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 -) [In 0.9.6c-engine release:]
16 Add support for Broadcom crypto accelerator cards, backported
18 [Broadcom, Nalin Dahyabhai <nalin@redhat.com>, Mark Cox]
20 -) [In 0.9.6c-engine release:]
21 Add support for SureWare crypto accelerator cards from
22 Baltimore Technologies. (Use engine 'sureware')
23 [Baltimore Technologies and Mark Cox]
25 -) [In 0.9.6c-engine release:]
26 Add support for crypto accelerator cards from Accelerated
27 Encryption Processing, www.aep.ie. (Use engine 'aep')
28 [AEP Inc. and Mark Cox]
30 *) Add a configuration entry for gcc on UnixWare.
31 [Gary Benson <gbenson@redhat.com>]
33 +) New functions/macros
35 SSL_CTX_set_msg_callback(ctx, cb)
36 SSL_CTX_set_msg_callback_arg(ctx, arg)
37 SSL_set_msg_callback(ssl, cb)
38 SSL_set_msg_callback_arg(ssl, arg)
40 to request calling a callback function
42 void cb(int write_p, int version, int content_type,
43 const void *buf, size_t len, SSL *ssl, void *arg)
45 whenever a protocol message has been completely received
46 (write_p == 0) or sent (write_p == 1). Here 'version' is the
47 protocol version according to which the SSL library interprets
48 the current protocol message (SSL2_VERSION, SSL3_VERSION, or
49 TLS1_VERSION). 'content_type' is 0 in the case of SSL 2.0, or
50 the content type as defined in the SSL 3.0/TLS 1.0 protocol
51 specification (change_cipher_spec(20), alert(21), handshake(22)).
52 'buf' and 'len' point to the actual message, 'ssl' to the
53 SSL object, and 'arg' is the application-defined value set by
54 SSL[_CTX]_set_msg_callback_arg().
56 'openssl s_client' and 'openssl s_server' have new '-msg' options
57 to enable a callback that displays all protocol messages.
60 *) Change ssl/s2_clnt.c and ssl/s2_srvr.c so that received handshake
61 messages are stored in a single piece (fixed-length part and
62 variable-length part combined) and fix various bugs found on the way.
65 +) Change the shared library support so shared libraries are built as
66 soon as the corresponding static library is finished, and thereby get
67 openssl and the test programs linked against the shared library.
68 This still only happens when the keyword "shard" has been given to
69 the configuration scripts.
71 NOTE: shared library support is still an experimental thing, and
72 backward binary compatibility is still not guaranteed.
73 ["Maciej W. Rozycki" <macro@ds2.pg.gda.pl> and Richard Levitte]
75 +) Add support for Subject Information Access extension.
76 [Peter Sylvester <Peter.Sylvester@EdelWeb.fr>]
78 +) Make BUF_MEM_grow() behaviour more consistent: Initialise to zero
79 additional bytes when new memory had to be allocated, not just
80 when reusing an existing buffer.
83 *) Disable caching in BIO_gethostbyname(), directly use gethostbyname()
84 instead. BIO_gethostbyname() does not know what timeouts are
85 appropriate, so entries would stay in cache even when they have
87 [Bodo Moeller; problem pointed out by Rich Salz <rsalz@zolera.com>
89 +) New command line and configuration option 'utf8' for the req command.
90 This allows field values to be specified as UTF8 strings.
93 +) Add -multi and -mr options to "openssl speed" - giving multiple parallel
94 runs for the former and machine-readable output for the latter.
97 +) Add '-noemailDN' option to 'openssl ca'. This prevents inclusion
98 of the e-mail address in the DN (i.e., it will go into a certificate
99 extension only). The new configuration file option 'email_in_dn = no'
101 [Massimiliano Pala madwolf@openca.org]
103 *) Change ssl23_get_client_hello (ssl/s23_srvr.c) behaviour when
104 faced with a pathologically small ClientHello fragment that does
105 not contain client_version: Instead of aborting with an error,
106 simply choose the highest available protocol version (i.e.,
107 TLS 1.0 unless it is disabled). In practice, ClientHello
108 messages are never sent like this, but this change gives us
109 strictly correct behaviour at least for TLS.
112 +) Change all functions with names starting with des_ to be starting
113 with DES_ instead. This because there are increasing clashes with
114 libdes and other des libraries that are currently used by other
115 projects. The old libdes interface is provided, as well as crypt(),
116 if openssl/des_old.h is included. Note that crypt() is no longer
117 declared in openssl/des.h.
119 NOTE: This is a major break of an old API into a new one. Software
120 authors are encouraged to switch to the DES_ style functions. Some
121 time in the future, des_old.h and the libdes compatibility functions
122 will be completely removed.
125 *) Fix SSL handshake functions and SSL_clear() such that SSL_clear()
126 never resets s->method to s->ctx->method when called from within
127 one of the SSL handshake functions.
128 [Bodo Moeller; problem pointed out by Niko Baric]
130 +) Test for certificates which contain unsupported critical extensions.
131 If such a certificate is found during a verify operation it is
132 rejected by default: this behaviour can be overridden by either
133 handling the new error X509_V_ERR_UNHANDLED_CRITICAL_EXTENSION or
134 by setting the verify flag X509_V_FLAG_IGNORE_CRITICAL. A new function
135 X509_supported_extension() has also been added which returns 1 if a
136 particular extension is supported.
139 *) In ssl3_get_client_hello (ssl/s3_srvr.c), generate a fatal alert
140 (sent using the client's version number) if client_version is
141 smaller than the protocol version in use. Also change
142 ssl23_get_client_hello (ssl/s23_srvr.c) to select TLS 1.0 if
143 the client demanded SSL 3.0 but only TLS 1.0 is enabled; then
144 the client will at least see that alert.
147 +) Modify the behaviour of EVP cipher functions in similar way to digests
148 to retain compatibility with existing code.
151 +) Modify the behaviour of EVP_DigestInit() and EVP_DigestFinal() to retain
152 compatibility with existing code. In particular the 'ctx' parameter does
153 not have to be to be initialized before the call to EVP_DigestInit() and
154 it is tidied up after a call to EVP_DigestFinal(). New function
155 EVP_DigestFinal_ex() which does not tidy up the ctx. Similarly function
156 EVP_MD_CTX_copy() changed to not require the destination to be
157 initialized valid and new function EVP_MD_CTX_copy_ex() added which
158 requires the destination to be valid.
160 Modify all the OpenSSL digest calls to use EVP_DigestInit_ex(),
161 EVP_DigestFinal_ex() and EVP_MD_CTX_copy_ex().
164 +) Change ssl3_get_message (ssl/s3_both.c) and the functions using it
165 so that complete 'Handshake' protocol structures are kept in memory
166 instead of overwriting 'msg_type' and 'length' with 'body' data.
169 *) Fix ssl3_get_message (ssl/s3_both.c) to handle message fragmentation
173 +) Add an implementation of SSL_add_dir_cert_subjects_to_stack for Win32.
174 [Massimo Santin via Richard Levitte]
176 +) Major restructuring to the underlying ENGINE code. This includes
177 reduction of linker bloat, separation of pure "ENGINE" manipulation
178 (initialisation, etc) from functionality dealing with implementations
179 of specific crypto iterfaces. This change also introduces integrated
180 support for symmetric ciphers and digest implementations - so ENGINEs
181 can now accelerate these by providing EVP_CIPHER and EVP_MD
182 implementations of their own. This is detailed in crypto/engine/README
183 as it couldn't be adequately described here. However, there are a few
184 API changes worth noting - some RSA, DSA, DH, and RAND functions that
185 were changed in the original introduction of ENGINE code have now
186 reverted back - the hooking from this code to ENGINE is now a good
187 deal more passive and at run-time, operations deal directly with
188 RSA_METHODs, DSA_METHODs (etc) as they did before, rather than
189 dereferencing through an ENGINE pointer any more. Also, the ENGINE
190 functions dealing with BN_MOD_EXP[_CRT] handlers have been removed -
191 they were not being used by the framework as there is no concept of a
192 BIGNUM_METHOD and they could not be generalised to the new
193 'ENGINE_TABLE' mechanism that underlies the new code. Similarly,
194 ENGINE_cpy() has been removed as it cannot be consistently defined in
198 +) Change ASN1_GENERALIZEDTIME_check() to allow fractional seconds.
201 +) Change mkdef.pl to sort symbols that get the same entry number,
202 and make sure the automatically generated functions ERR_load_*
203 become part of libeay.num as well.
206 *) Avoid infinite loop in ssl3_get_message (ssl/s3_both.c) if a
207 client receives HelloRequest while in a handshake.
208 [Bodo Moeller; bug noticed by Andy Schneider <andy.schneider@bjss.co.uk>]
210 +) New function SSL_renegotiate_pending(). This returns true once
211 renegotiation has been requested (either SSL_renegotiate() call
212 or HelloRequest/ClientHello receveived from the peer) and becomes
213 false once a handshake has been completed.
214 (For servers, SSL_renegotiate() followed by SSL_do_handshake()
215 sends a HelloRequest, but does not ensure that a handshake takes
216 place. SSL_renegotiate_pending() is useful for checking if the
217 client has followed the request.)
220 +) New SSL option SSL_OP_NO_SESSION_RESUMPTION_ON_RENEGOTIATION.
221 By default, clients may request session resumption even during
222 renegotiation (if session ID contexts permit); with this option,
223 session resumption is possible only in the first handshake.
226 *) Bugfix in ssl3_accept (ssl/s3_srvr.c): Case SSL3_ST_SW_HELLO_REQ_C
227 should end in 'break', not 'goto end' which circuments various
228 cleanups done in state SSL_ST_OK. But session related stuff
229 must be disabled for SSL_ST_OK in the case that we just sent a
232 Also avoid some overhead by not calling ssl_init_wbio_buffer()
233 before just sending a HelloRequest.
234 [Bodo Moeller, Eric Rescorla <ekr@rtfm.com>]
236 *) Fix ssl/s3_enc.c, ssl/t1_enc.c and ssl/s3_pkt.c so that we don't
237 reveal whether illegal block cipher padding was found or a MAC
238 verification error occured. (Neither SSLerr() codes nor alerts
239 are directly visible to potential attackers, but the information
240 may leak via logfiles.)
242 Similar changes are not required for the SSL 2.0 implementation
243 because the number of padding bytes is sent in clear for SSL 2.0,
244 and the extra bytes are just ignored. However ssl/s2_pkt.c
245 failed to verify that the purported number of padding bytes is in
249 +) Add some demos for certificate and certificate request creation.
252 +) Make maximum certificate chain size accepted from the peer application
253 settable (SSL*_get/set_max_cert_list()), as proposed by
254 "Douglas E. Engert" <deengert@anl.gov>.
257 +) Add support for shared libraries for Unixware-7
258 (Boyd Lynn Gerber <gerberb@zenez.com>).
261 *) Add OpenUNIX-8 support including shared libraries
262 (Boyd Lynn Gerber <gerberb@zenez.com>).
265 *) Improve RSA_padding_check_PKCS1_OAEP() check again to avoid
266 'wristwatch attack' using huge encoding parameters (cf.
267 James H. Manger's CRYPTO 2001 paper). Note that the
268 RSA_PKCS1_OAEP_PADDING case of RSA_private_decrypt() does not use
269 encoding parameters and hence was not vulnerable.
272 +) Add a "destroy" handler to ENGINEs that allows structural cleanup to
273 be done prior to destruction. Use this to unload error strings from
274 ENGINEs that load their own error strings. NB: This adds two new API
275 functions to "get" and "set" this destroy handler in an ENGINE.
278 +) Alter all existing ENGINE implementations (except "openssl" and
279 "openbsd") to dynamically instantiate their own error strings. This
280 makes them more flexible to be built both as statically-linked ENGINEs
281 and self-contained shared-libraries loadable via the "dynamic" ENGINE.
282 Also, add stub code to each that makes building them as self-contained
283 shared-libraries easier (see README.ENGINE).
286 +) Add a "dynamic" ENGINE that provides a mechanism for binding ENGINE
287 implementations into applications that are completely implemented in
288 self-contained shared-libraries. The "dynamic" ENGINE exposes control
289 commands that can be used to configure what shared-library to load and
290 to control aspects of the way it is handled. Also, made an update to
291 the README.ENGINE file that brings its information up-to-date and
292 provides some information and instructions on the "dynamic" ENGINE
293 (ie. how to use it, how to build "dynamic"-loadable ENGINEs, etc).
297 [Ulf Möller, reported by Jim Ellis <jim.ellis@cavium.com>]
299 *) Make it possible to unload ranges of ERR strings with a new
300 "ERR_unload_strings" function.
303 *) Rabin-Miller test analyses assume uniformly distributed witnesses,
304 so use BN_pseudo_rand_range() instead of using BN_pseudo_rand()
305 followed by modular reduction.
306 [Bodo Moeller; pointed out by Adam Young <AYoung1@NCSUS.JNJ.COM>]
308 *) Add BN_pseudo_rand_range() with obvious functionality: BN_rand_range()
309 equivalent based on BN_pseudo_rand() instead of BN_rand().
312 +) Add a copy() function to EVP_MD.
315 +) Make EVP_MD routines take a context pointer instead of just the
316 md_data void pointer.
319 +) Add flags to EVP_MD and EVP_MD_CTX. EVP_MD_FLAG_ONESHOT indicates
320 that the digest can only process a single chunk of data
321 (typically because it is provided by a piece of
322 hardware). EVP_MD_CTX_FLAG_ONESHOT indicates that the application
323 is only going to provide a single chunk of data, and hence the
324 framework needn't accumulate the data for oneshot drivers.
327 +) As with "ERR", make it possible to replace the underlying "ex_data"
328 functions. This change also alters the storage and management of global
329 ex_data state - it's now all inside ex_data.c and all "class" code (eg.
330 RSA, BIO, SSL_CTX, etc) no longer stores its own STACKS and per-class
331 index counters. The API functions that use this state have been changed
332 to take a "class_index" rather than pointers to the class's local STACK
333 and counter, and there is now an API function to dynamically create new
334 classes. This centralisation allows us to (a) plug a lot of the
335 thread-safety problems that existed, and (b) makes it possible to clean
336 up all allocated state using "CRYPTO_cleanup_all_ex_data()". W.r.t. (b)
337 such data would previously have always leaked in application code and
338 workarounds were in place to make the memory debugging turn a blind eye
339 to it. Application code that doesn't use this new function will still
340 leak as before, but their memory debugging output will announce it now
341 rather than letting it slide.
343 Besides the addition of CRYPTO_cleanup_all_ex_data(), another API change
344 induced by the "ex_data" overhaul is that X509_STORE_CTX_init() now
345 has a return value to indicate success or failure.
348 +) Make it possible to replace the underlying "ERR" functions such that the
349 global state (2 LHASH tables and 2 locks) is only used by the "default"
350 implementation. This change also adds two functions to "get" and "set"
351 the implementation prior to it being automatically set the first time
352 any other ERR function takes place. Ie. an application can call "get",
353 pass the return value to a module it has just loaded, and that module
354 can call its own "set" function using that value. This means the
355 module's "ERR" operations will use (and modify) the error state in the
356 application and not in its own statically linked copy of OpenSSL code.
359 +) Give DH, DSA, and RSA types their own "**_up_ref()" function to increment
360 reference counts. This performs normal REF_PRINT/REF_CHECK macros on
361 the operation, and provides a more encapsulated way for external code
362 (crypto/evp/ and ssl/) to do this. Also changed the evp and ssl code
363 to use these functions rather than manually incrementing the counts.
365 Also rename "DSO_up()" function to more descriptive "DSO_up_ref()".
368 *) s3_srvr.c: allow sending of large client certificate lists (> 16 kB).
369 This function was broken, as the check for a new client hello message
370 to handle SGC did not allow these large messages.
371 (Tracked down by "Douglas E. Engert" <deengert@anl.gov>.)
374 *) Add alert descriptions for TLSv1 to SSL_alert_desc_string[_long]().
377 +) Add EVP test program.
380 +) Add symmetric cipher support to ENGINE. Expect the API to change!
383 +) New CRL functions: X509_CRL_set_version(), X509_CRL_set_issuer_name()
384 X509_CRL_set_lastUpdate(), X509_CRL_set_nextUpdate(), X509_CRL_sort(),
385 X509_REVOKED_set_serialNumber(), and X509_REVOKED_set_revocationDate().
386 These allow a CRL to be built without having to access X509_CRL fields
387 directly. Modify 'ca' application to use new functions.
390 *) Fix buggy behaviour of BIO_get_num_renegotiates() and BIO_ctrl()
391 for BIO_C_GET_WRITE_BUF_SIZE ("Stephen Hinton" <shinton@netopia.com>).
394 *) Rework the configuration and shared library support for Tru64 Unix.
395 The configuration part makes use of modern compiler features and
396 still retains old compiler behavior for those that run older versions
397 of the OS. The shared library support part includes a variant that
398 uses the RPATH feature, and is available through the special
399 configuration target "alpha-cc-rpath", which will never be selected
401 [Tim Mooney <mooney@dogbert.cc.ndsu.NoDak.edu> via Richard Levitte]
403 *) In ssl3_get_key_exchange (ssl/s3_clnt.c), call ssl3_get_message()
404 with the same message size as in ssl3_get_certificate_request().
405 Otherwise, if no ServerKeyExchange message occurs, CertificateRequest
406 messages might inadvertently be reject as too long.
407 [Petr Lampa <lampa@fee.vutbr.cz>]
409 +) Move SSL_OP_TLS_ROLLBACK_BUG out of the SSL_OP_ALL list of recommended
410 bug workarounds. Rollback attack detection is a security feature.
411 The problem will only arise on OpenSSL servers when TLSv1 is not
412 available (sslv3_server_method() or SSL_OP_NO_TLSv1).
413 Software authors not wanting to support TLSv1 will have special reasons
414 for their choice and can explicitly enable this option.
415 [Bodo Moeller, Lutz Jaenicke]
417 +) Rationalise EVP so it can be extended: don't include a union of
418 cipher/digest structures, add init/cleanup functions. This also reduces
419 the number of header dependencies.
