1662 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1663 entries for variables.
1666 *) Fix a deadlock in CRYPTO_mem_leaks().
1669 +) Add functionality to apps/openssl.c for detecting locking
1670 problems: As the program is single-threaded, all we have
1671 to do is register a locking callback using an array for
1672 storing which locks are currently held by the program.
1675 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1676 SSL_get_ex_data_X509_STORE_idx(), which is used in
1677 ssl_verify_cert_chain() and thus can be called at any time
1678 during TLS/SSL handshakes so that thread-safety is essential.
1679 Unfortunately, the ex_data design is not at all suited
1680 for multi-threaded use, so it probably should be abolished.
1683 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1684 [Broadcom, tweaked and integrated by Geoff Thorpe]
1686 +) Move common extension printing code to new function
1687 X509V3_print_extensions(). Reorganise OCSP print routines and
1688 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1691 +) New function X509_signature_print() to remove duplication in some
1695 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1696 set (this was treated exactly the same as SET OF previously). This
1697 is used to reorder the STACK representing the structure to match the
1698 encoding. This will be used to get round a problem where a PKCS7
1699 structure which was signed could not be verified because the STACK
1700 order did not reflect the encoded order.
1703 +) Reimplement the OCSP ASN1 module using the new code.
1706 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1707 for its ASN1 operations. The old style function pointers still exist
1708 for now but they will eventually go away.
1711 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1712 completely replaces the old ASN1 functionality with a table driven
1713 encoder and decoder which interprets an ASN1_ITEM structure describing
1714 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1715 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1716 has also been converted to the new form.
1719 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1720 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1721 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1722 for negative moduli.
1725 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1726 of not touching the result's sign bit.
1729 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1733 +) Changed the LHASH code to use prototypes for callbacks, and created
1734 macros to declare and implement thin (optionally static) functions
1735 that provide type-safety and avoid function pointer casting for the
1736 type-specific callbacks.
1739 *) Use better test patterns in bntest.
1742 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1744 [Veers Staats <staatsvr@asc.hpc.mil>,
1745 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1747 *) rand_win.c fix for Borland C.
1750 *) BN_rshift bugfix for n == 0.
1753 +) Reformat the FAQ so the different questions and answers can be divided
1754 in sections depending on the subject.
1757 +) Have the zlib compression code load ZLIB.DLL dynamically under
1761 +) New function BN_mod_sqrt for computing square roots modulo a prime
1762 (using the probabilistic Tonelli-Shanks algorithm unless
1763 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1764 be handled deterministically).
1765 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1767 *) Add a 'bctest' script that checks for some known 'bc' bugs
1768 so that 'make test' does not abort just because 'bc' is broken.
1771 *) Store verify_result within SSL_SESSION also for client side to
1772 avoid potential security hole. (Re-used sessions on the client side
1773 always resulted in verify_result==X509_V_OK, not using the original
1774 result of the server certificate verification.)
1777 +) Make BN_mod_inverse faster by explicitly handling small quotients
1778 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1779 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1782 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1783 SSL3_RT_APPLICATION_DATA, return 0.
1784 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1788 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1789 releases, have been re-implemented by renaming the previous
1790 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1791 and ssl3_read_internal, respectively, and adding 'peek' parameters
1792 to them. The new ssl[23]_{read,peek} functions are calls to
1793 ssl[23]_read_internal with the 'peek' flag set appropriately.
1794 A 'peek' parameter has also been added to ssl3_read_bytes, which
1795 does the actual work for ssl3_read_internal.
1798 +) New function BN_kronecker.
1801 +) Fix BN_gcd so that it works on negative inputs; the result is
1802 positive unless both parameters are zero.
1803 Previously something reasonably close to an infinite loop was
1804 possible because numbers could be growing instead of shrinking
1805 in the implementation of Euclid's algorithm.
1808 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1809 sign of the number in question.
1811 Fix BN_is_word(a,w) to work correctly for w == 0.
1813 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1814 because its test if the absolute value of 'a' equals 'w'.
1815 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1816 it exists mostly for use in the implementations of BN_is_zero(),
1817 BN_is_one(), and BN_is_word().
1820 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1821 the method-specific "init()" handler. Also clean up ex_data after
1822 calling the method-specific "finish()" handler. Previously, this was
1823 happening the other way round.
1826 +) New function BN_swap.
1829 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1830 the exponentiation functions are more likely to produce reasonable
1831 results on negative inputs.
1834 +) Change BN_mod_mul so that the result is always non-negative.
1835 Previously, it could be negative if one of the factors was negative;
1836 I don't think anyone really wanted that behaviour.
1839 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1840 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1841 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1842 and add new functions:
1851 BN_mod_lshift1_quick
1855 These functions always generate non-negative results.
1857 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1858 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1860 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1861 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1862 be reduced modulo m.
1863 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1865 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1866 was actually never needed) and in BN_mul(). The removal in BN_mul()
1867 required a small change in bn_mul_part_recursive() and the addition
1868 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1869 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1870 bn_sub_words() and bn_add_words() except they take arrays with
1874 +) In 'openssl passwd', verify passwords read from the terminal
1875 unless the '-salt' option is used (which usually means that
1876 verification would just waste user's time since the resulting
1877 hash is going to be compared with some given password hash)
1878 or the new '-noverify' option is used.
1880 This is an incompatible change, but it does not affect
1881 non-interactive use of 'openssl passwd' (passwords on the command
1882 line, '-stdin' option, '-in ...' option) and thus should not
1886 +) Remove all references to RSAref, since there's no more need for it.
1889 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
1890 The previous value, 12, was not always sufficient for BN_mod_exp().
1893 +) Make DSO load along a path given through an environment variable
1894 (SHLIB_PATH) with shl_load().
1897 +) Constify the ENGINE code as a result of BIGNUM constification.
1898 Also constify the RSA code and most things related to it. In a
1899 few places, most notable in the depth of the ASN.1 code, ugly
1900 casts back to non-const were required (to be solved at a later
1904 +) Make it so the openssl application has all engines loaded by default.
1907 +) Constify the BIGNUM routines a little more.
1910 *) Make sure that shared libraries get the internal name engine with
1911 the full version number and not just 0. This should mark the
1912 shared libraries as not backward compatible. Of course, this should
1913 be changed again when we can guarantee backward binary compatibility.
1916 +) Add the following functions:
1918 ENGINE_load_cswift()
1920 ENGINE_load_atalla()
1922 ENGINE_load_builtin_engines()
1924 That way, an application can itself choose if external engines that
1925 are built-in in OpenSSL shall ever be used or not. The benefit is
1926 that applications won't have to be linked with libdl or other dso
1927 libraries unless it's really needed.
1929 Changed 'openssl engine' to load all engines on demand.
1930 Changed the engine header files to avoid the duplication of some
1931 declarations (they differed!).
1934 +) 'openssl engine' can now list capabilities.
1937 +) Better error reporting in 'openssl engine'.
1940 +) Never call load_dh_param(NULL) in s_server.
1943 +) Add engine application. It can currently list engines by name and
1944 identity, and test if they are actually available.
1947 *) Add support for shared libraries under Irix.
1948 [Albert Chin-A-Young <china@thewrittenword.com>]
1950 +) Improve RPM specification file by forcing symbolic linking and making
1951 sure the installed documentation is also owned by root.root.
1952 [Damien Miller <djm@mindrot.org>]
1954 *) Add configuration option to build on Linux on both big-endian and
1956 [Ralf Baechle <ralf@uni-koblenz.de>]
1958 +) Give the OpenSSL applications more possibilities to make use of
1959 keys (public as well as private) handled by engines.
1962 +) Add OCSP code that comes from CertCo.
1965 +) Add VMS support for the Rijndael code.
1968 +) Added untested support for Nuron crypto accelerator.