422 +) Make DES key schedule conform to the usual scheme, as well as
423 correcting its structure. This means that calls to DES functions
424 now have to pass a pointer to a des_key_schedule instead of a
425 plain des_key_schedule (which was actually always a pointer
429 *) Enhanced support for IA-64 Unix platforms (well, Linux and HP-UX).
432 *) Modified SSL library such that the verify_callback that has been set
433 specificly for an SSL object with SSL_set_verify() is actually being
434 used. Before the change, a verify_callback set with this function was
435 ignored and the verify_callback() set in the SSL_CTX at the time of
436 the call was used. New function X509_STORE_CTX_set_verify_cb() introduced
437 to allow the necessary settings.
440 +) Initial reduction of linker bloat: the use of some functions, such as
441 PEM causes large amounts of unused functions to be linked in due to
442 poor organisation. For example pem_all.c contains every PEM function
443 which has a knock on effect of linking in large amounts of (unused)
444 ASN1 code. Grouping together similar functions and splitting unrelated
445 functions prevents this.
448 *) Initialize static variable in crypto/dsa/dsa_lib.c and crypto/dh/dh_lib.c
449 explicitly to NULL, as at least on Solaris 8 this seems not always to be
450 done automatically (in contradiction to the requirements of the C
451 standard). This made problems when used from OpenSSH.
454 *) In OpenSSL 0.9.6a and 0.9.6b, crypto/dh/dh_key.c ignored
455 dh->length and always used
457 BN_rand_range(priv_key, dh->p).
459 BN_rand_range() is not necessary for Diffie-Hellman, and this
460 specific range makes Diffie-Hellman unnecessarily inefficient if
461 dh->length (recommended exponent length) is much smaller than the
462 length of dh->p. We could use BN_rand_range() if the order of
463 the subgroup was stored in the DH structure, but we only have
468 BN_rand(priv_key, l, ...)
470 where 'l' is dh->length if this is defined, or BN_num_bits(dh->p)-1
476 RSA_eay_public_encrypt
477 RSA_eay_private_decrypt
478 RSA_eay_private_encrypt (signing)
479 RSA_eay_public_decrypt (signature verification)
481 (default implementations for RSA_public_encrypt,
482 RSA_private_decrypt, RSA_private_encrypt, RSA_public_decrypt),
483 always reject numbers >= n.
486 *) In crypto/rand/md_rand.c, use a new short-time lock CRYPTO_LOCK_RAND2
487 to synchronize access to 'locking_thread'. This is necessary on
488 systems where access to 'locking_thread' (an 'unsigned long'
489 variable) is not atomic.
492 *) In crypto/rand/md_rand.c, set 'locking_thread' to current thread's ID
493 *before* setting the 'crypto_lock_rand' flag. The previous code had
494 a race condition if 0 is a valid thread ID.
495 [Travis Vitek <vitek@roguewave.com>]
497 +) Cleanup of EVP macros.
500 +) Change historical references to {NID,SN,LN}_des_ede and ede3 to add the
504 +) Add initial OCSP responder support to ocsp application. The
505 revocation information is handled using the text based index
506 use by the ca application. The responder can either handle
507 requests generated internally, supplied in files (for example
508 via a CGI script) or using an internal minimal server.
511 +) Add configuration choices to get zlib compression for TLS.
514 +) Changes to Kerberos SSL for RFC 2712 compliance:
515 1. Implemented real KerberosWrapper, instead of just using
516 KRB5 AP_REQ message. [Thanks to Simon Wilkinson <sxw@sxw.org.uk>]
517 2. Implemented optional authenticator field of KerberosWrapper.
519 Added openssl-style ASN.1 macros for Kerberos ticket, ap_req,
520 and authenticator structs; see crypto/krb5/.
522 Generalized Kerberos calls to support multiple Kerberos libraries.
523 [Vern Staats <staatsvr@asc.hpc.mil>,
524 Jeffrey Altman <jaltman@columbia.edu>
527 +) Cause 'openssl speed' to use fully hard-coded DSA keys as it
528 already does with RSA. testdsa.h now has 'priv_key/pub_key'
529 values for each of the key sizes rather than having just
530 parameters (and 'speed' generating keys each time).
533 -) OpenSSL 0.9.6b released [9 July 2001]
535 *) Change ssleay_rand_bytes (crypto/rand/md_rand.c)
536 to avoid a SSLeay/OpenSSL PRNG weakness pointed out by
537 Markku-Juhani O. Saarinen <markku-juhani.saarinen@nokia.com>:
538 PRNG state recovery was possible based on the output of
539 one PRNG request appropriately sized to gain knowledge on
540 'md' followed by enough consecutive 1-byte PRNG requests
541 to traverse all of 'state'.
543 1. When updating 'md_local' (the current thread's copy of 'md')
544 during PRNG output generation, hash all of the previous
545 'md_local' value, not just the half used for PRNG output.
547 2. Make the number of bytes from 'state' included into the hash
548 independent from the number of PRNG bytes requested.
550 The first measure alone would be sufficient to avoid
551 Markku-Juhani's attack. (Actually it had never occurred
552 to me that the half of 'md_local' used for chaining was the
553 half from which PRNG output bytes were taken -- I had always
554 assumed that the secret half would be used.) The second
555 measure makes sure that additional data from 'state' is never
556 mixed into 'md_local' in small portions; this heuristically
557 further strengthens the PRNG.
560 +) Speed up EVP routines.
563 type 8 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes
564 des-cbc 4408.85k 5560.51k 5778.46k 5862.20k 5825.16k
565 des-cbc 4389.55k 5571.17k 5792.23k 5846.91k 5832.11k
566 des-cbc 4394.32k 5575.92k 5807.44k 5848.37k 5841.30k
568 des-cbc 3482.66k 5069.49k 5496.39k 5614.16k 5639.28k
569 des-cbc 3480.74k 5068.76k 5510.34k 5609.87k 5635.52k
570 des-cbc 3483.72k 5067.62k 5504.60k 5708.01k 5724.80k
573 des-cbc 4660.16k 5650.19k 5807.19k 5827.13k 5783.32k
575 des-cbc 3624.96k 5258.21k 5530.91k 5624.30k 5628.26k
578 *) Fix crypto/bn/asm/mips3.s.
581 *) When only the key is given to "enc", the IV is undefined. Print out
582 an error message in this case.
585 +) Added the OS2-EMX target.
586 ["Brian Havard" <brianh@kheldar.apana.org.au> and Richard Levitte]
588 +) Rewrite apps to use NCONF routines instead of the old CONF. New functions
589 to support NCONF routines in extension code. New function CONF_set_nconf()
590 to allow functions which take an NCONF to also handle the old LHASH
591 structure: this means that the old CONF compatible routines can be
592 retained (in particular wrt extensions) without having to duplicate the
593 code. New function X509V3_add_ext_nconf_sk to add extensions to a stack.
596 *) Handle special case when X509_NAME is empty in X509 printing routines.
599 *) In dsa_do_verify (crypto/dsa/dsa_ossl.c), verify that r and s are
600 positive and less than q.
603 +) Enhance the general user interface with mechanisms for inner control
604 and with possibilities to have yes/no kind of prompts.
607 +) Change all calls to low level digest routines in the library and
608 applications to use EVP. Add missing calls to HMAC_cleanup() and
609 don't assume HMAC_CTX can be copied using memcpy().
610 [Verdon Walker <VWalker@novell.com>, Steve Henson]
612 +) Add the possibility to control engines through control names but with
613 arbitrary arguments instead of just a string.
614 Change the key loaders to take a UI_METHOD instead of a callback
615 function pointer. NOTE: this breaks binary compatibility with earlier
616 versions of OpenSSL [engine].
617 Adapt the nCipher code for these new conditions and add a card insertion
621 +) Enhance the general user interface with mechanisms to better support
622 dialog box interfaces, application-defined prompts, the possibility
623 to use defaults (for example default passwords from somewhere else)
624 and interrupts/cancellations.
627 *) Don't change *pointer in CRYPTO_add_lock() is add_lock_callback is
628 used: it isn't thread safe and the add_lock_callback should handle
630 [Paul Rose <Paul.Rose@bridge.com>]
632 *) Verify that incoming data obeys the block size in
633 ssl3_enc (ssl/s3_enc.c) and tls1_enc (ssl/t1_enc.c).
636 +) Tidy up PKCS#12 attribute handling. Add support for the CSP name
637 attribute in PKCS#12 files, add new -CSP option to pkcs12 utility.
641 [Ulf Möller, Bodo Möller]
643 *) The countermeasure against Bleichbacher's attack on PKCS #1 v1.5
644 RSA encryption was accidentally removed in s3_srvr.c in OpenSSL 0.9.5
645 when fixing the server behaviour for backwards-compatible 'client
646 hello' messages. (Note that the attack is impractical against
647 SSL 3.0 and TLS 1.0 anyway because length and version checking
648 means that the probability of guessing a valid ciphertext is
649 around 2^-40; see section 5 in Bleichenbacher's CRYPTO '98
652 Before 0.9.5, the countermeasure (hide the error by generating a
653 random 'decryption result') did not work properly because
654 ERR_clear_error() was missing, meaning that SSL_get_error() would
655 detect the supposedly ignored error.
657 Both problems are now fixed.
660 *) In crypto/bio/bf_buff.c, increase DEFAULT_BUFFER_SIZE to 4096
661 (previously it was 1024).
664 +) Fix a memory leak in 'sk_dup()' in the case reallocation fails. (Also
665 tidy up some unnecessarily weird code in 'sk_new()').
666 [Geoff, reported by Diego Tartara <dtartara@novamens.com>]
668 +) Change the key loading routines for ENGINEs to use the same kind
669 callback (pem_password_cb) as all other routines that need this
673 *) Fix for compatibility mode trust settings: ignore trust settings
674 unless some valid trust or reject settings are present.
677 *) Fix for blowfish EVP: its a variable length cipher.
680 +) Increase ENTROPY_NEEDED to 32 bytes, as Rijndael can operate with
681 256 bit (=32 byte) keys. Of course seeding with more entropy bytes
682 than this minimum value is recommended.
685 +) New random seeder for OpenVMS, using the system process statistics
686 that are easily reachable.
689 +) Windows apparently can't transparently handle global
690 variables defined in DLLs. Initialisations such as:
692 const ASN1_ITEM *it = &ASN1_INTEGER_it;
694 wont compile. This is used by the any applications that need to
695 declare their own ASN1 modules. This was fixed by adding the option
696 EXPORT_VAR_AS_FN to all Win32 platforms, although this isn't strictly
697 needed for static libraries under Win32.
700 +) New functions X509_PURPOSE_set() and X509_TRUST_set() to handle
701 setting of purpose and trust fields. New X509_STORE trust and
702 purpose functions and tidy up setting in other SSL functions.
705 +) Add copies of X509_STORE_CTX fields and callbacks to X509_STORE
706 structure. These are inherited by X509_STORE_CTX when it is
707 initialised. This allows various defaults to be set in the
708 X509_STORE structure (such as flags for CRL checking and custom
709 purpose or trust settings) for functions which only use X509_STORE_CTX
710 internally such as S/MIME.
712 Modify X509_STORE_CTX_purpose_inherit() so it only sets purposes and
713 trust settings if they are not set in X509_STORE. This allows X509_STORE
714 purposes and trust (in S/MIME for example) to override any set by default.
716 Add command line options for CRL checking to smime, s_client and s_server
720 +) Initial CRL based revocation checking. If the CRL checking flag(s)
721 are set then the CRL is looked up in the X509_STORE structure and
722 its validity and signature checked, then if the certificate is found
723 in the CRL the verify fails with a revoked error.
725 Various new CRL related callbacks added to X509_STORE_CTX structure.
727 Command line options added to 'verify' application to support this.
729 This needs some additional work, such as being able to handle multiple
730 CRLs with different times, extension based lookup (rather than just
731 by subject name) and ultimately more complete V2 CRL extension
735 +) Add a general user interface API (crypto/ui/). This is designed
736 to replace things like des_read_password and friends (backward
737 compatibility functions using this new API are provided).
738 The purpose is to remove prompting functions from the DES code
739 section as well as provide for prompting through dialog boxes in
740 a window system and the like.
743 *) Fix various bugs related to DSA S/MIME verification. Handle missing
744 parameters in DSA public key structures and return an error in the
745 DSA routines if parameters are absent.
748 *) In versions up to 0.9.6, RAND_file_name() resorted to file ".rnd"
749 in the current directory if neither $RANDFILE nor $HOME was set.
750 RAND_file_name() in 0.9.6a returned NULL in this case. This has
751 caused some confusion to Windows users who haven't defined $HOME.
752 Thus RAND_file_name() is changed again: e_os.h can define a
753 DEFAULT_HOME, which will be used if $HOME is not set.
754 For Windows, we use "C:"; on other platforms, we still require
755 environment variables.
757 +) Add "ex_data" support to ENGINE so implementations can add state at a
758 per-structure level rather than having to store it globally.
761 +) Make it possible for ENGINE structures to be copied when retrieved by
762 ENGINE_by_id() if the ENGINE specifies a new flag: ENGINE_FLAGS_BY_ID_COPY.
763 This causes the "original" ENGINE structure to act like a template,
764 analogous to the RSA vs. RSA_METHOD type of separation. Because of this
765 operational state can be localised to each ENGINE structure, despite the
766 fact they all share the same "methods". New ENGINE structures returned in
767 this case have no functional references and the return value is the single
768 structural reference. This matches the single structural reference returned
769 by ENGINE_by_id() normally, when it is incremented on the pre-existing
773 +) Fix ASN1 decoder when decoding type ANY and V_ASN1_OTHER: since this
774 needs to match any other type at all we need to manually clear the
778 +) Changes to the "openssl engine" utility to include;
779 - verbosity levels ('-v', '-vv', and '-vvv') that provide information
780 about an ENGINE's available control commands.
781 - executing control commands from command line arguments using the
782 '-pre' and '-post' switches. '-post' is only used if '-t' is
783 specified and the ENGINE is successfully initialised. The syntax for
784 the individual commands are colon-separated, for example;
785 openssl engine chil -pre FORK_CHECK:0 -pre SO_PATH:/lib/test.so
788 +) New dynamic control command support for ENGINEs. ENGINEs can now
789 declare their own commands (numbers), names (strings), descriptions,
790 and input types for run-time discovery by calling applications. A
791 subset of these commands are implicitly classed as "executable"
792 depending on their input type, and only these can be invoked through
793 the new string-based API function ENGINE_ctrl_cmd_string(). (Eg. this
794 can be based on user input, config files, etc). The distinction is
795 that "executable" commands cannot return anything other than a boolean
796 result and can only support numeric or string input, whereas some
797 discoverable commands may only be for direct use through
798 ENGINE_ctrl(), eg. supporting the exchange of binary data, function
799 pointers, or other custom uses. The "executable" commands are to
800 support parameterisations of ENGINE behaviour that can be
801 unambiguously defined by ENGINEs and used consistently across any
802 OpenSSL-based application. Commands have been added to all the
803 existing hardware-supporting ENGINEs, noticeably "SO_PATH" to allow
804 control over shared-library paths without source code alterations.
807 +) Changed all ENGINE implementations to dynamically allocate their
808 ENGINEs rather than declaring them statically. Apart from this being
809 necessary with the removal of the ENGINE_FLAGS_MALLOCED distinction,
810 this also allows the implementations to compile without using the
811 internal engine_int.h header.
814 +) Minor adjustment to "rand" code. RAND_get_rand_method() now returns a
815 'const' value. Any code that should be able to modify a RAND_METHOD
816 should already have non-const pointers to it (ie. they should only
817 modify their own ones).
820 +) Made a variety of little tweaks to the ENGINE code.
821 - "atalla" and "ubsec" string definitions were moved from header files
822 to C code. "nuron" string definitions were placed in variables
823 rather than hard-coded - allowing parameterisation of these values
824 later on via ctrl() commands.
825 - Removed unused "#if 0"'d code.
826 - Fixed engine list iteration code so it uses ENGINE_free() to release
827 structural references.
828 - Constified the RAND_METHOD element of ENGINE structures.
829 - Constified various get/set functions as appropriate and added
830 missing functions (including a catch-all ENGINE_cpy that duplicates
831 all ENGINE values onto a new ENGINE except reference counts/state).
832 - Removed NULL parameter checks in get/set functions. Setting a method
833 or function to NULL is a way of cancelling out a previously set
834 value. Passing a NULL ENGINE parameter is just plain stupid anyway
835 and doesn't justify the extra error symbols and code.
836 - Deprecate the ENGINE_FLAGS_MALLOCED define and move the area for
837 flags from engine_int.h to engine.h.
838 - Changed prototypes for ENGINE handler functions (init(), finish(),
839 ctrl(), key-load functions, etc) to take an (ENGINE*) parameter.
842 *) Move 'if (!initialized) RAND_poll()' into regions protected by
843 CRYPTO_LOCK_RAND. This is not strictly necessary, but avoids
844 having multiple threads call RAND_poll() concurrently.
847 *) In crypto/rand/md_rand.c, replace 'add_do_not_lock' flag by a
848 combination of a flag and a thread ID variable.
849 Otherwise while one thread is in ssleay_rand_bytes (which sets the
850 flag), *other* threads can enter ssleay_add_bytes without obeying
851 the CRYPTO_LOCK_RAND lock (and may even illegally release the lock
852 that they do not hold after the first thread unsets add_do_not_lock).
855 +) Implement binary inversion algorithm for BN_mod_inverse in addition
856 to the algorithm using long division. The binary algorithm can be
857 used only if the modulus is odd. On 32-bit systems, it is faster
858 only for relatively small moduli (roughly 20-30% for 128-bit moduli,
859 roughly 5-15% for 256-bit moduli), so we use it only for moduli
860 up to 450 bits. In 64-bit environments, the binary algorithm
861 appears to be advantageous for much longer moduli; here we use it
862 for moduli up to 2048 bits.