1971 +) Add support for external cryptographic devices. This code was
1972 previously distributed separately as the "engine" branch.
1973 [Geoff Thorpe, Richard Levitte]
1975 +) Rework the filename-translation in the DSO code. It is now possible to
1976 have far greater control over how a "name" is turned into a filename
1977 depending on the operating environment and any oddities about the
1978 different shared library filenames on each system.
1981 +) Support threads on FreeBSD-elf in Configure.
1984 *) Add the possibility to create shared libraries on HP-UX
1987 +) Fix for SHA1 assembly problem with MASM: it produces
1988 warnings about corrupt line number information when assembling
1989 with debugging information. This is caused by the overlapping
1991 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
1994 NCONF_get_number() has no error checking at all. As a replacement,
1995 NCONF_get_number_e() is defined (_e for "error checking") and is
1996 promoted strongly. The old NCONF_get_number is kept around for
1997 binary backward compatibility.
1998 Make it possible for methods to load from something other than a BIO,
1999 by providing a function pointer that is given a name instead of a BIO.
2000 For example, this could be used to load configuration data from an
2004 *) Fix typo in get_cert_by_subject() in by_dir.c
2005 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
2007 *) Rework the system to generate shared libraries:
2009 - Make note of the expected extension for the shared libraries and
2010 if there is a need for symbolic links from for example libcrypto.so.0
2011 to libcrypto.so.0.9.7. There is extended info in Configure for
2014 - Make as few rebuilds of the shared libraries as possible.
2016 - Still avoid linking the OpenSSL programs with the shared libraries.
2018 - When installing, install the shared libraries separately from the
2022 +) Fix for non blocking accept BIOs. Added new I/O special reason
2023 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
2024 with non blocking I/O was not possible because no retry code was
2025 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
2029 +) Added the beginnings of Rijndael support.
2032 +) Fix for bug in DirectoryString mask setting. Add support for
2033 X509_NAME_print_ex() in 'req' and X509_print_ex() function
2034 to allow certificate printing to more controllable, additional
2035 'certopt' option to 'x509' to allow new printing options to be
2039 +) Clean old EAY MD5 hack from e_os.h.
2042 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
2044 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
2045 and not in SSL_clear because the latter is also used by the
2046 accept/connect functions; previously, the settings made by
2047 SSL_set_read_ahead would be lost during the handshake.
2048 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
2050 *) Correct util/mkdef.pl to be selective about disabled algorithms.
2051 Previously, it would create entries for disableed algorithms no
2055 *) Added several new manual pages for SSL_* function.
2058 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
2060 *) In ssl23_get_client_hello, generate an error message when faced
2061 with an initial SSL 3.0/TLS record that is too small to contain the
2062 first two bytes of the ClientHello message, i.e. client_version.
2063 (Note that this is a pathologic case that probably has never happened
2064 in real life.) The previous approach was to use the version number
2065 from the record header as a substitute; but our protocol choice
2066 should not depend on that one because it is not authenticated
2067 by the Finished messages.
2070 *) More robust randomness gathering functions for Windows.
2071 [Jeffrey Altman <jaltman@columbia.edu>]
2073 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
2074 not set then we don't setup the error code for issuer check errors
2075 to avoid possibly overwriting other errors which the callback does
2076 handle. If an application does set the flag then we assume it knows
2077 what it is doing and can handle the new informational codes
2081 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
2082 a general "ANY" type, as such it should be able to decode anything
2083 including tagged types. However it didn't check the class so it would
2084 wrongly interpret tagged types in the same way as their universal
2085 counterpart and unknown types were just rejected. Changed so that the
2086 tagged and unknown types are handled in the same way as a SEQUENCE:
2087 that is the encoding is stored intact. There is also a new type
2088 "V_ASN1_OTHER" which is used when the class is not universal, in this
2089 case we have no idea what the actual type is so we just lump them all
2093 *) On VMS, stdout may very well lead to a file that is written to
2094 in a record-oriented fashion. That means that every write() will
2095 write a separate record, which will be read separately by the
2096 programs trying to read from it. This can be very confusing.
2098 The solution is to put a BIO filter in the way that will buffer
2099 text until a linefeed is reached, and then write everything a
2100 line at a time, so every record written will be an actual line,
2101 not chunks of lines and not (usually doesn't happen, but I've
2102 seen it once) several lines in one record. BIO_f_linebuffer() is
2105 Currently, it's a VMS-only method, because that's where it has
2106 been tested well enough.
2109 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
2110 it can return incorrect results.
2111 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
2112 but it was in 0.9.6-beta[12].)
2115 *) Disable the check for content being present when verifying detached
2116 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
2117 include zero length content when signing messages.
2120 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
2121 BIO_ctrl (for BIO pairs).
2124 *) Add DSO method for VMS.
2127 *) Bug fix: Montgomery multiplication could produce results with the
2131 *) Add RPM specification openssl.spec and modify it to build three
2132 packages. The default package contains applications, application
2133 documentation and run-time libraries. The devel package contains
2134 include files, static libraries and function documentation. The
2135 doc package contains the contents of the doc directory. The original
2136 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
2139 *) Add a large number of documentation files for many SSL routines.
2140 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2142 *) Add a configuration entry for Sony News 4.
2143 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
2145 *) Don't set the two most significant bits to one when generating a
2146 random number < q in the DSA library.
2149 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
2150 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
2151 the underlying transport is blocking) if a handshake took place.
2152 (The default behaviour is needed by applications such as s_client
2153 and s_server that use select() to determine when to use SSL_read;
2154 but for applications that know in advance when to expect data, it
2155 just makes things more complicated.)
2158 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
2162 *) Add a few more EBCDIC conditionals that make `req' and `x509'
2163 work better on such systems.
2164 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
2166 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
2167 Update PKCS12_parse() so it copies the friendlyName and the
2168 keyid to the certificates aux info.
2171 *) Fix bug in PKCS7_verify() which caused an infinite loop
2172 if there was more than one signature.
2173 [Sven Uszpelkat <su@celocom.de>]
2175 *) Major change in util/mkdef.pl to include extra information
2176 about each symbol, as well as presentig variables as well
2177 as functions. This change means that there's n more need
2178 to rebuild the .num files when some algorithms are excluded.
2181 *) Allow the verify time to be set by an application,
2182 rather than always using the current time.
2185 *) Phase 2 verify code reorganisation. The certificate
2186 verify code now looks up an issuer certificate by a
2187 number of criteria: subject name, authority key id
2188 and key usage. It also verifies self signed certificates
2189 by the same criteria. The main comparison function is
2190 X509_check_issued() which performs these checks.
2192 Lot of changes were necessary in order to support this
2193 without completely rewriting the lookup code.
2195 Authority and subject key identifier are now cached.
2197 The LHASH 'certs' is X509_STORE has now been replaced
2198 by a STACK_OF(X509_OBJECT). This is mainly because an
2199 LHASH can't store or retrieve multiple objects with
2200 the same hash value.
2202 As a result various functions (which were all internal
2203 use only) have changed to handle the new X509_STORE
2204 structure. This will break anything that messed round
2205 with X509_STORE internally.
2207 The functions X509_STORE_add_cert() now checks for an
2208 exact match, rather than just subject name.
2210 The X509_STORE API doesn't directly support the retrieval
2211 of multiple certificates matching a given criteria, however
2212 this can be worked round by performing a lookup first
2213 (which will fill the cache with candidate certificates)
2214 and then examining the cache for matches. This is probably
2215 the best we can do without throwing out X509_LOOKUP
2216 entirely (maybe later...).