865 *) Change bctest again: '-x' expressions are not available in all
869 -) OpenSSL 0.9.6a released [5 Apr 2001]
871 *) Fix a couple of memory leaks in PKCS7_dataDecode()
872 [Steve Henson, reported by Heyun Zheng <hzheng@atdsprint.com>]
874 *) Change Configure and Makefiles to provide EXE_EXT, which will contain
875 the default extension for executables, if any. Also, make the perl
876 scripts that use symlink() to test if it really exists and use "cp"
877 if it doesn't. All this made OpenSSL compilable and installable in
881 +) Rewrite CHOICE field setting in ASN1_item_ex_d2i(). The old code
882 could not support the combine flag in choice fields.
885 -) Fix for asn1_GetSequence() for indefinite length constructed data.
886 If SEQUENCE is length is indefinite just set c->slen to the total
887 amount of data available.
888 [Steve Henson, reported by shige@FreeBSD.org]
889 [This change does not apply to 0.9.7.]
891 *) Change bctest to avoid here-documents inside command substitution
892 (workaround for FreeBSD /bin/sh bug).
893 For compatibility with Ultrix, avoid shell functions (introduced
894 in the bctest version that searches along $PATH).
897 *) Rename 'des_encrypt' to 'des_encrypt1'. This avoids the clashes
898 with des_encrypt() defined on some operating systems, like Solaris
902 *) Check the result of RSA-CRT (see D. Boneh, R. DeMillo, R. Lipton:
903 On the Importance of Eliminating Errors in Cryptographic
904 Computations, J. Cryptology 14 (2001) 2, 101-119,
905 http://theory.stanford.edu/~dabo/papers/faults.ps.gz).
908 *) MIPS assembler BIGNUM division bug fix.
911 *) Disabled incorrect Alpha assembler code.
914 -) Fix PKCS#7 decode routines so they correctly update the length
915 after reading an EOC for the EXPLICIT tag.
917 [This change does not apply to 0.9.7.]
919 *) Fix bug in PKCS#12 key generation routines. This was triggered
920 if a 3DES key was generated with a 0 initial byte. Include
921 PKCS12_BROKEN_KEYGEN compilation option to retain the old
922 (but broken) behaviour.
925 *) Enhance bctest to search for a working bc along $PATH and print
927 [Tim Rice <tim@multitalents.net> via Richard Levitte]
929 +) Add a 'copy_extensions' option to the 'ca' utility. This copies
930 extensions from a certificate request to the certificate.
933 +) Allow multiple 'certopt' and 'nameopt' options to be separated
934 by commas. Add 'namopt' and 'certopt' options to the 'ca' config
935 file: this allows the display of the certificate about to be
936 signed to be customised, to allow certain fields to be included
937 or excluded and extension details. The old system didn't display
938 multicharacter strings properly, omitted fields not in the policy
939 and couldn't display additional details such as extensions.
942 *) Fix memory leaks in err.c: free err_data string if necessary;
943 don't write to the wrong index in ERR_set_error_data.
946 +) Function EC_POINTs_mul for simultaneous scalar multiplication
947 of an arbitrary number of elliptic curve points, optionally
948 including the generator defined for the EC_GROUP.
949 EC_POINT_mul is a simple wrapper function for the typical case
950 that the point list has just one item (besides the optional
954 +) First EC_METHODs for curves over GF(p):
956 EC_GFp_simple_method() uses the basic BN_mod_mul and BN_mod_sqr
957 operations and provides various method functions that can also
958 operate with faster implementations of modular arithmetic.
960 EC_GFp_mont_method() reuses most functions that are part of
961 EC_GFp_simple_method, but uses Montgomery arithmetic.
963 [Bodo Moeller; point addition and point doubling
964 implementation directly derived from source code provided by
965 Lenka Fibikova <fibikova@exp-math.uni-essen.de>]
967 +) Framework for elliptic curves (crypto/ec/ec.h, crypto/ec/ec_lcl.h,
970 Curves are EC_GROUP objects (with an optional group generator)
971 based on EC_METHODs that are built into the library.
973 Points are EC_POINT objects based on EC_GROUP objects.
975 Most of the framework would be able to handle curves over arbitrary
976 finite fields, but as there are no obvious types for fields other
977 than GF(p), some functions are limited to that for now.
980 +) Add the -HTTP option to s_server. It is similar to -WWW, but requires
981 that the file contains a complete HTTP response.
984 +) Add the ec directory to mkdef.pl and mkfiles.pl. In mkdef.pl
985 change the def and num file printf format specifier from "%-40sXXX"
986 to "%-39s XXX". The latter will always guarantee a space after the
987 field while the former will cause them to run together if the field
988 is 40 of more characters long.
991 +) Constify the cipher and digest 'method' functions and structures
992 and modify related functions to take constant EVP_MD and EVP_CIPHER
996 *) Implement ssl23_peek (analogous to ssl23_read), which previously
1000 *) Replace rdtsc with _emit statements for VC++ version 5.
1001 [Jeremy Cooper <jeremy@baymoo.org>]
1003 +) Hide BN_CTX structure details in bn_lcl.h instead of publishing them
1004 in <openssl/bn.h>. Also further increase BN_CTX_NUM to 32.
1007 +) Modify EVP_Digest*() routines so they now return values. Although the
1008 internal software routines can never fail additional hardware versions
1012 +) Clean up crypto/err/err.h and change some error codes to avoid conflicts:
1014 Previously ERR_R_FATAL was too small and coincided with ERR_LIB_PKCS7
1015 (= ERR_R_PKCS7_LIB); it is now 64 instead of 32.
1018 ERR_R_NESTED_ASN1_ERROR
1020 ERR_R_MISSING_ASN1_EOS
1021 were 4 .. 9, conflicting with
1022 ERR_LIB_RSA (= ERR_R_RSA_LIB)
1024 ERR_LIB_PEM (= ERR_R_PEM_LIB).
1025 They are now 58 .. 63 (i.e., just below ERR_R_FATAL).
1027 Add new error code 'ERR_R_INTERNAL_ERROR'.
1030 +) Don't overuse locks in crypto/err/err.c: For data retrieval, CRYPTO_r_lock
1034 +) New option '-subj arg' for 'openssl req' and 'openssl ca'. This
1035 sets the subject name for a new request or supersedes the
1036 subject name in a given request. Formats that can be parsed are
1037 'CN=Some Name, OU=myOU, C=IT'
1039 'CN=Some Name/OU=myOU/C=IT'.
1041 Add options '-batch' and '-verbose' to 'openssl req'.
1042 [Massimiliano Pala <madwolf@hackmasters.net>]
1044 *) Make it possible to reuse SSLv2 sessions.
1047 +) Introduce the possibility to access global variables through
1048 functions on platform were that's the best way to handle exporting
1049 global variables in shared libraries. To enable this functionality,
1050 one must configure with "EXPORT_VAR_AS_FN" or defined the C macro
1051 "OPENSSL_EXPORT_VAR_AS_FUNCTION" in crypto/opensslconf.h (the latter
1052 is normally done by Configure or something similar).
1054 To implement a global variable, use the macro OPENSSL_IMPLEMENT_GLOBAL
1055 in the source file (foo.c) like this:
1057 OPENSSL_IMPLEMENT_GLOBAL(int,foo)=1;
1058 OPENSSL_IMPLEMENT_GLOBAL(double,bar);
1060 To declare a global variable, use the macros OPENSSL_DECLARE_GLOBAL
1061 and OPENSSL_GLOBAL_REF in the header file (foo.h) like this:
1063 OPENSSL_DECLARE_GLOBAL(int,foo);
1064 #define foo OPENSSL_GLOBAL_REF(foo)
1065 OPENSSL_DECLARE_GLOBAL(double,bar);
1066 #define bar OPENSSL_GLOBAL_REF(bar)
1068 The #defines are very important, and therefore so is including the
1069 header file everywhere where the defined globals are used.
1071 The macro OPENSSL_EXPORT_VAR_AS_FUNCTION also affects the definition
1072 of ASN.1 items, but that structure is a bit different.
1074 The largest change is in util/mkdef.pl which has been enhanced with
1075 better and easier to understand logic to choose which symbols should
1076 go into the Windows .def files as well as a number of fixes and code
1077 cleanup (among others, algorithm keywords are now sorted
1078 lexicographically to avoid constant rewrites).
1081 *) In copy_email() check for >= 0 as a return value for
1082 X509_NAME_get_index_by_NID() since 0 is a valid index.
1083 [Steve Henson reported by Massimiliano Pala <madwolf@opensca.org>]
1085 +) In BN_div() keep a copy of the sign of 'num' before writing the
1086 result to 'rm' because if rm==num the value will be overwritten
1087 and produce the wrong result if 'num' is negative: this caused
1088 problems with BN_mod() and BN_nnmod().
1091 +) Function OCSP_request_verify(). This checks the signature on an
1092 OCSP request and verifies the signer certificate. The signer
1093 certificate is just checked for a generic purpose and OCSP request
1097 +) Add OCSP_check_validity() function to check the validity of OCSP
1098 responses. OCSP responses are prepared in real time and may only
1099 be a few seconds old. Simply checking that the current time lies
1100 between thisUpdate and nextUpdate max reject otherwise valid responses
1101 caused by either OCSP responder or client clock inaccuracy. Instead
1102 we allow thisUpdate and nextUpdate to fall within a certain period of
1103 the current time. The age of the response can also optionally be
1104 checked. Two new options -validity_period and -status_age added to
1108 +) If signature or public key algorithm is unrecognized print out its
1109 OID rather that just UNKNOWN.
1112 *) Avoid coredump with unsupported or invalid public keys by checking if
1113 X509_get_pubkey() fails in PKCS7_verify(). Fix memory leak when
1114 PKCS7_verify() fails with non detached data.
1117 +) Change OCSP_cert_to_id() to tolerate a NULL subject certificate and
1118 OCSP_cert_id_new() a NULL serialNumber. This allows a partial certificate
1119 ID to be generated from the issuer certificate alone which can then be
1120 passed to OCSP_id_issuer_cmp().
1123 +) New compilation option ASN1_ITEM_FUNCTIONS. This causes the new
1124 ASN1 modules to export functions returning ASN1_ITEM pointers
1125 instead of the ASN1_ITEM structures themselves. This adds several
1126 new macros which allow the underlying ASN1 function/structure to
1127 be accessed transparently. As a result code should not use ASN1_ITEM
1128 references directly (such as &X509_it) but instead use the relevant
1129 macros (such as ASN1_ITEM_rptr(X509)). This option is to allow
1130 use of the new ASN1 code on platforms where exporting structures
1131 is problematical (for example in shared libraries) but exporting
1132 functions returning pointers to structures is not.
1135 +) Add support for overriding the generation of SSL/TLS session IDs.
1136 These callbacks can be registered either in an SSL_CTX or per SSL.
1137 The purpose of this is to allow applications to control, if they wish,
1138 the arbitrary values chosen for use as session IDs, particularly as it
1139 can be useful for session caching in multiple-server environments. A
1140 command-line switch for testing this (and any client code that wishes
1141 to use such a feature) has been added to "s_server".
1142 [Geoff Thorpe, Lutz Jaenicke]
1144 +) Modify mkdef.pl to recognise and parse preprocessor conditionals
1145 of the form '#if defined(...) || defined(...) || ...' and
1146 '#if !defined(...) && !defined(...) && ...'. This also avoids
1147 the growing number of special cases it was previously handling.
1150 *) Don't use getenv in library functions when run as setuid/setgid.
1151 New function OPENSSL_issetugid().
1154 +) Make all configuration macros available for application by making
1155 sure they are available in opensslconf.h, by giving them names starting
1156 with "OPENSSL_" to avoid conflicts with other packages and by making
1157 sure e_os2.h will cover all platform-specific cases together with
1159 Additionally, it is now possible to define configuration/platform-
1160 specific names (called "system identities"). In the C code, these
1161 are prefixed with "OPENSSL_SYSNAME_". e_os2.h will create another
1162 macro with the name beginning with "OPENSSL_SYS_", which is determined
1163 from "OPENSSL_SYSNAME_*" or compiler-specific macros depending on
1167 +) New option -set_serial to 'req' and 'x509' this allows the serial
1168 number to use to be specified on the command line. Previously self
1169 signed certificates were hard coded with serial number 0 and the
1170 CA options of 'x509' had to use a serial number in a file which was
1174 *) Avoid false positives in memory leak detection code (crypto/mem_dbg.c)
1175 due to incorrect handling of multi-threading:
1177 1. Fix timing glitch in the MemCheck_off() portion of CRYPTO_mem_ctrl().
1179 2. Fix logical glitch in is_MemCheck_on() aka CRYPTO_is_mem_check_on().
1181 3. Count how many times MemCheck_off() has been called so that
1182 nested use can be treated correctly. This also avoids
1183 inband-signalling in the previous code (which relied on the
1184 assumption that thread ID 0 is impossible).
1187 +) New options to 'ca' utility to support V2 CRL entry extensions.
1188 Currently CRL reason, invalidity date and hold instruction are
1189 supported. Add new CRL extensions to V3 code and some new objects.
1192 *) Add "-rand" option also to s_client and s_server.
1195 +) New function EVP_CIPHER_CTX_set_padding() this is used to
1196 disable standard block padding (aka PKCS#5 padding) in the EVP
1197 API, which was previously mandatory. This means that the data is
1198 not padded in any way and so the total length much be a multiple
1199 of the block size, otherwise an error occurs.
1202 +) Initial (incomplete) OCSP SSL support.
1205 *) Fix CPU detection on Irix 6.x.
1206 [Kurt Hockenbury <khockenb@stevens-tech.edu> and
1207 "Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1209 +) New function OCSP_parse_url(). This splits up a URL into its host,
1210 port and path components: primarily to parse OCSP URLs. New -url
1211 option to ocsp utility.
1214 +) New nonce behavior. The return value of OCSP_check_nonce() now
1215 reflects the various checks performed. Applications can decide
1216 whether to tolerate certain situations such as an absent nonce
1217 in a response when one was present in a request: the ocsp application
1218 just prints out a warning. New function OCSP_add1_basic_nonce()
1219 this is to allow responders to include a nonce in a response even if
1220 the request is nonce-less.
1223 *) Fix X509_NAME bug which produced incorrect encoding if X509_NAME
1226 [This change does not apply to 0.9.7.]
1228 *) Use the cached encoding of an X509_NAME structure rather than
1229 copying it. This is apparently the reason for the libsafe "errors"
1230 but the code is actually correct.
1233 +) Disable stdin buffering in load_cert (apps/apps.c) so that no certs are
1234 skipped when using openssl x509 multiple times on a single input file,
1235 e.g. "(openssl x509 -out cert1; openssl x509 -out cert2) <certs".
1238 +) Make ASN1_UTCTIME_set_string() and ASN1_GENERALIZEDTIME_set_string()
1239 set string type: to handle setting ASN1_TIME structures. Fix ca
1240 utility to correctly initialize revocation date of CRLs.
1243 +) New option SSL_OP_CIPHER_SERVER_PREFERENCE allows the server to override
1244 the clients preferred ciphersuites and rather use its own preferences.
1245 Should help to work around M$ SGC (Server Gated Cryptography) bug in
1246 Internet Explorer by ensuring unchanged hash method during stepup.
1247 (Also replaces the broken/deactivated SSL_OP_NON_EXPORT_FIRST option.)
1250 +) Make mkdef.pl recognise all DECLARE_ASN1 macros, change rijndael
1251 to aes and add a new 'exist' option to print out symbols that don't
1255 +) Additional options to ocsp utility to allow flags to be set and
1256 additional certificates supplied.
1259 +) Add the option -VAfile to 'openssl ocsp', so the user can give the
1260 OCSP client a number of certificate to only verify the response
1264 *) Add new function BN_rand_range(), and fix DSA_sign_setup() to prevent
1265 Bleichenbacher's DSA attack.
1266 Extend BN_[pseudo_]rand: As before, top=1 forces the highest two bits
1267 to be set and top=0 forces the highest bit to be set; top=-1 is new
1268 and leaves the highest bit random.
1269 [Ulf Moeller, Bodo Moeller]
1271 +) Update Rijndael code to version 3.0 and change EVP AES ciphers to
1272 handle the new API. Currently only ECB, CBC modes supported. Add new
1273 AES OIDs. Add TLS AES ciphersuites as described in the "AES Ciphersuites
1274 for TLS" draft-ietf-tls-ciphersuite-03.txt.
1275 [Ben Laurie, Steve Henson]
1277 *) In the NCONF_...-based implementations for CONF_... queries
1278 (crypto/conf/conf_lib.c), if the input LHASH is NULL, avoid using
1279 a temporary CONF structure with the data component set to NULL
1280 (which gives segmentation faults in lh_retrieve).
1281 Instead, use NULL for the CONF pointer in CONF_get_string and
1282 CONF_get_number (which may use environment variables) and directly
1283 return NULL from CONF_get_section.
1286 *) Fix potential buffer overrun for EBCDIC.
1289 +) New function OCSP_copy_nonce() to copy nonce value (if present) from
1290 request to response.
1293 +) Functions for OCSP responders. OCSP_request_onereq_count(),
1294 OCSP_request_onereq_get0(), OCSP_onereq_get0_id() and OCSP_id_get0_info()
1295 extract information from a certificate request. OCSP_response_create()
1296 creates a response and optionally adds a basic response structure.
1297 OCSP_basic_add1_status() adds a complete single response to a basic
1298 response and returns the OCSP_SINGLERESP structure just added (to allow
1299 extensions to be included for example). OCSP_basic_add1_cert() adds a
1300 certificate to a basic response and OCSP_basic_sign() signs a basic
1301 response with various flags. New helper functions ASN1_TIME_check()
1302 (checks validity of ASN1_TIME structure) and ASN1_TIME_to_generalizedtime()
1303 (converts ASN1_TIME to GeneralizedTime).
1306 +) Various new functions. EVP_Digest() combines EVP_Digest{Init,Update,Final}()
1307 in a single operation. X509_get0_pubkey_bitstr() extracts the public_key
1308 structure from a certificate. X509_pubkey_digest() digests the public_key
1309 contents: this is used in various key identifiers.
1312 *) Tolerate nonRepudiation as being valid for S/MIME signing and certSign
1313 keyUsage if basicConstraints absent for a CA.
1316 *) Make SMIME_write_PKCS7() write mail header values with a format that
1317 is more generally accepted (no spaces before the semicolon), since
1318 some programs can't parse those values properly otherwise. Also make
1319 sure BIO's that break lines after each write do not create invalid
1323 +) Make sk_sort() tolerate a NULL argument.
1324 [Steve Henson reported by Massimiliano Pala <madwolf@comune.modena.it>]
1326 +) New OCSP verify flag OCSP_TRUSTOTHER. When set the "other" certificates
1327 passed by the function are trusted implicitly. If any of them signed the
1328 response then it is assumed to be valid and is not verified.
1331 -) Make the CRL encoding routines work with empty SEQUENCE OF. The
1332 macros previously used would not encode an empty SEQUENCE OF
1333 and break the signature.
1335 [This change does not apply to 0.9.7.]
1337 *) Zero the premaster secret after deriving the master secret in
1341 +) In PKCS7_set_type() initialise content_type in PKCS7_ENC_CONTENT
1342 to data. This was previously part of the PKCS7 ASN1 code. This
1343 was causing problems with OpenSSL created PKCS#12 and PKCS#7 structures.
1344 [Steve Henson, reported by Kenneth R. Robinette
1345 <support@securenetterm.com>]
1347 +) Add CRYPTO_push_info() and CRYPTO_pop_info() calls to new ASN1
1348 routines: without these tracing memory leaks is very painful.
1349 Fix leaks in PKCS12 and PKCS7 routines.
1352 *) Add some EVP_add_digest_alias registrations (as found in
1353 OpenSSL_add_all_digests()) to SSL_library_init()
1354 aka OpenSSL_add_ssl_algorithms(). This provides improved
1355 compatibility with peers using X.509 certificates
1356 with unconventional AlgorithmIdentifier OIDs.
1359 *) Fix for Irix with NO_ASM.
1360 ["Bruce W. Forsberg" <bruce.forsberg@baesystems.com>]
1362 *) ./config script fixes.
1363 [Ulf Moeller, Richard Levitte]
1365 +) Make X509_time_adj() cope with the new behaviour of ASN1_TIME_new().
1366 Previously it initialised the 'type' argument to V_ASN1_UTCTIME which
1367 effectively meant GeneralizedTime would never be used. Now it
1368 is initialised to -1 but X509_time_adj() now has to check the value
1369 and use ASN1_TIME_set() if the value is not V_ASN1_UTCTIME or
1370 V_ASN1_GENERALIZEDTIME, without this it always uses GeneralizedTime.
1371 [Steve Henson, reported by Kenneth R. Robinette
1372 <support@securenetterm.com>]
1374 +) Fixes to BN_to_ASN1_INTEGER when bn is zero. This would previously
1375 result in a zero length in the ASN1_INTEGER structure which was
1376 not consistent with the structure when d2i_ASN1_INTEGER() was used
1377 and would cause ASN1_INTEGER_cmp() to fail. Enhance s2i_ASN1_INTEGER()
1378 to cope with hex and negative integers. Fix bug in i2a_ASN1_INTEGER()
1379 where it did not print out a minus for negative ASN1_INTEGER.
1382 *) Fix 'openssl passwd -1'.
1385 +) Add summary printout to ocsp utility. The various functions which
1386 convert status values to strings have been renamed to:
1387 OCSP_response_status_str(), OCSP_cert_status_str() and
1388 OCSP_crl_reason_str() and are no longer static. New options
1389 to verify nonce values and to disable verification. OCSP response
1390 printout format cleaned up.
1393 +) Add additional OCSP certificate checks. These are those specified
1394 in RFC2560. This consists of two separate checks: the CA of the
1395 certificate being checked must either be the OCSP signer certificate
1396 or the issuer of the OCSP signer certificate. In the latter case the
1397 OCSP signer certificate must contain the OCSP signing extended key
1398 usage. This check is performed by attempting to match the OCSP
1399 signer or the OCSP signer CA to the issuerNameHash and issuerKeyHash
1400 in the OCSP_CERTID structures of the response.
1403 +) Initial OCSP certificate verification added to OCSP_basic_verify()
1404 and related routines. This uses the standard OpenSSL certificate
1405 verify routines to perform initial checks (just CA validity) and
1406 to obtain the certificate chain. Then additional checks will be
1407 performed on the chain. Currently the root CA is checked to see
1408 if it is explicitly trusted for OCSP signing. This is used to set
1409 a root CA as a global signing root: that is any certificate that
1410 chains to that CA is an acceptable OCSP signing certificate.
1413 +) New '-extfile ...' option to 'openssl ca' for reading X.509v3
1414 extensions from a separate configuration file.
1415 As when reading extensions from the main configuration file,
1416 the '-extensions ...' option may be used for specifying the
1418 [Massimiliano Pala <madwolf@comune.modena.it>]
1420 *) Change PKCS12_key_gen_asc() so it can cope with non null
1421 terminated strings whose length is passed in the passlen
1422 parameter, for example from PEM callbacks. This was done
1423 by adding an extra length parameter to asc2uni().
1424 [Steve Henson, reported by <oddissey@samsung.co.kr>]
1426 +) New OCSP utility. Allows OCSP requests to be generated or
1427 read. The request can be sent to a responder and the output
1428 parsed, outputed or printed in text form. Not complete yet:
1429 still needs to check the OCSP response validity.
1432 +) New subcommands for 'openssl ca':
1433 'openssl ca -status <serial>' prints the status of the cert with
1434 the given serial number (according to the index file).
1435 'openssl ca -updatedb' updates the expiry status of certificates
1437 [Massimiliano Pala <madwolf@comune.modena.it>]
1439 +) New '-newreq-nodes' command option to CA.pl. This is like
1440 '-newreq', but calls 'openssl req' with the '-nodes' option
1441 so that the resulting key is not encrypted.
1442 [Damien Miller <djm@mindrot.org>]
1444 +) New configuration for the GNU Hurd.
1445 [Jonathan Bartlett <johnnyb@wolfram.com> via Richard Levitte]
1447 +) Initial code to implement OCSP basic response verify. This
1448 is currently incomplete. Currently just finds the signer's
1449 certificate and verifies the signature on the response.
1452 +) New SSLeay_version code SSLEAY_DIR to determine the compiled-in
1453 value of OPENSSLDIR. This is available via the new '-d' option
1454 to 'openssl version', and is also included in 'openssl version -a'.
1457 *) Fix C code generated by 'openssl dsaparam -C': If a BN_bin2bn
1458 call failed, free the DSA structure.
1461 +) Allowing defining memory allocation callbacks that will be given
1462 file name and line number information in additional arguments
1463 (a const char* and an int). The basic functionality remains, as
1464 well as the original possibility to just replace malloc(),
1465 realloc() and free() by functions that do not know about these
1466 additional arguments. To register and find out the current
1467 settings for extended allocation functions, the following
1468 functions are provided:
1470 CRYPTO_set_mem_ex_functions
1471 CRYPTO_set_locked_mem_ex_functions
1472 CRYPTO_get_mem_ex_functions
1473 CRYPTO_get_locked_mem_ex_functions
1475 These work the same way as CRYPTO_set_mem_functions and friends.
1476 CRYPTO_get_[locked_]mem_functions now writes 0 where such an
1477 extended allocation function is enabled.
1478 Similarly, CRYPTO_get_[locked_]mem_ex_functions writes 0 where
1479 a conventional allocation function is enabled.
1480 [Richard Levitte, Bodo Moeller]
1482 *) Fix to uni2asc() to cope with zero length Unicode strings.
1483 These are present in some PKCS#12 files.
1486 +) Finish off removing the remaining LHASH function pointer casts.
1487 There should no longer be any prototype-casting required when using
1488 the LHASH abstraction, and any casts that remain are "bugs". See
1489 the callback types and macros at the head of lhash.h for details
1490 (and "OBJ_cleanup" in crypto/objects/obj_dat.c as an example).
1493 +) Add automatic query of EGD sockets in RAND_poll() for the unix variant.
1494 If /dev/[u]random devices are not available or do not return enough
1495 entropy, EGD style sockets (served by EGD or PRNGD) will automatically
1497 The locations /var/run/egd-pool, /dev/egd-pool, /etc/egd-pool, and
1498 /etc/entropy will be queried once each in this sequence, quering stops
1499 when enough entropy was collected without querying more sockets.
1502 +) Change the Unix RAND_poll() variant to be able to poll several
1503 random devices, as specified by DEVRANDOM, until a sufficient amount
1504 of data has been collected. We spend at most 10 ms on each file
1505 (select timeout) and read in non-blocking mode. DEVRANDOM now
1506 defaults to the list "/dev/urandom", "/dev/random", "/dev/srandom"
1507 (previously it was just the string "/dev/urandom"), so on typical
1508 platforms the 10 ms delay will never occur.
1509 Also separate out the Unix variant to its own file, rand_unix.c.
1510 For VMS, there's a currently-empty rand_vms.c.
1513 +) Move OCSP client related routines to ocsp_cl.c. These
1514 provide utility functions which an application needing
1515 to issue a request to an OCSP responder and analyse the
1516 response will typically need: as opposed to those which an
1517 OCSP responder itself would need which will be added later.
1519 OCSP_request_sign() signs an OCSP request with an API similar
1520 to PKCS7_sign(). OCSP_response_status() returns status of OCSP
1521 response. OCSP_response_get1_basic() extracts basic response
1522 from response. OCSP_resp_find_status(): finds and extracts status
1523 information from an OCSP_CERTID structure (which will be created
1524 when the request structure is built). These are built from lower
1525 level functions which work on OCSP_SINGLERESP structures but
1526 wont normally be used unless the application wishes to examine
1527 extensions in the OCSP response for example.
1529 Replace nonce routines with a pair of functions.
1530 OCSP_request_add1_nonce() adds a nonce value and optionally
1531 generates a random value. OCSP_check_nonce() checks the
1532 validity of the nonce in an OCSP response.
1535 +) Change function OCSP_request_add() to OCSP_request_add0_id().
1536 This doesn't copy the supplied OCSP_CERTID and avoids the
1537 need to free up the newly created id. Change return type
1538 to OCSP_ONEREQ to return the internal OCSP_ONEREQ structure.
1539 This can then be used to add extensions to the request.
1540 Deleted OCSP_request_new(), since most of its functionality
1541 is now in OCSP_REQUEST_new() (and the case insensitive name
1542 clash) apart from the ability to set the request name which
1543 will be added elsewhere.
1546 +) Update OCSP API. Remove obsolete extensions argument from
1547 various functions. Extensions are now handled using the new
1548 OCSP extension code. New simple OCSP HTTP function which
1549 can be used to send requests and parse the response.
1552 +) Fix the PKCS#7 (S/MIME) code to work with new ASN1. Two new
1553 ASN1_ITEM structures help with sign and verify. PKCS7_ATTR_SIGN
1554 uses the special reorder version of SET OF to sort the attributes
1555 and reorder them to match the encoded order. This resolves a long
1556 standing problem: a verify on a PKCS7 structure just after signing
1557 it used to fail because the attribute order did not match the
1558 encoded order. PKCS7_ATTR_VERIFY does not reorder the attributes:
1559 it uses the received order. This is necessary to tolerate some broken
1560 software that does not order SET OF. This is handled by encoding
1561 as a SEQUENCE OF but using implicit tagging (with UNIVERSAL class)
1562 to produce the required SET OF.
1565 +) Have mk1mf.pl generate the macros OPENSSL_BUILD_SHLIBCRYPTO and
1566 OPENSSL_BUILD_SHLIBSSL and use them appropriately in the header
1567 files to get correct declarations of the ASN.1 item variables.
1570 +) Rewrite of PKCS#12 code to use new ASN1 functionality. Replace many
1571 PKCS#12 macros with real functions. Fix two unrelated ASN1 bugs:
1572 asn1_check_tlen() would sometimes attempt to use 'ctx' when it was
1573 NULL and ASN1_TYPE was not dereferenced properly in asn1_ex_c2i().
1574 New ASN1 macro: DECLARE_ASN1_ITEM() which just declares the relevant
1575 ASN1_ITEM and no wrapper functions.
1578 +) New functions or ASN1_item_d2i_fp() and ASN1_item_d2i_bio(). These
1579 replace the old function pointer based I/O routines. Change most of
1580 the *_d2i_bio() and *_d2i_fp() functions to use these.
1583 +) Enhance mkdef.pl to be more accepting about spacing in C preprocessor
1584 lines, recognice more "algorithms" that can be deselected, and make
1585 it complain about algorithm deselection that isn't recognised.
1588 +) New ASN1 functions to handle dup, sign, verify, digest, pack and
1589 unpack operations in terms of ASN1_ITEM. Modify existing wrappers
1590 to use new functions. Add NO_ASN1_OLD which can be set to remove
1591 some old style ASN1 functions: this can be used to determine if old
1592 code will still work when these eventually go away.
1595 +) New extension functions for OCSP structures, these follow the
1596 same conventions as certificates and CRLs.
1599 +) New function X509V3_add1_i2d(). This automatically encodes and
1600 adds an extension. Its behaviour can be customised with various
1601 flags to append, replace or delete. Various wrappers added for
1602 certifcates and CRLs.
1605 +) Fix to avoid calling the underlying ASN1 print routine when
1606 an extension cannot be parsed. Correct a typo in the
1607 OCSP_SERVICELOC extension. Tidy up print OCSP format.
1610 *) Increase s2->wbuf allocation by one byte in ssl2_new (ssl/s2_lib.c).
1611 Otherwise do_ssl_write (ssl/s2_pkt.c) will write beyond buffer limits
1612 when writing a 32767 byte record.
1613 [Bodo Moeller; problem reported by Eric Day <eday@concentric.net>]
1615 *) In RSA_eay_public_{en,ed}crypt and RSA_eay_mod_exp (rsa_eay.c),
1616 obtain lock CRYPTO_LOCK_RSA before setting rsa->_method_mod_{n,p,q}.
1618 (RSA objects have a reference count access to which is protected
1619 by CRYPTO_LOCK_RSA [see rsa_lib.c, s3_srvr.c, ssl_cert.c, ssl_rsa.c],
1620 so they are meant to be shared between threads.)
1621 [Bodo Moeller, Geoff Thorpe; original patch submitted by
1622 "Reddie, Steven" <Steven.Reddie@ca.com>]
1624 +) Make mkdef.pl parse some of the ASN1 macros and add apropriate
1625 entries for variables.
1628 *) Fix a deadlock in CRYPTO_mem_leaks().
1631 +) Add functionality to apps/openssl.c for detecting locking
1632 problems: As the program is single-threaded, all we have
1633 to do is register a locking callback using an array for
1634 storing which locks are currently held by the program.
1637 +) Use a lock around the call to CRYPTO_get_ex_new_index() in
1638 SSL_get_ex_data_X509_STORE_idx(), which is used in
1639 ssl_verify_cert_chain() and thus can be called at any time
1640 during TLS/SSL handshakes so that thread-safety is essential.
1641 Unfortunately, the ex_data design is not at all suited
1642 for multi-threaded use, so it probably should be abolished.
1645 +) Added Broadcom "ubsec" ENGINE to OpenSSL.
1646 [Broadcom, tweaked and integrated by Geoff Thorpe]
1648 +) Move common extension printing code to new function
1649 X509V3_print_extensions(). Reorganise OCSP print routines and
1650 implement some needed OCSP ASN1 functions. Add OCSP extensions.
1653 +) New function X509_signature_print() to remove duplication in some
1657 +) Add a special meaning when SET OF and SEQUENCE OF flags are both
1658 set (this was treated exactly the same as SET OF previously). This
1659 is used to reorder the STACK representing the structure to match the
1660 encoding. This will be used to get round a problem where a PKCS7
1661 structure which was signed could not be verified because the STACK
1662 order did not reflect the encoded order.
1665 +) Reimplement the OCSP ASN1 module using the new code.
1668 +) Update the X509V3 code to permit the use of an ASN1_ITEM structure
1669 for its ASN1 operations. The old style function pointers still exist
1670 for now but they will eventually go away.
1673 +) Merge in replacement ASN1 code from the ASN1 branch. This almost
1674 completely replaces the old ASN1 functionality with a table driven
1675 encoder and decoder which interprets an ASN1_ITEM structure describing
1676 the ASN1 module. Compatibility with the existing ASN1 API (i2d,d2i) is
1677 largely maintained. Almost all of the old asn1_mac.h macro based ASN1
1678 has also been converted to the new form.
1681 +) Change BN_mod_exp_recp so that negative moduli are tolerated
1682 (the sign is ignored). Similarly, ignore the sign in BN_MONT_CTX_set
1683 so that BN_mod_exp_mont and BN_mod_exp_mont_word work
1684 for negative moduli.
1687 +) Fix BN_uadd and BN_usub: Always return non-negative results instead
1688 of not touching the result's sign bit.
1691 +) BN_div bugfix: If the result is 0, the sign (res->neg) must not be
1695 +) Changed the LHASH code to use prototypes for callbacks, and created
1696 macros to declare and implement thin (optionally static) functions
1697 that provide type-safety and avoid function pointer casting for the
1698 type-specific callbacks.
1701 *) Use better test patterns in bntest.
1704 +) Added Kerberos Cipher Suites to be used with TLS, as written in
1706 [Veers Staats <staatsvr@asc.hpc.mil>,
1707 Jeffrey Altman <jaltman@columbia.edu>, via Richard Levitte]
1709 *) rand_win.c fix for Borland C.
1712 *) BN_rshift bugfix for n == 0.
1715 +) Reformat the FAQ so the different questions and answers can be divided
1716 in sections depending on the subject.
1719 +) Have the zlib compression code load ZLIB.DLL dynamically under
1723 +) New function BN_mod_sqrt for computing square roots modulo a prime
1724 (using the probabilistic Tonelli-Shanks algorithm unless
1725 p == 3 (mod 4) or p == 5 (mod 8), which are cases that can
1726 be handled deterministically).
1727 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1729 *) Add a 'bctest' script that checks for some known 'bc' bugs
1730 so that 'make test' does not abort just because 'bc' is broken.
1733 *) Store verify_result within SSL_SESSION also for client side to
1734 avoid potential security hole. (Re-used sessions on the client side
1735 always resulted in verify_result==X509_V_OK, not using the original
1736 result of the server certificate verification.)