2218 The X509_VERIFY_CTX structure has been enhanced considerably.
2220 All certificate lookup operations now go via a get_issuer()
2221 callback. Although this currently uses an X509_STORE it
2222 can be replaced by custom lookups. This is a simple way
2223 to bypass the X509_STORE hackery necessary to make this
2224 work and makes it possible to use more efficient techniques
2225 in future. A very simple version which uses a simple
2226 STACK for its trusted certificate store is also provided
2227 using X509_STORE_CTX_trusted_stack().
2229 The verify_cb() and verify() callbacks now have equivalents
2230 in the X509_STORE_CTX structure.
2232 X509_STORE_CTX also has a 'flags' field which can be used
2233 to customise the verify behaviour.
2236 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
2237 excludes S/MIME capabilities.
2240 *) When a certificate request is read in keep a copy of the
2241 original encoding of the signed data and use it when outputing
2242 again. Signatures then use the original encoding rather than
2243 a decoded, encoded version which may cause problems if the
2244 request is improperly encoded.
2247 *) For consistency with other BIO_puts implementations, call
2248 buffer_write(b, ...) directly in buffer_puts instead of calling
2251 In BIO_puts, increment b->num_write as in BIO_write.
2252 [Peter.Sylvester@EdelWeb.fr]
2254 *) Fix BN_mul_word for the case where the word is 0. (We have to use
2255 BN_zero, we may not return a BIGNUM with an array consisting of
2259 *) Avoid calling abort() from within the library when problems are
2260 detected, except if preprocessor symbols have been defined
2261 (such as REF_CHECK, BN_DEBUG etc.).
2264 *) New openssl application 'rsautl'. This utility can be
2265 used for low level RSA operations. DER public key
2266 BIO/fp routines also added.
2269 *) New Configure entry and patches for compiling on QNX 4.
2270 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
2272 *) A demo state-machine implementation was sponsored by
2273 Nuron (http://www.nuron.com/) and is now available in
2274 demos/state_machine.
2277 *) New options added to the 'dgst' utility for signature
2278 generation and verification.
2281 *) Unrecognized PKCS#7 content types are now handled via a
2282 catch all ASN1_TYPE structure. This allows unsupported
2283 types to be stored as a "blob" and an application can
2284 encode and decode it manually.
2287 *) Fix various signed/unsigned issues to make a_strex.c
2289 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
2291 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
2292 length if passed a buffer. ASN1_INTEGER_to_BN failed
2293 if passed a NULL BN and its argument was negative.
2294 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
2296 *) Modification to PKCS#7 encoding routines to output definite
2297 length encoding. Since currently the whole structures are in
2298 memory there's not real point in using indefinite length
2299 constructed encoding. However if OpenSSL is compiled with
2300 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
2303 *) Added BIO_vprintf() and BIO_vsnprintf().
2306 *) Added more prefixes to parse for in the the strings written
2307 through a logging bio, to cover all the levels that are available
2308 through syslog. The prefixes are now:
2310 PANIC, EMERG, EMR => LOG_EMERG
2311 ALERT, ALR => LOG_ALERT
2312 CRIT, CRI => LOG_CRIT
2313 ERROR, ERR => LOG_ERR
2314 WARNING, WARN, WAR => LOG_WARNING
2315 NOTICE, NOTE, NOT => LOG_NOTICE
2316 INFO, INF => LOG_INFO
2317 DEBUG, DBG => LOG_DEBUG
2319 and as before, if none of those prefixes are present at the
2320 beginning of the string, LOG_ERR is chosen.
2322 On Win32, the LOG_* levels are mapped according to this:
2324 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
2325 LOG_WARNING => EVENTLOG_WARNING_TYPE
2326 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
2330 *) Made it possible to reconfigure with just the configuration
2331 argument "reconf" or "reconfigure". The command line arguments
2332 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
2333 and are retrieved from there when reconfiguring.
2337 [Assar Westerlund <assar@sics.se>, Richard Levitte]
2339 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
2342 *) The obj_dat.pl script was messing up the sorting of object
2343 names. The reason was that it compared the quoted version
2344 of strings as a result "OCSP" > "OCSP Signing" because
2345 " > SPACE. Changed script to store unquoted versions of
2346 names and add quotes on output. It was also omitting some
2347 names from the lookup table if they were given a default
2348 value (that is if SN is missing it is given the same
2349 value as LN and vice versa), these are now added on the
2350 grounds that if an object has a name we should be able to
2351 look it up. Finally added warning output when duplicate
2352 short or long names are found.
2355 *) Changes needed for Tandem NSK.
2356 [Scott Uroff <scott@xypro.com>]
2358 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
2359 RSA_padding_check_SSLv23(), special padding was never detected
2360 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
2361 version rollback attacks was not effective.
2363 In s23_clnt.c, don't use special rollback-attack detection padding
2364 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
2365 client; similarly, in s23_srvr.c, don't do the rollback check if
2366 SSL 2.0 is the only protocol enabled in the server.
2369 *) Make it possible to get hexdumps of unprintable data with 'openssl
2370 asn1parse'. By implication, the functions ASN1_parse_dump() and
2371 BIO_dump_indent() are added.
2374 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
2375 these print out strings and name structures based on various
2376 flags including RFC2253 support and proper handling of
2377 multibyte characters. Added options to the 'x509' utility
2378 to allow the various flags to be set.
2381 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
2382 Also change the functions X509_cmp_current_time() and
2383 X509_gmtime_adj() work with an ASN1_TIME structure,
2384 this will enable certificates using GeneralizedTime in validity
2385 dates to be checked.
2388 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
2389 negative public key encodings) on by default,
2390 NO_NEG_PUBKEY_BUG can be set to disable it.
2393 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
2394 content octets. An i2c_ASN1_OBJECT is unnecessary because
2395 the encoding can be trivially obtained from the structure.
2398 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
2399 not read locks (CRYPTO_r_[un]lock).
2402 *) A first attempt at creating official support for shared
2403 libraries through configuration. I've kept it so the
2404 default is static libraries only, and the OpenSSL programs
2405 are always statically linked for now, but there are
2406 preparations for dynamic linking in place.
2407 This has been tested on Linux and Tru64.
2410 *) Randomness polling function for Win9x, as described in:
2411 Peter Gutmann, Software Generation of Practically Strong
2415 *) Fix so PRNG is seeded in req if using an already existing
2419 *) New options to smime application. -inform and -outform
2420 allow alternative formats for the S/MIME message including
2421 PEM and DER. The -content option allows the content to be
2422 specified separately. This should allow things like Netscape
2423 form signing output easier to verify.
2426 *) Fix the ASN1 encoding of tags using the 'long form'.
2429 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
2430 STRING types. These convert content octets to and from the
2431 underlying type. The actual tag and length octets are
2432 already assumed to have been read in and checked. These
2433 are needed because all other string types have virtually
2434 identical handling apart from the tag. By having versions
2435 of the ASN1 functions that just operate on content octets
2436 IMPLICIT tagging can be handled properly. It also allows
2437 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
2438 and ASN1_INTEGER are identical apart from the tag.
2441 *) Change the handling of OID objects as follows:
2443 - New object identifiers are inserted in objects.txt, following
2444 the syntax given in objects.README.
2445 - objects.pl is used to process obj_mac.num and create a new
2447 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2450 This is currently kind of a hack, and the perl code in objects.pl
2451 isn't very elegant, but it works as I intended. The simplest way
2452 to check that it worked correctly is to look in obj_dat.h and
2453 check the array nid_objs and make sure the objects haven't moved
2454 around (this is important!). Additions are OK, as well as
2455 consistent name changes.
2458 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2461 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2462 The given file adds to whatever has already been seeded into the
2463 random pool through the RANDFILE configuration file option or
2464 environment variable, or the default random state file.
2467 *) mkstack.pl now sorts each macro group into lexical order.