1739 +) Make BN_mod_inverse faster by explicitly handling small quotients
1740 in the Euclid loop. (Speed gain about 20% for small moduli [256 or
1741 512 bits], about 30% for larger ones [1024 or 2048 bits].)
1744 *) Fix ssl3_pending: If the record in s->s3->rrec is not of type
1745 SSL3_RT_APPLICATION_DATA, return 0.
1746 Similarly, change ssl2_pending to return 0 if SSL_in_init(s) is true.
1750 Both ssl2_peek and ssl3_peek, which were totally broken in earlier
1751 releases, have been re-implemented by renaming the previous
1752 implementations of ssl2_read and ssl3_read to ssl2_read_internal
1753 and ssl3_read_internal, respectively, and adding 'peek' parameters
1754 to them. The new ssl[23]_{read,peek} functions are calls to
1755 ssl[23]_read_internal with the 'peek' flag set appropriately.
1756 A 'peek' parameter has also been added to ssl3_read_bytes, which
1757 does the actual work for ssl3_read_internal.
1760 +) New function BN_kronecker.
1763 +) Fix BN_gcd so that it works on negative inputs; the result is
1764 positive unless both parameters are zero.
1765 Previously something reasonably close to an infinite loop was
1766 possible because numbers could be growing instead of shrinking
1767 in the implementation of Euclid's algorithm.
1770 +) Fix BN_is_word() and BN_is_one() macros to take into account the
1771 sign of the number in question.
1773 Fix BN_is_word(a,w) to work correctly for w == 0.
1775 The old BN_is_word(a,w) macro is now called BN_abs_is_word(a,w)
1776 because its test if the absolute value of 'a' equals 'w'.
1777 Note that BN_abs_is_word does *not* handle w == 0 reliably;
1778 it exists mostly for use in the implementations of BN_is_zero(),
1779 BN_is_one(), and BN_is_word().
1782 *) Initialise "ex_data" member of RSA/DSA/DH structures prior to calling
1783 the method-specific "init()" handler. Also clean up ex_data after
1784 calling the method-specific "finish()" handler. Previously, this was
1785 happening the other way round.
1788 +) New function BN_swap.
1791 +) Use BN_nnmod instead of BN_mod in crypto/bn/bn_exp.c so that
1792 the exponentiation functions are more likely to produce reasonable
1793 results on negative inputs.
1796 +) Change BN_mod_mul so that the result is always non-negative.
1797 Previously, it could be negative if one of the factors was negative;
1798 I don't think anyone really wanted that behaviour.
1801 +) Move BN_mod_... functions into new file crypto/bn/bn_mod.c
1802 (except for exponentiation, which stays in crypto/bn/bn_exp.c,
1803 and BN_mod_mul_reciprocal, which stays in crypto/bn/bn_recp.c)
1804 and add new functions:
1813 BN_mod_lshift1_quick
1817 These functions always generate non-negative results.
1819 BN_nnmod otherwise is like BN_mod (if BN_mod computes a remainder r
1820 such that |m| < r < 0, BN_nnmod will output rem + |m| instead).
1822 BN_mod_XXX_quick(r, a, [b,] m) generates the same result as
1823 BN_mod_XXX(r, a, [b,] m, ctx), but requires that a [and b]
1824 be reduced modulo m.
1825 [Lenka Fibikova <fibikova@exp-math.uni-essen.de>, Bodo Moeller]
1827 +) Remove a few calls to bn_wexpand() in BN_sqr() (the one in there
1828 was actually never needed) and in BN_mul(). The removal in BN_mul()
1829 required a small change in bn_mul_part_recursive() and the addition
1830 of the functions bn_cmp_part_words(), bn_sub_part_words() and
1831 bn_add_part_words(), which do the same thing as bn_cmp_words(),
1832 bn_sub_words() and bn_add_words() except they take arrays with
1836 +) In 'openssl passwd', verify passwords read from the terminal
1837 unless the '-salt' option is used (which usually means that
1838 verification would just waste user's time since the resulting
1839 hash is going to be compared with some given password hash)
1840 or the new '-noverify' option is used.
1842 This is an incompatible change, but it does not affect
1843 non-interactive use of 'openssl passwd' (passwords on the command
1844 line, '-stdin' option, '-in ...' option) and thus should not
1848 +) Remove all references to RSAref, since there's no more need for it.
1851 *) Increase BN_CTX_NUM (the number of BIGNUMs in a BN_CTX) to 16.
1852 The previous value, 12, was not always sufficient for BN_mod_exp().
1855 +) Make DSO load along a path given through an environment variable
1856 (SHLIB_PATH) with shl_load().
1859 +) Constify the ENGINE code as a result of BIGNUM constification.
1860 Also constify the RSA code and most things related to it. In a
1861 few places, most notable in the depth of the ASN.1 code, ugly
1862 casts back to non-const were required (to be solved at a later
1866 +) Make it so the openssl application has all engines loaded by default.
1869 +) Constify the BIGNUM routines a little more.
1872 *) Make sure that shared libraries get the internal name engine with
1873 the full version number and not just 0. This should mark the
1874 shared libraries as not backward compatible. Of course, this should
1875 be changed again when we can guarantee backward binary compatibility.
1878 +) Add the following functions:
1880 ENGINE_load_cswift()
1882 ENGINE_load_atalla()
1884 ENGINE_load_builtin_engines()
1886 That way, an application can itself choose if external engines that
1887 are built-in in OpenSSL shall ever be used or not. The benefit is
1888 that applications won't have to be linked with libdl or other dso
1889 libraries unless it's really needed.
1891 Changed 'openssl engine' to load all engines on demand.
1892 Changed the engine header files to avoid the duplication of some
1893 declarations (they differed!).
1896 +) 'openssl engine' can now list capabilities.
1899 +) Better error reporting in 'openssl engine'.
1902 +) Never call load_dh_param(NULL) in s_server.
1905 +) Add engine application. It can currently list engines by name and
1906 identity, and test if they are actually available.
1909 *) Add support for shared libraries under Irix.
1910 [Albert Chin-A-Young <china@thewrittenword.com>]
1912 +) Improve RPM specification file by forcing symbolic linking and making
1913 sure the installed documentation is also owned by root.root.
1914 [Damien Miller <djm@mindrot.org>]
1916 *) Add configuration option to build on Linux on both big-endian and
1918 [Ralf Baechle <ralf@uni-koblenz.de>]
1920 +) Give the OpenSSL applications more possibilities to make use of
1921 keys (public as well as private) handled by engines.
1924 +) Add OCSP code that comes from CertCo.
1927 +) Add VMS support for the Rijndael code.
1930 +) Added untested support for Nuron crypto accelerator.
1933 +) Add support for external cryptographic devices. This code was
1934 previously distributed separately as the "engine" branch.
1935 [Geoff Thorpe, Richard Levitte]
1937 +) Rework the filename-translation in the DSO code. It is now possible to
1938 have far greater control over how a "name" is turned into a filename
1939 depending on the operating environment and any oddities about the
1940 different shared library filenames on each system.
1943 +) Support threads on FreeBSD-elf in Configure.
1946 *) Add the possibility to create shared libraries on HP-UX
1949 +) Fix for SHA1 assembly problem with MASM: it produces
1950 warnings about corrupt line number information when assembling
1951 with debugging information. This is caused by the overlapping
1953 [Bernd Matthes <mainbug@celocom.de>, Steve Henson]
1956 NCONF_get_number() has no error checking at all. As a replacement,
1957 NCONF_get_number_e() is defined (_e for "error checking") and is
1958 promoted strongly. The old NCONF_get_number is kept around for
1959 binary backward compatibility.
1960 Make it possible for methods to load from something other than a BIO,
1961 by providing a function pointer that is given a name instead of a BIO.
1962 For example, this could be used to load configuration data from an
1966 *) Fix typo in get_cert_by_subject() in by_dir.c
1967 [Jean-Marc Desperrier <jean-marc.desperrier@certplus.com>]
1969 *) Rework the system to generate shared libraries:
1971 - Make note of the expected extension for the shared libraries and
1972 if there is a need for symbolic links from for example libcrypto.so.0
1973 to libcrypto.so.0.9.7. There is extended info in Configure for
1976 - Make as few rebuilds of the shared libraries as possible.
1978 - Still avoid linking the OpenSSL programs with the shared libraries.
1980 - When installing, install the shared libraries separately from the
1984 +) Fix for non blocking accept BIOs. Added new I/O special reason
1985 BIO_RR_ACCEPT to cover this case. Previously use of accept BIOs
1986 with non blocking I/O was not possible because no retry code was
1987 implemented. Also added new SSL code SSL_WANT_ACCEPT to cover
1991 +) Added the beginnings of Rijndael support.
1994 +) Fix for bug in DirectoryString mask setting. Add support for
1995 X509_NAME_print_ex() in 'req' and X509_print_ex() function
1996 to allow certificate printing to more controllable, additional
1997 'certopt' option to 'x509' to allow new printing options to be
2001 +) Clean old EAY MD5 hack from e_os.h.
2004 *) Fix SSL_CTX_set_read_ahead macro to actually use its argument.
2006 Copy SSL_CTX's read_ahead flag to SSL object directly in SSL_new
2007 and not in SSL_clear because the latter is also used by the
2008 accept/connect functions; previously, the settings made by
2009 SSL_set_read_ahead would be lost during the handshake.
2010 [Bodo Moeller; problems reported by Anders Gertz <gertz@epact.se>]
2012 *) Correct util/mkdef.pl to be selective about disabled algorithms.
2013 Previously, it would create entries for disableed algorithms no
2017 *) Added several new manual pages for SSL_* function.
2020 Changes between 0.9.5a and 0.9.6 [24 Sep 2000]
2022 *) In ssl23_get_client_hello, generate an error message when faced
2023 with an initial SSL 3.0/TLS record that is too small to contain the
2024 first two bytes of the ClientHello message, i.e. client_version.
2025 (Note that this is a pathologic case that probably has never happened
2026 in real life.) The previous approach was to use the version number
2027 from the record header as a substitute; but our protocol choice
2028 should not depend on that one because it is not authenticated
2029 by the Finished messages.
2032 *) More robust randomness gathering functions for Windows.
2033 [Jeffrey Altman <jaltman@columbia.edu>]
2035 *) For compatibility reasons if the flag X509_V_FLAG_ISSUER_CHECK is
2036 not set then we don't setup the error code for issuer check errors
2037 to avoid possibly overwriting other errors which the callback does
2038 handle. If an application does set the flag then we assume it knows
2039 what it is doing and can handle the new informational codes
2043 *) Fix for a nasty bug in ASN1_TYPE handling. ASN1_TYPE is used for
2044 a general "ANY" type, as such it should be able to decode anything
2045 including tagged types. However it didn't check the class so it would
2046 wrongly interpret tagged types in the same way as their universal
2047 counterpart and unknown types were just rejected. Changed so that the
2048 tagged and unknown types are handled in the same way as a SEQUENCE:
2049 that is the encoding is stored intact. There is also a new type
2050 "V_ASN1_OTHER" which is used when the class is not universal, in this
2051 case we have no idea what the actual type is so we just lump them all
2055 *) On VMS, stdout may very well lead to a file that is written to
2056 in a record-oriented fashion. That means that every write() will
2057 write a separate record, which will be read separately by the
2058 programs trying to read from it. This can be very confusing.
2060 The solution is to put a BIO filter in the way that will buffer
2061 text until a linefeed is reached, and then write everything a
2062 line at a time, so every record written will be an actual line,
2063 not chunks of lines and not (usually doesn't happen, but I've
2064 seen it once) several lines in one record. BIO_f_linebuffer() is
2067 Currently, it's a VMS-only method, because that's where it has
2068 been tested well enough.
2071 *) Remove 'optimized' squaring variant in BN_mod_mul_montgomery,
2072 it can return incorrect results.
2073 (Note: The buggy variant was not enabled in OpenSSL 0.9.5a,
2074 but it was in 0.9.6-beta[12].)
2077 *) Disable the check for content being present when verifying detached
2078 signatures in pk7_smime.c. Some versions of Netscape (wrongly)
2079 include zero length content when signing messages.
2082 *) New BIO_shutdown_wr macro, which invokes the BIO_C_SHUTDOWN_WR
2083 BIO_ctrl (for BIO pairs).
2086 *) Add DSO method for VMS.
2089 *) Bug fix: Montgomery multiplication could produce results with the
2093 *) Add RPM specification openssl.spec and modify it to build three
2094 packages. The default package contains applications, application
2095 documentation and run-time libraries. The devel package contains
2096 include files, static libraries and function documentation. The
2097 doc package contains the contents of the doc directory. The original
2098 openssl.spec was provided by Damien Miller <djm@mindrot.org>.
2101 *) Add a large number of documentation files for many SSL routines.
2102 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
2104 *) Add a configuration entry for Sony News 4.
2105 [NAKAJI Hiroyuki <nakaji@tutrp.tut.ac.jp>]
2107 *) Don't set the two most significant bits to one when generating a
2108 random number < q in the DSA library.
2111 *) New SSL API mode 'SSL_MODE_AUTO_RETRY'. This disables the default
2112 behaviour that SSL_read may result in SSL_ERROR_WANT_READ (even if
2113 the underlying transport is blocking) if a handshake took place.
2114 (The default behaviour is needed by applications such as s_client
2115 and s_server that use select() to determine when to use SSL_read;
2116 but for applications that know in advance when to expect data, it
2117 just makes things more complicated.)
2120 *) Add RAND_egd_bytes(), which gives control over the number of bytes read
2124 *) Add a few more EBCDIC conditionals that make `req' and `x509'
2125 work better on such systems.
2126 [Martin Kraemer <Martin.Kraemer@MchP.Siemens.De>]
2128 *) Add two demo programs for PKCS12_parse() and PKCS12_create().
2129 Update PKCS12_parse() so it copies the friendlyName and the
2130 keyid to the certificates aux info.
2133 *) Fix bug in PKCS7_verify() which caused an infinite loop
2134 if there was more than one signature.
2135 [Sven Uszpelkat <su@celocom.de>]
2137 *) Major change in util/mkdef.pl to include extra information
2138 about each symbol, as well as presentig variables as well
2139 as functions. This change means that there's n more need
2140 to rebuild the .num files when some algorithms are excluded.
2143 *) Allow the verify time to be set by an application,
2144 rather than always using the current time.
2147 *) Phase 2 verify code reorganisation. The certificate
2148 verify code now looks up an issuer certificate by a
2149 number of criteria: subject name, authority key id
2150 and key usage. It also verifies self signed certificates
2151 by the same criteria. The main comparison function is
2152 X509_check_issued() which performs these checks.
2154 Lot of changes were necessary in order to support this
2155 without completely rewriting the lookup code.
2157 Authority and subject key identifier are now cached.
2159 The LHASH 'certs' is X509_STORE has now been replaced
2160 by a STACK_OF(X509_OBJECT). This is mainly because an
2161 LHASH can't store or retrieve multiple objects with
2162 the same hash value.
2164 As a result various functions (which were all internal
2165 use only) have changed to handle the new X509_STORE
2166 structure. This will break anything that messed round
2167 with X509_STORE internally.
2169 The functions X509_STORE_add_cert() now checks for an
2170 exact match, rather than just subject name.
2172 The X509_STORE API doesn't directly support the retrieval
2173 of multiple certificates matching a given criteria, however
2174 this can be worked round by performing a lookup first
2175 (which will fill the cache with candidate certificates)
2176 and then examining the cache for matches. This is probably
2177 the best we can do without throwing out X509_LOOKUP
2178 entirely (maybe later...).
2180 The X509_VERIFY_CTX structure has been enhanced considerably.
2182 All certificate lookup operations now go via a get_issuer()
2183 callback. Although this currently uses an X509_STORE it
2184 can be replaced by custom lookups. This is a simple way
2185 to bypass the X509_STORE hackery necessary to make this
2186 work and makes it possible to use more efficient techniques
2187 in future. A very simple version which uses a simple
2188 STACK for its trusted certificate store is also provided
2189 using X509_STORE_CTX_trusted_stack().
2191 The verify_cb() and verify() callbacks now have equivalents
2192 in the X509_STORE_CTX structure.
2194 X509_STORE_CTX also has a 'flags' field which can be used
2195 to customise the verify behaviour.
2198 *) Add new PKCS#7 signing option PKCS7_NOSMIMECAP which
2199 excludes S/MIME capabilities.
2202 *) When a certificate request is read in keep a copy of the
2203 original encoding of the signed data and use it when outputing
2204 again. Signatures then use the original encoding rather than
2205 a decoded, encoded version which may cause problems if the
2206 request is improperly encoded.
2209 *) For consistency with other BIO_puts implementations, call
2210 buffer_write(b, ...) directly in buffer_puts instead of calling
2213 In BIO_puts, increment b->num_write as in BIO_write.
2214 [Peter.Sylvester@EdelWeb.fr]
2216 *) Fix BN_mul_word for the case where the word is 0. (We have to use
2217 BN_zero, we may not return a BIGNUM with an array consisting of
2221 *) Avoid calling abort() from within the library when problems are
2222 detected, except if preprocessor symbols have been defined
2223 (such as REF_CHECK, BN_DEBUG etc.).
2226 *) New openssl application 'rsautl'. This utility can be
2227 used for low level RSA operations. DER public key
2228 BIO/fp routines also added.
2231 *) New Configure entry and patches for compiling on QNX 4.
2232 [Andreas Schneider <andreas@ds3.etech.fh-hamburg.de>]
2234 *) A demo state-machine implementation was sponsored by
2235 Nuron (http://www.nuron.com/) and is now available in
2236 demos/state_machine.
2239 *) New options added to the 'dgst' utility for signature
2240 generation and verification.
2243 *) Unrecognized PKCS#7 content types are now handled via a
2244 catch all ASN1_TYPE structure. This allows unsupported
2245 types to be stored as a "blob" and an application can
2246 encode and decode it manually.
2249 *) Fix various signed/unsigned issues to make a_strex.c
2251 [Oscar Jacobsson <oscar.jacobsson@celocom.com>]
2253 *) ASN1 fixes. i2d_ASN1_OBJECT was not returning the correct
2254 length if passed a buffer. ASN1_INTEGER_to_BN failed
2255 if passed a NULL BN and its argument was negative.