2468 Previously the output order depended on the order the files
2469 appeared in the directory, resulting in needless rewriting
2473 *) Patches to make OpenSSL compile under Win32 again. Mostly
2474 work arounds for the VC++ problem that it treats func() as
2475 func(void). Also stripped out the parts of mkdef.pl that
2476 added extra typesafe functions: these no longer exist.
2479 *) Reorganisation of the stack code. The macros are now all
2480 collected in safestack.h . Each macro is defined in terms of
2481 a "stack macro" of the form SKM_<name>(type, a, b). The
2482 DEBUG_SAFESTACK is now handled in terms of function casts,
2483 this has the advantage of retaining type safety without the
2484 use of additional functions. If DEBUG_SAFESTACK is not defined
2485 then the non typesafe macros are used instead. Also modified the
2486 mkstack.pl script to handle the new form. Needs testing to see
2487 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2488 the default if no major problems. Similar behaviour for ASN1_SET_OF
2489 and PKCS12_STACK_OF.
2492 *) When some versions of IIS use the 'NET' form of private key the
2493 key derivation algorithm is different. Normally MD5(password) is
2494 used as a 128 bit RC4 key. In the modified case
2495 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2496 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2497 as the old Netscape_RSA functions except they have an additional
2498 'sgckey' parameter which uses the modified algorithm. Also added
2499 an -sgckey command line option to the rsa utility. Thanks to
2500 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2501 algorithm to openssl-dev.
2504 *) The evp_local.h macros were using 'c.##kname' which resulted in
2505 invalid expansion on some systems (SCO 5.0.5 for example).
2506 Corrected to 'c.kname'.
2507 [Phillip Porch <root@theporch.com>]
2509 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2510 a STACK of email addresses from a certificate or request, these look
2511 in the subject name and the subject alternative name extensions and
2512 omit any duplicate addresses.
2515 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2516 This makes DSA verification about 2 % faster.
2519 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2520 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2521 plus overhead for 1024 bit moduli).
2522 This makes exponentiations about 0.5 % faster for 1024 bit
2523 exponents (as measured by "openssl speed rsa2048").
2526 *) Rename memory handling macros to avoid conflicts with other
2528 Malloc => OPENSSL_malloc
2529 Malloc_locked => OPENSSL_malloc_locked
2530 Realloc => OPENSSL_realloc
2531 Free => OPENSSL_free
2534 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2535 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2538 *) CygWin32 support.
2539 [John Jarvie <jjarvie@newsguy.com>]
2541 *) The type-safe stack code has been rejigged. It is now only compiled
2542 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2543 by default all type-specific stack functions are "#define"d back to
2544 standard stack functions. This results in more streamlined output
2545 but retains the type-safety checking possibilities of the original
2549 *) The STACK code has been cleaned up, and certain type declarations
2550 that didn't make a lot of sense have been brought in line. This has
2551 also involved a cleanup of sorts in safestack.h to more correctly
2552 map type-safe stack functions onto their plain stack counterparts.
2553 This work has also resulted in a variety of "const"ifications of
2554 lots of the code, especially "_cmp" operations which should normally
2555 be prototyped with "const" parameters anyway.
2558 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2559 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2560 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2561 where all of 'md' is used each time the PRNG is used, but 'state'
2562 is used only indexed by a cyclic counter. As entropy may not be
2563 well distributed from the beginning, 'md' is important as a
2564 chaining variable. However, the output function chains only half
2565 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2566 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2567 in all of 'state' being rewritten, with the new values depending
2568 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2571 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2572 the handshake is continued after ssl_verify_cert_chain();
2573 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2574 can lead to 'unexplainable' connection aborts later.
2575 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2577 *) Major EVP API cipher revision.
2578 Add hooks for extra EVP features. This allows various cipher
2579 parameters to be set in the EVP interface. Support added for variable
2580 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2581 setting of RC2 and RC5 parameters.
2583 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2586 Remove lots of duplicated code from the EVP library. For example *every*
2587 cipher init() function handles the 'iv' in the same way according to the
2588 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2589 for CFB and OFB modes they zero ctx->num.
2591 New functionality allows removal of S/MIME code RC2 hack.
2593 Most of the routines have the same form and so can be declared in terms
2596 By shifting this to the top level EVP_CipherInit() it can be removed from
2597 all individual ciphers. If the cipher wants to handle IVs or keys
2598 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2601 Change lots of functions like EVP_EncryptUpdate() to now return a
2602 value: although software versions of the algorithms cannot fail
2603 any installed hardware versions can.
2606 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2607 this option is set, tolerate broken clients that send the negotiated
2608 protocol version number instead of the requested protocol version
2612 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2613 i.e. non-zero for export ciphersuites, zero otherwise.
2614 Previous versions had this flag inverted, inconsistent with
2615 rsa_tmp_cb (..._TMP_RSA_CB).
2616 [Bodo Moeller; problem reported by Amit Chopra]
2618 *) Add missing DSA library text string. Work around for some IIS
2619 key files with invalid SEQUENCE encoding.
2622 *) Add a document (doc/standards.txt) that list all kinds of standards
2623 and so on that are implemented in OpenSSL.
2626 *) Enhance c_rehash script. Old version would mishandle certificates
2627 with the same subject name hash and wouldn't handle CRLs at all.
2628 Added -fingerprint option to crl utility, to support new c_rehash
2632 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2635 *) Fix for SSL server purpose checking. Server checking was
2636 rejecting certificates which had extended key usage present
2637 but no ssl client purpose.
2638 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2640 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2641 is a little unclear about how a blank password is handled.
2642 Since the password in encoded as a BMPString with terminating
2643 double NULL a zero length password would end up as just the
2644 double NULL. However no password at all is different and is
2645 handled differently in the PKCS#12 key generation code. NS
2646 treats a blank password as zero length. MSIE treats it as no
2647 password on export: but it will try both on import. We now do
2648 the same: PKCS12_parse() tries zero length and no password if
2649 the password is set to "" or NULL (NULL is now a valid password:
2650 it wasn't before) as does the pkcs12 application.
2653 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2654 perror when PEM_read_bio_X509_REQ fails, the error message must
2655 be obtained from the error queue.
2658 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2659 it in ERR_remove_state if appropriate, and change ERR_get_state
2660 accordingly to avoid race conditions (this is necessary because
2661 thread_hash is no longer constant once set).
2664 *) Bugfix for linux-elf makefile.one.
2667 *) RSA_get_default_method() will now cause a default
2668 RSA_METHOD to be chosen if one doesn't exist already.
2669 Previously this was only set during a call to RSA_new()
2670 or RSA_new_method(NULL) meaning it was possible for
2671 RSA_get_default_method() to return NULL.
2674 *) Added native name translation to the existing DSO code
2675 that will convert (if the flag to do so is set) filenames
2676 that are sufficiently small and have no path information
2677 into a canonical native form. Eg. "blah" converted to
2678 "libblah.so" or "blah.dll" etc.
2681 *) New function ERR_error_string_n(e, buf, len) which is like
2682 ERR_error_string(e, buf), but writes at most 'len' bytes
2683 including the 0 terminator. For ERR_error_string_n, 'buf'
2685 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2687 *) CONF library reworked to become more general. A new CONF
2688 configuration file reader "class" is implemented as well as a
2689 new functions (NCONF_*, for "New CONF") to handle it. The now
2690 old CONF_* functions are still there, but are reimplemented to
2691 work in terms of the new functions. Also, a set of functions
2692 to handle the internal storage of the configuration data is
2693 provided to make it easier to write new configuration file
2694 reader "classes" (I can definitely see something reading a
2695 configuration file in XML format, for example), called _CONF_*,
2696 or "the configuration storage API"...
2698 The new configuration file reading functions are:
2700 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2701 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2703 NCONF_default, NCONF_WIN32
2705 NCONF_dump_fp, NCONF_dump_bio
2707 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2708 NCONF_new creates a new CONF object. This works in the same way
2709 as other interfaces in OpenSSL, like the BIO interface.
2710 NCONF_dump_* dump the internal storage of the configuration file,
2711 which is useful for debugging. All other functions take the same
2712 arguments as the old CONF_* functions wth the exception of the
2713 first that must be a `CONF *' instead of a `LHASH *'.
2715 To make it easer to use the new classes with the old CONF_* functions,
2716 the function CONF_set_default_method is provided.
2719 *) Add '-tls1' option to 'openssl ciphers', which was already
2720 mentioned in the documentation but had not been implemented.
2721 (This option is not yet really useful because even the additional
2722 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2725 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2726 OpenSSL-based applications) load shared libraries and bind to
2727 them in a portable way.
2728 [Geoff Thorpe, with contributions from Richard Levitte]
2730 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2732 *) Make sure _lrotl and _lrotr are only used with MSVC.
2734 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2735 (the default implementation of RAND_status).
2737 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2738 to '-clrext' (= clear extensions), as intended and documented.
2739 [Bodo Moeller; inconsistency pointed out by Michael Attili
2742 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2743 was larger than the MD block size.
2744 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2746 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2747 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2748 using the passed key: if the passed key was a private key the result
2749 of X509_print(), for example, would be to print out all the private key
2753 *) des_quad_cksum() byte order bug fix.
2754 [Ulf Möller, using the problem description in krb4-0.9.7, where
2755 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2757 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2759 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2761 *) For easily testing in shell scripts whether some command
2762 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2763 returns with exit code 0 iff no command of the given name is available.
2764 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2765 the output goes to stdout and nothing is printed to stderr.
2766 Additional arguments are always ignored.
2768 Since for each cipher there is a command of the same name,
2769 the 'no-cipher' compilation switches can be tested this way.
2771 ('openssl no-XXX' is not able to detect pseudo-commands such
2772 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2775 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2778 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2779 is set; it will be thrown away anyway because each handshake creates
2781 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2782 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2783 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2784 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2787 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2788 'Q' and 'R' lose their special meanings (quit/renegotiate).
2789 This is part of what -quiet does; unlike -quiet, -ign_eof
2790 does not suppress any output.
2793 *) Add compatibility options to the purpose and trust code. The
2794 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2795 accepts a certificate or CA, this was the previous behaviour,
2796 with all the associated security issues.
2798 X509_TRUST_COMPAT is the old trust behaviour: only and
2799 automatically trust self signed roots in certificate store. A
2800 new trust setting X509_TRUST_DEFAULT is used to specify that
2801 a purpose has no associated trust setting and it should instead
2802 use the value in the default purpose.
2805 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2806 and fix a memory leak.
2809 *) In util/mkerr.pl (which implements 'make errors'), preserve
2810 reason strings from the previous version of the .c file, as
2811 the default to have only downcase letters (and digits) in
2812 automatically generated reasons codes is not always appropriate.
2815 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2816 using strerror. Previously, ERR_reason_error_string() returned
2817 library names as reason strings for SYSerr; but SYSerr is a special
2818 case where small numbers are errno values, not library numbers.
2821 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2822 converts DSA parameters into DH parameters. (When creating parameters,
2823 DSA_generate_parameters is used.)
2826 *) Include 'length' (recommended exponent length) in C code generated
2827 by 'openssl dhparam -C'.
2830 *) The second argument to set_label in perlasm was already being used
2831 so couldn't be used as a "file scope" flag. Moved to third argument
2835 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2836 instead of RAND_bytes for encryption IVs and salts.
2839 *) Include RAND_status() into RAND_METHOD instead of implementing
2840 it only for md_rand.c Otherwise replacing the PRNG by calling
2841 RAND_set_rand_method would be impossible.
2844 *) Don't let DSA_generate_key() enter an infinite loop if the random
2845 number generation fails.
2848 *) New 'rand' application for creating pseudo-random output.
2851 *) Added configuration support for Linux/IA64
2852 [Rolf Haberrecker <rolf@suse.de>]
2854 *) Assembler module support for Mingw32.
2857 *) Shared library support for HPUX (in shlib/).
2858 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2860 *) Shared library support for Solaris gcc.
2861 [Lutz Behnke <behnke@trustcenter.de>]
2863 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2865 *) PKCS7_encrypt() was adding text MIME headers twice because they
2866 were added manually and by SMIME_crlf_copy().
2869 *) In bntest.c don't call BN_rand with zero bits argument.
2870 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
2872 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
2873 case was implemented. This caused BN_div_recp() to fail occasionally.
2876 *) Add an optional second argument to the set_label() in the perl
2877 assembly language builder. If this argument exists and is set
2878 to 1 it signals that the assembler should use a symbol whose
2879 scope is the entire file, not just the current function. This
2880 is needed with MASM which uses the format label:: for this scope.
2881 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
2883 *) Change the ASN1 types so they are typedefs by default. Before
2884 almost all types were #define'd to ASN1_STRING which was causing
2885 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
2889 *) Change names of new functions to the new get1/get0 naming
2890 convention: After 'get1', the caller owns a reference count
2891 and has to call ..._free; 'get0' returns a pointer to some
2892 data structure without incrementing reference counters.
2893 (Some of the existing 'get' functions increment a reference
2894 counter, some don't.)
2895 Similarly, 'set1' and 'add1' functions increase reference
2896 counters or duplicate objects.
2899 *) Allow for the possibility of temp RSA key generation failure:
2900 the code used to assume it always worked and crashed on failure.
2903 *) Fix potential buffer overrun problem in BIO_printf().
2904 [Ulf Möller, using public domain code by Patrick Powell; problem
2905 pointed out by David Sacerdote <das33@cornell.edu>]
2907 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
2908 RAND_egd() and RAND_status(). In the command line application,
2909 the EGD socket can be specified like a seed file using RANDFILE
2913 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
2914 Some CAs (e.g. Verisign) distribute certificates in this form.
2917 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
2918 list to exclude them. This means that no special compilation option
2919 is needed to use anonymous DH: it just needs to be included in the
2923 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
2924 EVP_MD_type. The old functionality is available in a new macro called
2925 EVP_MD_md(). Change code that uses it and update docs.
2928 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
2929 where the 'void *' argument is replaced by a function pointer argument.
2930 Previously 'void *' was abused to point to functions, which works on
2931 many platforms, but is not correct. As these functions are usually
2932 called by macros defined in OpenSSL header files, most source code
2933 should work without changes.
2936 *) <openssl/opensslconf.h> (which is created by Configure) now contains
2937 sections with information on -D... compiler switches used for
2938 compiling the library so that applications can see them. To enable
2939 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
2940 must be defined. E.g.,
2941 #define OPENSSL_ALGORITHM_DEFINES
2942 #include <openssl/opensslconf.h>
2943 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
2944 [Richard Levitte, Ulf and Bodo Möller]
2946 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
2950 *) Change the 'other' type in certificate aux info to a STACK_OF
2951 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
2952 the required ASN1 format: arbitrary types determined by an OID.
2955 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
2956 argument to 'req'. This is not because the function is newer or
2957 better than others it just uses the work 'NEW' in the certificate
2958 request header lines. Some software needs this.
2961 *) Reorganise password command line arguments: now passwords can be
2962 obtained from various sources. Delete the PEM_cb function and make
2963 it the default behaviour: i.e. if the callback is NULL and the
2964 usrdata argument is not NULL interpret it as a null terminated pass
2965 phrase. If usrdata and the callback are NULL then the pass phrase
2966 is prompted for as usual.
2969 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
2970 the support is automatically enabled. The resulting binaries will
2971 autodetect the card and use it if present.
2972 [Ben Laurie and Compaq Inc.]