2256 [Steve Henson, pointed out by Sven Heiberg <sven@tartu.cyber.ee>]
2258 *) Modification to PKCS#7 encoding routines to output definite
2259 length encoding. Since currently the whole structures are in
2260 memory there's not real point in using indefinite length
2261 constructed encoding. However if OpenSSL is compiled with
2262 the flag PKCS7_INDEFINITE_ENCODING the old form is used.
2265 *) Added BIO_vprintf() and BIO_vsnprintf().
2268 *) Added more prefixes to parse for in the the strings written
2269 through a logging bio, to cover all the levels that are available
2270 through syslog. The prefixes are now:
2272 PANIC, EMERG, EMR => LOG_EMERG
2273 ALERT, ALR => LOG_ALERT
2274 CRIT, CRI => LOG_CRIT
2275 ERROR, ERR => LOG_ERR
2276 WARNING, WARN, WAR => LOG_WARNING
2277 NOTICE, NOTE, NOT => LOG_NOTICE
2278 INFO, INF => LOG_INFO
2279 DEBUG, DBG => LOG_DEBUG
2281 and as before, if none of those prefixes are present at the
2282 beginning of the string, LOG_ERR is chosen.
2284 On Win32, the LOG_* levels are mapped according to this:
2286 LOG_EMERG, LOG_ALERT, LOG_CRIT, LOG_ERR => EVENTLOG_ERROR_TYPE
2287 LOG_WARNING => EVENTLOG_WARNING_TYPE
2288 LOG_NOTICE, LOG_INFO, LOG_DEBUG => EVENTLOG_INFORMATION_TYPE
2292 *) Made it possible to reconfigure with just the configuration
2293 argument "reconf" or "reconfigure". The command line arguments
2294 are stored in Makefile.ssl in the variable CONFIGURE_ARGS,
2295 and are retrieved from there when reconfiguring.
2299 [Assar Westerlund <assar@sics.se>, Richard Levitte]
2301 *) Add the arguments -CAfile and -CApath to the pkcs12 utility.
2304 *) The obj_dat.pl script was messing up the sorting of object
2305 names. The reason was that it compared the quoted version
2306 of strings as a result "OCSP" > "OCSP Signing" because
2307 " > SPACE. Changed script to store unquoted versions of
2308 names and add quotes on output. It was also omitting some
2309 names from the lookup table if they were given a default
2310 value (that is if SN is missing it is given the same
2311 value as LN and vice versa), these are now added on the
2312 grounds that if an object has a name we should be able to
2313 look it up. Finally added warning output when duplicate
2314 short or long names are found.
2317 *) Changes needed for Tandem NSK.
2318 [Scott Uroff <scott@xypro.com>]
2320 *) Fix SSL 2.0 rollback checking: Due to an off-by-one error in
2321 RSA_padding_check_SSLv23(), special padding was never detected
2322 and thus the SSL 3.0/TLS 1.0 countermeasure against protocol
2323 version rollback attacks was not effective.
2325 In s23_clnt.c, don't use special rollback-attack detection padding
2326 (RSA_SSLV23_PADDING) if SSL 2.0 is the only protocol enabled in the
2327 client; similarly, in s23_srvr.c, don't do the rollback check if
2328 SSL 2.0 is the only protocol enabled in the server.
2331 *) Make it possible to get hexdumps of unprintable data with 'openssl
2332 asn1parse'. By implication, the functions ASN1_parse_dump() and
2333 BIO_dump_indent() are added.
2336 *) New functions ASN1_STRING_print_ex() and X509_NAME_print_ex()
2337 these print out strings and name structures based on various
2338 flags including RFC2253 support and proper handling of
2339 multibyte characters. Added options to the 'x509' utility
2340 to allow the various flags to be set.
2343 *) Various fixes to use ASN1_TIME instead of ASN1_UTCTIME.
2344 Also change the functions X509_cmp_current_time() and
2345 X509_gmtime_adj() work with an ASN1_TIME structure,
2346 this will enable certificates using GeneralizedTime in validity
2347 dates to be checked.
2350 *) Make the NEG_PUBKEY_BUG code (which tolerates invalid
2351 negative public key encodings) on by default,
2352 NO_NEG_PUBKEY_BUG can be set to disable it.
2355 *) New function c2i_ASN1_OBJECT() which acts on ASN1_OBJECT
2356 content octets. An i2c_ASN1_OBJECT is unnecessary because
2357 the encoding can be trivially obtained from the structure.
2360 *) crypto/err.c locking bugfix: Use write locks (CRYPTO_w_[un]lock),
2361 not read locks (CRYPTO_r_[un]lock).
2364 *) A first attempt at creating official support for shared
2365 libraries through configuration. I've kept it so the
2366 default is static libraries only, and the OpenSSL programs
2367 are always statically linked for now, but there are
2368 preparations for dynamic linking in place.
2369 This has been tested on Linux and Tru64.
2372 *) Randomness polling function for Win9x, as described in:
2373 Peter Gutmann, Software Generation of Practically Strong
2377 *) Fix so PRNG is seeded in req if using an already existing
2381 *) New options to smime application. -inform and -outform
2382 allow alternative formats for the S/MIME message including
2383 PEM and DER. The -content option allows the content to be
2384 specified separately. This should allow things like Netscape
2385 form signing output easier to verify.
2388 *) Fix the ASN1 encoding of tags using the 'long form'.
2391 *) New ASN1 functions, i2c_* and c2i_* for INTEGER and BIT
2392 STRING types. These convert content octets to and from the
2393 underlying type. The actual tag and length octets are
2394 already assumed to have been read in and checked. These
2395 are needed because all other string types have virtually
2396 identical handling apart from the tag. By having versions
2397 of the ASN1 functions that just operate on content octets
2398 IMPLICIT tagging can be handled properly. It also allows
2399 the ASN1_ENUMERATED code to be cut down because ASN1_ENUMERATED
2400 and ASN1_INTEGER are identical apart from the tag.
2403 *) Change the handling of OID objects as follows:
2405 - New object identifiers are inserted in objects.txt, following
2406 the syntax given in objects.README.
2407 - objects.pl is used to process obj_mac.num and create a new
2409 - obj_dat.pl is used to create a new obj_dat.h, using the data in
2412 This is currently kind of a hack, and the perl code in objects.pl
2413 isn't very elegant, but it works as I intended. The simplest way
2414 to check that it worked correctly is to look in obj_dat.h and
2415 check the array nid_objs and make sure the objects haven't moved
2416 around (this is important!). Additions are OK, as well as
2417 consistent name changes.
2420 *) Add BSD-style MD5-based passwords to 'openssl passwd' (option '-1').
2423 *) Addition of the command line parameter '-rand file' to 'openssl req'.
2424 The given file adds to whatever has already been seeded into the
2425 random pool through the RANDFILE configuration file option or
2426 environment variable, or the default random state file.
2429 *) mkstack.pl now sorts each macro group into lexical order.
2430 Previously the output order depended on the order the files
2431 appeared in the directory, resulting in needless rewriting
2435 *) Patches to make OpenSSL compile under Win32 again. Mostly
2436 work arounds for the VC++ problem that it treats func() as
2437 func(void). Also stripped out the parts of mkdef.pl that
2438 added extra typesafe functions: these no longer exist.
2441 *) Reorganisation of the stack code. The macros are now all
2442 collected in safestack.h . Each macro is defined in terms of
2443 a "stack macro" of the form SKM_<name>(type, a, b). The
2444 DEBUG_SAFESTACK is now handled in terms of function casts,
2445 this has the advantage of retaining type safety without the
2446 use of additional functions. If DEBUG_SAFESTACK is not defined
2447 then the non typesafe macros are used instead. Also modified the
2448 mkstack.pl script to handle the new form. Needs testing to see
2449 if which (if any) compilers it chokes and maybe make DEBUG_SAFESTACK
2450 the default if no major problems. Similar behaviour for ASN1_SET_OF
2451 and PKCS12_STACK_OF.
2454 *) When some versions of IIS use the 'NET' form of private key the
2455 key derivation algorithm is different. Normally MD5(password) is
2456 used as a 128 bit RC4 key. In the modified case
2457 MD5(MD5(password) + "SGCKEYSALT") is used insted. Added some
2458 new functions i2d_RSA_NET(), d2i_RSA_NET() etc which are the same
2459 as the old Netscape_RSA functions except they have an additional
2460 'sgckey' parameter which uses the modified algorithm. Also added
2461 an -sgckey command line option to the rsa utility. Thanks to
2462 Adrian Peck <bertie@ncipher.com> for posting details of the modified
2463 algorithm to openssl-dev.
2466 *) The evp_local.h macros were using 'c.##kname' which resulted in
2467 invalid expansion on some systems (SCO 5.0.5 for example).
2468 Corrected to 'c.kname'.
2469 [Phillip Porch <root@theporch.com>]
2471 *) New X509_get1_email() and X509_REQ_get1_email() functions that return
2472 a STACK of email addresses from a certificate or request, these look
2473 in the subject name and the subject alternative name extensions and
2474 omit any duplicate addresses.
2477 *) Re-implement BN_mod_exp2_mont using independent (and larger) windows.
2478 This makes DSA verification about 2 % faster.
2481 *) Increase maximum window size in BN_mod_exp_... to 6 bits instead of 5
2482 (meaning that now 2^5 values will be precomputed, which is only 4 KB
2483 plus overhead for 1024 bit moduli).
2484 This makes exponentiations about 0.5 % faster for 1024 bit
2485 exponents (as measured by "openssl speed rsa2048").
2488 *) Rename memory handling macros to avoid conflicts with other
2490 Malloc => OPENSSL_malloc
2491 Malloc_locked => OPENSSL_malloc_locked
2492 Realloc => OPENSSL_realloc
2493 Free => OPENSSL_free
2496 *) New function BN_mod_exp_mont_word for small bases (roughly 15%
2497 faster than BN_mod_exp_mont, i.e. 7% for a full DH exchange).
2500 *) CygWin32 support.
2501 [John Jarvie <jjarvie@newsguy.com>]
2503 *) The type-safe stack code has been rejigged. It is now only compiled
2504 in when OpenSSL is configured with the DEBUG_SAFESTACK option and
2505 by default all type-specific stack functions are "#define"d back to
2506 standard stack functions. This results in more streamlined output
2507 but retains the type-safety checking possibilities of the original
2511 *) The STACK code has been cleaned up, and certain type declarations
2512 that didn't make a lot of sense have been brought in line. This has
2513 also involved a cleanup of sorts in safestack.h to more correctly
2514 map type-safe stack functions onto their plain stack counterparts.
2515 This work has also resulted in a variety of "const"ifications of
2516 lots of the code, especially "_cmp" operations which should normally
2517 be prototyped with "const" parameters anyway.
2520 *) When generating bytes for the first time in md_rand.c, 'stir the pool'
2521 by seeding with STATE_SIZE dummy bytes (with zero entropy count).
2522 (The PRNG state consists of two parts, the large pool 'state' and 'md',
2523 where all of 'md' is used each time the PRNG is used, but 'state'
2524 is used only indexed by a cyclic counter. As entropy may not be
2525 well distributed from the beginning, 'md' is important as a
2526 chaining variable. However, the output function chains only half
2527 of 'md', i.e. 80 bits. ssleay_rand_add, on the other hand, chains
2528 all of 'md', and seeding with STATE_SIZE dummy bytes will result
2529 in all of 'state' being rewritten, with the new values depending
2530 on virtually all of 'md'. This overcomes the 80 bit limitation.)
2533 *) In ssl/s2_clnt.c and ssl/s3_clnt.c, call ERR_clear_error() when
2534 the handshake is continued after ssl_verify_cert_chain();
2535 otherwise, if SSL_VERIFY_NONE is set, remaining error codes
2536 can lead to 'unexplainable' connection aborts later.
2537 [Bodo Moeller; problem tracked down by Lutz Jaenicke]
2539 *) Major EVP API cipher revision.
2540 Add hooks for extra EVP features. This allows various cipher
2541 parameters to be set in the EVP interface. Support added for variable
2542 key length ciphers via the EVP_CIPHER_CTX_set_key_length() function and
2543 setting of RC2 and RC5 parameters.
2545 Modify EVP_OpenInit() and EVP_SealInit() to cope with variable key length
2548 Remove lots of duplicated code from the EVP library. For example *every*
2549 cipher init() function handles the 'iv' in the same way according to the
2550 cipher mode. They also all do nothing if the 'key' parameter is NULL and
2551 for CFB and OFB modes they zero ctx->num.
2553 New functionality allows removal of S/MIME code RC2 hack.
2555 Most of the routines have the same form and so can be declared in terms
2558 By shifting this to the top level EVP_CipherInit() it can be removed from
2559 all individual ciphers. If the cipher wants to handle IVs or keys
2560 differently it can set the EVP_CIPH_CUSTOM_IV or EVP_CIPH_ALWAYS_CALL_INIT
2563 Change lots of functions like EVP_EncryptUpdate() to now return a
2564 value: although software versions of the algorithms cannot fail
2565 any installed hardware versions can.
2568 *) Implement SSL_OP_TLS_ROLLBACK_BUG: In ssl3_get_client_key_exchange, if
2569 this option is set, tolerate broken clients that send the negotiated
2570 protocol version number instead of the requested protocol version
2574 *) Call dh_tmp_cb (set by ..._TMP_DH_CB) with correct 'is_export' flag;
2575 i.e. non-zero for export ciphersuites, zero otherwise.
2576 Previous versions had this flag inverted, inconsistent with
2577 rsa_tmp_cb (..._TMP_RSA_CB).
2578 [Bodo Moeller; problem reported by Amit Chopra]
2580 *) Add missing DSA library text string. Work around for some IIS
2581 key files with invalid SEQUENCE encoding.
2584 *) Add a document (doc/standards.txt) that list all kinds of standards
2585 and so on that are implemented in OpenSSL.
2588 *) Enhance c_rehash script. Old version would mishandle certificates
2589 with the same subject name hash and wouldn't handle CRLs at all.
2590 Added -fingerprint option to crl utility, to support new c_rehash
2594 *) Eliminate non-ANSI declarations in crypto.h and stack.h.
2597 *) Fix for SSL server purpose checking. Server checking was
2598 rejecting certificates which had extended key usage present
2599 but no ssl client purpose.
2600 [Steve Henson, reported by Rene Grosser <grosser@hisolutions.com>]
2602 *) Make PKCS#12 code work with no password. The PKCS#12 spec
2603 is a little unclear about how a blank password is handled.
2604 Since the password in encoded as a BMPString with terminating
2605 double NULL a zero length password would end up as just the
2606 double NULL. However no password at all is different and is
2607 handled differently in the PKCS#12 key generation code. NS
2608 treats a blank password as zero length. MSIE treats it as no
2609 password on export: but it will try both on import. We now do
2610 the same: PKCS12_parse() tries zero length and no password if
2611 the password is set to "" or NULL (NULL is now a valid password:
2612 it wasn't before) as does the pkcs12 application.
2615 *) Bugfixes in apps/x509.c: Avoid a memory leak; and don't use
2616 perror when PEM_read_bio_X509_REQ fails, the error message must
2617 be obtained from the error queue.
2620 *) Avoid 'thread_hash' memory leak in crypto/err/err.c by freeing
2621 it in ERR_remove_state if appropriate, and change ERR_get_state
2622 accordingly to avoid race conditions (this is necessary because
2623 thread_hash is no longer constant once set).
2626 *) Bugfix for linux-elf makefile.one.
2629 *) RSA_get_default_method() will now cause a default
2630 RSA_METHOD to be chosen if one doesn't exist already.
2631 Previously this was only set during a call to RSA_new()
2632 or RSA_new_method(NULL) meaning it was possible for
2633 RSA_get_default_method() to return NULL.
2636 *) Added native name translation to the existing DSO code
2637 that will convert (if the flag to do so is set) filenames
2638 that are sufficiently small and have no path information
2639 into a canonical native form. Eg. "blah" converted to
2640 "libblah.so" or "blah.dll" etc.
2643 *) New function ERR_error_string_n(e, buf, len) which is like
2644 ERR_error_string(e, buf), but writes at most 'len' bytes
2645 including the 0 terminator. For ERR_error_string_n, 'buf'
2647 [Damien Miller <djm@mindrot.org>, Bodo Moeller]
2649 *) CONF library reworked to become more general. A new CONF
2650 configuration file reader "class" is implemented as well as a
2651 new functions (NCONF_*, for "New CONF") to handle it. The now
2652 old CONF_* functions are still there, but are reimplemented to
2653 work in terms of the new functions. Also, a set of functions
2654 to handle the internal storage of the configuration data is
2655 provided to make it easier to write new configuration file
2656 reader "classes" (I can definitely see something reading a
2657 configuration file in XML format, for example), called _CONF_*,
2658 or "the configuration storage API"...
2660 The new configuration file reading functions are:
2662 NCONF_new, NCONF_free, NCONF_load, NCONF_load_fp, NCONF_load_bio,
2663 NCONF_get_section, NCONF_get_string, NCONF_get_numbre
2665 NCONF_default, NCONF_WIN32
2667 NCONF_dump_fp, NCONF_dump_bio
2669 NCONF_default and NCONF_WIN32 are method (or "class") choosers,
2670 NCONF_new creates a new CONF object. This works in the same way
2671 as other interfaces in OpenSSL, like the BIO interface.
2672 NCONF_dump_* dump the internal storage of the configuration file,
2673 which is useful for debugging. All other functions take the same
2674 arguments as the old CONF_* functions wth the exception of the
2675 first that must be a `CONF *' instead of a `LHASH *'.
2677 To make it easer to use the new classes with the old CONF_* functions,
2678 the function CONF_set_default_method is provided.
2681 *) Add '-tls1' option to 'openssl ciphers', which was already
2682 mentioned in the documentation but had not been implemented.
2683 (This option is not yet really useful because even the additional
2684 experimental TLS 1.0 ciphers are currently treated as SSL 3.0 ciphers.)
2687 *) Initial DSO code added into libcrypto for letting OpenSSL (and
2688 OpenSSL-based applications) load shared libraries and bind to
2689 them in a portable way.