2974 *) Work around for Netscape hang bug. This sends certificate request
2975 and server done in one record. Since this is perfectly legal in the
2976 SSL/TLS protocol it isn't a "bug" option and is on by default. See
2977 the bugs/SSLv3 entry for more info.
2980 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
2983 *) Add -rand argument to smime and pkcs12 applications and read/write
2987 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
2990 *) Add command line password options to the remaining applications.
2993 *) Bug fix for BN_div_recp() for numerators with an even number of
2997 *) More tests in bntest.c, and changed test_bn output.
3000 *) ./config recognizes MacOS X now.
3003 *) Bug fix for BN_div() when the first words of num and divsor are
3004 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
3007 *) Add support for various broken PKCS#8 formats, and command line
3008 options to produce them.
3011 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
3012 get temporary BIGNUMs from a BN_CTX.
3015 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
3019 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
3020 include a #define from the old name to the new. The original intent
3021 was that statically linked binaries could for example just call
3022 SSLeay_add_all_ciphers() to just add ciphers to the table and not
3023 link with digests. This never worked becayse SSLeay_add_all_digests()
3024 and SSLeay_add_all_ciphers() were in the same source file so calling
3025 one would link with the other. They are now in separate source files.
3028 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
3031 *) Use a less unusual form of the Miller-Rabin primality test (it used
3032 a binary algorithm for exponentiation integrated into the Miller-Rabin
3033 loop, our standard modexp algorithms are faster).
3036 *) Support for the EBCDIC character set completed.
3037 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
3039 *) Source code cleanups: use const where appropriate, eliminate casts,
3040 use void * instead of char * in lhash.
3043 *) Bugfix: ssl3_send_server_key_exchange was not restartable
3044 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
3045 this the server could overwrite ephemeral keys that the client
3049 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
3050 using 50 iterations of the Rabin-Miller test.
3052 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
3053 iterations of the Rabin-Miller test as required by the appendix
3054 to FIPS PUB 186[-1]) instead of DSA_is_prime.
3055 As BN_is_prime_fasttest includes trial division, DSA parameter
3056 generation becomes much faster.
3058 This implies a change for the callback functions in DSA_is_prime
3059 and DSA_generate_parameters: The callback function is called once
3060 for each positive witness in the Rabin-Miller test, not just
3061 occasionally in the inner loop; and the parameters to the
3062 callback function now provide an iteration count for the outer
3063 loop rather than for the current invocation of the inner loop.
3064 DSA_generate_parameters additionally can call the callback
3065 function with an 'iteration count' of -1, meaning that a
3066 candidate has passed the trial division test (when q is generated
3067 from an application-provided seed, trial division is skipped).
3070 *) New function BN_is_prime_fasttest that optionally does trial
3071 division before starting the Rabin-Miller test and has
3072 an additional BN_CTX * argument (whereas BN_is_prime always
3073 has to allocate at least one BN_CTX).
3074 'callback(1, -1, cb_arg)' is called when a number has passed the
3075 trial division stage.
3078 *) Fix for bug in CRL encoding. The validity dates weren't being handled
3082 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
3085 *) New function BN_pseudo_rand().
3088 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
3089 bignum version of BN_from_montgomery() with the working code from
3090 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
3094 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
3095 made it impossible to use the same SSL_SESSION data structure in
3096 SSL2 clients in multiple threads.
3099 *) The return value of RAND_load_file() no longer counts bytes obtained
3100 by stat(). RAND_load_file(..., -1) is new and uses the complete file
3101 to seed the PRNG (previously an explicit byte count was required).
3102 [Ulf Möller, Bodo Möller]
3104 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
3105 used (char *) instead of (void *) and had casts all over the place.
3108 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
3111 *) Retain source code compatibility for BN_prime_checks macro:
3112 BN_is_prime(..., BN_prime_checks, ...) now uses
3113 BN_prime_checks_for_size to determine the appropriate number of
3114 Rabin-Miller iterations.
3117 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
3118 DH_CHECK_P_NOT_SAFE_PRIME.
3119 (Check if this is true? OpenPGP calls them "strong".)
3122 *) Merge the functionality of "dh" and "gendh" programs into a new program
3123 "dhparam". The old programs are retained for now but will handle DH keys
3124 (instead of parameters) in future.
3127 *) Make the ciphers, s_server and s_client programs check the return values
3128 when a new cipher list is set.
3131 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
3132 ciphers. Before when the 56bit ciphers were enabled the sorting was
3135 The syntax for the cipher sorting has been extended to support sorting by
3136 cipher-strength (using the strength_bits hard coded in the tables).
3137 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
3139 Fix a bug in the cipher-command parser: when supplying a cipher command
3140 string with an "undefined" symbol (neither command nor alphanumeric
3141 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
3142 an error is flagged.
3144 Due to the strength-sorting extension, the code of the
3145 ssl_create_cipher_list() function was completely rearranged. I hope that
3146 the readability was also increased :-)
3147 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3149 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
3150 for the first serial number and places 2 in the serial number file. This
3151 avoids problems when the root CA is created with serial number zero and
3152 the first user certificate has the same issuer name and serial number
3156 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
3157 the new code. Add documentation for this stuff.
3160 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
3161 X509_*() to X509at_*() on the grounds that they don't handle X509
3162 structures and behave in an analagous way to the X509v3 functions:
3163 they shouldn't be called directly but wrapper functions should be used
3166 So we also now have some wrapper functions that call the X509at functions
3167 when passed certificate requests. (TO DO: similar things can be done with
3168 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
3169 things. Some of these need some d2i or i2d and print functionality
3170 because they handle more complex structures.)
3173 *) Add missing #ifndefs that caused missing symbols when building libssl
3174 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
3175 NO_RSA in ssl/s2*.c.
3176 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
3178 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
3179 has a return value which indicates the quality of the random data
3180 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
3181 error queue. New function RAND_pseudo_bytes() generates output that is
3182 guaranteed to be unique but not unpredictable. RAND_add is like
3183 RAND_seed, but takes an extra argument for an entropy estimate
3184 (RAND_seed always assumes full entropy).
3187 *) Do more iterations of Rabin-Miller probable prime test (specifically,
3188 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
3189 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
3190 in crypto/bn/bn_prime.c for the complete table). This guarantees a
3191 false-positive rate of at most 2^-80 for random input.
3194 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
3197 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
3198 in the 0.9.5 release), this returns the chain
3199 from an X509_CTX structure with a dup of the stack and all
3200 the X509 reference counts upped: so the stack will exist
3201 after X509_CTX_cleanup() has been called. Modify pkcs12.c
3204 Also make SSL_SESSION_print() print out the verify return
3208 *) Add manpage for the pkcs12 command. Also change the default
3209 behaviour so MAC iteration counts are used unless the new
3210 -nomaciter option is used. This improves file security and
3211 only older versions of MSIE (4.0 for example) need it.
3214 *) Honor the no-xxx Configure options when creating .DEF files.
3217 *) Add PKCS#10 attributes to field table: challengePassword,
3218 unstructuredName and unstructuredAddress. These are taken from
3219 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
3220 international characters are used.
3222 More changes to X509_ATTRIBUTE code: allow the setting of types
3223 based on strings. Remove the 'loc' parameter when adding
3224 attributes because these will be a SET OF encoding which is sorted
3228 *) Initial changes to the 'req' utility to allow request generation
3229 automation. This will allow an application to just generate a template
3230 file containing all the field values and have req construct the
3233 Initial support for X509_ATTRIBUTE handling. Stacks of these are
3234 used all over the place including certificate requests and PKCS#7
3235 structures. They are currently handled manually where necessary with
3236 some primitive wrappers for PKCS#7. The new functions behave in a
3237 manner analogous to the X509 extension functions: they allow
3238 attributes to be looked up by NID and added.
3240 Later something similar to the X509V3 code would be desirable to
3241 automatically handle the encoding, decoding and printing of the
3242 more complex types. The string types like challengePassword can
3243 be handled by the string table functions.
3245 Also modified the multi byte string table handling. Now there is
3246 a 'global mask' which masks out certain types. The table itself
3247 can use the flag STABLE_NO_MASK to ignore the mask setting: this
3248 is useful when for example there is only one permissible type
3249 (as in countryName) and using the mask might result in no valid
3253 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
3254 SSL_get_peer_finished to allow applications to obtain the latest
3255 Finished messages sent to the peer or expected from the peer,
3256 respectively. (SSL_get_peer_finished is usually the Finished message
3257 actually received from the peer, otherwise the protocol will be aborted.)
3259 As the Finished message are message digests of the complete handshake
3260 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
3261 be used for external authentication procedures when the authentication
3262 provided by SSL/TLS is not desired or is not enough.
3265 *) Enhanced support for Alpha Linux is added. Now ./config checks if
3266 the host supports BWX extension and if Compaq C is present on the
3267 $PATH. Just exploiting of the BWX extension results in 20-30%
3268 performance kick for some algorithms, e.g. DES and RC4 to mention
3269 a couple. Compaq C in turn generates ~20% faster code for MD5 and
3273 *) Add support for MS "fast SGC". This is arguably a violation of the
3274 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
3275 weak crypto and after checking the certificate is SGC a second one
3276 with strong crypto. MS SGC stops the first handshake after receiving
3277 the server certificate message and sends a second client hello. Since
3278 a server will typically do all the time consuming operations before
3279 expecting any further messages from the client (server key exchange
3280 is the most expensive) there is little difference between the two.
3282 To get OpenSSL to support MS SGC we have to permit a second client
3283 hello message after we have sent server done. In addition we have to
3284 reset the MAC if we do get this second client hello.
3287 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
3288 if a DER encoded private key is RSA or DSA traditional format. Changed
3289 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
3290 format DER encoded private key. Newer code should use PKCS#8 format which
3291 has the key type encoded in the ASN1 structure. Added DER private key
3292 support to pkcs8 application.
3295 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
3296 ciphersuites has been selected (as required by the SSL 3/TLS 1
3297 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
3298 is set, we interpret this as a request to violate the specification
3299 (the worst that can happen is a handshake failure, and 'correct'
3300 behaviour would result in a handshake failure anyway).
3303 *) In SSL_CTX_add_session, take into account that there might be multiple
3304 SSL_SESSION structures with the same session ID (e.g. when two threads
3305 concurrently obtain them from an external cache).
3306 The internal cache can handle only one SSL_SESSION with a given ID,
3307 so if there's a conflict, we now throw out the old one to achieve
3311 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
3312 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
3313 some routines that use cipher OIDs: some ciphers do not have OIDs
3314 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
3318 *) Simplify the trust setting structure and code. Now we just have
3319 two sequences of OIDs for trusted and rejected settings. These will
3320 typically have values the same as the extended key usage extension
3321 and any application specific purposes.
3323 The trust checking code now has a default behaviour: it will just
3324 check for an object with the same NID as the passed id. Functions can
3325 be provided to override either the default behaviour or the behaviour
3326 for a given id. SSL client, server and email already have functions
3327 in place for compatibility: they check the NID and also return "trusted"
3328 if the certificate is self signed.
3331 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
3332 traditional format into an EVP_PKEY structure.
3335 *) Add a password callback function PEM_cb() which either prompts for
3336 a password if usr_data is NULL or otherwise assumes it is a null
3337 terminated password. Allow passwords to be passed on command line
3338 environment or config files in a few more utilities.
3341 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
3342 keys. Add some short names for PKCS#8 PBE algorithms and allow them
3343 to be specified on the command line for the pkcs8 and pkcs12 utilities.
3344 Update documentation.
3347 *) Support for ASN1 "NULL" type. This could be handled before by using
3348 ASN1_TYPE but there wasn't any function that would try to read a NULL
3349 and produce an error if it couldn't. For compatibility we also have
3350 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
3351 don't allocate anything because they don't need to.
3354 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
3356 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
3358 *) Rebuild of the memory allocation routines used by OpenSSL code and
3359 possibly others as well. The purpose is to make an interface that
3360 provide hooks so anyone can build a separate set of allocation and
3361 deallocation routines to be used by OpenSSL, for example memory
3362 pool implementations, or something else, which was previously hard
3363 since Malloc(), Realloc() and Free() were defined as macros having
3364 the values malloc, realloc and free, respectively (except for Win32
3365 compilations). The same is provided for memory debugging code.
3366 OpenSSL already comes with functionality to find memory leaks, but
3367 this gives people a chance to debug other memory problems.
3369 With these changes, a new set of functions and macros have appeared:
3371 CRYPTO_set_mem_debug_functions() [F]
3372 CRYPTO_get_mem_debug_functions() [F]
3373 CRYPTO_dbg_set_options() [F]
3374 CRYPTO_dbg_get_options() [F]
3375 CRYPTO_malloc_debug_init() [M]
3377 The memory debug functions are NULL by default, unless the library
3378 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
3379 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
3380 gives the standard debugging functions that come with OpenSSL) or
3381 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
3382 provided by the library user) must be used. When the standard
3383 debugging functions are used, CRYPTO_dbg_set_options can be used to
3384 request additional information:
3385 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
3386 the CRYPTO_MDEBUG_xxx macro when compiling the library.
3388 Also, things like CRYPTO_set_mem_functions will always give the
3389 expected result (the new set of functions is used for allocation
3390 and deallocation) at all times, regardless of platform and compiler
3393 To finish it up, some functions that were never use in any other
3394 way than through macros have a new API and new semantic:
3397 CRYPTO_dbg_realloc()
3400 All macros of value have retained their old syntax.
3401 [Richard Levitte and Bodo Moeller]
3403 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
3404 ordering of SMIMECapabilities wasn't in "strength order" and there
3405 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
3409 *) Some ASN1 types with illegal zero length encoding (INTEGER,
3410 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
3411 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
3413 *) Merge in my S/MIME library for OpenSSL. This provides a simple
3414 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
3415 functionality to handle multipart/signed properly) and a utility
3416 called 'smime' to call all this stuff. This is based on code I
3417 originally wrote for Celo who have kindly allowed it to be
3418 included in OpenSSL.
3421 *) Add variants des_set_key_checked and des_set_key_unchecked of
3422 des_set_key (aka des_key_sched). Global variable des_check_key
3423 decides which of these is called by des_set_key; this way
3424 des_check_key behaves as it always did, but applications and
3425 the library itself, which was buggy for des_check_key == 1,
3426 have a cleaner way to pick the version they need.
3429 *) New function PKCS12_newpass() which changes the password of a
3433 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
3434 dynamic mix. In both cases the ids can be used as an index into the
3435 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
3436 functions so they accept a list of the field values and the
3437 application doesn't need to directly manipulate the X509_TRUST
3441 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
3445 *) Modify the way the V3 extension code looks up extensions. This now
3446 works in a similar way to the object code: we have some "standard"
3447 extensions in a static table which is searched with OBJ_bsearch()
3448 and the application can add dynamic ones if needed. The file
3449 crypto/x509v3/ext_dat.h now has the info: this file needs to be
3450 updated whenever a new extension is added to the core code and kept
3451 in ext_nid order. There is a simple program 'tabtest.c' which checks
3452 this. New extensions are not added too often so this file can readily
3453 be maintained manually.