2690 [Geoff Thorpe, with contributions from Richard Levitte]
2692 Changes between 0.9.5 and 0.9.5a [1 Apr 2000]
2694 *) Make sure _lrotl and _lrotr are only used with MSVC.
2696 *) Use lock CRYPTO_LOCK_RAND correctly in ssleay_rand_status
2697 (the default implementation of RAND_status).
2699 *) Rename openssl x509 option '-crlext', which was added in 0.9.5,
2700 to '-clrext' (= clear extensions), as intended and documented.
2701 [Bodo Moeller; inconsistency pointed out by Michael Attili
2704 *) Fix for HMAC. It wasn't zeroing the rest of the block if the key length
2705 was larger than the MD block size.
2706 [Steve Henson, pointed out by Yost William <YostW@tce.com>]
2708 *) Modernise PKCS12_parse() so it uses STACK_OF(X509) for its ca argument
2709 fix a leak when the ca argument was passed as NULL. Stop X509_PUBKEY_set()
2710 using the passed key: if the passed key was a private key the result
2711 of X509_print(), for example, would be to print out all the private key
2715 *) des_quad_cksum() byte order bug fix.
2716 [Ulf Möller, using the problem description in krb4-0.9.7, where
2717 the solution is attributed to Derrick J Brashear <shadow@DEMENTIA.ORG>]
2719 *) Fix so V_ASN1_APP_CHOOSE works again: however its use is strongly
2721 [Steve Henson, pointed out by Brian Korver <briank@cs.stanford.edu>]
2723 *) For easily testing in shell scripts whether some command
2724 'openssl XXX' exists, the new pseudo-command 'openssl no-XXX'
2725 returns with exit code 0 iff no command of the given name is available.
2726 'no-XXX' is printed in this case, 'XXX' otherwise. In both cases,
2727 the output goes to stdout and nothing is printed to stderr.
2728 Additional arguments are always ignored.
2730 Since for each cipher there is a command of the same name,
2731 the 'no-cipher' compilation switches can be tested this way.
2733 ('openssl no-XXX' is not able to detect pseudo-commands such
2734 as 'quit', 'list-XXX-commands', or 'no-XXX' itself.)
2737 *) Update test suite so that 'make test' succeeds in 'no-rsa' configuration.
2740 *) For SSL_[CTX_]set_tmp_dh, don't create a DH key if SSL_OP_SINGLE_DH_USE
2741 is set; it will be thrown away anyway because each handshake creates
2743 ssl_cert_dup, which is used by SSL_new, now copies DH keys in addition
2744 to parameters -- in previous versions (since OpenSSL 0.9.3) the
2745 'default key' from SSL_CTX_set_tmp_dh would always be lost, meanining
2746 you effectivly got SSL_OP_SINGLE_DH_USE when using this macro.
2749 *) New s_client option -ign_eof: EOF at stdin is ignored, and
2750 'Q' and 'R' lose their special meanings (quit/renegotiate).
2751 This is part of what -quiet does; unlike -quiet, -ign_eof
2752 does not suppress any output.
2755 *) Add compatibility options to the purpose and trust code. The
2756 purpose X509_PURPOSE_ANY is "any purpose" which automatically
2757 accepts a certificate or CA, this was the previous behaviour,
2758 with all the associated security issues.
2760 X509_TRUST_COMPAT is the old trust behaviour: only and
2761 automatically trust self signed roots in certificate store. A
2762 new trust setting X509_TRUST_DEFAULT is used to specify that
2763 a purpose has no associated trust setting and it should instead
2764 use the value in the default purpose.
2767 *) Fix the PKCS#8 DSA private key code so it decodes keys again
2768 and fix a memory leak.
2771 *) In util/mkerr.pl (which implements 'make errors'), preserve
2772 reason strings from the previous version of the .c file, as
2773 the default to have only downcase letters (and digits) in
2774 automatically generated reasons codes is not always appropriate.
2777 *) In ERR_load_ERR_strings(), build an ERR_LIB_SYS error reason table
2778 using strerror. Previously, ERR_reason_error_string() returned
2779 library names as reason strings for SYSerr; but SYSerr is a special
2780 case where small numbers are errno values, not library numbers.
2783 *) Add '-dsaparam' option to 'openssl dhparam' application. This
2784 converts DSA parameters into DH parameters. (When creating parameters,
2785 DSA_generate_parameters is used.)
2788 *) Include 'length' (recommended exponent length) in C code generated
2789 by 'openssl dhparam -C'.
2792 *) The second argument to set_label in perlasm was already being used
2793 so couldn't be used as a "file scope" flag. Moved to third argument
2797 *) In PEM_ASN1_write_bio and some other functions, use RAND_pseudo_bytes
2798 instead of RAND_bytes for encryption IVs and salts.
2801 *) Include RAND_status() into RAND_METHOD instead of implementing
2802 it only for md_rand.c Otherwise replacing the PRNG by calling
2803 RAND_set_rand_method would be impossible.
2806 *) Don't let DSA_generate_key() enter an infinite loop if the random
2807 number generation fails.
2810 *) New 'rand' application for creating pseudo-random output.
2813 *) Added configuration support for Linux/IA64
2814 [Rolf Haberrecker <rolf@suse.de>]
2816 *) Assembler module support for Mingw32.
2819 *) Shared library support for HPUX (in shlib/).
2820 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE> and Anonymous]
2822 *) Shared library support for Solaris gcc.
2823 [Lutz Behnke <behnke@trustcenter.de>]
2825 Changes between 0.9.4 and 0.9.5 [28 Feb 2000]
2827 *) PKCS7_encrypt() was adding text MIME headers twice because they
2828 were added manually and by SMIME_crlf_copy().
2831 *) In bntest.c don't call BN_rand with zero bits argument.
2832 [Steve Henson, pointed out by Andrew W. Gray <agray@iconsinc.com>]
2834 *) BN_mul bugfix: In bn_mul_part_recursion() only the a>a[n] && b>b[n]
2835 case was implemented. This caused BN_div_recp() to fail occasionally.
2838 *) Add an optional second argument to the set_label() in the perl
2839 assembly language builder. If this argument exists and is set
2840 to 1 it signals that the assembler should use a symbol whose
2841 scope is the entire file, not just the current function. This
2842 is needed with MASM which uses the format label:: for this scope.
2843 [Steve Henson, pointed out by Peter Runestig <peter@runestig.com>]
2845 *) Change the ASN1 types so they are typedefs by default. Before
2846 almost all types were #define'd to ASN1_STRING which was causing
2847 STACK_OF() problems: you couldn't declare STACK_OF(ASN1_UTF8STRING)
2851 *) Change names of new functions to the new get1/get0 naming
2852 convention: After 'get1', the caller owns a reference count
2853 and has to call ..._free; 'get0' returns a pointer to some
2854 data structure without incrementing reference counters.
2855 (Some of the existing 'get' functions increment a reference
2856 counter, some don't.)
2857 Similarly, 'set1' and 'add1' functions increase reference
2858 counters or duplicate objects.
2861 *) Allow for the possibility of temp RSA key generation failure:
2862 the code used to assume it always worked and crashed on failure.
2865 *) Fix potential buffer overrun problem in BIO_printf().
2866 [Ulf Möller, using public domain code by Patrick Powell; problem
2867 pointed out by David Sacerdote <das33@cornell.edu>]
2869 *) Support EGD <http://www.lothar.com/tech/crypto/>. New functions
2870 RAND_egd() and RAND_status(). In the command line application,
2871 the EGD socket can be specified like a seed file using RANDFILE
2875 *) Allow the string CERTIFICATE to be tolerated in PKCS#7 structures.
2876 Some CAs (e.g. Verisign) distribute certificates in this form.
2879 *) Remove the SSL_ALLOW_ADH compile option and set the default cipher
2880 list to exclude them. This means that no special compilation option
2881 is needed to use anonymous DH: it just needs to be included in the
2885 *) Change the EVP_MD_CTX_type macro so its meaning consistent with
2886 EVP_MD_type. The old functionality is available in a new macro called
2887 EVP_MD_md(). Change code that uses it and update docs.
2890 *) ..._ctrl functions now have corresponding ..._callback_ctrl functions
2891 where the 'void *' argument is replaced by a function pointer argument.
2892 Previously 'void *' was abused to point to functions, which works on
2893 many platforms, but is not correct. As these functions are usually
2894 called by macros defined in OpenSSL header files, most source code
2895 should work without changes.
2898 *) <openssl/opensslconf.h> (which is created by Configure) now contains
2899 sections with information on -D... compiler switches used for
2900 compiling the library so that applications can see them. To enable
2901 one of these sections, a pre-processor symbol OPENSSL_..._DEFINES
2902 must be defined. E.g.,
2903 #define OPENSSL_ALGORITHM_DEFINES
2904 #include <openssl/opensslconf.h>
2905 defines all pertinent NO_<algo> symbols, such as NO_IDEA, NO_RSA, etc.
2906 [Richard Levitte, Ulf and Bodo Möller]
2908 *) Bugfix: Tolerate fragmentation and interleaving in the SSL 3/TLS
2912 *) Change the 'other' type in certificate aux info to a STACK_OF
2913 X509_ALGOR. Although not an AlgorithmIdentifier as such it has
2914 the required ASN1 format: arbitrary types determined by an OID.
2917 *) Add some PEM_write_X509_REQ_NEW() functions and a command line
2918 argument to 'req'. This is not because the function is newer or
2919 better than others it just uses the work 'NEW' in the certificate
2920 request header lines. Some software needs this.
2923 *) Reorganise password command line arguments: now passwords can be
2924 obtained from various sources. Delete the PEM_cb function and make
2925 it the default behaviour: i.e. if the callback is NULL and the
2926 usrdata argument is not NULL interpret it as a null terminated pass
2927 phrase. If usrdata and the callback are NULL then the pass phrase
2928 is prompted for as usual.
2931 *) Add support for the Compaq Atalla crypto accelerator. If it is installed,
2932 the support is automatically enabled. The resulting binaries will
2933 autodetect the card and use it if present.
2934 [Ben Laurie and Compaq Inc.]
2936 *) Work around for Netscape hang bug. This sends certificate request
2937 and server done in one record. Since this is perfectly legal in the
2938 SSL/TLS protocol it isn't a "bug" option and is on by default. See
2939 the bugs/SSLv3 entry for more info.
2942 *) HP-UX tune-up: new unified configs, HP C compiler bug workaround.
2945 *) Add -rand argument to smime and pkcs12 applications and read/write
2949 *) New 'passwd' tool for crypt(3) and apr1 password hashes.
2952 *) Add command line password options to the remaining applications.
2955 *) Bug fix for BN_div_recp() for numerators with an even number of
2959 *) More tests in bntest.c, and changed test_bn output.
2962 *) ./config recognizes MacOS X now.
2965 *) Bug fix for BN_div() when the first words of num and divsor are
2966 equal (it gave wrong results if (rem=(n1-q*d0)&BN_MASK2) < d0).
2969 *) Add support for various broken PKCS#8 formats, and command line
2970 options to produce them.
2973 *) New functions BN_CTX_start(), BN_CTX_get() and BT_CTX_end() to
2974 get temporary BIGNUMs from a BN_CTX.
2977 *) Correct return values in BN_mod_exp_mont() and BN_mod_exp2_mont()
2981 *) Change the SSLeay_add_all_*() functions to OpenSSL_add_all_*() and
2982 include a #define from the old name to the new. The original intent
2983 was that statically linked binaries could for example just call
2984 SSLeay_add_all_ciphers() to just add ciphers to the table and not
2985 link with digests. This never worked becayse SSLeay_add_all_digests()
2986 and SSLeay_add_all_ciphers() were in the same source file so calling
2987 one would link with the other. They are now in separate source files.
2990 *) Add a new -notext option to 'ca' and a -pubkey option to 'spkac'.
2993 *) Use a less unusual form of the Miller-Rabin primality test (it used
2994 a binary algorithm for exponentiation integrated into the Miller-Rabin
2995 loop, our standard modexp algorithms are faster).
2998 *) Support for the EBCDIC character set completed.
2999 [Martin Kraemer <Martin.Kraemer@Mch.SNI.De>]
3001 *) Source code cleanups: use const where appropriate, eliminate casts,
3002 use void * instead of char * in lhash.
3005 *) Bugfix: ssl3_send_server_key_exchange was not restartable
3006 (the state was not changed to SSL3_ST_SW_KEY_EXCH_B, and because of
3007 this the server could overwrite ephemeral keys that the client
3011 *) Turn DSA_is_prime into a macro that calls BN_is_prime,
3012 using 50 iterations of the Rabin-Miller test.
3014 DSA_generate_parameters now uses BN_is_prime_fasttest (with 50
3015 iterations of the Rabin-Miller test as required by the appendix
3016 to FIPS PUB 186[-1]) instead of DSA_is_prime.
3017 As BN_is_prime_fasttest includes trial division, DSA parameter
3018 generation becomes much faster.
3020 This implies a change for the callback functions in DSA_is_prime
3021 and DSA_generate_parameters: The callback function is called once
3022 for each positive witness in the Rabin-Miller test, not just
3023 occasionally in the inner loop; and the parameters to the
3024 callback function now provide an iteration count for the outer
3025 loop rather than for the current invocation of the inner loop.
3026 DSA_generate_parameters additionally can call the callback
3027 function with an 'iteration count' of -1, meaning that a
3028 candidate has passed the trial division test (when q is generated
3029 from an application-provided seed, trial division is skipped).
3032 *) New function BN_is_prime_fasttest that optionally does trial
3033 division before starting the Rabin-Miller test and has
3034 an additional BN_CTX * argument (whereas BN_is_prime always
3035 has to allocate at least one BN_CTX).
3036 'callback(1, -1, cb_arg)' is called when a number has passed the
3037 trial division stage.
3040 *) Fix for bug in CRL encoding. The validity dates weren't being handled
3044 *) New -pkcs12 option to CA.pl script to write out a PKCS#12 file.
3047 *) New function BN_pseudo_rand().
3050 *) Clean up BN_mod_mul_montgomery(): replace the broken (and unreadable)
3051 bignum version of BN_from_montgomery() with the working code from
3052 SSLeay 0.9.0 (the word based version is faster anyway), and clean up
3056 *) Avoid a race condition in s2_clnt.c (function get_server_hello) that
3057 made it impossible to use the same SSL_SESSION data structure in
3058 SSL2 clients in multiple threads.
3061 *) The return value of RAND_load_file() no longer counts bytes obtained
3062 by stat(). RAND_load_file(..., -1) is new and uses the complete file
3063 to seed the PRNG (previously an explicit byte count was required).
3064 [Ulf Möller, Bodo Möller]
3066 *) Clean up CRYPTO_EX_DATA functions, some of these didn't have prototypes
3067 used (char *) instead of (void *) and had casts all over the place.
3070 *) Make BN_generate_prime() return NULL on error if ret!=NULL.
3073 *) Retain source code compatibility for BN_prime_checks macro:
3074 BN_is_prime(..., BN_prime_checks, ...) now uses
3075 BN_prime_checks_for_size to determine the appropriate number of
3076 Rabin-Miller iterations.
3079 *) Diffie-Hellman uses "safe" primes: DH_check() return code renamed to
3080 DH_CHECK_P_NOT_SAFE_PRIME.
3081 (Check if this is true? OpenPGP calls them "strong".)
3084 *) Merge the functionality of "dh" and "gendh" programs into a new program
3085 "dhparam". The old programs are retained for now but will handle DH keys
3086 (instead of parameters) in future.
3089 *) Make the ciphers, s_server and s_client programs check the return values
3090 when a new cipher list is set.
3093 *) Enhance the SSL/TLS cipher mechanism to correctly handle the TLS 56bit
3094 ciphers. Before when the 56bit ciphers were enabled the sorting was
3097 The syntax for the cipher sorting has been extended to support sorting by
3098 cipher-strength (using the strength_bits hard coded in the tables).
3099 The new command is "@STRENGTH" (see also doc/apps/ciphers.pod).
3101 Fix a bug in the cipher-command parser: when supplying a cipher command
3102 string with an "undefined" symbol (neither command nor alphanumeric
3103 [A-Za-z0-9], ssl_set_cipher_list used to hang in an endless loop. Now
3104 an error is flagged.
3106 Due to the strength-sorting extension, the code of the
3107 ssl_create_cipher_list() function was completely rearranged. I hope that
3108 the readability was also increased :-)
3109 [Lutz Jaenicke <Lutz.Jaenicke@aet.TU-Cottbus.DE>]
3111 *) Minor change to 'x509' utility. The -CAcreateserial option now uses 1
3112 for the first serial number and places 2 in the serial number file. This
3113 avoids problems when the root CA is created with serial number zero and
3114 the first user certificate has the same issuer name and serial number
3118 *) Fixes to X509_ATTRIBUTE utilities, change the 'req' program so it uses
3119 the new code. Add documentation for this stuff.
3122 *) Changes to X509_ATTRIBUTE utilities. These have been renamed from
3123 X509_*() to X509at_*() on the grounds that they don't handle X509
3124 structures and behave in an analagous way to the X509v3 functions:
3125 they shouldn't be called directly but wrapper functions should be used
3128 So we also now have some wrapper functions that call the X509at functions
3129 when passed certificate requests. (TO DO: similar things can be done with
3130 PKCS#7 signed and unsigned attributes, PKCS#12 attributes and a few other
3131 things. Some of these need some d2i or i2d and print functionality
3132 because they handle more complex structures.)
3135 *) Add missing #ifndefs that caused missing symbols when building libssl
3136 as a shared library without RSA. Use #ifndef NO_SSL2 instead of
3137 NO_RSA in ssl/s2*.c.
3138 [Kris Kennaway <kris@hub.freebsd.org>, modified by Ulf Möller]
3140 *) Precautions against using the PRNG uninitialized: RAND_bytes() now
3141 has a return value which indicates the quality of the random data
3142 (1 = ok, 0 = not seeded). Also an error is recorded on the thread's
3143 error queue. New function RAND_pseudo_bytes() generates output that is
3144 guaranteed to be unique but not unpredictable. RAND_add is like
3145 RAND_seed, but takes an extra argument for an entropy estimate
3146 (RAND_seed always assumes full entropy).
3149 *) Do more iterations of Rabin-Miller probable prime test (specifically,
3150 3 for 1024-bit primes, 6 for 512-bit primes, 12 for 256-bit primes
3151 instead of only 2 for all lengths; see BN_prime_checks_for_size definition
3152 in crypto/bn/bn_prime.c for the complete table). This guarantees a
3153 false-positive rate of at most 2^-80 for random input.
3156 *) Rewrite ssl3_read_n (ssl/s3_pkt.c) avoiding a couple of bugs.
3159 *) New function X509_CTX_rget_chain() (renamed to X509_CTX_get1_chain
3160 in the 0.9.5 release), this returns the chain
3161 from an X509_CTX structure with a dup of the stack and all
3162 the X509 reference counts upped: so the stack will exist
3163 after X509_CTX_cleanup() has been called. Modify pkcs12.c
3166 Also make SSL_SESSION_print() print out the verify return
3170 *) Add manpage for the pkcs12 command. Also change the default
3171 behaviour so MAC iteration counts are used unless the new
3172 -nomaciter option is used. This improves file security and
3173 only older versions of MSIE (4.0 for example) need it.
3176 *) Honor the no-xxx Configure options when creating .DEF files.
3179 *) Add PKCS#10 attributes to field table: challengePassword,
3180 unstructuredName and unstructuredAddress. These are taken from
3181 draft PKCS#9 v2.0 but are compatible with v1.2 provided no
3182 international characters are used.
3184 More changes to X509_ATTRIBUTE code: allow the setting of types
3185 based on strings. Remove the 'loc' parameter when adding
3186 attributes because these will be a SET OF encoding which is sorted
3190 *) Initial changes to the 'req' utility to allow request generation
3191 automation. This will allow an application to just generate a template
3192 file containing all the field values and have req construct the
3195 Initial support for X509_ATTRIBUTE handling. Stacks of these are
3196 used all over the place including certificate requests and PKCS#7
3197 structures. They are currently handled manually where necessary with
3198 some primitive wrappers for PKCS#7. The new functions behave in a
3199 manner analogous to the X509 extension functions: they allow
3200 attributes to be looked up by NID and added.
3202 Later something similar to the X509V3 code would be desirable to
3203 automatically handle the encoding, decoding and printing of the
3204 more complex types. The string types like challengePassword can
3205 be handled by the string table functions.
3207 Also modified the multi byte string table handling. Now there is
3208 a 'global mask' which masks out certain types. The table itself
3209 can use the flag STABLE_NO_MASK to ignore the mask setting: this
3210 is useful when for example there is only one permissible type
3211 (as in countryName) and using the mask might result in no valid
3215 *) Clean up 'Finished' handling, and add functions SSL_get_finished and
3216 SSL_get_peer_finished to allow applications to obtain the latest
3217 Finished messages sent to the peer or expected from the peer,
3218 respectively. (SSL_get_peer_finished is usually the Finished message
3219 actually received from the peer, otherwise the protocol will be aborted.)
3221 As the Finished message are message digests of the complete handshake
3222 (with a total of 192 bits for TLS 1.0 and more for SSL 3.0), they can
3223 be used for external authentication procedures when the authentication
3224 provided by SSL/TLS is not desired or is not enough.
3227 *) Enhanced support for Alpha Linux is added. Now ./config checks if
3228 the host supports BWX extension and if Compaq C is present on the
3229 $PATH. Just exploiting of the BWX extension results in 20-30%
3230 performance kick for some algorithms, e.g. DES and RC4 to mention
3231 a couple. Compaq C in turn generates ~20% faster code for MD5 and
3235 *) Add support for MS "fast SGC". This is arguably a violation of the
3236 SSL3/TLS protocol. Netscape SGC does two handshakes: the first with
3237 weak crypto and after checking the certificate is SGC a second one
3238 with strong crypto. MS SGC stops the first handshake after receiving
3239 the server certificate message and sends a second client hello. Since
3240 a server will typically do all the time consuming operations before
3241 expecting any further messages from the client (server key exchange
3242 is the most expensive) there is little difference between the two.
3244 To get OpenSSL to support MS SGC we have to permit a second client
3245 hello message after we have sent server done. In addition we have to
3246 reset the MAC if we do get this second client hello.
3249 *) Add a function 'd2i_AutoPrivateKey()' this will automatically decide
3250 if a DER encoded private key is RSA or DSA traditional format. Changed
3251 d2i_PrivateKey_bio() to use it. This is only needed for the "traditional"
3252 format DER encoded private key. Newer code should use PKCS#8 format which
3253 has the key type encoded in the ASN1 structure. Added DER private key
3254 support to pkcs8 application.
3257 *) SSL 3/TLS 1 servers now don't request certificates when an anonymous
3258 ciphersuites has been selected (as required by the SSL 3/TLS 1
3259 specifications). Exception: When SSL_VERIFY_FAIL_IF_NO_PEER_CERT
3260 is set, we interpret this as a request to violate the specification
3261 (the worst that can happen is a handshake failure, and 'correct'
3262 behaviour would result in a handshake failure anyway).
3265 *) In SSL_CTX_add_session, take into account that there might be multiple
3266 SSL_SESSION structures with the same session ID (e.g. when two threads
3267 concurrently obtain them from an external cache).
3268 The internal cache can handle only one SSL_SESSION with a given ID,
3269 so if there's a conflict, we now throw out the old one to achieve
3273 *) Add OIDs for idea and blowfish in CBC mode. This will allow both
3274 to be used in PKCS#5 v2.0 and S/MIME. Also add checking to
3275 some routines that use cipher OIDs: some ciphers do not have OIDs
3276 defined and so they cannot be used for S/MIME and PKCS#5 v2.0 for
3280 *) Simplify the trust setting structure and code. Now we just have
3281 two sequences of OIDs for trusted and rejected settings. These will
3282 typically have values the same as the extended key usage extension
3283 and any application specific purposes.
3285 The trust checking code now has a default behaviour: it will just
3286 check for an object with the same NID as the passed id. Functions can
3287 be provided to override either the default behaviour or the behaviour
3288 for a given id. SSL client, server and email already have functions
3289 in place for compatibility: they check the NID and also return "trusted"
3290 if the certificate is self signed.
3293 *) Add d2i,i2d bio/fp functions for PrivateKey: these convert the
3294 traditional format into an EVP_PKEY structure.
3297 *) Add a password callback function PEM_cb() which either prompts for
3298 a password if usr_data is NULL or otherwise assumes it is a null
3299 terminated password. Allow passwords to be passed on command line
3300 environment or config files in a few more utilities.
3303 *) Add a bunch of DER and PEM functions to handle PKCS#8 format private
3304 keys. Add some short names for PKCS#8 PBE algorithms and allow them
3305 to be specified on the command line for the pkcs8 and pkcs12 utilities.
3306 Update documentation.
3309 *) Support for ASN1 "NULL" type. This could be handled before by using
3310 ASN1_TYPE but there wasn't any function that would try to read a NULL
3311 and produce an error if it couldn't. For compatibility we also have
3312 ASN1_NULL_new() and ASN1_NULL_free() functions but these are faked and
3313 don't allocate anything because they don't need to.
3316 *) Initial support for MacOS is now provided. Examine INSTALL.MacOS
3318 [Andy Polyakov, Roy Woods <roy@centicsystems.ca>]
3320 *) Rebuild of the memory allocation routines used by OpenSSL code and
3321 possibly others as well. The purpose is to make an interface that
3322 provide hooks so anyone can build a separate set of allocation and
3323 deallocation routines to be used by OpenSSL, for example memory
3324 pool implementations, or something else, which was previously hard
3325 since Malloc(), Realloc() and Free() were defined as macros having
3326 the values malloc, realloc and free, respectively (except for Win32
3327 compilations). The same is provided for memory debugging code.
3328 OpenSSL already comes with functionality to find memory leaks, but
3329 this gives people a chance to debug other memory problems.
3331 With these changes, a new set of functions and macros have appeared:
3333 CRYPTO_set_mem_debug_functions() [F]
3334 CRYPTO_get_mem_debug_functions() [F]
3335 CRYPTO_dbg_set_options() [F]
3336 CRYPTO_dbg_get_options() [F]
3337 CRYPTO_malloc_debug_init() [M]
3339 The memory debug functions are NULL by default, unless the library
3340 is compiled with CRYPTO_MDEBUG or friends is defined. If someone
3341 wants to debug memory anyway, CRYPTO_malloc_debug_init() (which
3342 gives the standard debugging functions that come with OpenSSL) or
3343 CRYPTO_set_mem_debug_functions() (tells OpenSSL to use functions
3344 provided by the library user) must be used. When the standard
3345 debugging functions are used, CRYPTO_dbg_set_options can be used to
3346 request additional information:
3347 CRYPTO_dbg_set_options(V_CYRPTO_MDEBUG_xxx) corresponds to setting
3348 the CRYPTO_MDEBUG_xxx macro when compiling the library.
3350 Also, things like CRYPTO_set_mem_functions will always give the
3351 expected result (the new set of functions is used for allocation
3352 and deallocation) at all times, regardless of platform and compiler
3355 To finish it up, some functions that were never use in any other
3356 way than through macros have a new API and new semantic:
3359 CRYPTO_dbg_realloc()
3362 All macros of value have retained their old syntax.
3363 [Richard Levitte and Bodo Moeller]
3365 *) Some S/MIME fixes. The OID for SMIMECapabilities was wrong, the
3366 ordering of SMIMECapabilities wasn't in "strength order" and there
3367 was a missing NULL in the AlgorithmIdentifier for the SHA1 signature
3371 *) Some ASN1 types with illegal zero length encoding (INTEGER,
3372 ENUMERATED and OBJECT IDENTIFIER) choked the ASN1 routines.
3373 [Frans Heymans <fheymans@isaserver.be>, modified by Steve Henson]
3375 *) Merge in my S/MIME library for OpenSSL. This provides a simple
3376 S/MIME API on top of the PKCS#7 code, a MIME parser (with enough
3377 functionality to handle multipart/signed properly) and a utility
3378 called 'smime' to call all this stuff. This is based on code I
3379 originally wrote for Celo who have kindly allowed it to be
3380 included in OpenSSL.
3383 *) Add variants des_set_key_checked and des_set_key_unchecked of
3384 des_set_key (aka des_key_sched). Global variable des_check_key
3385 decides which of these is called by des_set_key; this way
3386 des_check_key behaves as it always did, but applications and
3387 the library itself, which was buggy for des_check_key == 1,
3388 have a cleaner way to pick the version they need.
3391 *) New function PKCS12_newpass() which changes the password of a
3395 *) Modify X509_TRUST and X509_PURPOSE so it also uses a static and
3396 dynamic mix. In both cases the ids can be used as an index into the
3397 table. Also modified the X509_TRUST_add() and X509_PURPOSE_add()
3398 functions so they accept a list of the field values and the
3399 application doesn't need to directly manipulate the X509_TRUST
3403 *) Modify the ASN1_STRING_TABLE stuff so it also uses bsearch and doesn't
3407 *) Modify the way the V3 extension code looks up extensions. This now
3408 works in a similar way to the object code: we have some "standard"
3409 extensions in a static table which is searched with OBJ_bsearch()
3410 and the application can add dynamic ones if needed. The file
3411 crypto/x509v3/ext_dat.h now has the info: this file needs to be
3412 updated whenever a new extension is added to the core code and kept
3413 in ext_nid order. There is a simple program 'tabtest.c' which checks
3414 this. New extensions are not added too often so this file can readily
3415 be maintained manually.
3417 There are two big advantages in doing things this way. The extensions
3418 can be looked up immediately and no longer need to be "added" using
3419 X509V3_add_standard_extensions(): this function now does nothing.
3420 [Side note: I get *lots* of email saying the extension code doesn't
3421 work because people forget to call this function]
3422 Also no dynamic allocation is done unless new extensions are added:
3423 so if we don't add custom extensions there is no need to call
3424 X509V3_EXT_cleanup().
3427 *) Modify enc utility's salting as follows: make salting the default. Add a
3428 magic header, so unsalted files fail gracefully instead of just decrypting
3429 to garbage. This is because not salting is a big security hole, so people
3430 should be discouraged from doing it.
3433 *) Fixes and enhancements to the 'x509' utility. It allowed a message
3434 digest to be passed on the command line but it only used this
3435 parameter when signing a certificate. Modified so all relevant
3436 operations are affected by the digest parameter including the
3437 -fingerprint and -x509toreq options. Also -x509toreq choked if a
3438 DSA key was used because it didn't fix the digest.
3441 *) Initial certificate chain verify code. Currently tests the untrusted
3442 certificates for consistency with the verify purpose (which is set
3443 when the X509_STORE_CTX structure is set up) and checks the pathlength.
3445 There is a NO_CHAIN_VERIFY compilation option to keep the old behaviour:
3446 this is because it will reject chains with invalid extensions whereas
3447 every previous version of OpenSSL and SSLeay made no checks at all.
3449 Trust code: checks the root CA for the relevant trust settings. Trust
3450 settings have an initial value consistent with the verify purpose: e.g.
3451 if the verify purpose is for SSL client use it expects the CA to be
3452 trusted for SSL client use. However the default value can be changed to
3453 permit custom trust settings: one example of this would be to only trust
3454 certificates from a specific "secure" set of CAs.
3456 Also added X509_STORE_CTX_new() and X509_STORE_CTX_free() functions
3457 which should be used for version portability: especially since the
3458 verify structure is likely to change more often now.
3460 SSL integration. Add purpose and trust to SSL_CTX and SSL and functions
3461 to set them. If not set then assume SSL clients will verify SSL servers
3464 Two new options to the verify program: -untrusted allows a set of
3465 untrusted certificates to be passed in and -purpose which sets the
3466 intended purpose of the certificate. If a purpose is set then the
3467 new chain verify code is used to check extension consistency.
3470 *) Support for the authority information access extension.
3473 *) Modify RSA and DSA PEM read routines to transparently handle
3474 PKCS#8 format private keys. New *_PUBKEY_* functions that handle
3475 public keys in a format compatible with certificate
3476 SubjectPublicKeyInfo structures. Unfortunately there were already
3477 functions called *_PublicKey_* which used various odd formats so
3478 these are retained for compatibility: however the DSA variants were
3479 never in a public release so they have been deleted. Changed dsa/rsa
3480 utilities to handle the new format: note no releases ever handled public
3481 keys so we should be OK.
3483 The primary motivation for this change is to avoid the same fiasco
3484 that dogs private keys: there are several incompatible private key
3485 formats some of which are standard and some OpenSSL specific and
3486 require various evil hacks to allow partial transparent handling and
3487 even then it doesn't work with DER formats. Given the option anything
3488 other than PKCS#8 should be dumped: but the other formats have to
3489 stay in the name of compatibility.
3491 With public keys and the benefit of hindsight one standard format
3492 is used which works with EVP_PKEY, RSA or DSA structures: though
3493 it clearly returns an error if you try to read the wrong kind of key.
3495 Added a -pubkey option to the 'x509' utility to output the public key.
3496 Also rename the EVP_PKEY_get_*() to EVP_PKEY_rget_*()
3497 (renamed to EVP_PKEY_get1_*() in the OpenSSL 0.9.5 release) and add
3498 EVP_PKEY_rset_*() functions (renamed to EVP_PKEY_set1_*())
3499 that do the same as the EVP_PKEY_assign_*() except they up the
3500 reference count of the added key (they don't "swallow" the
3504 *) Fixes to crypto/x509/by_file.c the code to read in certificates and
3505 CRLs would fail if the file contained no certificates or no CRLs:
3506 added a new function to read in both types and return the number
3507 read: this means that if none are read it will be an error. The
3508 DER versions of the certificate and CRL reader would always fail
3509 because it isn't possible to mix certificates and CRLs in DER format
3510 without choking one or the other routine. Changed this to just read
3511 a certificate: this is the best we can do. Also modified the code
3512 in apps/verify.c to take notice of return codes: it was previously
3513 attempting to read in certificates from NULL pointers and ignoring
3514 any errors: this is one reason why the cert and CRL reader seemed
3515 to work. It doesn't check return codes from the default certificate
3516 routines: these may well fail if the certificates aren't installed.
3519 *) Code to support otherName option in GeneralName.
3522 *) First update to verify code. Change the verify utility
3523 so it warns if it is passed a self signed certificate:
3524 for consistency with the normal behaviour. X509_verify
3525 has been modified to it will now verify a self signed
3526 certificate if *exactly* the same certificate appears
3527 in the store: it was previously impossible to trust a
3528 single self signed certificate. This means that:
3529 openssl verify ss.pem
3530 now gives a warning about a self signed certificate but
3531 openssl verify -CAfile ss.pem ss.pem
3535 *) For servers, store verify_result in SSL_SESSION data structure
3536 (and add it to external session representation).
3537 This is needed when client certificate verifications fails,
3538 but an application-provided verification callback (set by
3539 SSL_CTX_set_cert_verify_callback) allows accepting the session
3540 anyway (i.e. leaves x509_store_ctx->error != X509_V_OK
3541 but returns 1): When the session is reused, we have to set
3542 ssl->verify_result to the appropriate error code to avoid
3544 [Bodo Moeller, problem pointed out by Lutz Jaenicke]
3546 *) Fix a bug in the new PKCS#7 code: it didn't consider the
3547 case in PKCS7_dataInit() where the signed PKCS7 structure
3548 didn't contain any existing data because it was being created.
3549 [Po-Cheng Chen <pocheng@nst.com.tw>, slightly modified by Steve Henson]
3551 *) Add a salt to the key derivation routines in enc.c. This
3552 forms the first 8 bytes of the encrypted file. Also add a
3553 -S option to allow a salt to be input on the command line.
3556 *) New function X509_cmp(). Oddly enough there wasn't a function
3557 to compare two certificates. We do this by working out the SHA1
3558 hash and comparing that. X509_cmp() will be needed by the trust
3562 *) SSL_get1_session() is like SSL_get_session(), but increments
3563 the reference count in the SSL_SESSION returned.
3564 [Geoff Thorpe <geoff@eu.c2.net>]
3566 *) Fix for 'req': it was adding a null to request attributes.
3567 Also change the X509_LOOKUP and X509_INFO code to handle
projects / openssl.git / blob