3455 There are two big advantages in doing things this way. The extensions
3456 can be looked up immediately and no longer need to be "added" using
3457 X509V3_add_standard_extensions(): this function now does nothing.
3458 [Side note: I get *lots* of email saying the extension code doesn't
3459 work because people forget to call this function]
3460 Also no dynamic allocation is done unless new extensions are added:
3461 so if we don't add custom extensions there is no need to call
3462 X509V3_EXT_cleanup().
3465 *) Modify enc utility's salting as follows: make salting the default. Add a
3466 magic header, so unsalted files fail gracefully instead of just decrypting
3467 to garbage. This is because not salting is a big security hole, so people
3468 should be discouraged from doing it.
3471 *) Fixes and enhancements to the 'x509' utility. It allowed a message
3472 digest to be passed on the command line but it only used this
3473 parameter when signing a certificate. Modified so all relevant
3474 operations are affected by the digest parameter including the
3475 -fingerprint and -x509toreq options. Also -x509toreq choked if a
3476 DSA key was used because it didn't fix the digest.
3479 *) Initial certificate chain verify code. Currently tests the untrusted
3480 certificates for consistency with the verify purpose (which is set
3481 when the X509_STORE_CTX structure is set up) and checks the pathlength.
3483 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
3484 this is because it will reject chains with invalid extensions whereas
3485 every previous version of OpenSSL and SSLeay made no checks at all.
3487 Trust code: checks the root CA for the relevant trust settings. Trust
3488 settings have an initial value consistent with the verify purpose: e.g.
3489 if the verify purpose is for SSL client use it expects the CA to be
3490 trusted for SSL client use. However the default value can be changed to
3491 permit custom trust settings: one example of this would be to only trust
3492 certificates from a specific "secure" set of CAs.
3494 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
3495 which should be used for version portability: especially since the
3496 verify structure is likely to change more often now.
3498 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
3499 to set them. If not set then assume SSL clients will verify SSL servers
3502 Two new options to the verify program: -untrusted allows a set of
3503 untrusted certificates to be passed in and -purpose which sets the
3504 intended purpose of the certificate. If a purpose is set then the
3505 new chain verify code is used to check extension consistency.
3508 *) Support for the authority information access extension.
3511 *) Modify RSA and DSA PEM read routines to transparently handle
3512 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
3513 public keys in a format compatible with certificate
3514 SubjectPublicKeyInfo structures. Unfortunately there were already
3515 functions called *_PublicKey_* which used various odd formats so
3516 these are retained for compatibility: however the DSA variants were
3517 never in a public release so they have been deleted. Changed dsa/rsa
3518 utilities to handle the new format: note no releases ever handled public
3519 keys so we should be OK.
3521 The primary motivation for this change is to avoid the same fiasco
3522 that dogs private keys: there are several incompatible private key
3523 formats some of which are standard and some OpenSSL specific and
3524 require various evil hacks to allow partial transparent handling and
3525 even then it doesn't work with DER formats. Given the option anything
3526 other than PKCS#8 should be dumped: but the other formats have to
3527 stay in the name of compatibility.
3529 With public keys and the benefit of hindsight one standard format
3530 is used which works with EVP_PKEY, RSA or DSA structures: though
3531 it clearly returns an error if you try to read the wrong kind of key.
3533 Added a -pubkey option to the 'x509' utility to output the public key.
3534 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
3535 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
3536 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
3537 that do the same as the EVP_PKEY_assign_*() except they up the
3538 reference count of the added key (they don't "swallow" the
3542 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
3543 CRLs would fail if the file contained no certificates or no CRLs:
3544 added a new function to read in both types and return the number
3545 read: this means that if none are read it will be an error. The
3546 DER versions of the certificate and CRL reader would always fail
3547 because it isn't possible to mix certificates and CRLs in DER format
3548 without choking one or the other routine. Changed this to just read
3549 a certificate: this is the best we can do. Also modified the code
3550 in apps/verify.c to take notice of return codes: it was previously
3551 attempting to read in certificates from NULL pointers and ignoring
3552 any errors: this is one reason why the cert and CRL reader seemed
3553 to work. It doesn't check return codes from the default certificate
3554 routines: these may well fail if the certificates aren't installed.
3557 *) Code to support otherName option in GeneralName.
3560 *) First update to verify code. Change the verify utility
3561 so it warns if it is passed a self signed certificate:
3562 for consistency with the normal behaviour. X509_verify
3563 has been modified to it will now verify a self signed
3564 certificate if *exactly* the same certificate appears
3565 in the store: it was previously impossible to trust a
3566 single self signed certificate. This means that:
3567 openssl verify ss.pem
3568 now gives a warning about a self signed certificate but
3569 openssl verify -CAfile ss.pem ss.pem
3573 *) For servers, store verify_result in SSL_SESSION data structure
3574 (and add it to external session representation).
3575 This is needed when client certificate verifications fails,
3576 but an application-provided verification callback (set by
3577 SSL_CTX_set_cert_verify_callback) allows accepting the session
3578 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
3579 but returns 1): When the session is reused, we have to set
3580 ssl->verify_result to the appropriate error code to avoid
3582 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
3584 *) Fix a bug in the new PKCS#7 code: it didn't consider the
3585 case in PKCS7_dataInit() where the signed PKCS7 structure
3586 didn't contain any existing data because it was being created.
3587 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
3589 *) Add a salt to the key derivation routines in enc.c. This
3590 forms the first 8 bytes of the encrypted file. Also add a
3591 -S option to allow a salt to be input on the command line.
3594 *) New function X509_cmp(). Oddly enough there wasn't a function
3595 to compare two certificates. We do this by working out the SHA1
3596 hash and comparing that. X509_cmp() will be needed by the trust
3600 *) SSL_get1_session() is like SSL_get_session(), but increments
3601 the reference count in the SSL_SESSION returned.
3602 [Geoff Thorpe <geoff@eu.c2.net>]
3604 *) Fix for 'req': it was adding a null to request attributes.
3605 Also change the X509_LOOKUP and X509_INFO code to handle
3606 certificate auxiliary information.
3609 *) Add support for 40 and 64 bit RC2 and RC4 algorithms: document
3613 *) Add the possibility to add extra information to the memory leak
3614 detecting output, to form tracebacks, showing from where each
3615 allocation was originated: CRYPTO_push_info("constant string") adds
3616 the string plus current file name and line number to a per-thread
3617 stack, CRYPTO_pop_info() does the obvious, CRYPTO_remove_all_info()
3618 is like calling CYRPTO_pop_info() until the stack is empty.
3619 Also updated memory leak detection code to be multi-thread-safe.
3622 *) Add options -text and -noout to pkcs7 utility and delete the
3623 encryption options which never did anything. Update docs.
3626 *) Add options to some of the utilities to allow the pass phrase
3627 to be included on either the command line (not recommended on
3628 OSes like Unix) or read from the environment. Update the
3629 manpages and fix a few bugs.
3632 *) Add a few manpages for some of the openssl commands.
3635 *) Fix the -revoke option in ca. It was freeing up memory twice,
3636 leaking and not finding already revoked certificates.
3639 *) Extensive changes to support certificate auxiliary information.
3640 This involves the use of X509_CERT_AUX structure and X509_AUX
3641 functions. An X509_AUX function such as PEM_read_X509_AUX()
3642 can still read in a certificate file in the usual way but it
3643 will also read in any additional "auxiliary information". By
3644 doing things this way a fair degree of compatibility can be
3645 retained: existing certificates can have this information added
3646 using the new 'x509' options.
3648 Current auxiliary information includes an "alias" and some trust
3649 settings. The trust settings will ultimately be used in enhanced
3650 certificate chain verification routines: currently a certificate
3651 can only be trusted if it is self signed and then it is trusted
3655 *) Fix assembler for Alpha (tested only on DEC OSF not Linux or *BSD).
3656 The problem was that one of the replacement routines had not been working
3657 since SSLeay releases. For now
projects / openssl.git / blob